Original Research

Empiric antimicrobial therapy often consists of the combination of gram-positive coverage with vancomycin (VAN) and gram-negative coverage, specifically an antipseudomonal beta-lactam such as piperacillin-tazobactam (PTZ). Literature from a variety of patient populations reports nephrotoxicity associated with VAN, targeting troughs greater than 15 µg/mL, that occur in 5% to 43% of patients.¹ In a study of critically ill patients, acute kidney injury (AKI) was found in 21% of patients receiving VAN, with increasing duration of VAN treatment, greater VAN levels, concomitant vasoactive medication administration, and intermittent infusion methods being associated with higher odds of AKI.² A recent report from adult internal medicine patients estimated the incidence of VAN-associated nephrotoxicity at 13.6% and implicated concomitant PTZ therapy as a key factor in these patients.³

Further studies have explored the interaction between empiric beta-lactam and VAN therapy, showing mixed results. Reports of AKI associated with the combination of VAN and PTZ range from 16.3% to 34.8%,^4-8 while the cefepime-VAN combination is reported to range from 12.5% to 13.3%.^5,6 While VAN monotherapy groups were well represented, only 1 study⁷ compared the PTZ-VAN combination to a control group of PTZ monotherapy.

The primary objective of this study was to evaluate the differences in AKI incidence between patients treated with VAN and with PTZ, alone and in combination.

This is a retrospective cohort study of adult patients conducted at the University of Kentucky Chandler Medical Center (UKMC) from September 1, 2010 through August 31, 2014. Patients were included if they were at least 18 years of age on admission; remained hospitalized for at least 48 hours; received VAN combined with PTZ (VAN/PTZ), VAN alone, or PTZ alone; and had at least 48 hours of therapy (and 48 hours of overlapping therapy in the VAN/PTZ group). Patients were excluded if they had underlying diagnosis of chronic kidney disease according to the International Classification of Diseases 9 (ICD-9) code, were receiving renal replacement therapy before admission, had a diagnosis of cystic fibrosis, or were pregnant. Additionally, patients were excluded if they presented with AKI, defined as an initial creatinine clearance less than 30 mL/min, or if baseline creatinine clearance was greater than 4 times the standard deviation from the mean; serum creatinine values were not obtained during admission; and if AKI occurred prior to therapy initiation, within 48 hours of initiation, or more than 7 days after treatment was discontinued. Patients were followed throughout their stay until time of discharge.

Data Source

Patient data were collected from the University of Kentucky Center for Clinical and Translational Science Enterprise Data Trust (EDT). The EDT contains clinical data from the inpatient population of UKMC from 2006 to present. Data stored and updated nightly by the EDT includes: demographics, financial classification (Medicare, Medicaid, private insurance), provider-level detail (service line), medical diagnosis (ICD-9 codes), medical procedures (Current Procedural Terminology [CPT] codes), lab tests and results, medication administration details, visit details (age, length of stay, etc), and vital signs. This study was approved by the UKMC Institutional Review Board.

Data collected for each patient included: demographic data, visit details (length of stay, admitting and primary diagnosis codes, etc.), severity of underlying illness as defined by the Charlson Comorbidity Index (CCI), all serum creatinine levels drawn per visit, medication administration information (dose, date, and time administered), all VAN trough levels, receipt of other nephrotoxic agents, blood pressures, and receipt of vasopressors.

Outcome Ascertainment

The definition of AKI was based on the RIFLE (Risk, Injury, Failure, Loss, End-stage) criteria,⁹ with risk defined as a 25% to 50% decrease in estimated glomerular filtration rate (GFR), injury as a 50% to 75% decrease in estimated GFR, and failure defined as a greater than 75% decrease in estimated GFR. Loss and end-stage classifications were not assessed because of this study’s follow-up period. The adjusted Cockcroft and Gault equation¹⁰ was used to estimate GFR due to the inconsistency of weight availability in the dataset and concordance with the institution’s practice. Baseline creatinine clearance was calculated with the first serum creatinine obtained, and the minimum creatinine clearance was calculated using the maximum serum creatinine during each patient’s visit. The percent decrease in creatinine clearance was calculated from these 2 values. AKI status was defined as meeting any of the RIFLE criteria. Mortality was assessed for all patients and defined as the composite of inhospital mortality and discharge or transfer to hospice care.

Exposure Ascertainment

Hypotension exposure was defined as experiencing 1 of the following: mean arterial blood pressure less than 60 mm Hg, a diagnosis of hypotension by a physician, or receipt of vasopressors or inotropic agents. Days of therapy for each drug were obtained and combination days of therapy were calculated by including only those days in which the patient received both medications. Total days of therapy were calculated by the sum of all days receiving at least 1 study agent. Exposure to other nephrotoxic agents (eg, acyclovir, angiotensin converting enzyme [ACE] inhibitors, angiotensin II receptor antagonists, aminoglycosides, amphotericin B, cyclosporine, foscarnet, loop diuretics, nonsteroidal anti-inflammatory drugs, sulfonamides, tacrolimus, and tenofovir) were defined as receipt of at least 1 dose of the agent during hospitalization.

Statistical Analysis

Characteristics between groups were described with basic descriptive statistics. Continuous variables were compared with 1-way analysis of variance (ANOVA) or the Kruskal-Wallis test. Categorical variables were compared with chi-square or Fisher exact test. Yearly AKI trends were assessed with Pearson correlation coefficient. To control for differences in underlying severity of illness between groups, a subanalysis was performed in which the cohort was split into 4 groups (0, 1, 2 to 4, and ≥5 points) based on CCI. Univariate models for all covariates were created with probability of AKI as the outcome. Covariates significant after univariate were incorporated into the multivariate model, which was subsequently adjusted to achieve the highest predictive accuracy by minimizing the Akaike information criterion (AIC). Nephrotoxic agent exposures were included in the final multivariate model regardless of statistical significance in univariate analysis. Model fit was assessed with a standardized Hosmer-Lemeshow goodness-of-fit test.¹¹ All statistical analyses were completed with RStudio v 0.98 running R v 3.1.2 (R Foundation for Statistical Computing, Vienna, Austria).¹² All tests were 2-tailed and significance was defined at an alpha of 0.05.

Of 17,879 patients initially screened, 11,650 patients were evaluated, of which 5,497 received VAN and PTZ (VAN/PTZ), 3,055 received VAN alone, and 3,098 received PTZ alone. Table 1 contains basic demographic information. The mean age of patients was 52.5 years ± 16.8 years with 6,242 (53.6%) males. Patients receiving VAN/PTZ had higher CCIs than either monotherapy group and had significantly increased length of hospitalization. While patients in the combination therapy group were more likely to experience hypotension, concomitant nephrotoxic agent exposure was more common in the VAN monotherapy group.

RIFLE-defined AKI occurred in 1,647 (14.1%) across the entire cohort. AKI occurred in 21% of VAN/PTZ patients, 8.3% of VAN patients, and 7.8% of PTZ patients (P < 0.0001). RIFLE-defined risk, injury, and failure occurred more frequently in the VAN/PTZ cohort compared to the VAN and PTZ monotherapy groups (Figure). There were no differences in AKI rates between years studied (r² = 0.4732, P = 0.2). Patients in the VAN/PTZ group experienced AKI on average of 8.0 days after treatment initiation, compared to 8.7 days and 5.2 days for VAN and PTZ monotherapy groups, respectively. The composite of inhospital mortality and transfer-to-hospice care was more common in VAN/PTZ patients (9.6%) compared to monotherapy groups (VAN, 3.9%; PTZ, 3.4%), most likely due to the increased severity of illness.

Acute kidney injury secondary to VAN therapy is a well-characterized adverse effect, while AKI incidence secondary to PTZ is less understood. Additionally, there appears to be an additive effect when these agents are used in combination. This is the largest review of AKI in patients receiving VAN,PTZ, or the combination of both agents.

There is increasing evidence suggesting greater nephrotoxicity in patients treated with the combination of VAN and antipseudomonal beta-lactams. The mechanism for the apparent increase in nephrotoxicity with this drug combination is not well understood and needs further study in both animal models and humans.

Acute kidney injury rates related to VAN vary widely, with recent studies in critically ill and internal medicine patients estimated at 21% and 13.6%, respectively.^2,3 In our VAN monotherapy cohort, the AKI rate was 8.3%, with 2.3% of patients experiencing a greater than 50% decrease in creatinine clearance. Piperacillin-tazobactam-related AKI rates are not well characterized; however, a small retrospective analysis estimated that 11.1% of PTZ patients experienced acute renal failure (defined as either increase in serum creatinine greater than 0.5 mg/dL or 50% increase from baseline).¹³ In the present study, we found the PTZ-related AKI rate to be 7.8%, which may be due to a more stringent definition of AKI. Additionally, Hellwig et al¹³ found that PTZ monotherapy was associated with higher AKI rates compared to VAN monotherapy (11.1% vs 4.9%; P = 0.014). This was not replicated in our study, with VAN and PTZ monotherapy having similar AKI rates (8.3% and 7.8%, respectively) and an adjusted aOR of 0.88 (95% CI 0.0.73-1.08) for AKI in PTZ- compared to VAN-treated patients. The estimated AKI incidence of 21% in the combination therapy group at our institution is consistent with literature that ranges from 16.3% to 34.8%.^4-8,13

To control for differences in baseline severity of illness, we performed a subgroup analysis of patients with similar CCI scores. The finding of increased AKI in patients receiving combination VAN and PTZ was consistent in each subgroup, suggesting that the increase in AKI is independent of illness severity.

This study is not without limitations. As with all retrospective studies, it is difficult to determine a causal link between VAN and PTZ combination therapy and increased AKI incidence due to confounding. We employed a rigorous study design that controlled for major confounders of AKI, such as concomitant nephrotoxic exposure, hypotension, and renal disease. Severity of illness was measured with CCI, which may not accurately capture the severity of illness at treatment initiation. Alternatives, such as acute physiology and chronic health evaluation (APACHE) and sequential organ failure assessment (SOFA) scores, may more accurately reflect critical illness on presentation; however, this study was not focused specifically on critically ill patients. In addition to baseline comorbidity, we controlled for hypotension and dehydration as a surrogate marker for critical illness. In the subgroup analysis of patients with similar CCI, the effect of VAN/PTZ on AKI compared to VAN or PTZ monotherapy was consistent in each group. Nephrotoxic potential of agents was assumed to be equal, which is not necessarily true. Additionally, the binary representation of nephrotoxic exposure does not describe the amount of the agent received; as such, our estimations of AKI odds may be artificially elevated. Approximately one-quarter of the patients in this study were transferred from an outside hospital, for which no data regarding initial treatment are available. This may lead to exposure misclassification. We attempted to control for this factor in the regression model and found that, after controlling for other covariates, hospital transfer was associated with increasing odds of AKI. Finally, data were collected retrospectively from the electronic medical record and are subject to inaccuracies documented in the chart; however, any bias introduced should be nondifferential.

In our large retrospective study of combination empiric therapy with VAN and PTZ, we found that combination therapy was associated with more than double the odds of AKI occurring compared to either monotherapy with VAN or PTZ. Increasing duration of therapy was also associated with increases in AKI. These findings demonstrate the need for judicious use of combination therapy and strengthen the need for antimicrobial de-escalation when appropriate to avoid deleterious effects.

Acknowledgments

The authors thank Chantal Le Rutter, MPA, for copyediting services.

Disclosures

This project was supported by the National Center for Advancing Translational Sciences, National Institutes of Health, through grant numbers UL1TR000117 and UL1TR001998. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The authors report no conflicts of interest.

Empiric antimicrobial therapy often consists of the combination of gram-positive coverage with vancomycin (VAN) and gram-negative coverage, specifically an antipseudomonal beta-lactam such as piperacillin-tazobactam (PTZ). Literature from a variety of patient populations reports nephrotoxicity associated with VAN, targeting troughs greater than 15 µg/mL, that occur in 5% to 43% of patients.¹ In a study of critically ill patients, acute kidney injury (AKI) was found in 21% of patients receiving VAN, with increasing duration of VAN treatment, greater VAN levels, concomitant vasoactive medication administration, and intermittent infusion methods being associated with higher odds of AKI.² A recent report from adult internal medicine patients estimated the incidence of VAN-associated nephrotoxicity at 13.6% and implicated concomitant PTZ therapy as a key factor in these patients.³

Further studies have explored the interaction between empiric beta-lactam and VAN therapy, showing mixed results. Reports of AKI associated with the combination of VAN and PTZ range from 16.3% to 34.8%,^4-8 while the cefepime-VAN combination is reported to range from 12.5% to 13.3%.^5,6 While VAN monotherapy groups were well represented, only 1 study⁷ compared the PTZ-VAN combination to a control group of PTZ monotherapy.

The primary objective of this study was to evaluate the differences in AKI incidence between patients treated with VAN and with PTZ, alone and in combination.

This is a retrospective cohort study of adult patients conducted at the University of Kentucky Chandler Medical Center (UKMC) from September 1, 2010 through August 31, 2014. Patients were included if they were at least 18 years of age on admission; remained hospitalized for at least 48 hours; received VAN combined with PTZ (VAN/PTZ), VAN alone, or PTZ alone; and had at least 48 hours of therapy (and 48 hours of overlapping therapy in the VAN/PTZ group). Patients were excluded if they had underlying diagnosis of chronic kidney disease according to the International Classification of Diseases 9 (ICD-9) code, were receiving renal replacement therapy before admission, had a diagnosis of cystic fibrosis, or were pregnant. Additionally, patients were excluded if they presented with AKI, defined as an initial creatinine clearance less than 30 mL/min, or if baseline creatinine clearance was greater than 4 times the standard deviation from the mean; serum creatinine values were not obtained during admission; and if AKI occurred prior to therapy initiation, within 48 hours of initiation, or more than 7 days after treatment was discontinued. Patients were followed throughout their stay until time of discharge.

Data Source

Patient data were collected from the University of Kentucky Center for Clinical and Translational Science Enterprise Data Trust (EDT). The EDT contains clinical data from the inpatient population of UKMC from 2006 to present. Data stored and updated nightly by the EDT includes: demographics, financial classification (Medicare, Medicaid, private insurance), provider-level detail (service line), medical diagnosis (ICD-9 codes), medical procedures (Current Procedural Terminology [CPT] codes), lab tests and results, medication administration details, visit details (age, length of stay, etc), and vital signs. This study was approved by the UKMC Institutional Review Board.

Data collected for each patient included: demographic data, visit details (length of stay, admitting and primary diagnosis codes, etc.), severity of underlying illness as defined by the Charlson Comorbidity Index (CCI), all serum creatinine levels drawn per visit, medication administration information (dose, date, and time administered), all VAN trough levels, receipt of other nephrotoxic agents, blood pressures, and receipt of vasopressors.

Outcome Ascertainment

The definition of AKI was based on the RIFLE (Risk, Injury, Failure, Loss, End-stage) criteria,⁹ with risk defined as a 25% to 50% decrease in estimated glomerular filtration rate (GFR), injury as a 50% to 75% decrease in estimated GFR, and failure defined as a greater than 75% decrease in estimated GFR. Loss and end-stage classifications were not assessed because of this study’s follow-up period. The adjusted Cockcroft and Gault equation¹⁰ was used to estimate GFR due to the inconsistency of weight availability in the dataset and concordance with the institution’s practice. Baseline creatinine clearance was calculated with the first serum creatinine obtained, and the minimum creatinine clearance was calculated using the maximum serum creatinine during each patient’s visit. The percent decrease in creatinine clearance was calculated from these 2 values. AKI status was defined as meeting any of the RIFLE criteria. Mortality was assessed for all patients and defined as the composite of inhospital mortality and discharge or transfer to hospice care.

Exposure Ascertainment

Hypotension exposure was defined as experiencing 1 of the following: mean arterial blood pressure less than 60 mm Hg, a diagnosis of hypotension by a physician, or receipt of vasopressors or inotropic agents. Days of therapy for each drug were obtained and combination days of therapy were calculated by including only those days in which the patient received both medications. Total days of therapy were calculated by the sum of all days receiving at least 1 study agent. Exposure to other nephrotoxic agents (eg, acyclovir, angiotensin converting enzyme [ACE] inhibitors, angiotensin II receptor antagonists, aminoglycosides, amphotericin B, cyclosporine, foscarnet, loop diuretics, nonsteroidal anti-inflammatory drugs, sulfonamides, tacrolimus, and tenofovir) were defined as receipt of at least 1 dose of the agent during hospitalization.

Statistical Analysis

Characteristics between groups were described with basic descriptive statistics. Continuous variables were compared with 1-way analysis of variance (ANOVA) or the Kruskal-Wallis test. Categorical variables were compared with chi-square or Fisher exact test. Yearly AKI trends were assessed with Pearson correlation coefficient. To control for differences in underlying severity of illness between groups, a subanalysis was performed in which the cohort was split into 4 groups (0, 1, 2 to 4, and ≥5 points) based on CCI. Univariate models for all covariates were created with probability of AKI as the outcome. Covariates significant after univariate were incorporated into the multivariate model, which was subsequently adjusted to achieve the highest predictive accuracy by minimizing the Akaike information criterion (AIC). Nephrotoxic agent exposures were included in the final multivariate model regardless of statistical significance in univariate analysis. Model fit was assessed with a standardized Hosmer-Lemeshow goodness-of-fit test.¹¹ All statistical analyses were completed with RStudio v 0.98 running R v 3.1.2 (R Foundation for Statistical Computing, Vienna, Austria).¹² All tests were 2-tailed and significance was defined at an alpha of 0.05.

Of 17,879 patients initially screened, 11,650 patients were evaluated, of which 5,497 received VAN and PTZ (VAN/PTZ), 3,055 received VAN alone, and 3,098 received PTZ alone. Table 1 contains basic demographic information. The mean age of patients was 52.5 years ± 16.8 years with 6,242 (53.6%) males. Patients receiving VAN/PTZ had higher CCIs than either monotherapy group and had significantly increased length of hospitalization. While patients in the combination therapy group were more likely to experience hypotension, concomitant nephrotoxic agent exposure was more common in the VAN monotherapy group.

RIFLE-defined AKI occurred in 1,647 (14.1%) across the entire cohort. AKI occurred in 21% of VAN/PTZ patients, 8.3% of VAN patients, and 7.8% of PTZ patients (P < 0.0001). RIFLE-defined risk, injury, and failure occurred more frequently in the VAN/PTZ cohort compared to the VAN and PTZ monotherapy groups (Figure). There were no differences in AKI rates between years studied (r² = 0.4732, P = 0.2). Patients in the VAN/PTZ group experienced AKI on average of 8.0 days after treatment initiation, compared to 8.7 days and 5.2 days for VAN and PTZ monotherapy groups, respectively. The composite of inhospital mortality and transfer-to-hospice care was more common in VAN/PTZ patients (9.6%) compared to monotherapy groups (VAN, 3.9%; PTZ, 3.4%), most likely due to the increased severity of illness.

Acute kidney injury secondary to VAN therapy is a well-characterized adverse effect, while AKI incidence secondary to PTZ is less understood. Additionally, there appears to be an additive effect when these agents are used in combination. This is the largest review of AKI in patients receiving VAN,PTZ, or the combination of both agents.

There is increasing evidence suggesting greater nephrotoxicity in patients treated with the combination of VAN and antipseudomonal beta-lactams. The mechanism for the apparent increase in nephrotoxicity with this drug combination is not well understood and needs further study in both animal models and humans.

Acute kidney injury rates related to VAN vary widely, with recent studies in critically ill and internal medicine patients estimated at 21% and 13.6%, respectively.^2,3 In our VAN monotherapy cohort, the AKI rate was 8.3%, with 2.3% of patients experiencing a greater than 50% decrease in creatinine clearance. Piperacillin-tazobactam-related AKI rates are not well characterized; however, a small retrospective analysis estimated that 11.1% of PTZ patients experienced acute renal failure (defined as either increase in serum creatinine greater than 0.5 mg/dL or 50% increase from baseline).¹³ In the present study, we found the PTZ-related AKI rate to be 7.8%, which may be due to a more stringent definition of AKI. Additionally, Hellwig et al¹³ found that PTZ monotherapy was associated with higher AKI rates compared to VAN monotherapy (11.1% vs 4.9%; P = 0.014). This was not replicated in our study, with VAN and PTZ monotherapy having similar AKI rates (8.3% and 7.8%, respectively) and an adjusted aOR of 0.88 (95% CI 0.0.73-1.08) for AKI in PTZ- compared to VAN-treated patients. The estimated AKI incidence of 21% in the combination therapy group at our institution is consistent with literature that ranges from 16.3% to 34.8%.^4-8,13

To control for differences in baseline severity of illness, we performed a subgroup analysis of patients with similar CCI scores. The finding of increased AKI in patients receiving combination VAN and PTZ was consistent in each subgroup, suggesting that the increase in AKI is independent of illness severity.

This study is not without limitations. As with all retrospective studies, it is difficult to determine a causal link between VAN and PTZ combination therapy and increased AKI incidence due to confounding. We employed a rigorous study design that controlled for major confounders of AKI, such as concomitant nephrotoxic exposure, hypotension, and renal disease. Severity of illness was measured with CCI, which may not accurately capture the severity of illness at treatment initiation. Alternatives, such as acute physiology and chronic health evaluation (APACHE) and sequential organ failure assessment (SOFA) scores, may more accurately reflect critical illness on presentation; however, this study was not focused specifically on critically ill patients. In addition to baseline comorbidity, we controlled for hypotension and dehydration as a surrogate marker for critical illness. In the subgroup analysis of patients with similar CCI, the effect of VAN/PTZ on AKI compared to VAN or PTZ monotherapy was consistent in each group. Nephrotoxic potential of agents was assumed to be equal, which is not necessarily true. Additionally, the binary representation of nephrotoxic exposure does not describe the amount of the agent received; as such, our estimations of AKI odds may be artificially elevated. Approximately one-quarter of the patients in this study were transferred from an outside hospital, for which no data regarding initial treatment are available. This may lead to exposure misclassification. We attempted to control for this factor in the regression model and found that, after controlling for other covariates, hospital transfer was associated with increasing odds of AKI. Finally, data were collected retrospectively from the electronic medical record and are subject to inaccuracies documented in the chart; however, any bias introduced should be nondifferential.

In our large retrospective study of combination empiric therapy with VAN and PTZ, we found that combination therapy was associated with more than double the odds of AKI occurring compared to either monotherapy with VAN or PTZ. Increasing duration of therapy was also associated with increases in AKI. These findings demonstrate the need for judicious use of combination therapy and strengthen the need for antimicrobial de-escalation when appropriate to avoid deleterious effects.

Acknowledgments

The authors thank Chantal Le Rutter, MPA, for copyediting services.

Disclosures

This project was supported by the National Center for Advancing Translational Sciences, National Institutes of Health, through grant numbers UL1TR000117 and UL1TR001998. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The authors report no conflicts of interest.

Empiric antimicrobial therapy often consists of the combination of gram-positive coverage with vancomycin (VAN) and gram-negative coverage, specifically an antipseudomonal beta-lactam such as piperacillin-tazobactam (PTZ). Literature from a variety of patient populations reports nephrotoxicity associated with VAN, targeting troughs greater than 15 µg/mL, that occur in 5% to 43% of patients.¹ In a study of critically ill patients, acute kidney injury (AKI) was found in 21% of patients receiving VAN, with increasing duration of VAN treatment, greater VAN levels, concomitant vasoactive medication administration, and intermittent infusion methods being associated with higher odds of AKI.² A recent report from adult internal medicine patients estimated the incidence of VAN-associated nephrotoxicity at 13.6% and implicated concomitant PTZ therapy as a key factor in these patients.³

Further studies have explored the interaction between empiric beta-lactam and VAN therapy, showing mixed results. Reports of AKI associated with the combination of VAN and PTZ range from 16.3% to 34.8%,^4-8 while the cefepime-VAN combination is reported to range from 12.5% to 13.3%.^5,6 While VAN monotherapy groups were well represented, only 1 study⁷ compared the PTZ-VAN combination to a control group of PTZ monotherapy.

The primary objective of this study was to evaluate the differences in AKI incidence between patients treated with VAN and with PTZ, alone and in combination.

This is a retrospective cohort study of adult patients conducted at the University of Kentucky Chandler Medical Center (UKMC) from September 1, 2010 through August 31, 2014. Patients were included if they were at least 18 years of age on admission; remained hospitalized for at least 48 hours; received VAN combined with PTZ (VAN/PTZ), VAN alone, or PTZ alone; and had at least 48 hours of therapy (and 48 hours of overlapping therapy in the VAN/PTZ group). Patients were excluded if they had underlying diagnosis of chronic kidney disease according to the International Classification of Diseases 9 (ICD-9) code, were receiving renal replacement therapy before admission, had a diagnosis of cystic fibrosis, or were pregnant. Additionally, patients were excluded if they presented with AKI, defined as an initial creatinine clearance less than 30 mL/min, or if baseline creatinine clearance was greater than 4 times the standard deviation from the mean; serum creatinine values were not obtained during admission; and if AKI occurred prior to therapy initiation, within 48 hours of initiation, or more than 7 days after treatment was discontinued. Patients were followed throughout their stay until time of discharge.

Data Source

Patient data were collected from the University of Kentucky Center for Clinical and Translational Science Enterprise Data Trust (EDT). The EDT contains clinical data from the inpatient population of UKMC from 2006 to present. Data stored and updated nightly by the EDT includes: demographics, financial classification (Medicare, Medicaid, private insurance), provider-level detail (service line), medical diagnosis (ICD-9 codes), medical procedures (Current Procedural Terminology [CPT] codes), lab tests and results, medication administration details, visit details (age, length of stay, etc), and vital signs. This study was approved by the UKMC Institutional Review Board.

Data collected for each patient included: demographic data, visit details (length of stay, admitting and primary diagnosis codes, etc.), severity of underlying illness as defined by the Charlson Comorbidity Index (CCI), all serum creatinine levels drawn per visit, medication administration information (dose, date, and time administered), all VAN trough levels, receipt of other nephrotoxic agents, blood pressures, and receipt of vasopressors.

Outcome Ascertainment

The definition of AKI was based on the RIFLE (Risk, Injury, Failure, Loss, End-stage) criteria,⁹ with risk defined as a 25% to 50% decrease in estimated glomerular filtration rate (GFR), injury as a 50% to 75% decrease in estimated GFR, and failure defined as a greater than 75% decrease in estimated GFR. Loss and end-stage classifications were not assessed because of this study’s follow-up period. The adjusted Cockcroft and Gault equation¹⁰ was used to estimate GFR due to the inconsistency of weight availability in the dataset and concordance with the institution’s practice. Baseline creatinine clearance was calculated with the first serum creatinine obtained, and the minimum creatinine clearance was calculated using the maximum serum creatinine during each patient’s visit. The percent decrease in creatinine clearance was calculated from these 2 values. AKI status was defined as meeting any of the RIFLE criteria. Mortality was assessed for all patients and defined as the composite of inhospital mortality and discharge or transfer to hospice care.

Exposure Ascertainment

Hypotension exposure was defined as experiencing 1 of the following: mean arterial blood pressure less than 60 mm Hg, a diagnosis of hypotension by a physician, or receipt of vasopressors or inotropic agents. Days of therapy for each drug were obtained and combination days of therapy were calculated by including only those days in which the patient received both medications. Total days of therapy were calculated by the sum of all days receiving at least 1 study agent. Exposure to other nephrotoxic agents (eg, acyclovir, angiotensin converting enzyme [ACE] inhibitors, angiotensin II receptor antagonists, aminoglycosides, amphotericin B, cyclosporine, foscarnet, loop diuretics, nonsteroidal anti-inflammatory drugs, sulfonamides, tacrolimus, and tenofovir) were defined as receipt of at least 1 dose of the agent during hospitalization.

Statistical Analysis

Characteristics between groups were described with basic descriptive statistics. Continuous variables were compared with 1-way analysis of variance (ANOVA) or the Kruskal-Wallis test. Categorical variables were compared with chi-square or Fisher exact test. Yearly AKI trends were assessed with Pearson correlation coefficient. To control for differences in underlying severity of illness between groups, a subanalysis was performed in which the cohort was split into 4 groups (0, 1, 2 to 4, and ≥5 points) based on CCI. Univariate models for all covariates were created with probability of AKI as the outcome. Covariates significant after univariate were incorporated into the multivariate model, which was subsequently adjusted to achieve the highest predictive accuracy by minimizing the Akaike information criterion (AIC). Nephrotoxic agent exposures were included in the final multivariate model regardless of statistical significance in univariate analysis. Model fit was assessed with a standardized Hosmer-Lemeshow goodness-of-fit test.¹¹ All statistical analyses were completed with RStudio v 0.98 running R v 3.1.2 (R Foundation for Statistical Computing, Vienna, Austria).¹² All tests were 2-tailed and significance was defined at an alpha of 0.05.

Of 17,879 patients initially screened, 11,650 patients were evaluated, of which 5,497 received VAN and PTZ (VAN/PTZ), 3,055 received VAN alone, and 3,098 received PTZ alone. Table 1 contains basic demographic information. The mean age of patients was 52.5 years ± 16.8 years with 6,242 (53.6%) males. Patients receiving VAN/PTZ had higher CCIs than either monotherapy group and had significantly increased length of hospitalization. While patients in the combination therapy group were more likely to experience hypotension, concomitant nephrotoxic agent exposure was more common in the VAN monotherapy group.

RIFLE-defined AKI occurred in 1,647 (14.1%) across the entire cohort. AKI occurred in 21% of VAN/PTZ patients, 8.3% of VAN patients, and 7.8% of PTZ patients (P < 0.0001). RIFLE-defined risk, injury, and failure occurred more frequently in the VAN/PTZ cohort compared to the VAN and PTZ monotherapy groups (Figure). There were no differences in AKI rates between years studied (r² = 0.4732, P = 0.2). Patients in the VAN/PTZ group experienced AKI on average of 8.0 days after treatment initiation, compared to 8.7 days and 5.2 days for VAN and PTZ monotherapy groups, respectively. The composite of inhospital mortality and transfer-to-hospice care was more common in VAN/PTZ patients (9.6%) compared to monotherapy groups (VAN, 3.9%; PTZ, 3.4%), most likely due to the increased severity of illness.

Acute kidney injury secondary to VAN therapy is a well-characterized adverse effect, while AKI incidence secondary to PTZ is less understood. Additionally, there appears to be an additive effect when these agents are used in combination. This is the largest review of AKI in patients receiving VAN,PTZ, or the combination of both agents.

There is increasing evidence suggesting greater nephrotoxicity in patients treated with the combination of VAN and antipseudomonal beta-lactams. The mechanism for the apparent increase in nephrotoxicity with this drug combination is not well understood and needs further study in both animal models and humans.

Acute kidney injury rates related to VAN vary widely, with recent studies in critically ill and internal medicine patients estimated at 21% and 13.6%, respectively.^2,3 In our VAN monotherapy cohort, the AKI rate was 8.3%, with 2.3% of patients experiencing a greater than 50% decrease in creatinine clearance. Piperacillin-tazobactam-related AKI rates are not well characterized; however, a small retrospective analysis estimated that 11.1% of PTZ patients experienced acute renal failure (defined as either increase in serum creatinine greater than 0.5 mg/dL or 50% increase from baseline).¹³ In the present study, we found the PTZ-related AKI rate to be 7.8%, which may be due to a more stringent definition of AKI. Additionally, Hellwig et al¹³ found that PTZ monotherapy was associated with higher AKI rates compared to VAN monotherapy (11.1% vs 4.9%; P = 0.014). This was not replicated in our study, with VAN and PTZ monotherapy having similar AKI rates (8.3% and 7.8%, respectively) and an adjusted aOR of 0.88 (95% CI 0.0.73-1.08) for AKI in PTZ- compared to VAN-treated patients. The estimated AKI incidence of 21% in the combination therapy group at our institution is consistent with literature that ranges from 16.3% to 34.8%.^4-8,13

To control for differences in baseline severity of illness, we performed a subgroup analysis of patients with similar CCI scores. The finding of increased AKI in patients receiving combination VAN and PTZ was consistent in each subgroup, suggesting that the increase in AKI is independent of illness severity.

This study is not without limitations. As with all retrospective studies, it is difficult to determine a causal link between VAN and PTZ combination therapy and increased AKI incidence due to confounding. We employed a rigorous study design that controlled for major confounders of AKI, such as concomitant nephrotoxic exposure, hypotension, and renal disease. Severity of illness was measured with CCI, which may not accurately capture the severity of illness at treatment initiation. Alternatives, such as acute physiology and chronic health evaluation (APACHE) and sequential organ failure assessment (SOFA) scores, may more accurately reflect critical illness on presentation; however, this study was not focused specifically on critically ill patients. In addition to baseline comorbidity, we controlled for hypotension and dehydration as a surrogate marker for critical illness. In the subgroup analysis of patients with similar CCI, the effect of VAN/PTZ on AKI compared to VAN or PTZ monotherapy was consistent in each group. Nephrotoxic potential of agents was assumed to be equal, which is not necessarily true. Additionally, the binary representation of nephrotoxic exposure does not describe the amount of the agent received; as such, our estimations of AKI odds may be artificially elevated. Approximately one-quarter of the patients in this study were transferred from an outside hospital, for which no data regarding initial treatment are available. This may lead to exposure misclassification. We attempted to control for this factor in the regression model and found that, after controlling for other covariates, hospital transfer was associated with increasing odds of AKI. Finally, data were collected retrospectively from the electronic medical record and are subject to inaccuracies documented in the chart; however, any bias introduced should be nondifferential.

In our large retrospective study of combination empiric therapy with VAN and PTZ, we found that combination therapy was associated with more than double the odds of AKI occurring compared to either monotherapy with VAN or PTZ. Increasing duration of therapy was also associated with increases in AKI. These findings demonstrate the need for judicious use of combination therapy and strengthen the need for antimicrobial de-escalation when appropriate to avoid deleterious effects.

Acknowledgments

The authors thank Chantal Le Rutter, MPA, for copyediting services.

Disclosures

This project was supported by the National Center for Advancing Translational Sciences, National Institutes of Health, through grant numbers UL1TR000117 and UL1TR001998. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. The authors report no conflicts of interest.

Take-Home Points

• Patients sustaining DRFs commonly have associated ligament injuries and chondral damage as well.
• Many of these associated injuries do not seem to affect outcomes up to 1 year after surgery.
• Plain radiographs have a 74% sensitivity and 73% specificity for detecting intra-articular fractures.
• ”Minor” injuries identified incidentally by arthroscopy during fixation of DRFs may not require dedicated treatment.
• The optimal treatment for high-grade ligament or chondral injuries in patients with DRFs remains incompletely understood.

Distal radius fracture (DRF) is one of the most common upper extremity injuries, with up to 20% to 50% requiring surgical fixation.¹ With increasing use of wrist arthroscopy to assist in managing these fractures,^2-6 it has become easier to accurately assess concomitant wrist ligament injuries. Reported injury rates are 18% to 86% for the scapholunate interosseous ligament (SLIL),^7,8 5% to 29% for the lunotriquetral ligament (LTL),^8,9 and 17% to 60% for the triangular fibrocartilage complex (TFCC).^10,11 Reported chondral injury rates range from 18% to 60%.^7,9,12 Despite the common occurrence of these injuries, it is unclear how they affect outcomes and how aggressively they should be treated when detected during fracture surgery.

As the use of arthroscopy in DRF management becomes more common, surgeons often must decide how to treat ligamentous/chondral injuries incidentally discovered during surgery. To date, only 1 study prospectively evaluated how these injuries affect DRF outcomes,⁸ though it did not use a validated, patient-based outcome measure.

We conducted a study to address a common clinical scenario: When arthroscopy is used to assist with intra-articular reduction during DRF fixation, how should the surgeon respond to incidentally identified ligament and chondral injuries? Specifically, we wanted to address 3 questions: What is the overall incidence of SLIL, TFCC, and chondral surface injuries in patients undergoing operative fracture fixation? On initial injury films, do any radiographic parameters predict specific soft-tissue injuries or ultimate functional outcomes? Do wrist ligament and chondral injuries affect patient-rated outcomes (disability, pain) and objective measures (range of motion [ROM], grip strength, pinch strength) up to 1 year after fracture surgery?

Materials and Methods

Patient Selection/Population

This observational, prognostic study was approved by our Institutional Review Board. Inclusion criteria were age over 18 years, isolated acute operatively treated DRF (surgery within 14 days of injury), and informed consent. All patients were treated by the same surgeon. Exclusion criteria were open DRF, dorsal shear pattern, fractures requiring dorsal arthrotomy for reduction because of significant intra-articular damage, prior ipsilateral DRF, and prior SLIL or TFCC injury.

Surgery was indicated according to general radiographic parameters as measured on postreduction films: radial height, <8 mm; radial inclination, <15°; positive ulnar variance, >3 mm, or 3 mm more than contralateral side; dorsal tilt, >10°; and volar tilt, >15°. With these parameters within acceptable limits, surgery was also indicated when fractures were deemed unstable and likely to displace because of dorsal tilt >20°, dorsal comminution, intra-articular step-off of ≥2 mm on the posterior-anterior (PA) film, associated ulnar fracture, and age >60 years.¹³Over a 2-year period, 42 patients (12 male, 30 female) met the inclusion criteria and were enrolled in the study. The dominant arm was affected in 17 patients (40%). Mean (SD) age at time of injury was 56.6 (16.4) years (median, 54 years; range, 20-85 years).

Operative Technique

During surgery, damage to the SLIL, the TFCC, and chondral surfaces (scaphoid, lunate, scaphoid fossa, lunate fossa) and to the intra-articular extension of the DRF was assessed and recorded. Wrist arthroscopy was performed with the 3, 4 portal as the primary portal. When significant damage to the TFCC warranted débridement, the 6R (radial) portal was used as an accessory portal. As a midcarpal portal was not used for SLIL assessment, we used a novel classification system: 0 = no injury, normal-appearing ligament without hemorrhage and smooth transition from scaphoid to lunate surface except for slight concave indentation at the ligament; 1 = attenuation, no visible tear with convex shape of ligament with or without hemorrhage; 2 = partial tear with or without step-off at junction between scaphoid and lunate, but 2.7-mm arthroscope cannot “drive through” to midcarpal joint; and 3 = complete tear with positive “drive-through” sign. TFCC injuries were classified according to the system described by Palmer¹⁴: Avulsions were central (1A), ulnar (1B), distal (1C), or radial (1D). The trampoline test was performed through a 6R portal by using a probe to evaluate ligament tension/laxity. In some cases, a 6R portal was deemed unnecessary, and a modified trampoline test was performed—tension/laxity/displacement was evaluated by manually palpating at the fovea and observing TFCC motion with the arthroscope. When appropriate, the TFCC was débrided with a shaver through the 6R portal. In cases of significant instability at the SLIL interval, two 0.062-inch K-wires were placed percutaneously through the scaphoid and lunate, and one was placed from the scaphoid to the capitate.

All DRFs underwent internal fixation with a locked volar plate. When necessary, K-wires and/or a locked radial column plate was used for additional fixation. External fixation was not used. The postoperative protocol began with a dorsal wrist splint placed on the patient in the operating room and worn for 10 to 14 days. At the first postoperative visit, the patient received a removable splint that was to be worn at all times except during showers, therapy, and home exercises. Occupational therapy, initiated the week of the first postoperative visit, consisted of active and passive ROM exercises. At 6 weeks, the splint was removed and strengthening initiated.

Outcome Measures

Our primary outcome measure was the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire at 1 year.¹⁵ Secondary outcome measures were visual analog scale (VAS) pain rating, ROM, and radiographic measurements. Patients returned for evaluation 2, 6, 12, 24, and 52 weeks after surgery. At each follow-up visit, the DASH questionnaire and the pain VAS were administered, and ROM and strength were measured. Patient-reported pain was recorded on a standard VAS and measured on a scale from 0 (no pain) to 10 (worst possible pain). Wrist flexion and extension and radioulnar deviation were assessed with a goniometer. Forearm supination and pronation were assessed with the elbow flexed 90° at the patient’s side. Grip strength was measured with a calibrated Jamar dynamometer (Sammons Preston Rolyan), and lateral pinch strength was measured with a hydraulic pinch gauge (Sammons Preston Rolyan). The average of 3 trials for both hands was recorded for all strength measurements.

Radiographs were obtained on presentation. When appropriate, the fracture was manually reduced with a hematoma block, and postreduction radiographs were obtained. Then, radiographs were obtained at each postoperative visit until union. Radial height, radial inclination, tilt, and ulnar variance were measured on preoperative and postoperative radiographs according to standard methods.¹⁶ Radiographs were used to classify the fracture patterns according to the AO/ASIF (Arbeitsgemeinschaft für Osteosynthesefragen/Association for the Study of Internal Fixation) classification. Union was determined by radiographic healing, absence of tenderness to palpation, absence of pain with motion, and continued functional improvement.

Data Analysis

To evaluate for relationships between patient injury parameters and outcome measures, we used a 1-way analysis of variance seeking statistically significant differences between groups. Patients were divided into 4 groups: no ligament injuries; isolated SLIL injuries; isolated TFCC injuries; and both SLIL and TFCC injuries. These injury classification categories were then evaluated independently against our chosen outcome measures, which included DASH and VAS pain scores, ROM, and grip/pinch strength.

To determine the optimal sample size, we performed a power analysis to estimate the number of patients required to detect a clinically significant difference in DASH scores at 1 year among the 4 groups. According to the literature, standard deviations of DASH scores in healthy volunteers range from 10 to 15,¹⁷ consistent with values found in other recent trials of patients with DRFs.¹⁸ The recent literature on DASH construct validity has established a DASH score difference of 19 as representing a disability change being “much better or much worse.”¹⁹ As such, power analysis for a 1-way analysis of variance among 4 categories, detecting a DASH score difference of 19 with a standard deviation ranging from 10 to 15, would require 28 to 60 patients to detect a difference with an α of 0.05 and a power of 0.8.

In addition, radiographic parameters at time of injury were compared with injury characteristics to assess for significant relationships. Multivariate linear regression analysis was performed to evaluate radial height, radial inclination, and volar tilt as possible predictors of SLIL injury, TFCC injury, and chondral surface damage. A statistically significant result was defined as a correlation with P < .05.

Results

Of the 42 patients included in the study, 11 (26%) had no ligament injuries, 10 (24%) had isolated SLIL injuries, 12 (29%) had isolated TFCC injuries, and 9 (21%) had injuries to both the SLIL and the TFCC. In addition, in 12 patients (29%), the articular cartilage had visible damage (Table 1).

In all patients, bony union occurred. After union, 1 patient underwent hardware removal for hardware-related pain. The same patient had a dorsal ulnar cutaneous nerve neurolysis at the ulnar styloid fixation site. Another patient developed a partial extensor pollicis longus tear from a prominent dorsal screw tip.

All patients returned for their 2- and 6-week follow-ups. At 1 year, 30 patients (71%) returned for follow-up, 11 could not be contacted, and 1 was removed because of an olecranon fracture from a subsequent fall.

Regarding the primary outcome measure, mean DASH score at 1-year follow-up was 30.8 for the group without injuries, 10.8 for the group with SLIL injuries, 14.7 for the group with TFCC injuries, and 21.9 for the group with SLIL and TFCC injuries (Table 2).

Discussion

Use of wrist arthroscopy in DRF management has allowed assessment of the incidence of intra-articular injuries, including ligament and chondral surface injuries. Although the literature on the incidence of these injuries has been expanding, their clinical significance remains unclear.

Authors have postulated that some patients do not do well after DRF repair because of undetected ligament injuries. With the current trend of internal fixation, locked plating, and early motion—contrasting with older trends of prolonged immobilization in a cast or external fixation—concerns have been raised that early mobilization results in inadequate treatment of ligament injuries. However, data from the present study suggest no significant morbidity from early mobilization despite the presence of ligament injuries in more than half of all operatively treated DRFs. It is possible morbidity was not appreciated, as most patients with DRFs end up with some stiffness, which masks the effects of ligament injuries during healing.

We found no correlation between injury radiographic parameters, observed soft-tissue injuries, or final subjective outcomes. Interestingly, in this study, there was some discordance between the appearance of intra-articular fractures on radiographs and the direct arthroscopic observation of intra-articular fracture extension. With the present data and with arthroscopic visualization as the gold standard, radiographs had 74% sensitivity and 73% specificity for detecting intra-articular fractures (the corresponding positive predictive value was 83%, and the negative predictive value was 61%). As we typically rely on radiographs as the primary tool in assessing the articular component of a fracture, these results should be taken into account when basing management decisions exclusively on static injury films.

Observational studies of arthroscopy in DRFs have revealed a wide range of injury rates: For SLILs, the average injury rate was 44%; for LTLs, 13%; for TFCCs, 43%; and for chondral surfaces, 32% (Table 4).

Am J Orthop. 2017;46(1):E41-E46. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• Patients sustaining DRFs commonly have associated ligament injuries and chondral damage as well.
• Many of these associated injuries do not seem to affect outcomes up to 1 year after surgery.
• Plain radiographs have a 74% sensitivity and 73% specificity for detecting intra-articular fractures.
• ”Minor” injuries identified incidentally by arthroscopy during fixation of DRFs may not require dedicated treatment.
• The optimal treatment for high-grade ligament or chondral injuries in patients with DRFs remains incompletely understood.

Distal radius fracture (DRF) is one of the most common upper extremity injuries, with up to 20% to 50% requiring surgical fixation.¹ With increasing use of wrist arthroscopy to assist in managing these fractures,^2-6 it has become easier to accurately assess concomitant wrist ligament injuries. Reported injury rates are 18% to 86% for the scapholunate interosseous ligament (SLIL),^7,8 5% to 29% for the lunotriquetral ligament (LTL),^8,9 and 17% to 60% for the triangular fibrocartilage complex (TFCC).^10,11 Reported chondral injury rates range from 18% to 60%.^7,9,12 Despite the common occurrence of these injuries, it is unclear how they affect outcomes and how aggressively they should be treated when detected during fracture surgery.

As the use of arthroscopy in DRF management becomes more common, surgeons often must decide how to treat ligamentous/chondral injuries incidentally discovered during surgery. To date, only 1 study prospectively evaluated how these injuries affect DRF outcomes,⁸ though it did not use a validated, patient-based outcome measure.

We conducted a study to address a common clinical scenario: When arthroscopy is used to assist with intra-articular reduction during DRF fixation, how should the surgeon respond to incidentally identified ligament and chondral injuries? Specifically, we wanted to address 3 questions: What is the overall incidence of SLIL, TFCC, and chondral surface injuries in patients undergoing operative fracture fixation? On initial injury films, do any radiographic parameters predict specific soft-tissue injuries or ultimate functional outcomes? Do wrist ligament and chondral injuries affect patient-rated outcomes (disability, pain) and objective measures (range of motion [ROM], grip strength, pinch strength) up to 1 year after fracture surgery?

Materials and Methods

Patient Selection/Population

This observational, prognostic study was approved by our Institutional Review Board. Inclusion criteria were age over 18 years, isolated acute operatively treated DRF (surgery within 14 days of injury), and informed consent. All patients were treated by the same surgeon. Exclusion criteria were open DRF, dorsal shear pattern, fractures requiring dorsal arthrotomy for reduction because of significant intra-articular damage, prior ipsilateral DRF, and prior SLIL or TFCC injury.

Surgery was indicated according to general radiographic parameters as measured on postreduction films: radial height, <8 mm; radial inclination, <15°; positive ulnar variance, >3 mm, or 3 mm more than contralateral side; dorsal tilt, >10°; and volar tilt, >15°. With these parameters within acceptable limits, surgery was also indicated when fractures were deemed unstable and likely to displace because of dorsal tilt >20°, dorsal comminution, intra-articular step-off of ≥2 mm on the posterior-anterior (PA) film, associated ulnar fracture, and age >60 years.¹³Over a 2-year period, 42 patients (12 male, 30 female) met the inclusion criteria and were enrolled in the study. The dominant arm was affected in 17 patients (40%). Mean (SD) age at time of injury was 56.6 (16.4) years (median, 54 years; range, 20-85 years).

Operative Technique

During surgery, damage to the SLIL, the TFCC, and chondral surfaces (scaphoid, lunate, scaphoid fossa, lunate fossa) and to the intra-articular extension of the DRF was assessed and recorded. Wrist arthroscopy was performed with the 3, 4 portal as the primary portal. When significant damage to the TFCC warranted débridement, the 6R (radial) portal was used as an accessory portal. As a midcarpal portal was not used for SLIL assessment, we used a novel classification system: 0 = no injury, normal-appearing ligament without hemorrhage and smooth transition from scaphoid to lunate surface except for slight concave indentation at the ligament; 1 = attenuation, no visible tear with convex shape of ligament with or without hemorrhage; 2 = partial tear with or without step-off at junction between scaphoid and lunate, but 2.7-mm arthroscope cannot “drive through” to midcarpal joint; and 3 = complete tear with positive “drive-through” sign. TFCC injuries were classified according to the system described by Palmer¹⁴: Avulsions were central (1A), ulnar (1B), distal (1C), or radial (1D). The trampoline test was performed through a 6R portal by using a probe to evaluate ligament tension/laxity. In some cases, a 6R portal was deemed unnecessary, and a modified trampoline test was performed—tension/laxity/displacement was evaluated by manually palpating at the fovea and observing TFCC motion with the arthroscope. When appropriate, the TFCC was débrided with a shaver through the 6R portal. In cases of significant instability at the SLIL interval, two 0.062-inch K-wires were placed percutaneously through the scaphoid and lunate, and one was placed from the scaphoid to the capitate.

All DRFs underwent internal fixation with a locked volar plate. When necessary, K-wires and/or a locked radial column plate was used for additional fixation. External fixation was not used. The postoperative protocol began with a dorsal wrist splint placed on the patient in the operating room and worn for 10 to 14 days. At the first postoperative visit, the patient received a removable splint that was to be worn at all times except during showers, therapy, and home exercises. Occupational therapy, initiated the week of the first postoperative visit, consisted of active and passive ROM exercises. At 6 weeks, the splint was removed and strengthening initiated.

Outcome Measures

Our primary outcome measure was the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire at 1 year.¹⁵ Secondary outcome measures were visual analog scale (VAS) pain rating, ROM, and radiographic measurements. Patients returned for evaluation 2, 6, 12, 24, and 52 weeks after surgery. At each follow-up visit, the DASH questionnaire and the pain VAS were administered, and ROM and strength were measured. Patient-reported pain was recorded on a standard VAS and measured on a scale from 0 (no pain) to 10 (worst possible pain). Wrist flexion and extension and radioulnar deviation were assessed with a goniometer. Forearm supination and pronation were assessed with the elbow flexed 90° at the patient’s side. Grip strength was measured with a calibrated Jamar dynamometer (Sammons Preston Rolyan), and lateral pinch strength was measured with a hydraulic pinch gauge (Sammons Preston Rolyan). The average of 3 trials for both hands was recorded for all strength measurements.

Radiographs were obtained on presentation. When appropriate, the fracture was manually reduced with a hematoma block, and postreduction radiographs were obtained. Then, radiographs were obtained at each postoperative visit until union. Radial height, radial inclination, tilt, and ulnar variance were measured on preoperative and postoperative radiographs according to standard methods.¹⁶ Radiographs were used to classify the fracture patterns according to the AO/ASIF (Arbeitsgemeinschaft für Osteosynthesefragen/Association for the Study of Internal Fixation) classification. Union was determined by radiographic healing, absence of tenderness to palpation, absence of pain with motion, and continued functional improvement.

Data Analysis

To evaluate for relationships between patient injury parameters and outcome measures, we used a 1-way analysis of variance seeking statistically significant differences between groups. Patients were divided into 4 groups: no ligament injuries; isolated SLIL injuries; isolated TFCC injuries; and both SLIL and TFCC injuries. These injury classification categories were then evaluated independently against our chosen outcome measures, which included DASH and VAS pain scores, ROM, and grip/pinch strength.

To determine the optimal sample size, we performed a power analysis to estimate the number of patients required to detect a clinically significant difference in DASH scores at 1 year among the 4 groups. According to the literature, standard deviations of DASH scores in healthy volunteers range from 10 to 15,¹⁷ consistent with values found in other recent trials of patients with DRFs.¹⁸ The recent literature on DASH construct validity has established a DASH score difference of 19 as representing a disability change being “much better or much worse.”¹⁹ As such, power analysis for a 1-way analysis of variance among 4 categories, detecting a DASH score difference of 19 with a standard deviation ranging from 10 to 15, would require 28 to 60 patients to detect a difference with an α of 0.05 and a power of 0.8.

In addition, radiographic parameters at time of injury were compared with injury characteristics to assess for significant relationships. Multivariate linear regression analysis was performed to evaluate radial height, radial inclination, and volar tilt as possible predictors of SLIL injury, TFCC injury, and chondral surface damage. A statistically significant result was defined as a correlation with P < .05.

Results

Of the 42 patients included in the study, 11 (26%) had no ligament injuries, 10 (24%) had isolated SLIL injuries, 12 (29%) had isolated TFCC injuries, and 9 (21%) had injuries to both the SLIL and the TFCC. In addition, in 12 patients (29%), the articular cartilage had visible damage (Table 1).

In all patients, bony union occurred. After union, 1 patient underwent hardware removal for hardware-related pain. The same patient had a dorsal ulnar cutaneous nerve neurolysis at the ulnar styloid fixation site. Another patient developed a partial extensor pollicis longus tear from a prominent dorsal screw tip.

All patients returned for their 2- and 6-week follow-ups. At 1 year, 30 patients (71%) returned for follow-up, 11 could not be contacted, and 1 was removed because of an olecranon fracture from a subsequent fall.

Regarding the primary outcome measure, mean DASH score at 1-year follow-up was 30.8 for the group without injuries, 10.8 for the group with SLIL injuries, 14.7 for the group with TFCC injuries, and 21.9 for the group with SLIL and TFCC injuries (Table 2).

Discussion

Use of wrist arthroscopy in DRF management has allowed assessment of the incidence of intra-articular injuries, including ligament and chondral surface injuries. Although the literature on the incidence of these injuries has been expanding, their clinical significance remains unclear.

Authors have postulated that some patients do not do well after DRF repair because of undetected ligament injuries. With the current trend of internal fixation, locked plating, and early motion—contrasting with older trends of prolonged immobilization in a cast or external fixation—concerns have been raised that early mobilization results in inadequate treatment of ligament injuries. However, data from the present study suggest no significant morbidity from early mobilization despite the presence of ligament injuries in more than half of all operatively treated DRFs. It is possible morbidity was not appreciated, as most patients with DRFs end up with some stiffness, which masks the effects of ligament injuries during healing.

We found no correlation between injury radiographic parameters, observed soft-tissue injuries, or final subjective outcomes. Interestingly, in this study, there was some discordance between the appearance of intra-articular fractures on radiographs and the direct arthroscopic observation of intra-articular fracture extension. With the present data and with arthroscopic visualization as the gold standard, radiographs had 74% sensitivity and 73% specificity for detecting intra-articular fractures (the corresponding positive predictive value was 83%, and the negative predictive value was 61%). As we typically rely on radiographs as the primary tool in assessing the articular component of a fracture, these results should be taken into account when basing management decisions exclusively on static injury films.

Observational studies of arthroscopy in DRFs have revealed a wide range of injury rates: For SLILs, the average injury rate was 44%; for LTLs, 13%; for TFCCs, 43%; and for chondral surfaces, 32% (Table 4).

Am J Orthop. 2017;46(1):E41-E46. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• Patients sustaining DRFs commonly have associated ligament injuries and chondral damage as well.
• Many of these associated injuries do not seem to affect outcomes up to 1 year after surgery.
• Plain radiographs have a 74% sensitivity and 73% specificity for detecting intra-articular fractures.
• ”Minor” injuries identified incidentally by arthroscopy during fixation of DRFs may not require dedicated treatment.
• The optimal treatment for high-grade ligament or chondral injuries in patients with DRFs remains incompletely understood.

Distal radius fracture (DRF) is one of the most common upper extremity injuries, with up to 20% to 50% requiring surgical fixation.¹ With increasing use of wrist arthroscopy to assist in managing these fractures,^2-6 it has become easier to accurately assess concomitant wrist ligament injuries. Reported injury rates are 18% to 86% for the scapholunate interosseous ligament (SLIL),^7,8 5% to 29% for the lunotriquetral ligament (LTL),^8,9 and 17% to 60% for the triangular fibrocartilage complex (TFCC).^10,11 Reported chondral injury rates range from 18% to 60%.^7,9,12 Despite the common occurrence of these injuries, it is unclear how they affect outcomes and how aggressively they should be treated when detected during fracture surgery.

As the use of arthroscopy in DRF management becomes more common, surgeons often must decide how to treat ligamentous/chondral injuries incidentally discovered during surgery. To date, only 1 study prospectively evaluated how these injuries affect DRF outcomes,⁸ though it did not use a validated, patient-based outcome measure.

We conducted a study to address a common clinical scenario: When arthroscopy is used to assist with intra-articular reduction during DRF fixation, how should the surgeon respond to incidentally identified ligament and chondral injuries? Specifically, we wanted to address 3 questions: What is the overall incidence of SLIL, TFCC, and chondral surface injuries in patients undergoing operative fracture fixation? On initial injury films, do any radiographic parameters predict specific soft-tissue injuries or ultimate functional outcomes? Do wrist ligament and chondral injuries affect patient-rated outcomes (disability, pain) and objective measures (range of motion [ROM], grip strength, pinch strength) up to 1 year after fracture surgery?

Materials and Methods

Patient Selection/Population

This observational, prognostic study was approved by our Institutional Review Board. Inclusion criteria were age over 18 years, isolated acute operatively treated DRF (surgery within 14 days of injury), and informed consent. All patients were treated by the same surgeon. Exclusion criteria were open DRF, dorsal shear pattern, fractures requiring dorsal arthrotomy for reduction because of significant intra-articular damage, prior ipsilateral DRF, and prior SLIL or TFCC injury.

Surgery was indicated according to general radiographic parameters as measured on postreduction films: radial height, <8 mm; radial inclination, <15°; positive ulnar variance, >3 mm, or 3 mm more than contralateral side; dorsal tilt, >10°; and volar tilt, >15°. With these parameters within acceptable limits, surgery was also indicated when fractures were deemed unstable and likely to displace because of dorsal tilt >20°, dorsal comminution, intra-articular step-off of ≥2 mm on the posterior-anterior (PA) film, associated ulnar fracture, and age >60 years.¹³Over a 2-year period, 42 patients (12 male, 30 female) met the inclusion criteria and were enrolled in the study. The dominant arm was affected in 17 patients (40%). Mean (SD) age at time of injury was 56.6 (16.4) years (median, 54 years; range, 20-85 years).

Operative Technique

During surgery, damage to the SLIL, the TFCC, and chondral surfaces (scaphoid, lunate, scaphoid fossa, lunate fossa) and to the intra-articular extension of the DRF was assessed and recorded. Wrist arthroscopy was performed with the 3, 4 portal as the primary portal. When significant damage to the TFCC warranted débridement, the 6R (radial) portal was used as an accessory portal. As a midcarpal portal was not used for SLIL assessment, we used a novel classification system: 0 = no injury, normal-appearing ligament without hemorrhage and smooth transition from scaphoid to lunate surface except for slight concave indentation at the ligament; 1 = attenuation, no visible tear with convex shape of ligament with or without hemorrhage; 2 = partial tear with or without step-off at junction between scaphoid and lunate, but 2.7-mm arthroscope cannot “drive through” to midcarpal joint; and 3 = complete tear with positive “drive-through” sign. TFCC injuries were classified according to the system described by Palmer¹⁴: Avulsions were central (1A), ulnar (1B), distal (1C), or radial (1D). The trampoline test was performed through a 6R portal by using a probe to evaluate ligament tension/laxity. In some cases, a 6R portal was deemed unnecessary, and a modified trampoline test was performed—tension/laxity/displacement was evaluated by manually palpating at the fovea and observing TFCC motion with the arthroscope. When appropriate, the TFCC was débrided with a shaver through the 6R portal. In cases of significant instability at the SLIL interval, two 0.062-inch K-wires were placed percutaneously through the scaphoid and lunate, and one was placed from the scaphoid to the capitate.

All DRFs underwent internal fixation with a locked volar plate. When necessary, K-wires and/or a locked radial column plate was used for additional fixation. External fixation was not used. The postoperative protocol began with a dorsal wrist splint placed on the patient in the operating room and worn for 10 to 14 days. At the first postoperative visit, the patient received a removable splint that was to be worn at all times except during showers, therapy, and home exercises. Occupational therapy, initiated the week of the first postoperative visit, consisted of active and passive ROM exercises. At 6 weeks, the splint was removed and strengthening initiated.

Outcome Measures

Our primary outcome measure was the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire at 1 year.¹⁵ Secondary outcome measures were visual analog scale (VAS) pain rating, ROM, and radiographic measurements. Patients returned for evaluation 2, 6, 12, 24, and 52 weeks after surgery. At each follow-up visit, the DASH questionnaire and the pain VAS were administered, and ROM and strength were measured. Patient-reported pain was recorded on a standard VAS and measured on a scale from 0 (no pain) to 10 (worst possible pain). Wrist flexion and extension and radioulnar deviation were assessed with a goniometer. Forearm supination and pronation were assessed with the elbow flexed 90° at the patient’s side. Grip strength was measured with a calibrated Jamar dynamometer (Sammons Preston Rolyan), and lateral pinch strength was measured with a hydraulic pinch gauge (Sammons Preston Rolyan). The average of 3 trials for both hands was recorded for all strength measurements.

Radiographs were obtained on presentation. When appropriate, the fracture was manually reduced with a hematoma block, and postreduction radiographs were obtained. Then, radiographs were obtained at each postoperative visit until union. Radial height, radial inclination, tilt, and ulnar variance were measured on preoperative and postoperative radiographs according to standard methods.¹⁶ Radiographs were used to classify the fracture patterns according to the AO/ASIF (Arbeitsgemeinschaft für Osteosynthesefragen/Association for the Study of Internal Fixation) classification. Union was determined by radiographic healing, absence of tenderness to palpation, absence of pain with motion, and continued functional improvement.

Data Analysis

To evaluate for relationships between patient injury parameters and outcome measures, we used a 1-way analysis of variance seeking statistically significant differences between groups. Patients were divided into 4 groups: no ligament injuries; isolated SLIL injuries; isolated TFCC injuries; and both SLIL and TFCC injuries. These injury classification categories were then evaluated independently against our chosen outcome measures, which included DASH and VAS pain scores, ROM, and grip/pinch strength.

To determine the optimal sample size, we performed a power analysis to estimate the number of patients required to detect a clinically significant difference in DASH scores at 1 year among the 4 groups. According to the literature, standard deviations of DASH scores in healthy volunteers range from 10 to 15,¹⁷ consistent with values found in other recent trials of patients with DRFs.¹⁸ The recent literature on DASH construct validity has established a DASH score difference of 19 as representing a disability change being “much better or much worse.”¹⁹ As such, power analysis for a 1-way analysis of variance among 4 categories, detecting a DASH score difference of 19 with a standard deviation ranging from 10 to 15, would require 28 to 60 patients to detect a difference with an α of 0.05 and a power of 0.8.

In addition, radiographic parameters at time of injury were compared with injury characteristics to assess for significant relationships. Multivariate linear regression analysis was performed to evaluate radial height, radial inclination, and volar tilt as possible predictors of SLIL injury, TFCC injury, and chondral surface damage. A statistically significant result was defined as a correlation with P < .05.

Results

Of the 42 patients included in the study, 11 (26%) had no ligament injuries, 10 (24%) had isolated SLIL injuries, 12 (29%) had isolated TFCC injuries, and 9 (21%) had injuries to both the SLIL and the TFCC. In addition, in 12 patients (29%), the articular cartilage had visible damage (Table 1).

In all patients, bony union occurred. After union, 1 patient underwent hardware removal for hardware-related pain. The same patient had a dorsal ulnar cutaneous nerve neurolysis at the ulnar styloid fixation site. Another patient developed a partial extensor pollicis longus tear from a prominent dorsal screw tip.

All patients returned for their 2- and 6-week follow-ups. At 1 year, 30 patients (71%) returned for follow-up, 11 could not be contacted, and 1 was removed because of an olecranon fracture from a subsequent fall.

Regarding the primary outcome measure, mean DASH score at 1-year follow-up was 30.8 for the group without injuries, 10.8 for the group with SLIL injuries, 14.7 for the group with TFCC injuries, and 21.9 for the group with SLIL and TFCC injuries (Table 2).

Discussion

Use of wrist arthroscopy in DRF management has allowed assessment of the incidence of intra-articular injuries, including ligament and chondral surface injuries. Although the literature on the incidence of these injuries has been expanding, their clinical significance remains unclear.

Authors have postulated that some patients do not do well after DRF repair because of undetected ligament injuries. With the current trend of internal fixation, locked plating, and early motion—contrasting with older trends of prolonged immobilization in a cast or external fixation—concerns have been raised that early mobilization results in inadequate treatment of ligament injuries. However, data from the present study suggest no significant morbidity from early mobilization despite the presence of ligament injuries in more than half of all operatively treated DRFs. It is possible morbidity was not appreciated, as most patients with DRFs end up with some stiffness, which masks the effects of ligament injuries during healing.

We found no correlation between injury radiographic parameters, observed soft-tissue injuries, or final subjective outcomes. Interestingly, in this study, there was some discordance between the appearance of intra-articular fractures on radiographs and the direct arthroscopic observation of intra-articular fracture extension. With the present data and with arthroscopic visualization as the gold standard, radiographs had 74% sensitivity and 73% specificity for detecting intra-articular fractures (the corresponding positive predictive value was 83%, and the negative predictive value was 61%). As we typically rely on radiographs as the primary tool in assessing the articular component of a fracture, these results should be taken into account when basing management decisions exclusively on static injury films.

Observational studies of arthroscopy in DRFs have revealed a wide range of injury rates: For SLILs, the average injury rate was 44%; for LTLs, 13%; for TFCCs, 43%; and for chondral surfaces, 32% (Table 4).

Am J Orthop. 2017;46(1):E41-E46. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• DVT and PE are uncommon complications following osteotomies about the knee.
• Use of oral contraceptives can increase the risk of a patient sustaining a postoperative DVT and PE following osteotomies about the knee.
• In the absence of significant risk factors, postoperative chemical DVT prophylaxis may be unnecessary in patients undergoing osteotomies about the knee.

High tibial osteotomy (HTO), distal femoral osteotomy (DFO), and tibial tubercle osteotomy (TTO) are viable treatment options for deformities about the knee and patella maltracking.^1-4 Although TTO can be performed in many ways (eg, anteriorization, anteromedialization, medialization), the basic idea is to move the tibial tubercle to improve patellar tracking or to offload a patellar facet that has sustained trauma or degenerated.² DFO is a surgical option for treating a valgus knee deformity (the lateral tibiofemoral compartment is offloaded) or for protecting a knee compartment after cartilage or meniscal restoration (medial closing wedge or lateral opening wedge).¹ Similarly, HTO is an option for treating a varus knee deformity or isolated medial compartment arthritis; the diseased compartment is offloaded, and any malalignment is corrected. Akin to DFO, HTO is often performed to protect a knee compartment, typically the medial tibiofemoral compartment, after cartilage or meniscal restoration.^2-4

Compared to most arthroscopic knee surgeries, these osteotomies are much more involved, have longer operative times, and restrict postoperative weight-bearing and range of motion.^2-4 The rates of deep vein thrombosis (DVT) and pulmonary embolism (PE) after these osteotomies are not well documented. In addition, there is no documentation of the risks in patients who smoke, are obese, or are using oral contraceptives (OCs) at time of surgery, despite the increased DVT and PE risks posed by smoking, obesity, and OC use in other surgical procedures.^5-7 Although the American Academy of Orthopaedic Surgeons (AAOS) issued clinical practice guidelines for DVT/PE prophylaxis after hip and knee arthroplasty, there is no standard prophylaxis guidelines for DVT/PE prevention after HTO, DFO, or TTO.^8,9 Last, rates of DVT after total knee arthroplasty (TKA) are well defined; they range from 2% to 12%.^10,11 These rates may be surrogates for osteotomies about the knee, but this is only conjecture.

We conducted a study to determine the rates of symptomatic DVT and PE after HTO, DFO, or TTO in patients who did not receive postoperative DVT/PE prophylaxis. We also wanted to determine if age, body mass index (BMI), and smoking status have associations with the risk of developing either DVT or PE after HTO, DFO, or TTO. We hypothesized that the DVT and PE rates would both be <1%.

Methods

After this study was approved by our university’s Institutional Review Board, we searched the surgical database of Dr. Cole, a sports medicine fellowship–trained surgeon, to identify all patients who had HTO, DFO, or TTO performed between September 1, 2009 and September 30, 2014. Current Procedural Terminology (CPT) codes were used for the search. The code for HTO was 27457: osteotomy, proximal tibia, including fibular excision or osteotomy (includes correction of genu varus [bowleg] or genu valgus [knock-knee]); after epiphyseal closure). The code for DFO was 27450: osteotomy, femur, shaft or supracondylar; with fixation. Last, the code for TTO was 27418: anterior tibial tubercleplasty (eg, Maquet-type procedure). The 141 patients identified in the search were treated by Dr. Cole at a single institution and were included in the study. Study inclusion did not require a minimum follow-up. Follow-up duration was defined as the time between surgery and the final clinic note in the patient chart. No patient was excluded for lack of follow-up clinic visits, and none was lost to follow-up.

Age, BMI, smoking status, and OC use were recorded for all patients. For each procedure, the surgeon’s technique remained the same throughout the study period: HTO, medial opening-wedge osteotomy with plate-and-screw fixation; DFO, lateral opening-wedge osteotomy with plate-and-screw fixation; and TTO, mostly anteromedialization with screw fixation (though this was dictated by patellar contact pressures). A tourniquet was used in all cases. Each patient’s hospital electronic medical record and outpatient office notes were reviewed to determine if symptomatic DVT or PE developed after surgery. The diagnosis of symptomatic DVT was based on clinical symptoms and confirmatory ultrasound, and the PE diagnosis was based on computed tomography. Doppler ultrasound was performed only in symptomatic patients (ie, it was not routinely performed).

Per surgeon protocol, postoperative DVT prophylaxis was not administered. Patients were encouraged to begin dorsiflexion and plantar flexion of the ankle (ankle pumps) immediately and to mobilize as soon as comfortable. Each patient received a cold therapy machine with compression sleeve. Patients were allowed toe-touch weight-bearing for 6 weeks, and then progressed 25% per week for 4 weeks to full weight-bearing by 10 weeks. After surgery, each patient was placed in a brace, which was kept locked in extension for 10 days; when the brace was unlocked, the patient was allowed to range the knee.

Continuous variable data are reported as weighted means and weighted standard deviations. Categorical variable data are reported as frequencies and percentages.

Results

Our database search identified 141 patients (44% male, 56% female) who underwent HTO (47 patients, 33.3%), DFO (13 patients, 9.2%), or TTO (81 patients, 57.5%). Mean (SD) age was 34.28 (9.86) years, mean (SD) BMI was 26.88 (5.11) kg/m², and mean (SD) follow-up was 17.1 (4.1) months. Of the female patients, 36.7% were using OCs at time of surgery. Of all patients, 13.48% were smokers.

Two patients (1.42%) had clinical symptoms consistent with DVT. In each case, the diagnosis was confirmed with Doppler ultrasound. The below-knee DVT was unilateral in 1 case and bilateral in the other.

Discussion

As the rates of DVT and PE after osteotomies about the knee have not been well studied, we wanted to determine these rates after HTO, DFO, and TTO in patients who did not receive postoperative DVT prophylaxis. We hypothesized that DVT and PE rates would both be <1%, and this hypothesis was partly confirmed: The rate of PE after HTO, DFO, and TTO was <1%, and the rate of symptomatic DVT was >1%. Similarly, the patients who developed these complications were nonsmokers and had a BMI no higher than that of the patients who did not develop DVT or PE. In addition, only 1 patient developed DVT and PE, and she was using OCs and had a family history of DVT. Last, the patients who developed these complications were on average 14 years older than the patients who did not develop DVT or PE.

Although there is a plethora of reports on the incidence of DVT and PE after TKA, there is little on the incidence after osteotomies about the knee.^8,12 The rate of DVT after TKA varies, but many studies place it between 2% and 12%, and routinely find a PE rate of <0.5%.^10,11,13,14 Although the AAOS issued a clinical practice guideline for postoperative DVT prophylaxis after TKA, and evaluated the best available evidence, it could not reach consensus on a specific type of DVT prophylaxis, though the workgroup did recommend that patients be administered postoperative DVT prophylaxis of some kind.^8,9 Similarly, the American College of Chest Physicians (ACCP) issued clinical practice guidelines for preventing DVT and PE after elective TKA and total hip arthroplasty.¹⁵ According to the ACCP guidelines, patients should receive prophylaxis—LMWH, fondaparinux, apixaban, dabigatran, rivaroxaban, low-dose unfractionated heparin, adjusted-dose vitamin K antagonist, aspirin, or an intermittent pneumatic compression device—for a minimum of 14 days. Unfortunately, though there are similarities between TKAs and peri-knee osteotomies, these procedures are markedly different, and it is difficult to extrapolate and adapt recommendations and produce a consensus statement for knee arthroplasties. In addition, guidelines exist for hospitalized patients who are being treated for medical conditions or have undergone surgery, but all the patients in the present study had their osteotomies performed on an outpatient basis.

Martin and colleagues¹⁶ reviewed 323 cases of medial opening-wedge HTO and found a DVT rate of 1.4% in the absence of routine DVT prophylaxis, except in patients with a history of DVT. Their rate is almost identical to ours, but we also included other osteotomies in our study. Miller and colleagues¹⁷ reviewed 46 cases of medial opening-wedge HTO and found a 4.3% DVT rate, despite routine prophylaxis with once-daily 325-mg aspirin and ankle pumps. This finding contrasts with our 1.42% DVT rate in the absence of postoperative chemical DVT prophylaxis. Motycka and colleagues¹⁸ reviewed 65 HTO cases in which DVT prophylaxis (oral anticoagulant) was given for 6 weeks, and they found a DVT rate of 9.7%. Turner and colleagues¹⁹ performed venous ultrasound on 81 consecutive patients who underwent HTO and received DVT prophylaxis (twice-daily subcutaneous heparin), and they found a DVT rate of 41% and a PE rate of 1.2%, though only 8.6% of the DVT cases were symptomatic. Of note, whereas the lowest postoperative DVT rate was for patients who did not receive postoperative DVT prophylaxis, the rate of symptomatic DVT after these osteotomies ranged from 1.4% to 8.6% in patients who received prophylaxis.^16,19 Given this evidence and our study results, it appears routine chemical DVT prophylaxis after osteotomies about the knee may not be necessary, though higher level evidence is needed in order to make definitive recommendations.

In the present study, the 2 patients who developed symptomatic DVT (1 subsequently developed PE) were nonsmokers in good health. The female patient (DVT plus PE) was using OCs at time of surgery. Studies have shown that patients who smoke and who use OCs are at increased risk for developing DVT or PE after surgery.^5,6,12 Given that only 2 of our patients developed DVT/PE, and neither was a smoker, smoking was not associated with increased DVT or PE risk in this study population, in which 13.48% of patients were smokers at time of surgery. In addition, given that the 1 female patient who developed DVT/PE was using OCs and that 36.7% of all female patients in the study were using OCs, it is difficult to conclude whether OC use increased the female patient’s risk for DVT or PE. Furthermore, neither the literature nor the AAOS consensus statement supports discontinuing OCs for this surgical procedure.

Patients in this study did not receive chemical or mechanical DVT prophylaxis after surgery. Regarding various post-TKA DVT prophylaxis regimens, aspirin is as effective as LMWH in preventing DVT, and the risk for postoperative blood loss and wound complications is lower with aspirin than with rivaroxaban.^20,21 Given that the present study’s postoperative rates of DVT (1.42%) and PE (0.71%) are equal to or less than rates already reported in the literature, routine DVT prophylaxis after osteotomies about the knee may be unnecessary in the absence of other significant risk factors.^16,19 However, our study considered only symptomatic DVT and PE, so it is possible that the number of asymptomatic DVT cases is higher in this patient population. Definitively answering our study’s clinical question will require a multicenter registry study (prospective cohort study).

Study Limitations

The strengths of this study include the large number of patients treated by a single surgeon using the same postoperative protocol. Limitations of this study include the lack of a control group. Although we found a DVT rate of 1.42% and a PE rate of 0.71%, the literature on the accepted risks for DVT and PE after HTO, DFO, and TTO is unclear. With our results stratified by procedure, the DVT rate was 2% in the HTO group, 0% in the DFO group, and 1% in the TTO group. However, we were unable to reliably stratify these results by each specific procedure, as the number of patients in each group would be too low. This study involved reviewing charts; as patients were not contacted, it is possible a patient developed DVT or PE, was treated at an outside facility, and then never followed up with the treating surgeon. Patients were identified by CPT codes, so, if a patient underwent HTO, DFO, or TTO that was recorded under a different CPT code, it is possible the patient was missed by our search. All patients were seen after surgery, and we reviewed the outpatient office notes that were taken, so unless the DVT or PE occurred after a patient’s final postoperative visit, it would have been recorded. Similarly, the DVT and PE rates reported here cannot be extrapolated to overall risks for DVT and PE after osteotomies about the knee in all patients—only in patients who did not receive DVT prophylaxis after surgery.

Conclusion

The rates of DVT and PE after HTO, DFO, and TTO in patients who did not receive chemical prophylaxis are low: 1.42% and 0.71%, respectively. After these osteotomies, DVT/PE prophylaxis in the absence of known risk factors may not be warranted.

Am J Orthop. 2017;46(1):E23-E27. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• DVT and PE are uncommon complications following osteotomies about the knee.
• Use of oral contraceptives can increase the risk of a patient sustaining a postoperative DVT and PE following osteotomies about the knee.
• In the absence of significant risk factors, postoperative chemical DVT prophylaxis may be unnecessary in patients undergoing osteotomies about the knee.

High tibial osteotomy (HTO), distal femoral osteotomy (DFO), and tibial tubercle osteotomy (TTO) are viable treatment options for deformities about the knee and patella maltracking.^1-4 Although TTO can be performed in many ways (eg, anteriorization, anteromedialization, medialization), the basic idea is to move the tibial tubercle to improve patellar tracking or to offload a patellar facet that has sustained trauma or degenerated.² DFO is a surgical option for treating a valgus knee deformity (the lateral tibiofemoral compartment is offloaded) or for protecting a knee compartment after cartilage or meniscal restoration (medial closing wedge or lateral opening wedge).¹ Similarly, HTO is an option for treating a varus knee deformity or isolated medial compartment arthritis; the diseased compartment is offloaded, and any malalignment is corrected. Akin to DFO, HTO is often performed to protect a knee compartment, typically the medial tibiofemoral compartment, after cartilage or meniscal restoration.^2-4

Compared to most arthroscopic knee surgeries, these osteotomies are much more involved, have longer operative times, and restrict postoperative weight-bearing and range of motion.^2-4 The rates of deep vein thrombosis (DVT) and pulmonary embolism (PE) after these osteotomies are not well documented. In addition, there is no documentation of the risks in patients who smoke, are obese, or are using oral contraceptives (OCs) at time of surgery, despite the increased DVT and PE risks posed by smoking, obesity, and OC use in other surgical procedures.^5-7 Although the American Academy of Orthopaedic Surgeons (AAOS) issued clinical practice guidelines for DVT/PE prophylaxis after hip and knee arthroplasty, there is no standard prophylaxis guidelines for DVT/PE prevention after HTO, DFO, or TTO.^8,9 Last, rates of DVT after total knee arthroplasty (TKA) are well defined; they range from 2% to 12%.^10,11 These rates may be surrogates for osteotomies about the knee, but this is only conjecture.

We conducted a study to determine the rates of symptomatic DVT and PE after HTO, DFO, or TTO in patients who did not receive postoperative DVT/PE prophylaxis. We also wanted to determine if age, body mass index (BMI), and smoking status have associations with the risk of developing either DVT or PE after HTO, DFO, or TTO. We hypothesized that the DVT and PE rates would both be <1%.

Methods

After this study was approved by our university’s Institutional Review Board, we searched the surgical database of Dr. Cole, a sports medicine fellowship–trained surgeon, to identify all patients who had HTO, DFO, or TTO performed between September 1, 2009 and September 30, 2014. Current Procedural Terminology (CPT) codes were used for the search. The code for HTO was 27457: osteotomy, proximal tibia, including fibular excision or osteotomy (includes correction of genu varus [bowleg] or genu valgus [knock-knee]); after epiphyseal closure). The code for DFO was 27450: osteotomy, femur, shaft or supracondylar; with fixation. Last, the code for TTO was 27418: anterior tibial tubercleplasty (eg, Maquet-type procedure). The 141 patients identified in the search were treated by Dr. Cole at a single institution and were included in the study. Study inclusion did not require a minimum follow-up. Follow-up duration was defined as the time between surgery and the final clinic note in the patient chart. No patient was excluded for lack of follow-up clinic visits, and none was lost to follow-up.

Age, BMI, smoking status, and OC use were recorded for all patients. For each procedure, the surgeon’s technique remained the same throughout the study period: HTO, medial opening-wedge osteotomy with plate-and-screw fixation; DFO, lateral opening-wedge osteotomy with plate-and-screw fixation; and TTO, mostly anteromedialization with screw fixation (though this was dictated by patellar contact pressures). A tourniquet was used in all cases. Each patient’s hospital electronic medical record and outpatient office notes were reviewed to determine if symptomatic DVT or PE developed after surgery. The diagnosis of symptomatic DVT was based on clinical symptoms and confirmatory ultrasound, and the PE diagnosis was based on computed tomography. Doppler ultrasound was performed only in symptomatic patients (ie, it was not routinely performed).

Per surgeon protocol, postoperative DVT prophylaxis was not administered. Patients were encouraged to begin dorsiflexion and plantar flexion of the ankle (ankle pumps) immediately and to mobilize as soon as comfortable. Each patient received a cold therapy machine with compression sleeve. Patients were allowed toe-touch weight-bearing for 6 weeks, and then progressed 25% per week for 4 weeks to full weight-bearing by 10 weeks. After surgery, each patient was placed in a brace, which was kept locked in extension for 10 days; when the brace was unlocked, the patient was allowed to range the knee.

Continuous variable data are reported as weighted means and weighted standard deviations. Categorical variable data are reported as frequencies and percentages.

Results

Our database search identified 141 patients (44% male, 56% female) who underwent HTO (47 patients, 33.3%), DFO (13 patients, 9.2%), or TTO (81 patients, 57.5%). Mean (SD) age was 34.28 (9.86) years, mean (SD) BMI was 26.88 (5.11) kg/m², and mean (SD) follow-up was 17.1 (4.1) months. Of the female patients, 36.7% were using OCs at time of surgery. Of all patients, 13.48% were smokers.

Two patients (1.42%) had clinical symptoms consistent with DVT. In each case, the diagnosis was confirmed with Doppler ultrasound. The below-knee DVT was unilateral in 1 case and bilateral in the other.

Discussion

As the rates of DVT and PE after osteotomies about the knee have not been well studied, we wanted to determine these rates after HTO, DFO, and TTO in patients who did not receive postoperative DVT prophylaxis. We hypothesized that DVT and PE rates would both be <1%, and this hypothesis was partly confirmed: The rate of PE after HTO, DFO, and TTO was <1%, and the rate of symptomatic DVT was >1%. Similarly, the patients who developed these complications were nonsmokers and had a BMI no higher than that of the patients who did not develop DVT or PE. In addition, only 1 patient developed DVT and PE, and she was using OCs and had a family history of DVT. Last, the patients who developed these complications were on average 14 years older than the patients who did not develop DVT or PE.

Although there is a plethora of reports on the incidence of DVT and PE after TKA, there is little on the incidence after osteotomies about the knee.^8,12 The rate of DVT after TKA varies, but many studies place it between 2% and 12%, and routinely find a PE rate of <0.5%.^10,11,13,14 Although the AAOS issued a clinical practice guideline for postoperative DVT prophylaxis after TKA, and evaluated the best available evidence, it could not reach consensus on a specific type of DVT prophylaxis, though the workgroup did recommend that patients be administered postoperative DVT prophylaxis of some kind.^8,9 Similarly, the American College of Chest Physicians (ACCP) issued clinical practice guidelines for preventing DVT and PE after elective TKA and total hip arthroplasty.¹⁵ According to the ACCP guidelines, patients should receive prophylaxis—LMWH, fondaparinux, apixaban, dabigatran, rivaroxaban, low-dose unfractionated heparin, adjusted-dose vitamin K antagonist, aspirin, or an intermittent pneumatic compression device—for a minimum of 14 days. Unfortunately, though there are similarities between TKAs and peri-knee osteotomies, these procedures are markedly different, and it is difficult to extrapolate and adapt recommendations and produce a consensus statement for knee arthroplasties. In addition, guidelines exist for hospitalized patients who are being treated for medical conditions or have undergone surgery, but all the patients in the present study had their osteotomies performed on an outpatient basis.

Martin and colleagues¹⁶ reviewed 323 cases of medial opening-wedge HTO and found a DVT rate of 1.4% in the absence of routine DVT prophylaxis, except in patients with a history of DVT. Their rate is almost identical to ours, but we also included other osteotomies in our study. Miller and colleagues¹⁷ reviewed 46 cases of medial opening-wedge HTO and found a 4.3% DVT rate, despite routine prophylaxis with once-daily 325-mg aspirin and ankle pumps. This finding contrasts with our 1.42% DVT rate in the absence of postoperative chemical DVT prophylaxis. Motycka and colleagues¹⁸ reviewed 65 HTO cases in which DVT prophylaxis (oral anticoagulant) was given for 6 weeks, and they found a DVT rate of 9.7%. Turner and colleagues¹⁹ performed venous ultrasound on 81 consecutive patients who underwent HTO and received DVT prophylaxis (twice-daily subcutaneous heparin), and they found a DVT rate of 41% and a PE rate of 1.2%, though only 8.6% of the DVT cases were symptomatic. Of note, whereas the lowest postoperative DVT rate was for patients who did not receive postoperative DVT prophylaxis, the rate of symptomatic DVT after these osteotomies ranged from 1.4% to 8.6% in patients who received prophylaxis.^16,19 Given this evidence and our study results, it appears routine chemical DVT prophylaxis after osteotomies about the knee may not be necessary, though higher level evidence is needed in order to make definitive recommendations.

In the present study, the 2 patients who developed symptomatic DVT (1 subsequently developed PE) were nonsmokers in good health. The female patient (DVT plus PE) was using OCs at time of surgery. Studies have shown that patients who smoke and who use OCs are at increased risk for developing DVT or PE after surgery.^5,6,12 Given that only 2 of our patients developed DVT/PE, and neither was a smoker, smoking was not associated with increased DVT or PE risk in this study population, in which 13.48% of patients were smokers at time of surgery. In addition, given that the 1 female patient who developed DVT/PE was using OCs and that 36.7% of all female patients in the study were using OCs, it is difficult to conclude whether OC use increased the female patient’s risk for DVT or PE. Furthermore, neither the literature nor the AAOS consensus statement supports discontinuing OCs for this surgical procedure.

Patients in this study did not receive chemical or mechanical DVT prophylaxis after surgery. Regarding various post-TKA DVT prophylaxis regimens, aspirin is as effective as LMWH in preventing DVT, and the risk for postoperative blood loss and wound complications is lower with aspirin than with rivaroxaban.^20,21 Given that the present study’s postoperative rates of DVT (1.42%) and PE (0.71%) are equal to or less than rates already reported in the literature, routine DVT prophylaxis after osteotomies about the knee may be unnecessary in the absence of other significant risk factors.^16,19 However, our study considered only symptomatic DVT and PE, so it is possible that the number of asymptomatic DVT cases is higher in this patient population. Definitively answering our study’s clinical question will require a multicenter registry study (prospective cohort study).

Study Limitations

The strengths of this study include the large number of patients treated by a single surgeon using the same postoperative protocol. Limitations of this study include the lack of a control group. Although we found a DVT rate of 1.42% and a PE rate of 0.71%, the literature on the accepted risks for DVT and PE after HTO, DFO, and TTO is unclear. With our results stratified by procedure, the DVT rate was 2% in the HTO group, 0% in the DFO group, and 1% in the TTO group. However, we were unable to reliably stratify these results by each specific procedure, as the number of patients in each group would be too low. This study involved reviewing charts; as patients were not contacted, it is possible a patient developed DVT or PE, was treated at an outside facility, and then never followed up with the treating surgeon. Patients were identified by CPT codes, so, if a patient underwent HTO, DFO, or TTO that was recorded under a different CPT code, it is possible the patient was missed by our search. All patients were seen after surgery, and we reviewed the outpatient office notes that were taken, so unless the DVT or PE occurred after a patient’s final postoperative visit, it would have been recorded. Similarly, the DVT and PE rates reported here cannot be extrapolated to overall risks for DVT and PE after osteotomies about the knee in all patients—only in patients who did not receive DVT prophylaxis after surgery.

Conclusion

The rates of DVT and PE after HTO, DFO, and TTO in patients who did not receive chemical prophylaxis are low: 1.42% and 0.71%, respectively. After these osteotomies, DVT/PE prophylaxis in the absence of known risk factors may not be warranted.

Am J Orthop. 2017;46(1):E23-E27. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• DVT and PE are uncommon complications following osteotomies about the knee.
• Use of oral contraceptives can increase the risk of a patient sustaining a postoperative DVT and PE following osteotomies about the knee.
• In the absence of significant risk factors, postoperative chemical DVT prophylaxis may be unnecessary in patients undergoing osteotomies about the knee.

High tibial osteotomy (HTO), distal femoral osteotomy (DFO), and tibial tubercle osteotomy (TTO) are viable treatment options for deformities about the knee and patella maltracking.^1-4 Although TTO can be performed in many ways (eg, anteriorization, anteromedialization, medialization), the basic idea is to move the tibial tubercle to improve patellar tracking or to offload a patellar facet that has sustained trauma or degenerated.² DFO is a surgical option for treating a valgus knee deformity (the lateral tibiofemoral compartment is offloaded) or for protecting a knee compartment after cartilage or meniscal restoration (medial closing wedge or lateral opening wedge).¹ Similarly, HTO is an option for treating a varus knee deformity or isolated medial compartment arthritis; the diseased compartment is offloaded, and any malalignment is corrected. Akin to DFO, HTO is often performed to protect a knee compartment, typically the medial tibiofemoral compartment, after cartilage or meniscal restoration.^2-4

Compared to most arthroscopic knee surgeries, these osteotomies are much more involved, have longer operative times, and restrict postoperative weight-bearing and range of motion.^2-4 The rates of deep vein thrombosis (DVT) and pulmonary embolism (PE) after these osteotomies are not well documented. In addition, there is no documentation of the risks in patients who smoke, are obese, or are using oral contraceptives (OCs) at time of surgery, despite the increased DVT and PE risks posed by smoking, obesity, and OC use in other surgical procedures.^5-7 Although the American Academy of Orthopaedic Surgeons (AAOS) issued clinical practice guidelines for DVT/PE prophylaxis after hip and knee arthroplasty, there is no standard prophylaxis guidelines for DVT/PE prevention after HTO, DFO, or TTO.^8,9 Last, rates of DVT after total knee arthroplasty (TKA) are well defined; they range from 2% to 12%.^10,11 These rates may be surrogates for osteotomies about the knee, but this is only conjecture.

We conducted a study to determine the rates of symptomatic DVT and PE after HTO, DFO, or TTO in patients who did not receive postoperative DVT/PE prophylaxis. We also wanted to determine if age, body mass index (BMI), and smoking status have associations with the risk of developing either DVT or PE after HTO, DFO, or TTO. We hypothesized that the DVT and PE rates would both be <1%.

Methods

After this study was approved by our university’s Institutional Review Board, we searched the surgical database of Dr. Cole, a sports medicine fellowship–trained surgeon, to identify all patients who had HTO, DFO, or TTO performed between September 1, 2009 and September 30, 2014. Current Procedural Terminology (CPT) codes were used for the search. The code for HTO was 27457: osteotomy, proximal tibia, including fibular excision or osteotomy (includes correction of genu varus [bowleg] or genu valgus [knock-knee]); after epiphyseal closure). The code for DFO was 27450: osteotomy, femur, shaft or supracondylar; with fixation. Last, the code for TTO was 27418: anterior tibial tubercleplasty (eg, Maquet-type procedure). The 141 patients identified in the search were treated by Dr. Cole at a single institution and were included in the study. Study inclusion did not require a minimum follow-up. Follow-up duration was defined as the time between surgery and the final clinic note in the patient chart. No patient was excluded for lack of follow-up clinic visits, and none was lost to follow-up.

Age, BMI, smoking status, and OC use were recorded for all patients. For each procedure, the surgeon’s technique remained the same throughout the study period: HTO, medial opening-wedge osteotomy with plate-and-screw fixation; DFO, lateral opening-wedge osteotomy with plate-and-screw fixation; and TTO, mostly anteromedialization with screw fixation (though this was dictated by patellar contact pressures). A tourniquet was used in all cases. Each patient’s hospital electronic medical record and outpatient office notes were reviewed to determine if symptomatic DVT or PE developed after surgery. The diagnosis of symptomatic DVT was based on clinical symptoms and confirmatory ultrasound, and the PE diagnosis was based on computed tomography. Doppler ultrasound was performed only in symptomatic patients (ie, it was not routinely performed).

Per surgeon protocol, postoperative DVT prophylaxis was not administered. Patients were encouraged to begin dorsiflexion and plantar flexion of the ankle (ankle pumps) immediately and to mobilize as soon as comfortable. Each patient received a cold therapy machine with compression sleeve. Patients were allowed toe-touch weight-bearing for 6 weeks, and then progressed 25% per week for 4 weeks to full weight-bearing by 10 weeks. After surgery, each patient was placed in a brace, which was kept locked in extension for 10 days; when the brace was unlocked, the patient was allowed to range the knee.

Continuous variable data are reported as weighted means and weighted standard deviations. Categorical variable data are reported as frequencies and percentages.

Results

Our database search identified 141 patients (44% male, 56% female) who underwent HTO (47 patients, 33.3%), DFO (13 patients, 9.2%), or TTO (81 patients, 57.5%). Mean (SD) age was 34.28 (9.86) years, mean (SD) BMI was 26.88 (5.11) kg/m², and mean (SD) follow-up was 17.1 (4.1) months. Of the female patients, 36.7% were using OCs at time of surgery. Of all patients, 13.48% were smokers.

Two patients (1.42%) had clinical symptoms consistent with DVT. In each case, the diagnosis was confirmed with Doppler ultrasound. The below-knee DVT was unilateral in 1 case and bilateral in the other.

Discussion

As the rates of DVT and PE after osteotomies about the knee have not been well studied, we wanted to determine these rates after HTO, DFO, and TTO in patients who did not receive postoperative DVT prophylaxis. We hypothesized that DVT and PE rates would both be <1%, and this hypothesis was partly confirmed: The rate of PE after HTO, DFO, and TTO was <1%, and the rate of symptomatic DVT was >1%. Similarly, the patients who developed these complications were nonsmokers and had a BMI no higher than that of the patients who did not develop DVT or PE. In addition, only 1 patient developed DVT and PE, and she was using OCs and had a family history of DVT. Last, the patients who developed these complications were on average 14 years older than the patients who did not develop DVT or PE.

Although there is a plethora of reports on the incidence of DVT and PE after TKA, there is little on the incidence after osteotomies about the knee.^8,12 The rate of DVT after TKA varies, but many studies place it between 2% and 12%, and routinely find a PE rate of <0.5%.^10,11,13,14 Although the AAOS issued a clinical practice guideline for postoperative DVT prophylaxis after TKA, and evaluated the best available evidence, it could not reach consensus on a specific type of DVT prophylaxis, though the workgroup did recommend that patients be administered postoperative DVT prophylaxis of some kind.^8,9 Similarly, the American College of Chest Physicians (ACCP) issued clinical practice guidelines for preventing DVT and PE after elective TKA and total hip arthroplasty.¹⁵ According to the ACCP guidelines, patients should receive prophylaxis—LMWH, fondaparinux, apixaban, dabigatran, rivaroxaban, low-dose unfractionated heparin, adjusted-dose vitamin K antagonist, aspirin, or an intermittent pneumatic compression device—for a minimum of 14 days. Unfortunately, though there are similarities between TKAs and peri-knee osteotomies, these procedures are markedly different, and it is difficult to extrapolate and adapt recommendations and produce a consensus statement for knee arthroplasties. In addition, guidelines exist for hospitalized patients who are being treated for medical conditions or have undergone surgery, but all the patients in the present study had their osteotomies performed on an outpatient basis.

Martin and colleagues¹⁶ reviewed 323 cases of medial opening-wedge HTO and found a DVT rate of 1.4% in the absence of routine DVT prophylaxis, except in patients with a history of DVT. Their rate is almost identical to ours, but we also included other osteotomies in our study. Miller and colleagues¹⁷ reviewed 46 cases of medial opening-wedge HTO and found a 4.3% DVT rate, despite routine prophylaxis with once-daily 325-mg aspirin and ankle pumps. This finding contrasts with our 1.42% DVT rate in the absence of postoperative chemical DVT prophylaxis. Motycka and colleagues¹⁸ reviewed 65 HTO cases in which DVT prophylaxis (oral anticoagulant) was given for 6 weeks, and they found a DVT rate of 9.7%. Turner and colleagues¹⁹ performed venous ultrasound on 81 consecutive patients who underwent HTO and received DVT prophylaxis (twice-daily subcutaneous heparin), and they found a DVT rate of 41% and a PE rate of 1.2%, though only 8.6% of the DVT cases were symptomatic. Of note, whereas the lowest postoperative DVT rate was for patients who did not receive postoperative DVT prophylaxis, the rate of symptomatic DVT after these osteotomies ranged from 1.4% to 8.6% in patients who received prophylaxis.^16,19 Given this evidence and our study results, it appears routine chemical DVT prophylaxis after osteotomies about the knee may not be necessary, though higher level evidence is needed in order to make definitive recommendations.

In the present study, the 2 patients who developed symptomatic DVT (1 subsequently developed PE) were nonsmokers in good health. The female patient (DVT plus PE) was using OCs at time of surgery. Studies have shown that patients who smoke and who use OCs are at increased risk for developing DVT or PE after surgery.^5,6,12 Given that only 2 of our patients developed DVT/PE, and neither was a smoker, smoking was not associated with increased DVT or PE risk in this study population, in which 13.48% of patients were smokers at time of surgery. In addition, given that the 1 female patient who developed DVT/PE was using OCs and that 36.7% of all female patients in the study were using OCs, it is difficult to conclude whether OC use increased the female patient’s risk for DVT or PE. Furthermore, neither the literature nor the AAOS consensus statement supports discontinuing OCs for this surgical procedure.

Patients in this study did not receive chemical or mechanical DVT prophylaxis after surgery. Regarding various post-TKA DVT prophylaxis regimens, aspirin is as effective as LMWH in preventing DVT, and the risk for postoperative blood loss and wound complications is lower with aspirin than with rivaroxaban.^20,21 Given that the present study’s postoperative rates of DVT (1.42%) and PE (0.71%) are equal to or less than rates already reported in the literature, routine DVT prophylaxis after osteotomies about the knee may be unnecessary in the absence of other significant risk factors.^16,19 However, our study considered only symptomatic DVT and PE, so it is possible that the number of asymptomatic DVT cases is higher in this patient population. Definitively answering our study’s clinical question will require a multicenter registry study (prospective cohort study).

Study Limitations

The strengths of this study include the large number of patients treated by a single surgeon using the same postoperative protocol. Limitations of this study include the lack of a control group. Although we found a DVT rate of 1.42% and a PE rate of 0.71%, the literature on the accepted risks for DVT and PE after HTO, DFO, and TTO is unclear. With our results stratified by procedure, the DVT rate was 2% in the HTO group, 0% in the DFO group, and 1% in the TTO group. However, we were unable to reliably stratify these results by each specific procedure, as the number of patients in each group would be too low. This study involved reviewing charts; as patients were not contacted, it is possible a patient developed DVT or PE, was treated at an outside facility, and then never followed up with the treating surgeon. Patients were identified by CPT codes, so, if a patient underwent HTO, DFO, or TTO that was recorded under a different CPT code, it is possible the patient was missed by our search. All patients were seen after surgery, and we reviewed the outpatient office notes that were taken, so unless the DVT or PE occurred after a patient’s final postoperative visit, it would have been recorded. Similarly, the DVT and PE rates reported here cannot be extrapolated to overall risks for DVT and PE after osteotomies about the knee in all patients—only in patients who did not receive DVT prophylaxis after surgery.

Conclusion

The rates of DVT and PE after HTO, DFO, and TTO in patients who did not receive chemical prophylaxis are low: 1.42% and 0.71%, respectively. After these osteotomies, DVT/PE prophylaxis in the absence of known risk factors may not be warranted.

Am J Orthop. 2017;46(1):E23-E27. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• Music therapists use patient-preferred live music, increasing neurologic cues that enhance movement—a seminal recovery function in postoperative spine patients.
• Music therapy is an evidence-based, integrative treatment addressing body, mind, and spirit.
• Tension release through music therapy can serve as a critical mechanism for building resilience related to pain management.
• Music therapy and music medicine are distinct forms of clinical practice that focus on mind-body integration in the healing process.
• Music therapists, board-certified and licensed by the state as recognized healthcare professionals, address pain management, which is an increasing subspecialty in postoperative care.

About 70% of people in the United States experience at least 1 episode of back pain in their lifetime,¹ and more than 5 million are temporarily or permanently disabled by spinal disorders.^2-4 Some require surgery, which may rectify injury, but pain during recovery is often inevitable, and the road to recovery is not guaranteed to be smooth.^5-20

Postoperative spine patients are at major risk for pain management challenges.^14,15,18,20 Treatment is primarily pharmacologic and based on the surgical team’s pain management orders. Nursing care consists of monitoring the airway, vital signs, and neurovascular status and having patients rate their pain on a visual analog scale (VAS; 0 = no pain, 10 = worst pain imaginable). Nurses have the challenge of monitoring and continually assessing to make sure patients are achieving the optimal outcomes, particularly during the immediate postoperative period, when pain and anxiety are prominently increased.

Variability in spine surgery outcomes can be explained at least partly on the basis of prognostic psychological factors, including hypochondriasis, hysteria, depression, and poor pain coping strategies (eg, catastrophizing).²¹ In spine surgery patients, kinesiophobia (fear of moving) is a common component of distress that can impede recuperation.^21-23Psychological interventions that assist with the secondary stressors associated with pain and loss during physical recuperation are recommended, with increased attention given to the importance of treating the whole person: body, mind, and spirit.^24-29 Conventional pain-alleviating medical interventions can be enhanced with integrative therapies that empower patients to marshal their inner resources during recovery.^25-28Music therapy may be particularly suited to this effort, as it is adaptable to the patient’s individual and culturally specific needs.^29-33

Rationale for Live Music

Pain is subjective and personal, and warrants an individualized approach to care. There is a body of music medicine research on the use of recorded music in modulating psychological and physiological factors in pain perception.^30,32,34-54 This research supports the unique relationship of music to well-being, and the understanding that controlling any of these factors affects the duration, intensity, and quality of that experience.^41,43,52

These findings provide incentive for breathing-entrained music therapy interventions, which enhance the relaxation response and release of pain-related tension;^32,55-58 empower patients to unlock physical and emotional tension;^32,57,58 provide a channel for expression and body movement; and enhance blood flow and/or alleviate pain by activating neurologic areas involved in the experience of pain.^59-62Studies have found that physical endurance may be enhanced when movement is rhythmically coordinated with a musical stimulus.^63-66 Music may prolong physical endurance by inhibiting psychological feedback associated with physical exertion related to fatigue, which may translate into accelerated recovery periods. When we listen to a rhythmic sound, our brains tend to automatically synchronize, or entrain, to external rhythmic cues that can stimulate increased motor control and coordination.⁶³ Sound can arouse and raise the excitability of spinal motor neurons mediated by auditory-motor neuronal connections on the brain stem and spinal cord level.^64-66 Rhythmically organized sounds serve as a neurological function in our capacity to organize predictable timing cues that are apparent in music, and may result in an effective treatment intervention in recovery.^63,64

Music Therapy in Recovery From Spine Surgery

In music therapy, music is used within a therapeutic relationship to support or affect change in the patient and the treatment regimen.^32,33,56-58 Research on music therapy with patients who are recovering from spine surgery is scant.^67-69 Kleiber and Adamek⁶⁷ studied perceptions of music therapy in 8 adolescents after spinal fusion surgery. In their study, a music therapist provided patients with a postoperative music therapy session focusing on the use of patient-preferred live music for relaxation and expression. Although their qualitative query was based on a therapeutic approach similar to that used in the present study, only 1 session was offered during the recovery period, and follow-up was conducted by survey invitation and telephone. In addition, the number of participants was small, and there was no quantitative measure of pain or other symptoms.

Another study focused on the effects of listening to music on pain intensity and distress after spine surgery.⁶⁸ Patients in the study’s music group made their selections from prerecorded classical music and domestic and international popular songs from various genres and listened to their chosen recordings 30 minutes a day. Although the study was not a music therapy study per se, it showed a positive impact of listening to music on anxiety and pain perception in 60 adults who were randomly assigned to the music group or to a non-music control group (n = 30 in each). Differences between the music and control groups’ VAS ratings of anxiety (Ps = .018-.001) and pain (P = .001) were statistically significant.

Different from our study, the aforementioned studies did not include tension release–focused live music offered within a therapeutic relationship. Our 1.5-year pilot study, conducted prior to the present study indicated that music therapy led to increased resilience and recovery mechanisms.⁵⁸

Methods

Our mixed-methods study design combined standard medical treatment with integrative music therapy interventions based on pain assessments to better understand the effects of music therapy on the recovery of patients after spine surgery.

The Spine Institute of New York within the Department of Orthopedic Surgery at Mount Sinai Beth Israel provides surgical treatment of common spinal cord conditions. Prioritizing patient satisfaction and positive outcomes,^27,28 the institute integrates music therapy through the Louis Armstrong Center for Music and Medicine to enhance treatment of pain symptoms.

Patients were recruited by the research team as per the daily surgical schedule, or through referral by the medical team or patient care navigator. Sixty patients (35 female, 25 male) ranging in age from 40 to 55 years underwent anterior, posterior, or anterior-posterior spinal fusion and were enrolled in the study after signing a participation consent form. Minorities, women, and patients with Medicaid and Medicare were included. Patients who received a diagnosis of clinical psychosis or depression prior to spine injury were excluded.

The experimental group received music therapy plus standard care (medical and nursing care with scheduled pharmacologic pain intervention), and a wait-listed control group received standard care only. A randomization chart created by a blinded statistician who did not have access to the patient census determined the intervention–nonintervention schedule. Patients in the music therapy group received one 30-minute music therapy session during an 8-hour period within 72 hours after surgery.

For both groups, measurements were completed before and after the study window. Control patients were offered music therapy after completion of the post-intervention surveys in order to minimize the ethical dilemma of denying potentially helpful pain intervention. For this same reason, both groups were given the option of receiving follow-up music therapy sessions for the duration of their hospitalization.

The research team consisted of 2 licensed, board-certified music therapists. In addition, Master’s-level music therapy interns completing clinical hours as part of the trajectory for board certification served on the research team over the 5-year period 2009 to 2014, and 13 blinded research assistants helped with enrolling and collecting data on patients.

Intervention

Each music therapy session included a warm-up phase of verbal or musical discourse. Next was the treatment phase, which was based on patient need as assessed during warm-up. Treatment options included use of patient-preferred live music that supported tension release/relaxation through incentive-based clinical improvisation, singing, and/or rhythmic drumming or through breathwork and visualization. Psychoeducation about mind–body awareness through the use of breath and imagery was introduced and explained by the therapist at this time.

The improvised music intervention was focused on making salient the natural harmonic tension-resolution cycles that occur in music and that were entrained to the patient’s presentation (respiratory rate, verbal report, clinical presentation). When patient-preferred precomposed songs were used, tension resolution was achieved by sustaining cadence and resolution, also entrained to the patient’s respiratory cycles.^32,57,58

After the music therapy intervention, a period of closure or integration was facilitated by the therapist contingent on the patient’s degree of alertness. If awake, the patient was supported in a reflexive process of thoughts, impressions, or issues that may have contributed to the overall experience. If the patient was asleep, the researcher returned within 30 minutes for post-intervention interviewing. Interview information was recorded in a qualitative post-participation survey. To prevent bias, researchers who were not the treating clinicians conducted the surveys.

Outcome Measures

Both primary and secondary outcome measures were collected before and after the intervention. The primary outcome measure was VAS pain ratings, and the secondary outcome measures were scores on the Hospital Anxiety and Depression Scale (HADS), the Tampa Scale for Kinesiophobia (TSK), and the Color Analysis Scale (CAS).

VAS. With the VAS, images are used to rate pain. The scale has points labeled 0 to 10 and corresponding faces representing progression in pain intensity. The scale is quickly rendered and can be interpreted according to the patient’s recovery phase at time of rendering.

HADS. The HADS⁷⁰ provides a specific baseline for anxiety and depression as an indicator of how the patient might fare during hospitalization (admission through recovery and discharge).

TSK. The TSK⁷¹ provides insight into the patient’s perception of fear-related movement, which is an important factor in this study because of the movement required for rehabilitation. We used a shortened version of the TSK to accommodate the sensitive threshold for pain tolerance and pharmacologic side effects commonly experienced by spine patients.

CAS. The CAS was developed at the Louis Armstrong Center for Music and Medicine to assess comorbidities and dynamic aspects of pain. Through a coloring exercise, patients illustrate their pain experience, which gives tangible form to the abstract experience of pain.

Coding

We collected patients’ demographic data, including age, sex, and diagnoses. Clinical indicators of the preoperative baseline included lifestyle, surgical history, and prior experience with music or other mind–body strategies for self-regulation.

As fundamental to qualitative methodology,^72,73 the reported responses to questions were grouped into themes that were peer-tested with members of the research team before and during the coding process.

Statistical Methods

Means and standard deviations were used for continuous variables, and frequencies (percentages) for categorical variables. All outcomes were analyzed on an intent-to-treat basis. Repeated-measures analysis of variance was used to compare changes in outcomes from before to after intervention for the music and control groups. In particular, a statistically significant Group (music vs control) × Time (before vs after intervention) interaction would support the hypothesis that there would be more benefit (less pain) in the music group as a result of the music therapy. For all tests, significance was set at P < .05. SPSS Version 20 (IBM) was used for all statistical analyses. Based on previously found differences in heart rate and mobility,³¹ we assumed an effect size of 0.71 for the difference between music and control (no music), which would require 32 patients per group to achieve a power of 0.8 with an α of 0.05.

Results

Of the 136 patients who were asked to participate in the study, 76 were not enrolled; the other 60 were equally assigned to either the control group or the music therapy group (n = 30 in each) according to randomization indicated by a blinded statistician (Figure 1).

Table 2 lists the pre-intervention and post-intervention comparisons of the main outcomes between groups.

Relationship with music was coded for significance and included reports of music as a resource accessed for stimulation and/or relaxation through listening; direct involvement with instrument playing; and history of music training. 

Perceptions of surgical outcome in patients’ responses were coded across 3 themes: (1) optimistic (belief and hope in returning to original baseline of functionality), (2) indifferent (neither hopeful nor cynical about results of surgery), and (3) pessimistic (belief that nothing will restore the quality of life that existed before the spinal condition).

The CAS helped us better understand the diversity and complexity of the pain experience.

Discussion

Our hospital has the unique capability of providing music therapy to postoperative and other hospitalized patients. In this study, we compared the impact of a structured postoperative music therapy program on spine patients relative to control patients who did not receive music therapy after spine surgery.

We found a significant benefit in VAS pain levels (>1 point) but no statistically significant differences in HADS Anxiety, HADS Depression, or TSK scores. Although a 2-point difference is usually considered clinically significant, the degree of change in the music group is notable for having been achieved by nonpharmacologic means with scant chance of adverse effects. We suspect the lack of significant change in HADS Anxiety, HADS Depression, and TSK scores is attributable to the narrow study window. Given the observational data from our pilot study⁵⁸ and ongoing results with spine patients,³² it seems clear that both mood state and resilience in coping are enhanced through an ongoing relationship with music therapy.

The study of a population as vulnerable as patients recovering from spine surgery raises many issues for providers and researchers. Although it is worthwhile to determine the efficacy of integrative modalities in serving these patients, the request for participation in a protocol at such a vulnerable time was often resisted. During our pilot work, it became clear that the ability of potential subjects to comprehend and complete protocol surveys was impacted by adverse effects, including sedation drowsiness; respiratory depression; nausea and vomiting; pruritus; and urinary retention caused by the medications used for postoperative pain management. Consequently, after piloting 5 cases before the main study, we extended the enrollment window to 72 hours.

Other unforeseen intrinsic or external obstacles were identified: Patient-related issues—including availability, level of interest in participation, and inability to participate because of the medication adverse effects mentioned.

Staff investment/education—addressed over the first 3 study years with several in-services, starting with the surgical team and continuing with nursing and support staff in various combinations. These meetings led to the creation of an Institutional Review Board (IRB) approved educational sheet for inclusion in the information packet given to surgical patients on registration.

Programming interruptions—caused by the convergence of several unanticipated factors, including a delay in expedited review of the IRB renewal during the year of Hurricane Sandy and an interruption in the spine team’s service for administrative and program modification.

Conclusion

Music therapy interventions (eg, use of patient-preferred live music) offered within a therapeutic relationship favorably affected pain perceptions in patients recovering from spine surgery. This effect was achieved through several therapeutic entry points, including support of expression and opportunities for emotional catharsis.

At the core of music therapy’s efficacy is individualized treatment, through which patients are supported in their recovery of “self.” Measurable benefits—including increased comfort; reduced pain; improved gait; increased range of motion, endurance, and ability to relax; and empowerment to actively participate in one’s own care through daily activities imbued with an enhanced sense of agency—are of cardinal importance, as they may lead to quicker recovery perceptions and enhanced quality of life.

Am J Orthop. 2017;46(1):E13-E22. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• Music therapists use patient-preferred live music, increasing neurologic cues that enhance movement—a seminal recovery function in postoperative spine patients.
• Music therapy is an evidence-based, integrative treatment addressing body, mind, and spirit.
• Tension release through music therapy can serve as a critical mechanism for building resilience related to pain management.
• Music therapy and music medicine are distinct forms of clinical practice that focus on mind-body integration in the healing process.
• Music therapists, board-certified and licensed by the state as recognized healthcare professionals, address pain management, which is an increasing subspecialty in postoperative care.

About 70% of people in the United States experience at least 1 episode of back pain in their lifetime,¹ and more than 5 million are temporarily or permanently disabled by spinal disorders.^2-4 Some require surgery, which may rectify injury, but pain during recovery is often inevitable, and the road to recovery is not guaranteed to be smooth.^5-20

Postoperative spine patients are at major risk for pain management challenges.^14,15,18,20 Treatment is primarily pharmacologic and based on the surgical team’s pain management orders. Nursing care consists of monitoring the airway, vital signs, and neurovascular status and having patients rate their pain on a visual analog scale (VAS; 0 = no pain, 10 = worst pain imaginable). Nurses have the challenge of monitoring and continually assessing to make sure patients are achieving the optimal outcomes, particularly during the immediate postoperative period, when pain and anxiety are prominently increased.

Variability in spine surgery outcomes can be explained at least partly on the basis of prognostic psychological factors, including hypochondriasis, hysteria, depression, and poor pain coping strategies (eg, catastrophizing).²¹ In spine surgery patients, kinesiophobia (fear of moving) is a common component of distress that can impede recuperation.^21-23Psychological interventions that assist with the secondary stressors associated with pain and loss during physical recuperation are recommended, with increased attention given to the importance of treating the whole person: body, mind, and spirit.^24-29 Conventional pain-alleviating medical interventions can be enhanced with integrative therapies that empower patients to marshal their inner resources during recovery.^25-28Music therapy may be particularly suited to this effort, as it is adaptable to the patient’s individual and culturally specific needs.^29-33

Rationale for Live Music

Pain is subjective and personal, and warrants an individualized approach to care. There is a body of music medicine research on the use of recorded music in modulating psychological and physiological factors in pain perception.^30,32,34-54 This research supports the unique relationship of music to well-being, and the understanding that controlling any of these factors affects the duration, intensity, and quality of that experience.^41,43,52

These findings provide incentive for breathing-entrained music therapy interventions, which enhance the relaxation response and release of pain-related tension;^32,55-58 empower patients to unlock physical and emotional tension;^32,57,58 provide a channel for expression and body movement; and enhance blood flow and/or alleviate pain by activating neurologic areas involved in the experience of pain.^59-62Studies have found that physical endurance may be enhanced when movement is rhythmically coordinated with a musical stimulus.^63-66 Music may prolong physical endurance by inhibiting psychological feedback associated with physical exertion related to fatigue, which may translate into accelerated recovery periods. When we listen to a rhythmic sound, our brains tend to automatically synchronize, or entrain, to external rhythmic cues that can stimulate increased motor control and coordination.⁶³ Sound can arouse and raise the excitability of spinal motor neurons mediated by auditory-motor neuronal connections on the brain stem and spinal cord level.^64-66 Rhythmically organized sounds serve as a neurological function in our capacity to organize predictable timing cues that are apparent in music, and may result in an effective treatment intervention in recovery.^63,64

Music Therapy in Recovery From Spine Surgery

In music therapy, music is used within a therapeutic relationship to support or affect change in the patient and the treatment regimen.^32,33,56-58 Research on music therapy with patients who are recovering from spine surgery is scant.^67-69 Kleiber and Adamek⁶⁷ studied perceptions of music therapy in 8 adolescents after spinal fusion surgery. In their study, a music therapist provided patients with a postoperative music therapy session focusing on the use of patient-preferred live music for relaxation and expression. Although their qualitative query was based on a therapeutic approach similar to that used in the present study, only 1 session was offered during the recovery period, and follow-up was conducted by survey invitation and telephone. In addition, the number of participants was small, and there was no quantitative measure of pain or other symptoms.

Another study focused on the effects of listening to music on pain intensity and distress after spine surgery.⁶⁸ Patients in the study’s music group made their selections from prerecorded classical music and domestic and international popular songs from various genres and listened to their chosen recordings 30 minutes a day. Although the study was not a music therapy study per se, it showed a positive impact of listening to music on anxiety and pain perception in 60 adults who were randomly assigned to the music group or to a non-music control group (n = 30 in each). Differences between the music and control groups’ VAS ratings of anxiety (Ps = .018-.001) and pain (P = .001) were statistically significant.

Different from our study, the aforementioned studies did not include tension release–focused live music offered within a therapeutic relationship. Our 1.5-year pilot study, conducted prior to the present study indicated that music therapy led to increased resilience and recovery mechanisms.⁵⁸

Methods

Our mixed-methods study design combined standard medical treatment with integrative music therapy interventions based on pain assessments to better understand the effects of music therapy on the recovery of patients after spine surgery.

The Spine Institute of New York within the Department of Orthopedic Surgery at Mount Sinai Beth Israel provides surgical treatment of common spinal cord conditions. Prioritizing patient satisfaction and positive outcomes,^27,28 the institute integrates music therapy through the Louis Armstrong Center for Music and Medicine to enhance treatment of pain symptoms.

Patients were recruited by the research team as per the daily surgical schedule, or through referral by the medical team or patient care navigator. Sixty patients (35 female, 25 male) ranging in age from 40 to 55 years underwent anterior, posterior, or anterior-posterior spinal fusion and were enrolled in the study after signing a participation consent form. Minorities, women, and patients with Medicaid and Medicare were included. Patients who received a diagnosis of clinical psychosis or depression prior to spine injury were excluded.

The experimental group received music therapy plus standard care (medical and nursing care with scheduled pharmacologic pain intervention), and a wait-listed control group received standard care only. A randomization chart created by a blinded statistician who did not have access to the patient census determined the intervention–nonintervention schedule. Patients in the music therapy group received one 30-minute music therapy session during an 8-hour period within 72 hours after surgery.

For both groups, measurements were completed before and after the study window. Control patients were offered music therapy after completion of the post-intervention surveys in order to minimize the ethical dilemma of denying potentially helpful pain intervention. For this same reason, both groups were given the option of receiving follow-up music therapy sessions for the duration of their hospitalization.

The research team consisted of 2 licensed, board-certified music therapists. In addition, Master’s-level music therapy interns completing clinical hours as part of the trajectory for board certification served on the research team over the 5-year period 2009 to 2014, and 13 blinded research assistants helped with enrolling and collecting data on patients.

Intervention

Each music therapy session included a warm-up phase of verbal or musical discourse. Next was the treatment phase, which was based on patient need as assessed during warm-up. Treatment options included use of patient-preferred live music that supported tension release/relaxation through incentive-based clinical improvisation, singing, and/or rhythmic drumming or through breathwork and visualization. Psychoeducation about mind–body awareness through the use of breath and imagery was introduced and explained by the therapist at this time.

The improvised music intervention was focused on making salient the natural harmonic tension-resolution cycles that occur in music and that were entrained to the patient’s presentation (respiratory rate, verbal report, clinical presentation). When patient-preferred precomposed songs were used, tension resolution was achieved by sustaining cadence and resolution, also entrained to the patient’s respiratory cycles.^32,57,58

After the music therapy intervention, a period of closure or integration was facilitated by the therapist contingent on the patient’s degree of alertness. If awake, the patient was supported in a reflexive process of thoughts, impressions, or issues that may have contributed to the overall experience. If the patient was asleep, the researcher returned within 30 minutes for post-intervention interviewing. Interview information was recorded in a qualitative post-participation survey. To prevent bias, researchers who were not the treating clinicians conducted the surveys.

Outcome Measures

Both primary and secondary outcome measures were collected before and after the intervention. The primary outcome measure was VAS pain ratings, and the secondary outcome measures were scores on the Hospital Anxiety and Depression Scale (HADS), the Tampa Scale for Kinesiophobia (TSK), and the Color Analysis Scale (CAS).

VAS. With the VAS, images are used to rate pain. The scale has points labeled 0 to 10 and corresponding faces representing progression in pain intensity. The scale is quickly rendered and can be interpreted according to the patient’s recovery phase at time of rendering.

HADS. The HADS⁷⁰ provides a specific baseline for anxiety and depression as an indicator of how the patient might fare during hospitalization (admission through recovery and discharge).

TSK. The TSK⁷¹ provides insight into the patient’s perception of fear-related movement, which is an important factor in this study because of the movement required for rehabilitation. We used a shortened version of the TSK to accommodate the sensitive threshold for pain tolerance and pharmacologic side effects commonly experienced by spine patients.

CAS. The CAS was developed at the Louis Armstrong Center for Music and Medicine to assess comorbidities and dynamic aspects of pain. Through a coloring exercise, patients illustrate their pain experience, which gives tangible form to the abstract experience of pain.

Coding

We collected patients’ demographic data, including age, sex, and diagnoses. Clinical indicators of the preoperative baseline included lifestyle, surgical history, and prior experience with music or other mind–body strategies for self-regulation.

As fundamental to qualitative methodology,^72,73 the reported responses to questions were grouped into themes that were peer-tested with members of the research team before and during the coding process.

Statistical Methods

Means and standard deviations were used for continuous variables, and frequencies (percentages) for categorical variables. All outcomes were analyzed on an intent-to-treat basis. Repeated-measures analysis of variance was used to compare changes in outcomes from before to after intervention for the music and control groups. In particular, a statistically significant Group (music vs control) × Time (before vs after intervention) interaction would support the hypothesis that there would be more benefit (less pain) in the music group as a result of the music therapy. For all tests, significance was set at P < .05. SPSS Version 20 (IBM) was used for all statistical analyses. Based on previously found differences in heart rate and mobility,³¹ we assumed an effect size of 0.71 for the difference between music and control (no music), which would require 32 patients per group to achieve a power of 0.8 with an α of 0.05.

Results

Of the 136 patients who were asked to participate in the study, 76 were not enrolled; the other 60 were equally assigned to either the control group or the music therapy group (n = 30 in each) according to randomization indicated by a blinded statistician (Figure 1).

Table 2 lists the pre-intervention and post-intervention comparisons of the main outcomes between groups.

Relationship with music was coded for significance and included reports of music as a resource accessed for stimulation and/or relaxation through listening; direct involvement with instrument playing; and history of music training. 

Perceptions of surgical outcome in patients’ responses were coded across 3 themes: (1) optimistic (belief and hope in returning to original baseline of functionality), (2) indifferent (neither hopeful nor cynical about results of surgery), and (3) pessimistic (belief that nothing will restore the quality of life that existed before the spinal condition).

The CAS helped us better understand the diversity and complexity of the pain experience.

Discussion

Our hospital has the unique capability of providing music therapy to postoperative and other hospitalized patients. In this study, we compared the impact of a structured postoperative music therapy program on spine patients relative to control patients who did not receive music therapy after spine surgery.

We found a significant benefit in VAS pain levels (>1 point) but no statistically significant differences in HADS Anxiety, HADS Depression, or TSK scores. Although a 2-point difference is usually considered clinically significant, the degree of change in the music group is notable for having been achieved by nonpharmacologic means with scant chance of adverse effects. We suspect the lack of significant change in HADS Anxiety, HADS Depression, and TSK scores is attributable to the narrow study window. Given the observational data from our pilot study⁵⁸ and ongoing results with spine patients,³² it seems clear that both mood state and resilience in coping are enhanced through an ongoing relationship with music therapy.

The study of a population as vulnerable as patients recovering from spine surgery raises many issues for providers and researchers. Although it is worthwhile to determine the efficacy of integrative modalities in serving these patients, the request for participation in a protocol at such a vulnerable time was often resisted. During our pilot work, it became clear that the ability of potential subjects to comprehend and complete protocol surveys was impacted by adverse effects, including sedation drowsiness; respiratory depression; nausea and vomiting; pruritus; and urinary retention caused by the medications used for postoperative pain management. Consequently, after piloting 5 cases before the main study, we extended the enrollment window to 72 hours.

Other unforeseen intrinsic or external obstacles were identified: Patient-related issues—including availability, level of interest in participation, and inability to participate because of the medication adverse effects mentioned.

Staff investment/education—addressed over the first 3 study years with several in-services, starting with the surgical team and continuing with nursing and support staff in various combinations. These meetings led to the creation of an Institutional Review Board (IRB) approved educational sheet for inclusion in the information packet given to surgical patients on registration.

Programming interruptions—caused by the convergence of several unanticipated factors, including a delay in expedited review of the IRB renewal during the year of Hurricane Sandy and an interruption in the spine team’s service for administrative and program modification.

Conclusion

Music therapy interventions (eg, use of patient-preferred live music) offered within a therapeutic relationship favorably affected pain perceptions in patients recovering from spine surgery. This effect was achieved through several therapeutic entry points, including support of expression and opportunities for emotional catharsis.

At the core of music therapy’s efficacy is individualized treatment, through which patients are supported in their recovery of “self.” Measurable benefits—including increased comfort; reduced pain; improved gait; increased range of motion, endurance, and ability to relax; and empowerment to actively participate in one’s own care through daily activities imbued with an enhanced sense of agency—are of cardinal importance, as they may lead to quicker recovery perceptions and enhanced quality of life.

Am J Orthop. 2017;46(1):E13-E22. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• Music therapists use patient-preferred live music, increasing neurologic cues that enhance movement—a seminal recovery function in postoperative spine patients.
• Music therapy is an evidence-based, integrative treatment addressing body, mind, and spirit.
• Tension release through music therapy can serve as a critical mechanism for building resilience related to pain management.
• Music therapy and music medicine are distinct forms of clinical practice that focus on mind-body integration in the healing process.
• Music therapists, board-certified and licensed by the state as recognized healthcare professionals, address pain management, which is an increasing subspecialty in postoperative care.

About 70% of people in the United States experience at least 1 episode of back pain in their lifetime,¹ and more than 5 million are temporarily or permanently disabled by spinal disorders.^2-4 Some require surgery, which may rectify injury, but pain during recovery is often inevitable, and the road to recovery is not guaranteed to be smooth.^5-20

Postoperative spine patients are at major risk for pain management challenges.^14,15,18,20 Treatment is primarily pharmacologic and based on the surgical team’s pain management orders. Nursing care consists of monitoring the airway, vital signs, and neurovascular status and having patients rate their pain on a visual analog scale (VAS; 0 = no pain, 10 = worst pain imaginable). Nurses have the challenge of monitoring and continually assessing to make sure patients are achieving the optimal outcomes, particularly during the immediate postoperative period, when pain and anxiety are prominently increased.

Variability in spine surgery outcomes can be explained at least partly on the basis of prognostic psychological factors, including hypochondriasis, hysteria, depression, and poor pain coping strategies (eg, catastrophizing).²¹ In spine surgery patients, kinesiophobia (fear of moving) is a common component of distress that can impede recuperation.^21-23Psychological interventions that assist with the secondary stressors associated with pain and loss during physical recuperation are recommended, with increased attention given to the importance of treating the whole person: body, mind, and spirit.^24-29 Conventional pain-alleviating medical interventions can be enhanced with integrative therapies that empower patients to marshal their inner resources during recovery.^25-28Music therapy may be particularly suited to this effort, as it is adaptable to the patient’s individual and culturally specific needs.^29-33

Rationale for Live Music

Pain is subjective and personal, and warrants an individualized approach to care. There is a body of music medicine research on the use of recorded music in modulating psychological and physiological factors in pain perception.^30,32,34-54 This research supports the unique relationship of music to well-being, and the understanding that controlling any of these factors affects the duration, intensity, and quality of that experience.^41,43,52

These findings provide incentive for breathing-entrained music therapy interventions, which enhance the relaxation response and release of pain-related tension;^32,55-58 empower patients to unlock physical and emotional tension;^32,57,58 provide a channel for expression and body movement; and enhance blood flow and/or alleviate pain by activating neurologic areas involved in the experience of pain.^59-62Studies have found that physical endurance may be enhanced when movement is rhythmically coordinated with a musical stimulus.^63-66 Music may prolong physical endurance by inhibiting psychological feedback associated with physical exertion related to fatigue, which may translate into accelerated recovery periods. When we listen to a rhythmic sound, our brains tend to automatically synchronize, or entrain, to external rhythmic cues that can stimulate increased motor control and coordination.⁶³ Sound can arouse and raise the excitability of spinal motor neurons mediated by auditory-motor neuronal connections on the brain stem and spinal cord level.^64-66 Rhythmically organized sounds serve as a neurological function in our capacity to organize predictable timing cues that are apparent in music, and may result in an effective treatment intervention in recovery.^63,64

Music Therapy in Recovery From Spine Surgery

In music therapy, music is used within a therapeutic relationship to support or affect change in the patient and the treatment regimen.^32,33,56-58 Research on music therapy with patients who are recovering from spine surgery is scant.^67-69 Kleiber and Adamek⁶⁷ studied perceptions of music therapy in 8 adolescents after spinal fusion surgery. In their study, a music therapist provided patients with a postoperative music therapy session focusing on the use of patient-preferred live music for relaxation and expression. Although their qualitative query was based on a therapeutic approach similar to that used in the present study, only 1 session was offered during the recovery period, and follow-up was conducted by survey invitation and telephone. In addition, the number of participants was small, and there was no quantitative measure of pain or other symptoms.

Another study focused on the effects of listening to music on pain intensity and distress after spine surgery.⁶⁸ Patients in the study’s music group made their selections from prerecorded classical music and domestic and international popular songs from various genres and listened to their chosen recordings 30 minutes a day. Although the study was not a music therapy study per se, it showed a positive impact of listening to music on anxiety and pain perception in 60 adults who were randomly assigned to the music group or to a non-music control group (n = 30 in each). Differences between the music and control groups’ VAS ratings of anxiety (Ps = .018-.001) and pain (P = .001) were statistically significant.

Different from our study, the aforementioned studies did not include tension release–focused live music offered within a therapeutic relationship. Our 1.5-year pilot study, conducted prior to the present study indicated that music therapy led to increased resilience and recovery mechanisms.⁵⁸

Methods

Our mixed-methods study design combined standard medical treatment with integrative music therapy interventions based on pain assessments to better understand the effects of music therapy on the recovery of patients after spine surgery.

The Spine Institute of New York within the Department of Orthopedic Surgery at Mount Sinai Beth Israel provides surgical treatment of common spinal cord conditions. Prioritizing patient satisfaction and positive outcomes,^27,28 the institute integrates music therapy through the Louis Armstrong Center for Music and Medicine to enhance treatment of pain symptoms.

Patients were recruited by the research team as per the daily surgical schedule, or through referral by the medical team or patient care navigator. Sixty patients (35 female, 25 male) ranging in age from 40 to 55 years underwent anterior, posterior, or anterior-posterior spinal fusion and were enrolled in the study after signing a participation consent form. Minorities, women, and patients with Medicaid and Medicare were included. Patients who received a diagnosis of clinical psychosis or depression prior to spine injury were excluded.

The experimental group received music therapy plus standard care (medical and nursing care with scheduled pharmacologic pain intervention), and a wait-listed control group received standard care only. A randomization chart created by a blinded statistician who did not have access to the patient census determined the intervention–nonintervention schedule. Patients in the music therapy group received one 30-minute music therapy session during an 8-hour period within 72 hours after surgery.

For both groups, measurements were completed before and after the study window. Control patients were offered music therapy after completion of the post-intervention surveys in order to minimize the ethical dilemma of denying potentially helpful pain intervention. For this same reason, both groups were given the option of receiving follow-up music therapy sessions for the duration of their hospitalization.

The research team consisted of 2 licensed, board-certified music therapists. In addition, Master’s-level music therapy interns completing clinical hours as part of the trajectory for board certification served on the research team over the 5-year period 2009 to 2014, and 13 blinded research assistants helped with enrolling and collecting data on patients.

Intervention

Each music therapy session included a warm-up phase of verbal or musical discourse. Next was the treatment phase, which was based on patient need as assessed during warm-up. Treatment options included use of patient-preferred live music that supported tension release/relaxation through incentive-based clinical improvisation, singing, and/or rhythmic drumming or through breathwork and visualization. Psychoeducation about mind–body awareness through the use of breath and imagery was introduced and explained by the therapist at this time.

The improvised music intervention was focused on making salient the natural harmonic tension-resolution cycles that occur in music and that were entrained to the patient’s presentation (respiratory rate, verbal report, clinical presentation). When patient-preferred precomposed songs were used, tension resolution was achieved by sustaining cadence and resolution, also entrained to the patient’s respiratory cycles.^32,57,58

After the music therapy intervention, a period of closure or integration was facilitated by the therapist contingent on the patient’s degree of alertness. If awake, the patient was supported in a reflexive process of thoughts, impressions, or issues that may have contributed to the overall experience. If the patient was asleep, the researcher returned within 30 minutes for post-intervention interviewing. Interview information was recorded in a qualitative post-participation survey. To prevent bias, researchers who were not the treating clinicians conducted the surveys.

Outcome Measures

Both primary and secondary outcome measures were collected before and after the intervention. The primary outcome measure was VAS pain ratings, and the secondary outcome measures were scores on the Hospital Anxiety and Depression Scale (HADS), the Tampa Scale for Kinesiophobia (TSK), and the Color Analysis Scale (CAS).

VAS. With the VAS, images are used to rate pain. The scale has points labeled 0 to 10 and corresponding faces representing progression in pain intensity. The scale is quickly rendered and can be interpreted according to the patient’s recovery phase at time of rendering.

HADS. The HADS⁷⁰ provides a specific baseline for anxiety and depression as an indicator of how the patient might fare during hospitalization (admission through recovery and discharge).

TSK. The TSK⁷¹ provides insight into the patient’s perception of fear-related movement, which is an important factor in this study because of the movement required for rehabilitation. We used a shortened version of the TSK to accommodate the sensitive threshold for pain tolerance and pharmacologic side effects commonly experienced by spine patients.

CAS. The CAS was developed at the Louis Armstrong Center for Music and Medicine to assess comorbidities and dynamic aspects of pain. Through a coloring exercise, patients illustrate their pain experience, which gives tangible form to the abstract experience of pain.

Coding

We collected patients’ demographic data, including age, sex, and diagnoses. Clinical indicators of the preoperative baseline included lifestyle, surgical history, and prior experience with music or other mind–body strategies for self-regulation.

As fundamental to qualitative methodology,^72,73 the reported responses to questions were grouped into themes that were peer-tested with members of the research team before and during the coding process.

Statistical Methods

Means and standard deviations were used for continuous variables, and frequencies (percentages) for categorical variables. All outcomes were analyzed on an intent-to-treat basis. Repeated-measures analysis of variance was used to compare changes in outcomes from before to after intervention for the music and control groups. In particular, a statistically significant Group (music vs control) × Time (before vs after intervention) interaction would support the hypothesis that there would be more benefit (less pain) in the music group as a result of the music therapy. For all tests, significance was set at P < .05. SPSS Version 20 (IBM) was used for all statistical analyses. Based on previously found differences in heart rate and mobility,³¹ we assumed an effect size of 0.71 for the difference between music and control (no music), which would require 32 patients per group to achieve a power of 0.8 with an α of 0.05.

Results

Of the 136 patients who were asked to participate in the study, 76 were not enrolled; the other 60 were equally assigned to either the control group or the music therapy group (n = 30 in each) according to randomization indicated by a blinded statistician (Figure 1).

Table 2 lists the pre-intervention and post-intervention comparisons of the main outcomes between groups.

Relationship with music was coded for significance and included reports of music as a resource accessed for stimulation and/or relaxation through listening; direct involvement with instrument playing; and history of music training. 

Perceptions of surgical outcome in patients’ responses were coded across 3 themes: (1) optimistic (belief and hope in returning to original baseline of functionality), (2) indifferent (neither hopeful nor cynical about results of surgery), and (3) pessimistic (belief that nothing will restore the quality of life that existed before the spinal condition).

The CAS helped us better understand the diversity and complexity of the pain experience.

Discussion

Our hospital has the unique capability of providing music therapy to postoperative and other hospitalized patients. In this study, we compared the impact of a structured postoperative music therapy program on spine patients relative to control patients who did not receive music therapy after spine surgery.

We found a significant benefit in VAS pain levels (>1 point) but no statistically significant differences in HADS Anxiety, HADS Depression, or TSK scores. Although a 2-point difference is usually considered clinically significant, the degree of change in the music group is notable for having been achieved by nonpharmacologic means with scant chance of adverse effects. We suspect the lack of significant change in HADS Anxiety, HADS Depression, and TSK scores is attributable to the narrow study window. Given the observational data from our pilot study⁵⁸ and ongoing results with spine patients,³² it seems clear that both mood state and resilience in coping are enhanced through an ongoing relationship with music therapy.

The study of a population as vulnerable as patients recovering from spine surgery raises many issues for providers and researchers. Although it is worthwhile to determine the efficacy of integrative modalities in serving these patients, the request for participation in a protocol at such a vulnerable time was often resisted. During our pilot work, it became clear that the ability of potential subjects to comprehend and complete protocol surveys was impacted by adverse effects, including sedation drowsiness; respiratory depression; nausea and vomiting; pruritus; and urinary retention caused by the medications used for postoperative pain management. Consequently, after piloting 5 cases before the main study, we extended the enrollment window to 72 hours.

Other unforeseen intrinsic or external obstacles were identified: Patient-related issues—including availability, level of interest in participation, and inability to participate because of the medication adverse effects mentioned.

Staff investment/education—addressed over the first 3 study years with several in-services, starting with the surgical team and continuing with nursing and support staff in various combinations. These meetings led to the creation of an Institutional Review Board (IRB) approved educational sheet for inclusion in the information packet given to surgical patients on registration.

Programming interruptions—caused by the convergence of several unanticipated factors, including a delay in expedited review of the IRB renewal during the year of Hurricane Sandy and an interruption in the spine team’s service for administrative and program modification.

Conclusion

Music therapy interventions (eg, use of patient-preferred live music) offered within a therapeutic relationship favorably affected pain perceptions in patients recovering from spine surgery. This effect was achieved through several therapeutic entry points, including support of expression and opportunities for emotional catharsis.

At the core of music therapy’s efficacy is individualized treatment, through which patients are supported in their recovery of “self.” Measurable benefits—including increased comfort; reduced pain; improved gait; increased range of motion, endurance, and ability to relax; and empowerment to actively participate in one’s own care through daily activities imbued with an enhanced sense of agency—are of cardinal importance, as they may lead to quicker recovery perceptions and enhanced quality of life.

Am J Orthop. 2017;46(1):E13-E22. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• Using the NSQIP database, the authors report that the overall complication rate was 1.21% after hip arthroscopy.
• The most common complications cited were bleeding requiring transfusion (0.45%), return to OR (0.23%), superficial infection (0.23%), and thrombophlebitis (0.15).
• Most common 10CPT code was arthroscopic débridement in 50% of cases, reflecting the types of cases being performed in the time period.
• FAI codes were less common in this database–labral repair in 24%, femoral osteochondroplasty in 16%, and acetabuloplasty in 9%.
• Use caution in patients over age 65 years as this appears to be a risk factor for morbidity.

Hip arthroscopy is a well-described method for treating a number of pathologies.^1-3 Surgical indications are wide-ranging and include femoral acetabular impingement (FAI), labral tears, loose bodies, osteochondral injuries, ruptured ligamentum teres, and synovitis, as well as extra-articular injuries, including hip abductor tears and sciatic nerve entrapment.^2,4-6 Authors have suggested that the advantages of hip arthroscopy over open procedures include less traumatic access to the hip joint and faster recovery,^7,8 and hip arthroscopy has been found cost-effective in select groups of patients.⁹

Overall complications have been reported in 1% to 20% of hip arthroscopy patients,^6,8,10,11 and a meta-analysis identified an overall complication rate of 4%.⁸ Complications include iatrogenic chondrolabral injury, nerve injury, superficial surgical-site infection, deep vein thrombosis (DVT), instrument failure, portal wound bleeding, soft-tissue injury, and intra-abdominal fluid extravasation.^6,8,10-13 Rates of major complications are relatively low, 0.3% to 0.58%, according to several recent systematic reviews.^8,12 Given the lack of universally accepted definitions, reports of minor complications (eg, iatrogenic chondrolabral injury, neuropraxia) in hip arthroscopy vary widely.⁸ Furthermore, many of the series with high complication rates represent early experience with the technique, and later authors suggested that complications should decrease with improvements in technique and technology.^12,14,15The literature is lacking in reports of risk factors for patient morbidity and large multi-institutional cohorts in the setting of hip arthroscopy. We conducted a study of elective hip arthroscopy patients to determine type and incidence of complications and rates of and risk factors for minor and major morbidity.

Materials and Methods

This retrospective study was deemed compliant with HIPAA (Health Insurance Portability and Accountability Act of 1996) and exempt from the need for Institutional Review Board approval. In the National Surgical Quality Improvement Program (NSQIP), academic and private medical institutions prospectively collect patient preoperative and operative data as well as 30-day outcome data from more than 500 hospitals throughout the United States.^16-21 Surgical clinical reviewers, who are responsible for data acquisition, prospectively collect morbidity data for 30 days after surgery through a chart review of patient progress notes, operative notes, and follow-up clinic visits. Patients may be contacted by a surgical clinical reviewer if they have not had a clinic visit within 30 days after a procedure to verify the presence or absence of complications or admissions at outside institutions, and in this way even outpatient complications should be captured. If the medical record is unclear, the reviewer may also contact the surgeon directly. In addition, NSQIP data are routinely audited; the interobserver disagreement rate is 1.56%.²²

We used Current Procedural Terminology (CPT) billing codes to retrospectively survey the NSQIP database for hip arthroscopies performed between 2006 and 2013. Excluding cases of compromised surgical wounds, emergent surgeries, surgeries involving fracture, hip dislocations, preoperative sepsis, septic joints, and osteomyelitis, we identified 1325 cases with CPT codes 29861 (hip arthroscopy), 29862 (arthroscopic hip débridement, shaving), 29914 (arthroscopic femoroplasty), 29915 (arthroscopic acetabuloplasty), and 29916 (arthroscopic labral repair). Postoperative outcomes were categorized as major morbidity or mortality, minor morbidity, and any complication. A major complication was a systemic life-threatening event or a substantial threat to a vital organ, whereas a minor complication did not pose a major systemic threat and was localized to the operative extremity (previously used definitions^23,24). We have used similar methods to report the rates of and risk factors for complications of knee arthroscopy, shoulder arthroscopy, and total shoulder arthroplasty.^16,20,21 For any-complication outcomes, we included both major and minor morbidities, and mortality. NSQIP applies strict definitions (listed in its user file¹⁷) to patient comorbidities and complications. Data points collected included patient demographics, medical comorbidities, laboratory values, and surgical characteristics.

Initially, we performed a univariate analysis that considered age, sex, race, body mass index, current alcohol abuse, current smoking status, recent weight loss, dyspnea, chronic obstructive pulmonary disease, CPT code, congestive heart failure, hypertension, diabetes, peripheral vascular disease, esophageal varices, disseminated cancer, steroid use, bleeding disorder, dialysis, chemotherapy within previous 30 days, radiation therapy within previous 90 days, operation within previous 30 days, American Society of Anesthesiologists class, operative time, resident involvement, and patient functional status. We also included mean preoperative sodium, blood urea nitrogen, and albumin levels; white blood cell count; hematocrit; platelet count; and international normalized ratio. The analysis revealed unadjusted differences between patients with and without complications (t test was used for continuous variables, χ² test for categorical variables). Any variable with P < .2 in the univariate analysis and more than 80% complete data was considered fit for our multivariate model. We controlled for confounders by performing a multivariate logistic regression analysis. Three separate analyses were performed; the outcome variables were major morbidity or mortality, minor morbidity, and any complication. P < .05 was used for statistical significance across all models. We used SAS Version 9.3 (SAS Institute) for statistical analysis. Model quality was evaluated for calibration (Hosmer-Lemeshow test) and discrimination (C statistics). The calibration test yielded a modified χ² statistic, and P > .05 indicated the model was appropriate and fit the data well. Good discrimination is commonly reported to be between 0.65 and 0.85.

Results

Of the 1325 patients who underwent hip arthroscopy, 60% were female. Regarding age, 52% were younger than 40 years, and 45% were between 45 years and 60 years. The most common diagnoses were articular cartilage disorder involving the pelvic region (15%), enthesopathy of the hip (12%), and joint pain involving the pelvic region or thigh (11%). The most common primary CPT code (50%) was for hip arthroscopic débridement (29862), followed by 24% for arthroscopic labral repair (29916), 16% for arthroscopic femoroplasty (29914), and 9% for arthroscopic acetabuloplasty (29915). Of the 16 complications found, 12 involved hip arthroscopic débridement, and 4 involved hip arthroscopic femoroplasty. There were no complications of arthroscopic acetabuloplasty (29915), arthroscopic labral repair (29916), or hip arthroscopy (29861).

Of the 1325 hip arthroscopy patients, 16 (1.21%) had at least 1 complication (Table 1).

Discussion

Earlier reports on hip arthroscopy did not consider risk factors for systemic morbidity and were mainly single-institution case series.^{3,10,11,13,25} Given a renewed focus on outcomes measurement and quality assessment in orthopedic surgery, we wanted to describe short-term complications of and risk factors for morbidity in hip arthroscopy. In this article, we report baseline data from a large multicenter cohort. For hip arthroscopy, we found low rates of short-term complications (1.21%) and major morbidities (0.45%). We considered many modifiable and nonmodifiable risk factors for complications and found age over 65 years to be an independent risk factor for any complication and minor morbidity. Several of our findings merit further discussion.

Other authors have reported hip arthroscopy complication rates of 1% to 20%, citing both systemic and local complications,^6,8,10-12 and major complication rates of 0.3% to 0.58%.^8,12 Minor complications of hip arthroscopy vary, and depend on definition, with long-term consequences unknown in some cases.⁸ Sensory neuropraxia, a relatively common minor complication in hip arthroscopy, is thought to be affected by the amount of traction against a perineal post and by increased operative time, with operative time under 2 hours previously suggested.^{3,6,10,11,13,25,26}

In the present study, the overall rate of any complication of hip arthroscopy was 1.21%, and the most common complications were bleeding resulting in transfusion, return to operating room, superficial surgical-site infection, and DVT/thrombophlebitis. When we excluded bleeding resulting in transfusion, the overall complication rate fell to 0.75%. Operative time was relatively short, <2 hours for 70% of patients. Last, there were no mortalities. As our data set did not include variables encompassing sensory neuropraxia or iatrogenic chondrolabral injury, we were unable to report on these data.

Surgeons and healthcare systems should be advised that rates of systemic complications in hip arthroscopy are low and that hip arthroscopy is a relatively safe procedure. Surgeons and healthcare systems can refer to our reported complication rates and risk factors when assessing quality and performing cost analysis in hip arthroscopy. For our 1325 patients, the major morbidity rate was 0.45%, within the range of previous reports.^8,12 There were no nerve injuries in our patient cohort, likely because of the strict NSQIP definitions of nerve injury. We cannot report on sensory neuropraxia and iatrogenic chondrolabral injury. We speculate that lack of these variables may have artificially lowered our minor complication rate.

Some authors have reported clinical benefits of hip arthroscopy in older patients,^27-29 whereas others have suggested age may be a negative prognostic factor.^27,30 Suggested indications for hip arthroscopy in an elderly population include chondral defects, labral tears, and FAI in the absence of significant arthritic changes.^28,29 Larson and colleagues,³⁰ who reported a 52% failure rate for osteoarthritis patients who underwent hip arthroscopy for FAI, concluded that arthroscopy should not be offered to patients with evidence of advanced radiographic joint space narrowing. Others have noted that patients who were under age 55 years and had minimal osteoarthritic changes had a longer interval between hip arthroscopy and total hip arthroplasty in comparison with patients over age 55 years.³¹ Previous work in knee arthroscopy found older age (40-65 years vs <40 years) was an independent predictor of short-term complications (1.5 times increased risk).²¹ In the present study, 7.69% of patients who were over age 65 years when they underwent hip arthroscopy had a complication, and we report age over 65 years as an independent risk factor for any complication (OR, 6.52) and minor morbidity (OR, 7.97). Surgeons should be aware that advanced age is an independent risk factor for complications in hip arthroscopy. Potential benefits of hip arthroscopy should be carefully weighed against the increased risk in this patient cohort, and surgeons should ascertain the scope of an elderly patient’s disease to determine if hip arthroscopy is indicated and worth the potential risks.

To our knowledge, bleeding resulting in transfusion was not previously described as a complication of hip arthroscopy. In the present study, bleeding resulting in transfusion was the most common complication (6 patients, 0.45%), and all the affected patients had a primary CPT code for arthroscopic débridement (29862). The 6 primary diagnoses were hip osteoarthrosis (3), thigh/pelvis pain (1), unspecified injury (1), and congenital hip deformity (1). The 6 transfusion patients also tended to be older (ages 30, 53, 64, 67, 76, and 90 years). Although drawing firm conclusions from so few patients would be inappropriate, we acknowledge that the majority who received a transfusion were older, underwent arthroscopic débridement of a hip, and had a primary diagnosis of osteoarthrosis or pain. As transfusion practices can differ between surgeons and groups, we conclude that the risk for bleeding requiring transfusion is low in hip arthroscopy. Patients who are older and who undergo arthroscopic débridement of an osteoarthritic hip may be at elevated risk for transfusion.

This study had several limitations. First, with use of the NSQIP database, follow-up was limited to 30 days. We speculate that longer follow-up might yield higher complication rates and additional risk factors. Second, we could not distinguish individual surgeon or site data and acknowledge complications might differ between surgeons and sites that perform hip arthroscopy more frequently. Third, as data were limited to medical and broadly applicable surgical variables included in the NSQIP database, they might not be specific to hip arthroscopy, and we cannot report on iatrogenic chondrolabral injury and neuropraxia, 2 previously reported minor complications in hip arthroscopy. We speculate that data collection focused on problems specific to hip arthroscopy would yield more complications and risk factors.

Conclusion

According to the NSQIP data, the rate of short-term morbidity after elective hip arthroscopy was low, 1.21%. Surgeons may use our reported complications and risk factors when counseling patients, and healthcare systems may use our data when assessing quality and performance in hip arthroscopy. Surgeons who perform elective hip arthroscopy should be aware that age over 65 years is an independent predictor of complications. Careful attention should be given to this patient group when indicating hip arthroscopy procedures.

Am J Orthop. 2017;46(1):E1-E9. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• Using the NSQIP database, the authors report that the overall complication rate was 1.21% after hip arthroscopy.
• The most common complications cited were bleeding requiring transfusion (0.45%), return to OR (0.23%), superficial infection (0.23%), and thrombophlebitis (0.15).
• Most common 10CPT code was arthroscopic débridement in 50% of cases, reflecting the types of cases being performed in the time period.
• FAI codes were less common in this database–labral repair in 24%, femoral osteochondroplasty in 16%, and acetabuloplasty in 9%.
• Use caution in patients over age 65 years as this appears to be a risk factor for morbidity.

Hip arthroscopy is a well-described method for treating a number of pathologies.^1-3 Surgical indications are wide-ranging and include femoral acetabular impingement (FAI), labral tears, loose bodies, osteochondral injuries, ruptured ligamentum teres, and synovitis, as well as extra-articular injuries, including hip abductor tears and sciatic nerve entrapment.^2,4-6 Authors have suggested that the advantages of hip arthroscopy over open procedures include less traumatic access to the hip joint and faster recovery,^7,8 and hip arthroscopy has been found cost-effective in select groups of patients.⁹

Overall complications have been reported in 1% to 20% of hip arthroscopy patients,^6,8,10,11 and a meta-analysis identified an overall complication rate of 4%.⁸ Complications include iatrogenic chondrolabral injury, nerve injury, superficial surgical-site infection, deep vein thrombosis (DVT), instrument failure, portal wound bleeding, soft-tissue injury, and intra-abdominal fluid extravasation.^6,8,10-13 Rates of major complications are relatively low, 0.3% to 0.58%, according to several recent systematic reviews.^8,12 Given the lack of universally accepted definitions, reports of minor complications (eg, iatrogenic chondrolabral injury, neuropraxia) in hip arthroscopy vary widely.⁸ Furthermore, many of the series with high complication rates represent early experience with the technique, and later authors suggested that complications should decrease with improvements in technique and technology.^12,14,15The literature is lacking in reports of risk factors for patient morbidity and large multi-institutional cohorts in the setting of hip arthroscopy. We conducted a study of elective hip arthroscopy patients to determine type and incidence of complications and rates of and risk factors for minor and major morbidity.

Materials and Methods

This retrospective study was deemed compliant with HIPAA (Health Insurance Portability and Accountability Act of 1996) and exempt from the need for Institutional Review Board approval. In the National Surgical Quality Improvement Program (NSQIP), academic and private medical institutions prospectively collect patient preoperative and operative data as well as 30-day outcome data from more than 500 hospitals throughout the United States.^16-21 Surgical clinical reviewers, who are responsible for data acquisition, prospectively collect morbidity data for 30 days after surgery through a chart review of patient progress notes, operative notes, and follow-up clinic visits. Patients may be contacted by a surgical clinical reviewer if they have not had a clinic visit within 30 days after a procedure to verify the presence or absence of complications or admissions at outside institutions, and in this way even outpatient complications should be captured. If the medical record is unclear, the reviewer may also contact the surgeon directly. In addition, NSQIP data are routinely audited; the interobserver disagreement rate is 1.56%.²²

We used Current Procedural Terminology (CPT) billing codes to retrospectively survey the NSQIP database for hip arthroscopies performed between 2006 and 2013. Excluding cases of compromised surgical wounds, emergent surgeries, surgeries involving fracture, hip dislocations, preoperative sepsis, septic joints, and osteomyelitis, we identified 1325 cases with CPT codes 29861 (hip arthroscopy), 29862 (arthroscopic hip débridement, shaving), 29914 (arthroscopic femoroplasty), 29915 (arthroscopic acetabuloplasty), and 29916 (arthroscopic labral repair). Postoperative outcomes were categorized as major morbidity or mortality, minor morbidity, and any complication. A major complication was a systemic life-threatening event or a substantial threat to a vital organ, whereas a minor complication did not pose a major systemic threat and was localized to the operative extremity (previously used definitions^23,24). We have used similar methods to report the rates of and risk factors for complications of knee arthroscopy, shoulder arthroscopy, and total shoulder arthroplasty.^16,20,21 For any-complication outcomes, we included both major and minor morbidities, and mortality. NSQIP applies strict definitions (listed in its user file¹⁷) to patient comorbidities and complications. Data points collected included patient demographics, medical comorbidities, laboratory values, and surgical characteristics.

Initially, we performed a univariate analysis that considered age, sex, race, body mass index, current alcohol abuse, current smoking status, recent weight loss, dyspnea, chronic obstructive pulmonary disease, CPT code, congestive heart failure, hypertension, diabetes, peripheral vascular disease, esophageal varices, disseminated cancer, steroid use, bleeding disorder, dialysis, chemotherapy within previous 30 days, radiation therapy within previous 90 days, operation within previous 30 days, American Society of Anesthesiologists class, operative time, resident involvement, and patient functional status. We also included mean preoperative sodium, blood urea nitrogen, and albumin levels; white blood cell count; hematocrit; platelet count; and international normalized ratio. The analysis revealed unadjusted differences between patients with and without complications (t test was used for continuous variables, χ² test for categorical variables). Any variable with P < .2 in the univariate analysis and more than 80% complete data was considered fit for our multivariate model. We controlled for confounders by performing a multivariate logistic regression analysis. Three separate analyses were performed; the outcome variables were major morbidity or mortality, minor morbidity, and any complication. P < .05 was used for statistical significance across all models. We used SAS Version 9.3 (SAS Institute) for statistical analysis. Model quality was evaluated for calibration (Hosmer-Lemeshow test) and discrimination (C statistics). The calibration test yielded a modified χ² statistic, and P > .05 indicated the model was appropriate and fit the data well. Good discrimination is commonly reported to be between 0.65 and 0.85.

Results

Of the 1325 patients who underwent hip arthroscopy, 60% were female. Regarding age, 52% were younger than 40 years, and 45% were between 45 years and 60 years. The most common diagnoses were articular cartilage disorder involving the pelvic region (15%), enthesopathy of the hip (12%), and joint pain involving the pelvic region or thigh (11%). The most common primary CPT code (50%) was for hip arthroscopic débridement (29862), followed by 24% for arthroscopic labral repair (29916), 16% for arthroscopic femoroplasty (29914), and 9% for arthroscopic acetabuloplasty (29915). Of the 16 complications found, 12 involved hip arthroscopic débridement, and 4 involved hip arthroscopic femoroplasty. There were no complications of arthroscopic acetabuloplasty (29915), arthroscopic labral repair (29916), or hip arthroscopy (29861).

Of the 1325 hip arthroscopy patients, 16 (1.21%) had at least 1 complication (Table 1).

Discussion

Earlier reports on hip arthroscopy did not consider risk factors for systemic morbidity and were mainly single-institution case series.^{3,10,11,13,25} Given a renewed focus on outcomes measurement and quality assessment in orthopedic surgery, we wanted to describe short-term complications of and risk factors for morbidity in hip arthroscopy. In this article, we report baseline data from a large multicenter cohort. For hip arthroscopy, we found low rates of short-term complications (1.21%) and major morbidities (0.45%). We considered many modifiable and nonmodifiable risk factors for complications and found age over 65 years to be an independent risk factor for any complication and minor morbidity. Several of our findings merit further discussion.

Other authors have reported hip arthroscopy complication rates of 1% to 20%, citing both systemic and local complications,^6,8,10-12 and major complication rates of 0.3% to 0.58%.^8,12 Minor complications of hip arthroscopy vary, and depend on definition, with long-term consequences unknown in some cases.⁸ Sensory neuropraxia, a relatively common minor complication in hip arthroscopy, is thought to be affected by the amount of traction against a perineal post and by increased operative time, with operative time under 2 hours previously suggested.^{3,6,10,11,13,25,26}

In the present study, the overall rate of any complication of hip arthroscopy was 1.21%, and the most common complications were bleeding resulting in transfusion, return to operating room, superficial surgical-site infection, and DVT/thrombophlebitis. When we excluded bleeding resulting in transfusion, the overall complication rate fell to 0.75%. Operative time was relatively short, <2 hours for 70% of patients. Last, there were no mortalities. As our data set did not include variables encompassing sensory neuropraxia or iatrogenic chondrolabral injury, we were unable to report on these data.

Surgeons and healthcare systems should be advised that rates of systemic complications in hip arthroscopy are low and that hip arthroscopy is a relatively safe procedure. Surgeons and healthcare systems can refer to our reported complication rates and risk factors when assessing quality and performing cost analysis in hip arthroscopy. For our 1325 patients, the major morbidity rate was 0.45%, within the range of previous reports.^8,12 There were no nerve injuries in our patient cohort, likely because of the strict NSQIP definitions of nerve injury. We cannot report on sensory neuropraxia and iatrogenic chondrolabral injury. We speculate that lack of these variables may have artificially lowered our minor complication rate.

Some authors have reported clinical benefits of hip arthroscopy in older patients,^27-29 whereas others have suggested age may be a negative prognostic factor.^27,30 Suggested indications for hip arthroscopy in an elderly population include chondral defects, labral tears, and FAI in the absence of significant arthritic changes.^28,29 Larson and colleagues,³⁰ who reported a 52% failure rate for osteoarthritis patients who underwent hip arthroscopy for FAI, concluded that arthroscopy should not be offered to patients with evidence of advanced radiographic joint space narrowing. Others have noted that patients who were under age 55 years and had minimal osteoarthritic changes had a longer interval between hip arthroscopy and total hip arthroplasty in comparison with patients over age 55 years.³¹ Previous work in knee arthroscopy found older age (40-65 years vs <40 years) was an independent predictor of short-term complications (1.5 times increased risk).²¹ In the present study, 7.69% of patients who were over age 65 years when they underwent hip arthroscopy had a complication, and we report age over 65 years as an independent risk factor for any complication (OR, 6.52) and minor morbidity (OR, 7.97). Surgeons should be aware that advanced age is an independent risk factor for complications in hip arthroscopy. Potential benefits of hip arthroscopy should be carefully weighed against the increased risk in this patient cohort, and surgeons should ascertain the scope of an elderly patient’s disease to determine if hip arthroscopy is indicated and worth the potential risks.

To our knowledge, bleeding resulting in transfusion was not previously described as a complication of hip arthroscopy. In the present study, bleeding resulting in transfusion was the most common complication (6 patients, 0.45%), and all the affected patients had a primary CPT code for arthroscopic débridement (29862). The 6 primary diagnoses were hip osteoarthrosis (3), thigh/pelvis pain (1), unspecified injury (1), and congenital hip deformity (1). The 6 transfusion patients also tended to be older (ages 30, 53, 64, 67, 76, and 90 years). Although drawing firm conclusions from so few patients would be inappropriate, we acknowledge that the majority who received a transfusion were older, underwent arthroscopic débridement of a hip, and had a primary diagnosis of osteoarthrosis or pain. As transfusion practices can differ between surgeons and groups, we conclude that the risk for bleeding requiring transfusion is low in hip arthroscopy. Patients who are older and who undergo arthroscopic débridement of an osteoarthritic hip may be at elevated risk for transfusion.

This study had several limitations. First, with use of the NSQIP database, follow-up was limited to 30 days. We speculate that longer follow-up might yield higher complication rates and additional risk factors. Second, we could not distinguish individual surgeon or site data and acknowledge complications might differ between surgeons and sites that perform hip arthroscopy more frequently. Third, as data were limited to medical and broadly applicable surgical variables included in the NSQIP database, they might not be specific to hip arthroscopy, and we cannot report on iatrogenic chondrolabral injury and neuropraxia, 2 previously reported minor complications in hip arthroscopy. We speculate that data collection focused on problems specific to hip arthroscopy would yield more complications and risk factors.

Conclusion

According to the NSQIP data, the rate of short-term morbidity after elective hip arthroscopy was low, 1.21%. Surgeons may use our reported complications and risk factors when counseling patients, and healthcare systems may use our data when assessing quality and performance in hip arthroscopy. Surgeons who perform elective hip arthroscopy should be aware that age over 65 years is an independent predictor of complications. Careful attention should be given to this patient group when indicating hip arthroscopy procedures.

Am J Orthop. 2017;46(1):E1-E9. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Take-Home Points

• Using the NSQIP database, the authors report that the overall complication rate was 1.21% after hip arthroscopy.
• The most common complications cited were bleeding requiring transfusion (0.45%), return to OR (0.23%), superficial infection (0.23%), and thrombophlebitis (0.15).
• Most common 10CPT code was arthroscopic débridement in 50% of cases, reflecting the types of cases being performed in the time period.
• FAI codes were less common in this database–labral repair in 24%, femoral osteochondroplasty in 16%, and acetabuloplasty in 9%.
• Use caution in patients over age 65 years as this appears to be a risk factor for morbidity.

Hip arthroscopy is a well-described method for treating a number of pathologies.^1-3 Surgical indications are wide-ranging and include femoral acetabular impingement (FAI), labral tears, loose bodies, osteochondral injuries, ruptured ligamentum teres, and synovitis, as well as extra-articular injuries, including hip abductor tears and sciatic nerve entrapment.^2,4-6 Authors have suggested that the advantages of hip arthroscopy over open procedures include less traumatic access to the hip joint and faster recovery,^7,8 and hip arthroscopy has been found cost-effective in select groups of patients.⁹

Overall complications have been reported in 1% to 20% of hip arthroscopy patients,^6,8,10,11 and a meta-analysis identified an overall complication rate of 4%.⁸ Complications include iatrogenic chondrolabral injury, nerve injury, superficial surgical-site infection, deep vein thrombosis (DVT), instrument failure, portal wound bleeding, soft-tissue injury, and intra-abdominal fluid extravasation.^6,8,10-13 Rates of major complications are relatively low, 0.3% to 0.58%, according to several recent systematic reviews.^8,12 Given the lack of universally accepted definitions, reports of minor complications (eg, iatrogenic chondrolabral injury, neuropraxia) in hip arthroscopy vary widely.⁸ Furthermore, many of the series with high complication rates represent early experience with the technique, and later authors suggested that complications should decrease with improvements in technique and technology.^12,14,15The literature is lacking in reports of risk factors for patient morbidity and large multi-institutional cohorts in the setting of hip arthroscopy. We conducted a study of elective hip arthroscopy patients to determine type and incidence of complications and rates of and risk factors for minor and major morbidity.

Materials and Methods

This retrospective study was deemed compliant with HIPAA (Health Insurance Portability and Accountability Act of 1996) and exempt from the need for Institutional Review Board approval. In the National Surgical Quality Improvement Program (NSQIP), academic and private medical institutions prospectively collect patient preoperative and operative data as well as 30-day outcome data from more than 500 hospitals throughout the United States.^16-21 Surgical clinical reviewers, who are responsible for data acquisition, prospectively collect morbidity data for 30 days after surgery through a chart review of patient progress notes, operative notes, and follow-up clinic visits. Patients may be contacted by a surgical clinical reviewer if they have not had a clinic visit within 30 days after a procedure to verify the presence or absence of complications or admissions at outside institutions, and in this way even outpatient complications should be captured. If the medical record is unclear, the reviewer may also contact the surgeon directly. In addition, NSQIP data are routinely audited; the interobserver disagreement rate is 1.56%.²²

We used Current Procedural Terminology (CPT) billing codes to retrospectively survey the NSQIP database for hip arthroscopies performed between 2006 and 2013. Excluding cases of compromised surgical wounds, emergent surgeries, surgeries involving fracture, hip dislocations, preoperative sepsis, septic joints, and osteomyelitis, we identified 1325 cases with CPT codes 29861 (hip arthroscopy), 29862 (arthroscopic hip débridement, shaving), 29914 (arthroscopic femoroplasty), 29915 (arthroscopic acetabuloplasty), and 29916 (arthroscopic labral repair). Postoperative outcomes were categorized as major morbidity or mortality, minor morbidity, and any complication. A major complication was a systemic life-threatening event or a substantial threat to a vital organ, whereas a minor complication did not pose a major systemic threat and was localized to the operative extremity (previously used definitions^23,24). We have used similar methods to report the rates of and risk factors for complications of knee arthroscopy, shoulder arthroscopy, and total shoulder arthroplasty.^16,20,21 For any-complication outcomes, we included both major and minor morbidities, and mortality. NSQIP applies strict definitions (listed in its user file¹⁷) to patient comorbidities and complications. Data points collected included patient demographics, medical comorbidities, laboratory values, and surgical characteristics.

Initially, we performed a univariate analysis that considered age, sex, race, body mass index, current alcohol abuse, current smoking status, recent weight loss, dyspnea, chronic obstructive pulmonary disease, CPT code, congestive heart failure, hypertension, diabetes, peripheral vascular disease, esophageal varices, disseminated cancer, steroid use, bleeding disorder, dialysis, chemotherapy within previous 30 days, radiation therapy within previous 90 days, operation within previous 30 days, American Society of Anesthesiologists class, operative time, resident involvement, and patient functional status. We also included mean preoperative sodium, blood urea nitrogen, and albumin levels; white blood cell count; hematocrit; platelet count; and international normalized ratio. The analysis revealed unadjusted differences between patients with and without complications (t test was used for continuous variables, χ² test for categorical variables). Any variable with P < .2 in the univariate analysis and more than 80% complete data was considered fit for our multivariate model. We controlled for confounders by performing a multivariate logistic regression analysis. Three separate analyses were performed; the outcome variables were major morbidity or mortality, minor morbidity, and any complication. P < .05 was used for statistical significance across all models. We used SAS Version 9.3 (SAS Institute) for statistical analysis. Model quality was evaluated for calibration (Hosmer-Lemeshow test) and discrimination (C statistics). The calibration test yielded a modified χ² statistic, and P > .05 indicated the model was appropriate and fit the data well. Good discrimination is commonly reported to be between 0.65 and 0.85.

Results

Of the 1325 patients who underwent hip arthroscopy, 60% were female. Regarding age, 52% were younger than 40 years, and 45% were between 45 years and 60 years. The most common diagnoses were articular cartilage disorder involving the pelvic region (15%), enthesopathy of the hip (12%), and joint pain involving the pelvic region or thigh (11%). The most common primary CPT code (50%) was for hip arthroscopic débridement (29862), followed by 24% for arthroscopic labral repair (29916), 16% for arthroscopic femoroplasty (29914), and 9% for arthroscopic acetabuloplasty (29915). Of the 16 complications found, 12 involved hip arthroscopic débridement, and 4 involved hip arthroscopic femoroplasty. There were no complications of arthroscopic acetabuloplasty (29915), arthroscopic labral repair (29916), or hip arthroscopy (29861).

Of the 1325 hip arthroscopy patients, 16 (1.21%) had at least 1 complication (Table 1).

Discussion

Earlier reports on hip arthroscopy did not consider risk factors for systemic morbidity and were mainly single-institution case series.^{3,10,11,13,25} Given a renewed focus on outcomes measurement and quality assessment in orthopedic surgery, we wanted to describe short-term complications of and risk factors for morbidity in hip arthroscopy. In this article, we report baseline data from a large multicenter cohort. For hip arthroscopy, we found low rates of short-term complications (1.21%) and major morbidities (0.45%). We considered many modifiable and nonmodifiable risk factors for complications and found age over 65 years to be an independent risk factor for any complication and minor morbidity. Several of our findings merit further discussion.

Other authors have reported hip arthroscopy complication rates of 1% to 20%, citing both systemic and local complications,^6,8,10-12 and major complication rates of 0.3% to 0.58%.^8,12 Minor complications of hip arthroscopy vary, and depend on definition, with long-term consequences unknown in some cases.⁸ Sensory neuropraxia, a relatively common minor complication in hip arthroscopy, is thought to be affected by the amount of traction against a perineal post and by increased operative time, with operative time under 2 hours previously suggested.^{3,6,10,11,13,25,26}

In the present study, the overall rate of any complication of hip arthroscopy was 1.21%, and the most common complications were bleeding resulting in transfusion, return to operating room, superficial surgical-site infection, and DVT/thrombophlebitis. When we excluded bleeding resulting in transfusion, the overall complication rate fell to 0.75%. Operative time was relatively short, <2 hours for 70% of patients. Last, there were no mortalities. As our data set did not include variables encompassing sensory neuropraxia or iatrogenic chondrolabral injury, we were unable to report on these data.

Surgeons and healthcare systems should be advised that rates of systemic complications in hip arthroscopy are low and that hip arthroscopy is a relatively safe procedure. Surgeons and healthcare systems can refer to our reported complication rates and risk factors when assessing quality and performing cost analysis in hip arthroscopy. For our 1325 patients, the major morbidity rate was 0.45%, within the range of previous reports.^8,12 There were no nerve injuries in our patient cohort, likely because of the strict NSQIP definitions of nerve injury. We cannot report on sensory neuropraxia and iatrogenic chondrolabral injury. We speculate that lack of these variables may have artificially lowered our minor complication rate.

Some authors have reported clinical benefits of hip arthroscopy in older patients,^27-29 whereas others have suggested age may be a negative prognostic factor.^27,30 Suggested indications for hip arthroscopy in an elderly population include chondral defects, labral tears, and FAI in the absence of significant arthritic changes.^28,29 Larson and colleagues,³⁰ who reported a 52% failure rate for osteoarthritis patients who underwent hip arthroscopy for FAI, concluded that arthroscopy should not be offered to patients with evidence of advanced radiographic joint space narrowing. Others have noted that patients who were under age 55 years and had minimal osteoarthritic changes had a longer interval between hip arthroscopy and total hip arthroplasty in comparison with patients over age 55 years.³¹ Previous work in knee arthroscopy found older age (40-65 years vs <40 years) was an independent predictor of short-term complications (1.5 times increased risk).²¹ In the present study, 7.69% of patients who were over age 65 years when they underwent hip arthroscopy had a complication, and we report age over 65 years as an independent risk factor for any complication (OR, 6.52) and minor morbidity (OR, 7.97). Surgeons should be aware that advanced age is an independent risk factor for complications in hip arthroscopy. Potential benefits of hip arthroscopy should be carefully weighed against the increased risk in this patient cohort, and surgeons should ascertain the scope of an elderly patient’s disease to determine if hip arthroscopy is indicated and worth the potential risks.

To our knowledge, bleeding resulting in transfusion was not previously described as a complication of hip arthroscopy. In the present study, bleeding resulting in transfusion was the most common complication (6 patients, 0.45%), and all the affected patients had a primary CPT code for arthroscopic débridement (29862). The 6 primary diagnoses were hip osteoarthrosis (3), thigh/pelvis pain (1), unspecified injury (1), and congenital hip deformity (1). The 6 transfusion patients also tended to be older (ages 30, 53, 64, 67, 76, and 90 years). Although drawing firm conclusions from so few patients would be inappropriate, we acknowledge that the majority who received a transfusion were older, underwent arthroscopic débridement of a hip, and had a primary diagnosis of osteoarthrosis or pain. As transfusion practices can differ between surgeons and groups, we conclude that the risk for bleeding requiring transfusion is low in hip arthroscopy. Patients who are older and who undergo arthroscopic débridement of an osteoarthritic hip may be at elevated risk for transfusion.

This study had several limitations. First, with use of the NSQIP database, follow-up was limited to 30 days. We speculate that longer follow-up might yield higher complication rates and additional risk factors. Second, we could not distinguish individual surgeon or site data and acknowledge complications might differ between surgeons and sites that perform hip arthroscopy more frequently. Third, as data were limited to medical and broadly applicable surgical variables included in the NSQIP database, they might not be specific to hip arthroscopy, and we cannot report on iatrogenic chondrolabral injury and neuropraxia, 2 previously reported minor complications in hip arthroscopy. We speculate that data collection focused on problems specific to hip arthroscopy would yield more complications and risk factors.

Conclusion

According to the NSQIP data, the rate of short-term morbidity after elective hip arthroscopy was low, 1.21%. Surgeons may use our reported complications and risk factors when counseling patients, and healthcare systems may use our data when assessing quality and performance in hip arthroscopy. Surgeons who perform elective hip arthroscopy should be aware that age over 65 years is an independent predictor of complications. Careful attention should be given to this patient group when indicating hip arthroscopy procedures.

Am J Orthop. 2017;46(1):E1-E9. Copyright Frontline Medical Communications Inc. 2017. All rights reserved.

Patients leave the hospital against medical advice (AMA) for a variety of reasons. The AMA rate is approximately 1% nationally but substantially higher at safety-net hospitals and has rapidly increased over the past decade.^1-5 The principle that patients have the right to make choices about their healthcare, up to and including whether to leave the hospital against the advice of medical staff, is well-established law and a foundation of medical ethics.⁶ In practice, however, AMA discharges are often emotionally charged for both patients and providers, and, in the high-stress setting of AMA discharge, providers may be confused about their roles.^7-9

The demographics of patients who leave AMA have been well described. Compared with conventionally discharged patients, AMA patients are younger, more likely to be male, and more likely a marginalized ethnic or racial minority.^10-14 Patients with mental illnesses and addiction issues are overrepresented in AMA discharges, and complicated capacity assessments and limited resources may strain providers.^7,8,15,16 Studies have repeatedly shown higher rates of readmission and mortality for AMA patients than for conventionally discharged patients.^17-21 Whether AMA discharge is a marker for other prognostic factors that bode poorly for patients or contributes to negative outcomes, data suggest this group of patients is vulnerable, having mortality rates up to 40% higher 1 year after discharge, relative to conventionally discharged patients.¹²

Several models of standardized best practice approaches for AMA have been proposed by bioethicists.^6,22,23 Although details of these approaches vary, all involve assessing the patient’s decision-making capacity, clarifying the risks of AMA discharge, addressing factors that might be prompting the discharge, formulating an alternative outpatient treatment plan or “next best” option, and documenting extensively. A recent study found patients often gave advance warning of an AMA discharge, but physicians rarely prepared by arranging follow-up care.⁸ The investigators hypothesized that providers might not have known what they were permitted to arrange for AMA patients, or might have thought that providing “second best” options went against their principles. The investigators noted that nurses might have become aware of AMA risk sooner than physicians did but could not act on this awareness by preparing medications and arranging follow-up.

Translating models of best practice care for AMA patients into clinical practice requires buy-in from bedside providers, not just bioethicists. Given the study findings that providers have misconceptions about their roles in the AMA discharge,⁷ it is prudent to investigate providers’ current practices, beliefs, and concerns about AMA discharges before introducing a new approach. 

The present authors conducted a mixed-methods cross-sectional study of the state of AMA discharges at Highland Hospital (Oakland, California), a 236-bed county hospital and trauma center serving a primarily underserved urban patient population. The aim of this study was to assess current provider practices for AMA discharges and provider perceptions and knowledge about AMA discharges, ultimately to help direct future educational interventions with medical providers or hospital policy changes needed to improve the quality of AMA discharges. 

Phase 1 of this study involved identifying AMA patients through a review of data from Highland Hospital’s electronic medical records for 2014. These data included discharge status (eg, AMA vs other discharge types). The hospital’s floor clerk distinguishes between absent without official leave (AWOL; the patient leaves without notifying a provider) and AMA discharge. Discharges designated AWOL were excluded from the analyses. 

In phase 2, a structured chart review (Appendix A) was performed for all patients identified during phase 1 as being discharged AMA in 2014. In these reviews, further assessment was made of patient and visit characteristics in hospitalizations that ended in AMA discharge, and of providers’ documentation of AMA discharges—that is, whether several factors were documented (capacity; predischarge indication that patient might leave AMA; reason for AMA; and indications that discharge medications, transportation, and follow-up were arranged). These visit factors were reviewed because the literature has identified them as being important markers for AMA discharge safety.^6,8 Two research assistants, under the guidance of Dr. Stearns, reviewed the charts. To ensure agreement across chart reviews with respect to subjective questions (eg, whether capacity was adequately documented), the group reviewed the first 10 consecutive charts together; there was full agreement on how to classify the data of interest. Throughout the study, whenever a research assistant asked how to classify particular patient data, Dr. Stearns reviewed the data, and the research team made a decision together. Additional data, for AMA patients and for all patients admitted to Highland Hospital, were obtained from the hospital’s data warehouse, which pools data from within the health system.

Phase 3 involved surveying healthcare providers who were involved in patient care on the internal medicine and trauma surgery services at the hospital. These providers were selected because chart review revealed that the vast majority of patients who left AMA in 2014 were on one of these services. Surveys (Appendix B) asked participant providers to identify their role at the hospital, to provide a self-assessment of competence in various aspects of AMA discharge, to voice opinions about provider responsibilities in arranging follow-up for AMA patients, and to make suggestions about the AMA process. The authors designed these surveys, which included questions about aspects of care that have been highlighted in the AMA discharge literature as being important for AMA discharge safety.^6,8,22,23 Surveys were distributed to providers at internal medicine and trauma surgery department meetings and nursing conferences. Data (without identifying information) were analyzed, and survey responses kept anonymous.

The Alameda Health System Institutional Review Board approved this project. Providers were given the option of writing their name and contact information at the top of the survey in order to be entered into a drawing to receive a prize for completion.

We performed statistical analyses of the patient charts and physician survey data using Stata (version 14.0, Stata Corp., College Station, Texas). We analyzed both patient- and encounter-level data. In demographic analyses, this approach prevented duplicate counting of patients who left AMA multiple times. Patient-level analyses compared the demographic characteristics of AMA patients and patients discharged conventionally from the hospital in 2014. In addition, patients with either 1 or multiple AMA discharges were compared to identify characteristics that might be linked to highest risk of recurrent AMA discharge in the hope that early identification of these patients might facilitate providers’ early awareness and preparation for follow-up care or hospitalization alternatives. We used ANOVAs for continuous variables and tests of proportions for categorical variables. On the encounter level, analyses examined data about each admission (eg, AMA forms signed, follow-up arrangements made, capacity documented, etc.) for all AMA discharges. We employed chi square tests to identify variations in healthcare provider survey responses. A P value < 0.05 was used as the significance cut-off point.

Staged logistic regression analyses, adjusted for demographic characteristics, were performed to assess the association between risk of leaving AMA (yes or no) and demographic characteristics and the association between risk of leaving AMA more than once (yes or no) and health-related characteristics.

Demographic, Clinical, and Utilization Characteristics

Of the 12,036 Highland Hospital admissions in 2014, 319 (2.7%) ended with an AMA discharge. Of the 8207 individual patients discharged, 268 left AMA once, and 29 left AMA multiple times. Further review of the Admissions, Discharges, and Transfers Report generated from the electronic medical record revealed that 15 AWOL discharges were misclassified as AMA discharges.

Compared with patients discharged conventionally, AMA patients were significantly younger; more likely to be male, to self-identify as Black/African American, and to be English-speaking; and less likely to self-identify as Asian/Pacific Islander or Hispanic/Latino or to be Chinese- or Spanish-speaking (Table 1). They were also more likely than all patients admitted to Highland to be homeless (15.7% vs 8.7%; P < 0.01). Multivariate regression analysis revealed persistent age and sex disparities, but racial disparities were mitigated in adjusted analyses (Appendix C). Language disparities persisted only for Spanish speakers, who had a significantly lower rate of AMA discharge, even in adjusted analyses. 

The majority of AMA patients were on the internal medicine service (63.5%) or the trauma surgery service (24.8%). Regarding admission diagnosis, 17.2% of AMA patients were admitted for infections, 5.0% for drug or alcohol intoxication or withdrawal, 38.9% for acute noninfectious illnesses, 16.7% for decompensation of chronic disease, 18.4% for injuries or trauma, and 3.8% for pregnancy complications or labor. Compared with patients who left AMA once, patients who left AMA multiple times had higher rates of heavy alcohol use (53.9% vs 30.9%; P = 0.01) and illicit drug use (88.5% vs 53.7%; P < 0.001) (Table 2). In multivariate analyses, the increased odds of leaving AMA more than once persisted for current heavy illicit drug users compared with patients who had never engaged in illicit drug use.

Discharge Characteristics and Documentation

Providers documented a patient’s plan to leave AMA before actual discharge 17.3% of the time. The documented plan to leave had to indicate that the patient was actually considering leaving. For example, “Patient is eager to go home” was not enough to qualify as a plan, but “Patient is thinking of leaving” qualified. For 84.3% of AMA discharges, the hospital’s AMA form was signed and was included in the medical record. Documentation showed that medications were prescribed for AMA patients 21.4% of the time, follow-up was arranged 25.7% of the time, and follow-up was pending arrangement 14.8% of the time. The majority of AMA patients (71.4%) left during daytime hours. In 29.6% of AMA discharges, providers documented AMA patients had decision-making capacity.

Readmission After AMA Discharge

Of the 268 AMA patients, 67.7% were not readmitted within the 6 months after AMA, 24.5% had 1 or 2 readmissions, and the rest had 3 or more readmissions (1 patient had 15). In addition, 35.8% returned to the emergency department within 30 days, and 16.4% were readmitted within 30 days. In 2014, the hospital’s overall 30-day readmission rate was 10.8%. Of the patients readmitted within 6 months after AMA, 23.5% left AMA again at the next visit, 9.4% left AWOL, and 67.1% were discharged conventionally.

Drivers of Premature Discharge

Qualitative analysis of the 35.5% of patient charts documenting a reason for leaving the hospital revealed 3 broad, interrelated themes (Figure 1). The first theme, dissatisfaction with hospital care, included chart notations such as “His wife couldn’t sleep in the hospital room” and “Not satisfied with all-liquid diet.” The second theme, urgent personal issues, included comments such as “He has a very important court date for his children” and “He needed to take care of immigration forms.” The third theme, mental health and substance abuse issues, included notations such as “He wants to go smoke” and “Severe anxiety and prison flashbacks.”

Provider Self-Assessment and Beliefs 

The survey was completed by 178 healthcare providers: 49.4% registered nurses, 19.1% trainee physicians, 20.8% attending physicians, and 10.7% other providers, including chaplains, social workers, and clerks. Regarding self-assessment of competency in AMA discharges, 94% of providers agreed they were comfortable assessing capacity, and 94% agreed they were comfortable talking with patients about the risks of leaving AMA (Figure 2). Nurses were more likely than trainee physicians to agree they knew what to do for patients who lacked capacity (74% vs 49%; P = 0.02). Most providers (70%) agreed they usually knew why their patients were leaving AMA; in this self-assessment, there were no significant differences between types of providers.

Regarding follow-up, attending physicians and trainee physicians demonstrated more agreement than nurses that AMA patients should receive medications and follow-up (94% and 84% vs 64%; P < 0.05). Nurses were more likely than attending physicians to say patients should lose their rights to hospital follow-up because of leaving AMA (38% vs 6%; P < 0.01). A minority of providers (37%) agreed transportation should be arranged. Addiction was the most common driver of AMA discharge (35%),  followed by familial obligations (19%), dissatisfaction with hospital care (16%), and financial concerns (15%).

The demographic characteristics of AMA patients in this study are similar to those identified in other studies, showing overrepresentation of young male patients.^12,14 Homeless patients were also overrepresented in the AMA discharge population at Highland Hospital—a finding that has not been consistently reported in prior studies, and that warrants further examination. In adjusted analyses, Spanish speakers had a lower rate of AMA discharge, and there were no racial variations. This is consistent with another study’s finding: that racial disparities in AMA discharge rates were largely attributable to confounders.²⁴ Language differences may result from failure of staff to fully explain the option of AMA discharge to non-English speakers, or from fear of immigration consequences after AMA discharge. Further investigation of patient experiences is needed to identify factors that contribute to demographic variations in AMA discharge rates.^25,26

Of the patients who left AMA multiple times, nearly all were actively using illicit drugs. In a recent study conducted at a safety-net hospital in Vancouver, Canada, 43% of patients with illicit drug use and at least 1 hospitalization left AMA at least once during the 6-year study period.¹¹ Many factors might explain this correlation—addiction itself, poor pain control for patients with addiction issues, fears about incarceration, and poor treatment of drug users by healthcare staff.¹⁵ Although the medical literature highlights deficits in pain control for patients addicted to opiates, proposed solutions are sparse and focus on perioperative pain control and physician prescribing practices.^27,28 At safety-net hospitals in which addiction is a factor in many hospitalizations, there is opportunity for new research in inpatient pain control for patients with substance dependence. In addition, harm reduction strategies—such as methadone maintenance for hospitalized patients with opiate dependence and abscess clinics as hospitalization alternatives for injection-associated infection treatment—may be key in improving safety for patients.^11,15,29

Comparing the provider survey and chart review results highlights discordance between provider beliefs and clinical practice. Healthcare providers at Highland Hospital considered themselves competent in assessing capacity and talking with patients about the risks of AMA discharge. In practice, however, capacity was documented in less than a third of AMA discharges. Although the majority of providers thought medications and follow-up should be arranged for patients, arrangements were seldom made. This may be partially attributable to limited resources for making these arrangements. Average time to “third next available” primary care appointment within the county health system that includes Highland was 44.6 days for established patients during the period of study; for new primary care patients, the average wait for an appointment was 2 to 3 months. Highland has a same-day clinic, but inpatient providers are discouraged from using it as a postdischarge clinic for patients who would be better served in primary care. Medications and transportation are easily arranged during daytime hours but are not immediately available at night. In addition, some of this discrepancy may be attributable to the limited documentation rather than to provider failure to achieve their own benchmarks of quality care for AMA patients. 

Documentation in AMA discharges is key for multiple reasons. Most AMA patients in this study signed an AMA form, and it could be that the rate of documenting decision-making capacity was low because providers thought a signed AMA form was adequate documentation of capacity and informed consent. In numerous court cases, however, these forms were found to be insufficient evidence of informed consent (lacking other supportive documentation) and possibly to go against the public good.³⁰ In addition, high rates of repeat emergency department visits and readmissions for AMA patients, demonstrated here and in other studies, highlight the importance of careful documentation in informing subsequent providers about hospital returnees’ ongoing issues.^17-19

This study also demonstrated differences between nurses and physicians in their beliefs about arranging follow-up for AMA patients. Nurses were less likely than physicians to think follow-up arrangements should be made for AMA patients and more likely to say these patients should lose the right to follow-up because of the AMA discharge. For conventional discharges, nurses provide patients with significantly more discharge education than interns or hospitalists do.³¹ This discrepancy highlights an urgent need for the education and involvement of nurses as stakeholders in the challenging AMA discharge process. Although the percentage of physicians who thought they were not obligated to provide medications and arrange follow-up for AMA patients was lower than the percentage of nurses, these beliefs contradict best practice guidelines for AMA discharges,^22,23 and this finding calls attention to the need for interventions to improve adherence to professional and ethical guidelines in this aspect of clinical practice. 

Providers showed a lack of familiarity with practice guidelines regarding certain aspects of the AMA discharge process. For example, most providers thought they should not have to arrange transportation for AMA patients, even though both the California Hospital Association Guidelines and the Highland Hospital internal policy on AMA discharges recommend arranging appropriate transportation.³² This finding suggests a need for educational interventions to ensure providers are informed about state and hospital policies, and a need to include both physicians and nurses in policymaking so theory can be tied to practice.

This study was limited to a single center with healthcare provider and patient populations that might not be generalizable to other settings. In the retrospective chart review, the authors were limited to information documented in the medical record, which might not accurately reflect the AMA discharge process. As they surveyed a limited number of social workers, case managers, and others who play an important role in the AMA discharge process, their data may lack varying viewpoints.

Overall, these data suggest providers at this county hospital generally agreed in principle with the best practice guidelines proposed by bioethicists for AMA discharges. In practice, however, providers were not reliably following these guidelines. Future interventions—including provider education on best practice guidelines for AMA discharge, provider involvement in policymaking, supportive templates for guiding documentation of AMA discharges, and improving access to follow-up care—will be key in improving the safety and health outcomes of AMA patients.

Acknowledgments

The authors thank Kelly Aguilar, Kethia Chheng, Irene Yen, and the Research Advancement and Coordination Initiative at Alameda Health System for important contributions to this project.

Disclosures

Highland Hospital Department of Medicine internal grant 2015.23 helped fund this research. A portion of the data was presented as a poster at the University of California San Francisco Health Disparities Symposium; October 2015; San Francisco, CA. Two posters from the data were presented at Hospital Medicine 2016, March 2016; San Diego, CA.

Patients leave the hospital against medical advice (AMA) for a variety of reasons. The AMA rate is approximately 1% nationally but substantially higher at safety-net hospitals and has rapidly increased over the past decade.^1-5 The principle that patients have the right to make choices about their healthcare, up to and including whether to leave the hospital against the advice of medical staff, is well-established law and a foundation of medical ethics.⁶ In practice, however, AMA discharges are often emotionally charged for both patients and providers, and, in the high-stress setting of AMA discharge, providers may be confused about their roles.^7-9

The demographics of patients who leave AMA have been well described. Compared with conventionally discharged patients, AMA patients are younger, more likely to be male, and more likely a marginalized ethnic or racial minority.^10-14 Patients with mental illnesses and addiction issues are overrepresented in AMA discharges, and complicated capacity assessments and limited resources may strain providers.^7,8,15,16 Studies have repeatedly shown higher rates of readmission and mortality for AMA patients than for conventionally discharged patients.^17-21 Whether AMA discharge is a marker for other prognostic factors that bode poorly for patients or contributes to negative outcomes, data suggest this group of patients is vulnerable, having mortality rates up to 40% higher 1 year after discharge, relative to conventionally discharged patients.¹²

Several models of standardized best practice approaches for AMA have been proposed by bioethicists.^6,22,23 Although details of these approaches vary, all involve assessing the patient’s decision-making capacity, clarifying the risks of AMA discharge, addressing factors that might be prompting the discharge, formulating an alternative outpatient treatment plan or “next best” option, and documenting extensively. A recent study found patients often gave advance warning of an AMA discharge, but physicians rarely prepared by arranging follow-up care.⁸ The investigators hypothesized that providers might not have known what they were permitted to arrange for AMA patients, or might have thought that providing “second best” options went against their principles. The investigators noted that nurses might have become aware of AMA risk sooner than physicians did but could not act on this awareness by preparing medications and arranging follow-up.

Translating models of best practice care for AMA patients into clinical practice requires buy-in from bedside providers, not just bioethicists. Given the study findings that providers have misconceptions about their roles in the AMA discharge,⁷ it is prudent to investigate providers’ current practices, beliefs, and concerns about AMA discharges before introducing a new approach. 

The present authors conducted a mixed-methods cross-sectional study of the state of AMA discharges at Highland Hospital (Oakland, California), a 236-bed county hospital and trauma center serving a primarily underserved urban patient population. The aim of this study was to assess current provider practices for AMA discharges and provider perceptions and knowledge about AMA discharges, ultimately to help direct future educational interventions with medical providers or hospital policy changes needed to improve the quality of AMA discharges. 

Phase 1 of this study involved identifying AMA patients through a review of data from Highland Hospital’s electronic medical records for 2014. These data included discharge status (eg, AMA vs other discharge types). The hospital’s floor clerk distinguishes between absent without official leave (AWOL; the patient leaves without notifying a provider) and AMA discharge. Discharges designated AWOL were excluded from the analyses. 

In phase 2, a structured chart review (Appendix A) was performed for all patients identified during phase 1 as being discharged AMA in 2014. In these reviews, further assessment was made of patient and visit characteristics in hospitalizations that ended in AMA discharge, and of providers’ documentation of AMA discharges—that is, whether several factors were documented (capacity; predischarge indication that patient might leave AMA; reason for AMA; and indications that discharge medications, transportation, and follow-up were arranged). These visit factors were reviewed because the literature has identified them as being important markers for AMA discharge safety.^6,8 Two research assistants, under the guidance of Dr. Stearns, reviewed the charts. To ensure agreement across chart reviews with respect to subjective questions (eg, whether capacity was adequately documented), the group reviewed the first 10 consecutive charts together; there was full agreement on how to classify the data of interest. Throughout the study, whenever a research assistant asked how to classify particular patient data, Dr. Stearns reviewed the data, and the research team made a decision together. Additional data, for AMA patients and for all patients admitted to Highland Hospital, were obtained from the hospital’s data warehouse, which pools data from within the health system.

Phase 3 involved surveying healthcare providers who were involved in patient care on the internal medicine and trauma surgery services at the hospital. These providers were selected because chart review revealed that the vast majority of patients who left AMA in 2014 were on one of these services. Surveys (Appendix B) asked participant providers to identify their role at the hospital, to provide a self-assessment of competence in various aspects of AMA discharge, to voice opinions about provider responsibilities in arranging follow-up for AMA patients, and to make suggestions about the AMA process. The authors designed these surveys, which included questions about aspects of care that have been highlighted in the AMA discharge literature as being important for AMA discharge safety.^6,8,22,23 Surveys were distributed to providers at internal medicine and trauma surgery department meetings and nursing conferences. Data (without identifying information) were analyzed, and survey responses kept anonymous.

The Alameda Health System Institutional Review Board approved this project. Providers were given the option of writing their name and contact information at the top of the survey in order to be entered into a drawing to receive a prize for completion.

We performed statistical analyses of the patient charts and physician survey data using Stata (version 14.0, Stata Corp., College Station, Texas). We analyzed both patient- and encounter-level data. In demographic analyses, this approach prevented duplicate counting of patients who left AMA multiple times. Patient-level analyses compared the demographic characteristics of AMA patients and patients discharged conventionally from the hospital in 2014. In addition, patients with either 1 or multiple AMA discharges were compared to identify characteristics that might be linked to highest risk of recurrent AMA discharge in the hope that early identification of these patients might facilitate providers’ early awareness and preparation for follow-up care or hospitalization alternatives. We used ANOVAs for continuous variables and tests of proportions for categorical variables. On the encounter level, analyses examined data about each admission (eg, AMA forms signed, follow-up arrangements made, capacity documented, etc.) for all AMA discharges. We employed chi square tests to identify variations in healthcare provider survey responses. A P value < 0.05 was used as the significance cut-off point.

Staged logistic regression analyses, adjusted for demographic characteristics, were performed to assess the association between risk of leaving AMA (yes or no) and demographic characteristics and the association between risk of leaving AMA more than once (yes or no) and health-related characteristics.

Demographic, Clinical, and Utilization Characteristics

Of the 12,036 Highland Hospital admissions in 2014, 319 (2.7%) ended with an AMA discharge. Of the 8207 individual patients discharged, 268 left AMA once, and 29 left AMA multiple times. Further review of the Admissions, Discharges, and Transfers Report generated from the electronic medical record revealed that 15 AWOL discharges were misclassified as AMA discharges.

Compared with patients discharged conventionally, AMA patients were significantly younger; more likely to be male, to self-identify as Black/African American, and to be English-speaking; and less likely to self-identify as Asian/Pacific Islander or Hispanic/Latino or to be Chinese- or Spanish-speaking (Table 1). They were also more likely than all patients admitted to Highland to be homeless (15.7% vs 8.7%; P < 0.01). Multivariate regression analysis revealed persistent age and sex disparities, but racial disparities were mitigated in adjusted analyses (Appendix C). Language disparities persisted only for Spanish speakers, who had a significantly lower rate of AMA discharge, even in adjusted analyses. 

The majority of AMA patients were on the internal medicine service (63.5%) or the trauma surgery service (24.8%). Regarding admission diagnosis, 17.2% of AMA patients were admitted for infections, 5.0% for drug or alcohol intoxication or withdrawal, 38.9% for acute noninfectious illnesses, 16.7% for decompensation of chronic disease, 18.4% for injuries or trauma, and 3.8% for pregnancy complications or labor. Compared with patients who left AMA once, patients who left AMA multiple times had higher rates of heavy alcohol use (53.9% vs 30.9%; P = 0.01) and illicit drug use (88.5% vs 53.7%; P < 0.001) (Table 2). In multivariate analyses, the increased odds of leaving AMA more than once persisted for current heavy illicit drug users compared with patients who had never engaged in illicit drug use.

Discharge Characteristics and Documentation

Providers documented a patient’s plan to leave AMA before actual discharge 17.3% of the time. The documented plan to leave had to indicate that the patient was actually considering leaving. For example, “Patient is eager to go home” was not enough to qualify as a plan, but “Patient is thinking of leaving” qualified. For 84.3% of AMA discharges, the hospital’s AMA form was signed and was included in the medical record. Documentation showed that medications were prescribed for AMA patients 21.4% of the time, follow-up was arranged 25.7% of the time, and follow-up was pending arrangement 14.8% of the time. The majority of AMA patients (71.4%) left during daytime hours. In 29.6% of AMA discharges, providers documented AMA patients had decision-making capacity.

Readmission After AMA Discharge

Of the 268 AMA patients, 67.7% were not readmitted within the 6 months after AMA, 24.5% had 1 or 2 readmissions, and the rest had 3 or more readmissions (1 patient had 15). In addition, 35.8% returned to the emergency department within 30 days, and 16.4% were readmitted within 30 days. In 2014, the hospital’s overall 30-day readmission rate was 10.8%. Of the patients readmitted within 6 months after AMA, 23.5% left AMA again at the next visit, 9.4% left AWOL, and 67.1% were discharged conventionally.

Drivers of Premature Discharge

Qualitative analysis of the 35.5% of patient charts documenting a reason for leaving the hospital revealed 3 broad, interrelated themes (Figure 1). The first theme, dissatisfaction with hospital care, included chart notations such as “His wife couldn’t sleep in the hospital room” and “Not satisfied with all-liquid diet.” The second theme, urgent personal issues, included comments such as “He has a very important court date for his children” and “He needed to take care of immigration forms.” The third theme, mental health and substance abuse issues, included notations such as “He wants to go smoke” and “Severe anxiety and prison flashbacks.”

Provider Self-Assessment and Beliefs 

The survey was completed by 178 healthcare providers: 49.4% registered nurses, 19.1% trainee physicians, 20.8% attending physicians, and 10.7% other providers, including chaplains, social workers, and clerks. Regarding self-assessment of competency in AMA discharges, 94% of providers agreed they were comfortable assessing capacity, and 94% agreed they were comfortable talking with patients about the risks of leaving AMA (Figure 2). Nurses were more likely than trainee physicians to agree they knew what to do for patients who lacked capacity (74% vs 49%; P = 0.02). Most providers (70%) agreed they usually knew why their patients were leaving AMA; in this self-assessment, there were no significant differences between types of providers.

Regarding follow-up, attending physicians and trainee physicians demonstrated more agreement than nurses that AMA patients should receive medications and follow-up (94% and 84% vs 64%; P < 0.05). Nurses were more likely than attending physicians to say patients should lose their rights to hospital follow-up because of leaving AMA (38% vs 6%; P < 0.01). A minority of providers (37%) agreed transportation should be arranged. Addiction was the most common driver of AMA discharge (35%),  followed by familial obligations (19%), dissatisfaction with hospital care (16%), and financial concerns (15%).

The demographic characteristics of AMA patients in this study are similar to those identified in other studies, showing overrepresentation of young male patients.^12,14 Homeless patients were also overrepresented in the AMA discharge population at Highland Hospital—a finding that has not been consistently reported in prior studies, and that warrants further examination. In adjusted analyses, Spanish speakers had a lower rate of AMA discharge, and there were no racial variations. This is consistent with another study’s finding: that racial disparities in AMA discharge rates were largely attributable to confounders.²⁴ Language differences may result from failure of staff to fully explain the option of AMA discharge to non-English speakers, or from fear of immigration consequences after AMA discharge. Further investigation of patient experiences is needed to identify factors that contribute to demographic variations in AMA discharge rates.^25,26

Of the patients who left AMA multiple times, nearly all were actively using illicit drugs. In a recent study conducted at a safety-net hospital in Vancouver, Canada, 43% of patients with illicit drug use and at least 1 hospitalization left AMA at least once during the 6-year study period.¹¹ Many factors might explain this correlation—addiction itself, poor pain control for patients with addiction issues, fears about incarceration, and poor treatment of drug users by healthcare staff.¹⁵ Although the medical literature highlights deficits in pain control for patients addicted to opiates, proposed solutions are sparse and focus on perioperative pain control and physician prescribing practices.^27,28 At safety-net hospitals in which addiction is a factor in many hospitalizations, there is opportunity for new research in inpatient pain control for patients with substance dependence. In addition, harm reduction strategies—such as methadone maintenance for hospitalized patients with opiate dependence and abscess clinics as hospitalization alternatives for injection-associated infection treatment—may be key in improving safety for patients.^11,15,29

Comparing the provider survey and chart review results highlights discordance between provider beliefs and clinical practice. Healthcare providers at Highland Hospital considered themselves competent in assessing capacity and talking with patients about the risks of AMA discharge. In practice, however, capacity was documented in less than a third of AMA discharges. Although the majority of providers thought medications and follow-up should be arranged for patients, arrangements were seldom made. This may be partially attributable to limited resources for making these arrangements. Average time to “third next available” primary care appointment within the county health system that includes Highland was 44.6 days for established patients during the period of study; for new primary care patients, the average wait for an appointment was 2 to 3 months. Highland has a same-day clinic, but inpatient providers are discouraged from using it as a postdischarge clinic for patients who would be better served in primary care. Medications and transportation are easily arranged during daytime hours but are not immediately available at night. In addition, some of this discrepancy may be attributable to the limited documentation rather than to provider failure to achieve their own benchmarks of quality care for AMA patients. 

Documentation in AMA discharges is key for multiple reasons. Most AMA patients in this study signed an AMA form, and it could be that the rate of documenting decision-making capacity was low because providers thought a signed AMA form was adequate documentation of capacity and informed consent. In numerous court cases, however, these forms were found to be insufficient evidence of informed consent (lacking other supportive documentation) and possibly to go against the public good.³⁰ In addition, high rates of repeat emergency department visits and readmissions for AMA patients, demonstrated here and in other studies, highlight the importance of careful documentation in informing subsequent providers about hospital returnees’ ongoing issues.^17-19

This study also demonstrated differences between nurses and physicians in their beliefs about arranging follow-up for AMA patients. Nurses were less likely than physicians to think follow-up arrangements should be made for AMA patients and more likely to say these patients should lose the right to follow-up because of the AMA discharge. For conventional discharges, nurses provide patients with significantly more discharge education than interns or hospitalists do.³¹ This discrepancy highlights an urgent need for the education and involvement of nurses as stakeholders in the challenging AMA discharge process. Although the percentage of physicians who thought they were not obligated to provide medications and arrange follow-up for AMA patients was lower than the percentage of nurses, these beliefs contradict best practice guidelines for AMA discharges,^22,23 and this finding calls attention to the need for interventions to improve adherence to professional and ethical guidelines in this aspect of clinical practice. 

Providers showed a lack of familiarity with practice guidelines regarding certain aspects of the AMA discharge process. For example, most providers thought they should not have to arrange transportation for AMA patients, even though both the California Hospital Association Guidelines and the Highland Hospital internal policy on AMA discharges recommend arranging appropriate transportation.³² This finding suggests a need for educational interventions to ensure providers are informed about state and hospital policies, and a need to include both physicians and nurses in policymaking so theory can be tied to practice.

This study was limited to a single center with healthcare provider and patient populations that might not be generalizable to other settings. In the retrospective chart review, the authors were limited to information documented in the medical record, which might not accurately reflect the AMA discharge process. As they surveyed a limited number of social workers, case managers, and others who play an important role in the AMA discharge process, their data may lack varying viewpoints.

Overall, these data suggest providers at this county hospital generally agreed in principle with the best practice guidelines proposed by bioethicists for AMA discharges. In practice, however, providers were not reliably following these guidelines. Future interventions—including provider education on best practice guidelines for AMA discharge, provider involvement in policymaking, supportive templates for guiding documentation of AMA discharges, and improving access to follow-up care—will be key in improving the safety and health outcomes of AMA patients.

Acknowledgments

The authors thank Kelly Aguilar, Kethia Chheng, Irene Yen, and the Research Advancement and Coordination Initiative at Alameda Health System for important contributions to this project.

Disclosures

Highland Hospital Department of Medicine internal grant 2015.23 helped fund this research. A portion of the data was presented as a poster at the University of California San Francisco Health Disparities Symposium; October 2015; San Francisco, CA. Two posters from the data were presented at Hospital Medicine 2016, March 2016; San Diego, CA.

Patients leave the hospital against medical advice (AMA) for a variety of reasons. The AMA rate is approximately 1% nationally but substantially higher at safety-net hospitals and has rapidly increased over the past decade.^1-5 The principle that patients have the right to make choices about their healthcare, up to and including whether to leave the hospital against the advice of medical staff, is well-established law and a foundation of medical ethics.⁶ In practice, however, AMA discharges are often emotionally charged for both patients and providers, and, in the high-stress setting of AMA discharge, providers may be confused about their roles.^7-9

The demographics of patients who leave AMA have been well described. Compared with conventionally discharged patients, AMA patients are younger, more likely to be male, and more likely a marginalized ethnic or racial minority.^10-14 Patients with mental illnesses and addiction issues are overrepresented in AMA discharges, and complicated capacity assessments and limited resources may strain providers.^7,8,15,16 Studies have repeatedly shown higher rates of readmission and mortality for AMA patients than for conventionally discharged patients.^17-21 Whether AMA discharge is a marker for other prognostic factors that bode poorly for patients or contributes to negative outcomes, data suggest this group of patients is vulnerable, having mortality rates up to 40% higher 1 year after discharge, relative to conventionally discharged patients.¹²

Several models of standardized best practice approaches for AMA have been proposed by bioethicists.^6,22,23 Although details of these approaches vary, all involve assessing the patient’s decision-making capacity, clarifying the risks of AMA discharge, addressing factors that might be prompting the discharge, formulating an alternative outpatient treatment plan or “next best” option, and documenting extensively. A recent study found patients often gave advance warning of an AMA discharge, but physicians rarely prepared by arranging follow-up care.⁸ The investigators hypothesized that providers might not have known what they were permitted to arrange for AMA patients, or might have thought that providing “second best” options went against their principles. The investigators noted that nurses might have become aware of AMA risk sooner than physicians did but could not act on this awareness by preparing medications and arranging follow-up.

Translating models of best practice care for AMA patients into clinical practice requires buy-in from bedside providers, not just bioethicists. Given the study findings that providers have misconceptions about their roles in the AMA discharge,⁷ it is prudent to investigate providers’ current practices, beliefs, and concerns about AMA discharges before introducing a new approach. 

The present authors conducted a mixed-methods cross-sectional study of the state of AMA discharges at Highland Hospital (Oakland, California), a 236-bed county hospital and trauma center serving a primarily underserved urban patient population. The aim of this study was to assess current provider practices for AMA discharges and provider perceptions and knowledge about AMA discharges, ultimately to help direct future educational interventions with medical providers or hospital policy changes needed to improve the quality of AMA discharges. 

Phase 1 of this study involved identifying AMA patients through a review of data from Highland Hospital’s electronic medical records for 2014. These data included discharge status (eg, AMA vs other discharge types). The hospital’s floor clerk distinguishes between absent without official leave (AWOL; the patient leaves without notifying a provider) and AMA discharge. Discharges designated AWOL were excluded from the analyses. 

In phase 2, a structured chart review (Appendix A) was performed for all patients identified during phase 1 as being discharged AMA in 2014. In these reviews, further assessment was made of patient and visit characteristics in hospitalizations that ended in AMA discharge, and of providers’ documentation of AMA discharges—that is, whether several factors were documented (capacity; predischarge indication that patient might leave AMA; reason for AMA; and indications that discharge medications, transportation, and follow-up were arranged). These visit factors were reviewed because the literature has identified them as being important markers for AMA discharge safety.^6,8 Two research assistants, under the guidance of Dr. Stearns, reviewed the charts. To ensure agreement across chart reviews with respect to subjective questions (eg, whether capacity was adequately documented), the group reviewed the first 10 consecutive charts together; there was full agreement on how to classify the data of interest. Throughout the study, whenever a research assistant asked how to classify particular patient data, Dr. Stearns reviewed the data, and the research team made a decision together. Additional data, for AMA patients and for all patients admitted to Highland Hospital, were obtained from the hospital’s data warehouse, which pools data from within the health system.

Phase 3 involved surveying healthcare providers who were involved in patient care on the internal medicine and trauma surgery services at the hospital. These providers were selected because chart review revealed that the vast majority of patients who left AMA in 2014 were on one of these services. Surveys (Appendix B) asked participant providers to identify their role at the hospital, to provide a self-assessment of competence in various aspects of AMA discharge, to voice opinions about provider responsibilities in arranging follow-up for AMA patients, and to make suggestions about the AMA process. The authors designed these surveys, which included questions about aspects of care that have been highlighted in the AMA discharge literature as being important for AMA discharge safety.^6,8,22,23 Surveys were distributed to providers at internal medicine and trauma surgery department meetings and nursing conferences. Data (without identifying information) were analyzed, and survey responses kept anonymous.

The Alameda Health System Institutional Review Board approved this project. Providers were given the option of writing their name and contact information at the top of the survey in order to be entered into a drawing to receive a prize for completion.

We performed statistical analyses of the patient charts and physician survey data using Stata (version 14.0, Stata Corp., College Station, Texas). We analyzed both patient- and encounter-level data. In demographic analyses, this approach prevented duplicate counting of patients who left AMA multiple times. Patient-level analyses compared the demographic characteristics of AMA patients and patients discharged conventionally from the hospital in 2014. In addition, patients with either 1 or multiple AMA discharges were compared to identify characteristics that might be linked to highest risk of recurrent AMA discharge in the hope that early identification of these patients might facilitate providers’ early awareness and preparation for follow-up care or hospitalization alternatives. We used ANOVAs for continuous variables and tests of proportions for categorical variables. On the encounter level, analyses examined data about each admission (eg, AMA forms signed, follow-up arrangements made, capacity documented, etc.) for all AMA discharges. We employed chi square tests to identify variations in healthcare provider survey responses. A P value < 0.05 was used as the significance cut-off point.

Staged logistic regression analyses, adjusted for demographic characteristics, were performed to assess the association between risk of leaving AMA (yes or no) and demographic characteristics and the association between risk of leaving AMA more than once (yes or no) and health-related characteristics.

Demographic, Clinical, and Utilization Characteristics

Of the 12,036 Highland Hospital admissions in 2014, 319 (2.7%) ended with an AMA discharge. Of the 8207 individual patients discharged, 268 left AMA once, and 29 left AMA multiple times. Further review of the Admissions, Discharges, and Transfers Report generated from the electronic medical record revealed that 15 AWOL discharges were misclassified as AMA discharges.

Compared with patients discharged conventionally, AMA patients were significantly younger; more likely to be male, to self-identify as Black/African American, and to be English-speaking; and less likely to self-identify as Asian/Pacific Islander or Hispanic/Latino or to be Chinese- or Spanish-speaking (Table 1). They were also more likely than all patients admitted to Highland to be homeless (15.7% vs 8.7%; P < 0.01). Multivariate regression analysis revealed persistent age and sex disparities, but racial disparities were mitigated in adjusted analyses (Appendix C). Language disparities persisted only for Spanish speakers, who had a significantly lower rate of AMA discharge, even in adjusted analyses. 

The majority of AMA patients were on the internal medicine service (63.5%) or the trauma surgery service (24.8%). Regarding admission diagnosis, 17.2% of AMA patients were admitted for infections, 5.0% for drug or alcohol intoxication or withdrawal, 38.9% for acute noninfectious illnesses, 16.7% for decompensation of chronic disease, 18.4% for injuries or trauma, and 3.8% for pregnancy complications or labor. Compared with patients who left AMA once, patients who left AMA multiple times had higher rates of heavy alcohol use (53.9% vs 30.9%; P = 0.01) and illicit drug use (88.5% vs 53.7%; P < 0.001) (Table 2). In multivariate analyses, the increased odds of leaving AMA more than once persisted for current heavy illicit drug users compared with patients who had never engaged in illicit drug use.

Discharge Characteristics and Documentation

Providers documented a patient’s plan to leave AMA before actual discharge 17.3% of the time. The documented plan to leave had to indicate that the patient was actually considering leaving. For example, “Patient is eager to go home” was not enough to qualify as a plan, but “Patient is thinking of leaving” qualified. For 84.3% of AMA discharges, the hospital’s AMA form was signed and was included in the medical record. Documentation showed that medications were prescribed for AMA patients 21.4% of the time, follow-up was arranged 25.7% of the time, and follow-up was pending arrangement 14.8% of the time. The majority of AMA patients (71.4%) left during daytime hours. In 29.6% of AMA discharges, providers documented AMA patients had decision-making capacity.

Readmission After AMA Discharge

Of the 268 AMA patients, 67.7% were not readmitted within the 6 months after AMA, 24.5% had 1 or 2 readmissions, and the rest had 3 or more readmissions (1 patient had 15). In addition, 35.8% returned to the emergency department within 30 days, and 16.4% were readmitted within 30 days. In 2014, the hospital’s overall 30-day readmission rate was 10.8%. Of the patients readmitted within 6 months after AMA, 23.5% left AMA again at the next visit, 9.4% left AWOL, and 67.1% were discharged conventionally.

Drivers of Premature Discharge

Qualitative analysis of the 35.5% of patient charts documenting a reason for leaving the hospital revealed 3 broad, interrelated themes (Figure 1). The first theme, dissatisfaction with hospital care, included chart notations such as “His wife couldn’t sleep in the hospital room” and “Not satisfied with all-liquid diet.” The second theme, urgent personal issues, included comments such as “He has a very important court date for his children” and “He needed to take care of immigration forms.” The third theme, mental health and substance abuse issues, included notations such as “He wants to go smoke” and “Severe anxiety and prison flashbacks.”

Provider Self-Assessment and Beliefs 

The survey was completed by 178 healthcare providers: 49.4% registered nurses, 19.1% trainee physicians, 20.8% attending physicians, and 10.7% other providers, including chaplains, social workers, and clerks. Regarding self-assessment of competency in AMA discharges, 94% of providers agreed they were comfortable assessing capacity, and 94% agreed they were comfortable talking with patients about the risks of leaving AMA (Figure 2). Nurses were more likely than trainee physicians to agree they knew what to do for patients who lacked capacity (74% vs 49%; P = 0.02). Most providers (70%) agreed they usually knew why their patients were leaving AMA; in this self-assessment, there were no significant differences between types of providers.

Regarding follow-up, attending physicians and trainee physicians demonstrated more agreement than nurses that AMA patients should receive medications and follow-up (94% and 84% vs 64%; P < 0.05). Nurses were more likely than attending physicians to say patients should lose their rights to hospital follow-up because of leaving AMA (38% vs 6%; P < 0.01). A minority of providers (37%) agreed transportation should be arranged. Addiction was the most common driver of AMA discharge (35%),  followed by familial obligations (19%), dissatisfaction with hospital care (16%), and financial concerns (15%).

The demographic characteristics of AMA patients in this study are similar to those identified in other studies, showing overrepresentation of young male patients.^12,14 Homeless patients were also overrepresented in the AMA discharge population at Highland Hospital—a finding that has not been consistently reported in prior studies, and that warrants further examination. In adjusted analyses, Spanish speakers had a lower rate of AMA discharge, and there were no racial variations. This is consistent with another study’s finding: that racial disparities in AMA discharge rates were largely attributable to confounders.²⁴ Language differences may result from failure of staff to fully explain the option of AMA discharge to non-English speakers, or from fear of immigration consequences after AMA discharge. Further investigation of patient experiences is needed to identify factors that contribute to demographic variations in AMA discharge rates.^25,26

Of the patients who left AMA multiple times, nearly all were actively using illicit drugs. In a recent study conducted at a safety-net hospital in Vancouver, Canada, 43% of patients with illicit drug use and at least 1 hospitalization left AMA at least once during the 6-year study period.¹¹ Many factors might explain this correlation—addiction itself, poor pain control for patients with addiction issues, fears about incarceration, and poor treatment of drug users by healthcare staff.¹⁵ Although the medical literature highlights deficits in pain control for patients addicted to opiates, proposed solutions are sparse and focus on perioperative pain control and physician prescribing practices.^27,28 At safety-net hospitals in which addiction is a factor in many hospitalizations, there is opportunity for new research in inpatient pain control for patients with substance dependence. In addition, harm reduction strategies—such as methadone maintenance for hospitalized patients with opiate dependence and abscess clinics as hospitalization alternatives for injection-associated infection treatment—may be key in improving safety for patients.^11,15,29

Comparing the provider survey and chart review results highlights discordance between provider beliefs and clinical practice. Healthcare providers at Highland Hospital considered themselves competent in assessing capacity and talking with patients about the risks of AMA discharge. In practice, however, capacity was documented in less than a third of AMA discharges. Although the majority of providers thought medications and follow-up should be arranged for patients, arrangements were seldom made. This may be partially attributable to limited resources for making these arrangements. Average time to “third next available” primary care appointment within the county health system that includes Highland was 44.6 days for established patients during the period of study; for new primary care patients, the average wait for an appointment was 2 to 3 months. Highland has a same-day clinic, but inpatient providers are discouraged from using it as a postdischarge clinic for patients who would be better served in primary care. Medications and transportation are easily arranged during daytime hours but are not immediately available at night. In addition, some of this discrepancy may be attributable to the limited documentation rather than to provider failure to achieve their own benchmarks of quality care for AMA patients. 

Documentation in AMA discharges is key for multiple reasons. Most AMA patients in this study signed an AMA form, and it could be that the rate of documenting decision-making capacity was low because providers thought a signed AMA form was adequate documentation of capacity and informed consent. In numerous court cases, however, these forms were found to be insufficient evidence of informed consent (lacking other supportive documentation) and possibly to go against the public good.³⁰ In addition, high rates of repeat emergency department visits and readmissions for AMA patients, demonstrated here and in other studies, highlight the importance of careful documentation in informing subsequent providers about hospital returnees’ ongoing issues.^17-19

This study also demonstrated differences between nurses and physicians in their beliefs about arranging follow-up for AMA patients. Nurses were less likely than physicians to think follow-up arrangements should be made for AMA patients and more likely to say these patients should lose the right to follow-up because of the AMA discharge. For conventional discharges, nurses provide patients with significantly more discharge education than interns or hospitalists do.³¹ This discrepancy highlights an urgent need for the education and involvement of nurses as stakeholders in the challenging AMA discharge process. Although the percentage of physicians who thought they were not obligated to provide medications and arrange follow-up for AMA patients was lower than the percentage of nurses, these beliefs contradict best practice guidelines for AMA discharges,^22,23 and this finding calls attention to the need for interventions to improve adherence to professional and ethical guidelines in this aspect of clinical practice. 

Providers showed a lack of familiarity with practice guidelines regarding certain aspects of the AMA discharge process. For example, most providers thought they should not have to arrange transportation for AMA patients, even though both the California Hospital Association Guidelines and the Highland Hospital internal policy on AMA discharges recommend arranging appropriate transportation.³² This finding suggests a need for educational interventions to ensure providers are informed about state and hospital policies, and a need to include both physicians and nurses in policymaking so theory can be tied to practice.

This study was limited to a single center with healthcare provider and patient populations that might not be generalizable to other settings. In the retrospective chart review, the authors were limited to information documented in the medical record, which might not accurately reflect the AMA discharge process. As they surveyed a limited number of social workers, case managers, and others who play an important role in the AMA discharge process, their data may lack varying viewpoints.

Overall, these data suggest providers at this county hospital generally agreed in principle with the best practice guidelines proposed by bioethicists for AMA discharges. In practice, however, providers were not reliably following these guidelines. Future interventions—including provider education on best practice guidelines for AMA discharge, provider involvement in policymaking, supportive templates for guiding documentation of AMA discharges, and improving access to follow-up care—will be key in improving the safety and health outcomes of AMA patients.

Acknowledgments

The authors thank Kelly Aguilar, Kethia Chheng, Irene Yen, and the Research Advancement and Coordination Initiative at Alameda Health System for important contributions to this project.

Disclosures

Highland Hospital Department of Medicine internal grant 2015.23 helped fund this research. A portion of the data was presented as a poster at the University of California San Francisco Health Disparities Symposium; October 2015; San Francisco, CA. Two posters from the data were presented at Hospital Medicine 2016, March 2016; San Diego, CA.