While a general consensus exists that surface wound cultures have less utility than deeper cultures, surface cultures are nevertheless routinely used to guide empiric antibiotic administration. This is due in part to the ease with which surface cultures are obtained and the delay in obtaining deeper wound and bone cultures. The Infectious Diseases Society of America (IDSA) recommends routine culture of all diabetic infections before initiating empiric antibiotic therapy, despite caveats regarding undebrided wounds.1 Further examination of 2 additional societies, the European Society of Clinical Microbiology and Infectious Diseases and the Australasian Society for Infectious Diseases, reveals that they also do not describe guidelines on the role of surface wound cultures in skin, and skin structure infection (SSSI) management.2, 3

Surface wound cultures are used to aid in diagnosis and appropriate antibiotic treatment of lower extremity foot ulcers.4 Contaminated cultures from other body locations have shown little utility and may be wasteful of resources.5, 6 We hypothesize that given commensal skin flora, coupled with the additional flora that colonizes (chronic) lower extremity wounds, surface wound cultures provide poor diagnostic accuracy for determining the etiology of acute infection. In contrast, many believe that deep tissue cultures obtained at time of debridement or surgical intervention may provide more relevant information to guide antibiotic therapy, thus serving as a gold standard.13, 7, 8 Nevertheless, with the ease of obtaining these superficial cultures and the promptness of the results, surface wound cultures are still used as a surrogate for the information derived from deeper cultures.

Purpose

The frequency at which superficial wound cultures correlate with the data obtained from deeper cultures is needed to interpret the posttest likelihood of infection. However, the sensitivity and specificity of superficial wound culture as a diagnostic test is unclear. The purpose of this study is to conduct a systematic review of the existing literature in order to investigate the relationship between superficial wound cultures and the etiology of SSSI. Accordingly, we aim to describe any role that surface wound cultures may play in the treatment of lower extremity ulcers.

Materials and Methods

Data Sources

We identified eligible articles through an electronic search of the following databases: Medline through PubMed, Excerpta Medica Database (EMBASE), Cumulative Index of Nursing and Allied Health Literature (CINAHL), and Scopus. We also hand searched the reference lists of key review articles identified by the electronic search and the reference lists of all eligible articles (Figure 1).

Figure 1

Results

Of 9032 unique citations, eight studies met all inclusion criteria (Figure 1). Inter‐rater reliability was substantial (Kappa = 0.78).12 Areas of initial disagreement generally involved whether a study adequately described an appropriate alternative culture method for comparison or whether data available in an article was sufficient for sensitivity and specificity calculation. Consensus was achieved once the full article was retrieved, reviewed and discussed.

The 8 studies evaluated in the review included a total number of 615 patients or samples (Table 1). Diabetic wounds were described in four studies.1316 Two studies described wounds associated with peripheral vascular disease,13, 17 while four involved traumatic wounds.13, 1719 One study did not identify the clinical circumstances concerning the wounds.20

The studies used several different methods for obtaining superficial cultures. Six studies obtained purulent wound drainage material through the application of sterile swabs.1316, 18, 19 One study obtained purulent drainage material using needle aspiration.18 Two studies obtained culture material from sinus tracts associated with the wounds, one through sinus tract washings17 and another by obtaining sinus tract discharge material.20

The types of comparison cultures used were equally divided between deep tissue biopsies1316 and bone biopsies,1720 each accounting for 50% (4 of 8) of studies.

In assessing the data from the eight studies, the pooled test sensitivity for superficial wound swabs was 49% (95% confidence interval [CI], 37‐61%) (Figure 2). The pooled specificity for superficial wound swabs was 62% (95% CI, 51‐74%), while the pooled positive and negative LRs were 1.1 (95% CI, 0.71‐1.5) and 0.67 (95% CI, 0.52‐0.82), respectively (Figure 2).

Figure 2

The median number of bacterial isolates reported for each culture type, superficial and comparison culture, was collected from each study (Table 2). The median value for number of bacterial isolates identified by superficial culture was 2.7 (interquartile range [IQR] 1.8‐3.2). The median value for number of bacterial isolates identified by comparison culture was 2.2 (IQR 1.7‐2.9). A Wilcoxon rank sum analysis showed that the number of isolates for surface wound cultures was not significantly different than the number of isolates for comparison cultures (P = 0.75) (Table 1).

Discussion

In performing this review, we discovered ambiguity in the literature regarding the utility of surface wound cultures. Some studies obtained findings to suggest the utility of surface wound cultures,8, 14, 17 while other studies in our review16, 18, 19 provided evidence against them. This variability confirmed the need for a meta‐analytic approach as provided by this review.

While we have tried to minimize bias through a well‐established methodology, we acknowledge that certain methodological limitations should be considered in interpreting the results. There may be publication bias in reviews that include only published articles; a funnel plot of sensitivity vs. sample size showed some asymmetry, suggesting bias. Our search strategy was limited to English‐only articles, which may result in publication bias.

Further, this review included a group of studies that were heterogeneous in several regards. Differences exist in culturing methods and laboratory technology, as exemplified by the variety of superficial culture methods used. We were not able to account for these laboratory differences, as methodologies in obtaining and isolating bacteria were not uniformly well described.

Additionally, the studies classified organisms in different ways. Three studies categorized organisms according to Gram's stain characteristics.13, 16, 18 One study described organisms primarily in terms of aerobic or anaerobic respiration.15 Two studies14, 19 discussed pathogens both in terms of respiration (aerobic/anaerobic) and Gram's stain characteristic, while another 2 studies17, 20 did not describe organisms in either terms. These inconsistencies limited our ability to provide sensitivity and specificity information for specific subclasses of organisms.

The clinical conditions in each study surrounding the wounds were also heterogeneous: most significantly in the issue of prior antibiotic administration. All but 1 study16 indicated that the patients had received antibiotics prior to having cultures obtained. The type of antibiotics (narrow‐spectrum or broad‐spectrum), the route of administration, and the cessation of antibiotics in relation to obtaining swabs and cultures all varied widely or were not well described. This degree of ambiguity will necessarily impact both the reliability of data regarding microbial growth as well as the component flora.

The inclusion of higher quality studies is likely to result in a more reliable meta‐analysis.21 We had hoped that antibiotic trials would contain uniform outcomes and thus strengthen our meta‐analysis through the inclusion of randomized‐controlled studies. Unfortunately, the majority of antibiotic trials did not use superficial wound cultures, did not report mean number of isolates, or did not provide microbiological data in sufficient detail to calculate concordance ratesand therefore, did not meet eligibility criteria. Randomized‐controlled trials were a minority among our included articles; the majority of study designs were retrospective cohorts and case‐controlled studies.

Despite these limitations, we were able to conclude that superficial wound culture provides mediocre sensitivity (49%) and specificity (62%). The positive LR of 1.1 is unhelpful in decision making, having a CI that includes 1. Interestingly, the negative LR of 0.67 could be somewhat helpful in medical decision making, modifying the pretest probability and assisting in ruling out a deeper bacterial infection. Although, according to Fagan's nomogram, a negative LR of 0.67 has only a mild effect on pretest odds.22

The bacterial bioburden assessed by the number of isolates obtained by culture method serves as a proxy for reliability of culture results14, 23 by suggesting that fewer organisms isolated from deep tissue or bone samples reflects a less contaminated specimen. Our assessment of the bioburden found that the median number of isolates was slightly higher in surface cultures than deeper cultures, though not to a significant degree (P = 0.75). This indicates that the degree of contamination in superficial cultures was neither significantly worse nor better than deep cultures.

We attempted to define a role for surface wound cultures; however, we found that these did not show any greater utility than deep cultures for identifying the microbiologic etiology of diabetic wound infections. While the negative LR provides some quantitative verification of the common clinical practice that a negative culture argues against infection, the finding is not especially robust.

Although for this meta‐analysis we grouped all organisms in the same way, we recognize that the sensitivity and specificity may differ according to various subclasses of bacteria. Interpretations of culture results also vary (eg, Gram positive vs. negative; aerobic vs. anaerobic); practitioners will not interpret superficial cultures of coagulase‐negative Staphylococcus in the same way as Pseudomonas. However, this study seeks to establish a reasonable starting point for the medical decision‐making process by providing quantitative values in an area with previously conflicting data. We anticipate that as laboratory techniques improve and research into superficial wounds continues, greater sensitivity of superficial wound cultures will result.

Ultimately, physicians use culture data to target therapy in an effort to use the least toxic and most effective antimicrobial agent possible to successfully treat infections. Clinical outcomes were not described in all included articles; in those that did, the endpoints were too dissimilar for meaningful comparison. Limiting our review to studies reporting treatment outcomes would have resulted in too few included studies. Thus, we were unable able to assess whether superficial wound cultures were associated with improved patient‐oriented outcomes in this meta‐analysis.

There is a significant paucity of trials evaluating the accurate concordance of superficial swabs to deep tissue culture. The current data shows poor sensitivity and specificity of superficial culture methods. The presumption that deeper cultures (such as a bone biopsy) should result in a less contaminated sample and more targeted culture results was also not borne out in our review. When presented with a patient with a wound infection, physicians mentally supply a pretest (or a pretreatment) probability as to the microbiologic etiology of the infection. Careful history will, of course, be critical in identifying extenuating circumstance or unusual exposures. From our meta‐analysis, we cannot recommend the routine use of superficial wound cultures to guide initial antibiotic therapy as this may result in poor resource utilization.5 While clinical outcomes from the use of routine superficial cultures are unclear, we suggest greater use of local antibiograms and methicillin‐resistant Staphylococcus aureus (MRSA) prevalence data to determine resistance patterns and guide the selection of empiric therapies.

While a general consensus exists that surface wound cultures have less utility than deeper cultures, surface cultures are nevertheless routinely used to guide empiric antibiotic administration. This is due in part to the ease with which surface cultures are obtained and the delay in obtaining deeper wound and bone cultures. The Infectious Diseases Society of America (IDSA) recommends routine culture of all diabetic infections before initiating empiric antibiotic therapy, despite caveats regarding undebrided wounds.1 Further examination of 2 additional societies, the European Society of Clinical Microbiology and Infectious Diseases and the Australasian Society for Infectious Diseases, reveals that they also do not describe guidelines on the role of surface wound cultures in skin, and skin structure infection (SSSI) management.2, 3

Surface wound cultures are used to aid in diagnosis and appropriate antibiotic treatment of lower extremity foot ulcers.4 Contaminated cultures from other body locations have shown little utility and may be wasteful of resources.5, 6 We hypothesize that given commensal skin flora, coupled with the additional flora that colonizes (chronic) lower extremity wounds, surface wound cultures provide poor diagnostic accuracy for determining the etiology of acute infection. In contrast, many believe that deep tissue cultures obtained at time of debridement or surgical intervention may provide more relevant information to guide antibiotic therapy, thus serving as a gold standard.13, 7, 8 Nevertheless, with the ease of obtaining these superficial cultures and the promptness of the results, surface wound cultures are still used as a surrogate for the information derived from deeper cultures.

Purpose

The frequency at which superficial wound cultures correlate with the data obtained from deeper cultures is needed to interpret the posttest likelihood of infection. However, the sensitivity and specificity of superficial wound culture as a diagnostic test is unclear. The purpose of this study is to conduct a systematic review of the existing literature in order to investigate the relationship between superficial wound cultures and the etiology of SSSI. Accordingly, we aim to describe any role that surface wound cultures may play in the treatment of lower extremity ulcers.

Materials and Methods

Data Sources

We identified eligible articles through an electronic search of the following databases: Medline through PubMed, Excerpta Medica Database (EMBASE), Cumulative Index of Nursing and Allied Health Literature (CINAHL), and Scopus. We also hand searched the reference lists of key review articles identified by the electronic search and the reference lists of all eligible articles (Figure 1).

Figure 1

Results

Of 9032 unique citations, eight studies met all inclusion criteria (Figure 1). Inter‐rater reliability was substantial (Kappa = 0.78).12 Areas of initial disagreement generally involved whether a study adequately described an appropriate alternative culture method for comparison or whether data available in an article was sufficient for sensitivity and specificity calculation. Consensus was achieved once the full article was retrieved, reviewed and discussed.

The 8 studies evaluated in the review included a total number of 615 patients or samples (Table 1). Diabetic wounds were described in four studies.1316 Two studies described wounds associated with peripheral vascular disease,13, 17 while four involved traumatic wounds.13, 1719 One study did not identify the clinical circumstances concerning the wounds.20

The studies used several different methods for obtaining superficial cultures. Six studies obtained purulent wound drainage material through the application of sterile swabs.1316, 18, 19 One study obtained purulent drainage material using needle aspiration.18 Two studies obtained culture material from sinus tracts associated with the wounds, one through sinus tract washings17 and another by obtaining sinus tract discharge material.20

The types of comparison cultures used were equally divided between deep tissue biopsies1316 and bone biopsies,1720 each accounting for 50% (4 of 8) of studies.

In assessing the data from the eight studies, the pooled test sensitivity for superficial wound swabs was 49% (95% confidence interval [CI], 37‐61%) (Figure 2). The pooled specificity for superficial wound swabs was 62% (95% CI, 51‐74%), while the pooled positive and negative LRs were 1.1 (95% CI, 0.71‐1.5) and 0.67 (95% CI, 0.52‐0.82), respectively (Figure 2).

Figure 2

The median number of bacterial isolates reported for each culture type, superficial and comparison culture, was collected from each study (Table 2). The median value for number of bacterial isolates identified by superficial culture was 2.7 (interquartile range [IQR] 1.8‐3.2). The median value for number of bacterial isolates identified by comparison culture was 2.2 (IQR 1.7‐2.9). A Wilcoxon rank sum analysis showed that the number of isolates for surface wound cultures was not significantly different than the number of isolates for comparison cultures (P = 0.75) (Table 1).

Discussion

In performing this review, we discovered ambiguity in the literature regarding the utility of surface wound cultures. Some studies obtained findings to suggest the utility of surface wound cultures,8, 14, 17 while other studies in our review16, 18, 19 provided evidence against them. This variability confirmed the need for a meta‐analytic approach as provided by this review.

While we have tried to minimize bias through a well‐established methodology, we acknowledge that certain methodological limitations should be considered in interpreting the results. There may be publication bias in reviews that include only published articles; a funnel plot of sensitivity vs. sample size showed some asymmetry, suggesting bias. Our search strategy was limited to English‐only articles, which may result in publication bias.

Further, this review included a group of studies that were heterogeneous in several regards. Differences exist in culturing methods and laboratory technology, as exemplified by the variety of superficial culture methods used. We were not able to account for these laboratory differences, as methodologies in obtaining and isolating bacteria were not uniformly well described.

Additionally, the studies classified organisms in different ways. Three studies categorized organisms according to Gram's stain characteristics.13, 16, 18 One study described organisms primarily in terms of aerobic or anaerobic respiration.15 Two studies14, 19 discussed pathogens both in terms of respiration (aerobic/anaerobic) and Gram's stain characteristic, while another 2 studies17, 20 did not describe organisms in either terms. These inconsistencies limited our ability to provide sensitivity and specificity information for specific subclasses of organisms.

The clinical conditions in each study surrounding the wounds were also heterogeneous: most significantly in the issue of prior antibiotic administration. All but 1 study16 indicated that the patients had received antibiotics prior to having cultures obtained. The type of antibiotics (narrow‐spectrum or broad‐spectrum), the route of administration, and the cessation of antibiotics in relation to obtaining swabs and cultures all varied widely or were not well described. This degree of ambiguity will necessarily impact both the reliability of data regarding microbial growth as well as the component flora.

The inclusion of higher quality studies is likely to result in a more reliable meta‐analysis.21 We had hoped that antibiotic trials would contain uniform outcomes and thus strengthen our meta‐analysis through the inclusion of randomized‐controlled studies. Unfortunately, the majority of antibiotic trials did not use superficial wound cultures, did not report mean number of isolates, or did not provide microbiological data in sufficient detail to calculate concordance ratesand therefore, did not meet eligibility criteria. Randomized‐controlled trials were a minority among our included articles; the majority of study designs were retrospective cohorts and case‐controlled studies.

Despite these limitations, we were able to conclude that superficial wound culture provides mediocre sensitivity (49%) and specificity (62%). The positive LR of 1.1 is unhelpful in decision making, having a CI that includes 1. Interestingly, the negative LR of 0.67 could be somewhat helpful in medical decision making, modifying the pretest probability and assisting in ruling out a deeper bacterial infection. Although, according to Fagan's nomogram, a negative LR of 0.67 has only a mild effect on pretest odds.22

The bacterial bioburden assessed by the number of isolates obtained by culture method serves as a proxy for reliability of culture results14, 23 by suggesting that fewer organisms isolated from deep tissue or bone samples reflects a less contaminated specimen. Our assessment of the bioburden found that the median number of isolates was slightly higher in surface cultures than deeper cultures, though not to a significant degree (P = 0.75). This indicates that the degree of contamination in superficial cultures was neither significantly worse nor better than deep cultures.

We attempted to define a role for surface wound cultures; however, we found that these did not show any greater utility than deep cultures for identifying the microbiologic etiology of diabetic wound infections. While the negative LR provides some quantitative verification of the common clinical practice that a negative culture argues against infection, the finding is not especially robust.

Although for this meta‐analysis we grouped all organisms in the same way, we recognize that the sensitivity and specificity may differ according to various subclasses of bacteria. Interpretations of culture results also vary (eg, Gram positive vs. negative; aerobic vs. anaerobic); practitioners will not interpret superficial cultures of coagulase‐negative Staphylococcus in the same way as Pseudomonas. However, this study seeks to establish a reasonable starting point for the medical decision‐making process by providing quantitative values in an area with previously conflicting data. We anticipate that as laboratory techniques improve and research into superficial wounds continues, greater sensitivity of superficial wound cultures will result.

Ultimately, physicians use culture data to target therapy in an effort to use the least toxic and most effective antimicrobial agent possible to successfully treat infections. Clinical outcomes were not described in all included articles; in those that did, the endpoints were too dissimilar for meaningful comparison. Limiting our review to studies reporting treatment outcomes would have resulted in too few included studies. Thus, we were unable able to assess whether superficial wound cultures were associated with improved patient‐oriented outcomes in this meta‐analysis.

There is a significant paucity of trials evaluating the accurate concordance of superficial swabs to deep tissue culture. The current data shows poor sensitivity and specificity of superficial culture methods. The presumption that deeper cultures (such as a bone biopsy) should result in a less contaminated sample and more targeted culture results was also not borne out in our review. When presented with a patient with a wound infection, physicians mentally supply a pretest (or a pretreatment) probability as to the microbiologic etiology of the infection. Careful history will, of course, be critical in identifying extenuating circumstance or unusual exposures. From our meta‐analysis, we cannot recommend the routine use of superficial wound cultures to guide initial antibiotic therapy as this may result in poor resource utilization.5 While clinical outcomes from the use of routine superficial cultures are unclear, we suggest greater use of local antibiograms and methicillin‐resistant Staphylococcus aureus (MRSA) prevalence data to determine resistance patterns and guide the selection of empiric therapies.

While a general consensus exists that surface wound cultures have less utility than deeper cultures, surface cultures are nevertheless routinely used to guide empiric antibiotic administration. This is due in part to the ease with which surface cultures are obtained and the delay in obtaining deeper wound and bone cultures. The Infectious Diseases Society of America (IDSA) recommends routine culture of all diabetic infections before initiating empiric antibiotic therapy, despite caveats regarding undebrided wounds.1 Further examination of 2 additional societies, the European Society of Clinical Microbiology and Infectious Diseases and the Australasian Society for Infectious Diseases, reveals that they also do not describe guidelines on the role of surface wound cultures in skin, and skin structure infection (SSSI) management.2, 3

Surface wound cultures are used to aid in diagnosis and appropriate antibiotic treatment of lower extremity foot ulcers.4 Contaminated cultures from other body locations have shown little utility and may be wasteful of resources.5, 6 We hypothesize that given commensal skin flora, coupled with the additional flora that colonizes (chronic) lower extremity wounds, surface wound cultures provide poor diagnostic accuracy for determining the etiology of acute infection. In contrast, many believe that deep tissue cultures obtained at time of debridement or surgical intervention may provide more relevant information to guide antibiotic therapy, thus serving as a gold standard.13, 7, 8 Nevertheless, with the ease of obtaining these superficial cultures and the promptness of the results, surface wound cultures are still used as a surrogate for the information derived from deeper cultures.

Purpose

The frequency at which superficial wound cultures correlate with the data obtained from deeper cultures is needed to interpret the posttest likelihood of infection. However, the sensitivity and specificity of superficial wound culture as a diagnostic test is unclear. The purpose of this study is to conduct a systematic review of the existing literature in order to investigate the relationship between superficial wound cultures and the etiology of SSSI. Accordingly, we aim to describe any role that surface wound cultures may play in the treatment of lower extremity ulcers.

Materials and Methods

Data Sources

We identified eligible articles through an electronic search of the following databases: Medline through PubMed, Excerpta Medica Database (EMBASE), Cumulative Index of Nursing and Allied Health Literature (CINAHL), and Scopus. We also hand searched the reference lists of key review articles identified by the electronic search and the reference lists of all eligible articles (Figure 1).

Figure 1

Results

Of 9032 unique citations, eight studies met all inclusion criteria (Figure 1). Inter‐rater reliability was substantial (Kappa = 0.78).12 Areas of initial disagreement generally involved whether a study adequately described an appropriate alternative culture method for comparison or whether data available in an article was sufficient for sensitivity and specificity calculation. Consensus was achieved once the full article was retrieved, reviewed and discussed.

The 8 studies evaluated in the review included a total number of 615 patients or samples (Table 1). Diabetic wounds were described in four studies.1316 Two studies described wounds associated with peripheral vascular disease,13, 17 while four involved traumatic wounds.13, 1719 One study did not identify the clinical circumstances concerning the wounds.20

The studies used several different methods for obtaining superficial cultures. Six studies obtained purulent wound drainage material through the application of sterile swabs.1316, 18, 19 One study obtained purulent drainage material using needle aspiration.18 Two studies obtained culture material from sinus tracts associated with the wounds, one through sinus tract washings17 and another by obtaining sinus tract discharge material.20

The types of comparison cultures used were equally divided between deep tissue biopsies1316 and bone biopsies,1720 each accounting for 50% (4 of 8) of studies.

In assessing the data from the eight studies, the pooled test sensitivity for superficial wound swabs was 49% (95% confidence interval [CI], 37‐61%) (Figure 2). The pooled specificity for superficial wound swabs was 62% (95% CI, 51‐74%), while the pooled positive and negative LRs were 1.1 (95% CI, 0.71‐1.5) and 0.67 (95% CI, 0.52‐0.82), respectively (Figure 2).

Figure 2

The median number of bacterial isolates reported for each culture type, superficial and comparison culture, was collected from each study (Table 2). The median value for number of bacterial isolates identified by superficial culture was 2.7 (interquartile range [IQR] 1.8‐3.2). The median value for number of bacterial isolates identified by comparison culture was 2.2 (IQR 1.7‐2.9). A Wilcoxon rank sum analysis showed that the number of isolates for surface wound cultures was not significantly different than the number of isolates for comparison cultures (P = 0.75) (Table 1).

Discussion

In performing this review, we discovered ambiguity in the literature regarding the utility of surface wound cultures. Some studies obtained findings to suggest the utility of surface wound cultures,8, 14, 17 while other studies in our review16, 18, 19 provided evidence against them. This variability confirmed the need for a meta‐analytic approach as provided by this review.

While we have tried to minimize bias through a well‐established methodology, we acknowledge that certain methodological limitations should be considered in interpreting the results. There may be publication bias in reviews that include only published articles; a funnel plot of sensitivity vs. sample size showed some asymmetry, suggesting bias. Our search strategy was limited to English‐only articles, which may result in publication bias.

Further, this review included a group of studies that were heterogeneous in several regards. Differences exist in culturing methods and laboratory technology, as exemplified by the variety of superficial culture methods used. We were not able to account for these laboratory differences, as methodologies in obtaining and isolating bacteria were not uniformly well described.

Additionally, the studies classified organisms in different ways. Three studies categorized organisms according to Gram's stain characteristics.13, 16, 18 One study described organisms primarily in terms of aerobic or anaerobic respiration.15 Two studies14, 19 discussed pathogens both in terms of respiration (aerobic/anaerobic) and Gram's stain characteristic, while another 2 studies17, 20 did not describe organisms in either terms. These inconsistencies limited our ability to provide sensitivity and specificity information for specific subclasses of organisms.

The clinical conditions in each study surrounding the wounds were also heterogeneous: most significantly in the issue of prior antibiotic administration. All but 1 study16 indicated that the patients had received antibiotics prior to having cultures obtained. The type of antibiotics (narrow‐spectrum or broad‐spectrum), the route of administration, and the cessation of antibiotics in relation to obtaining swabs and cultures all varied widely or were not well described. This degree of ambiguity will necessarily impact both the reliability of data regarding microbial growth as well as the component flora.

The inclusion of higher quality studies is likely to result in a more reliable meta‐analysis.21 We had hoped that antibiotic trials would contain uniform outcomes and thus strengthen our meta‐analysis through the inclusion of randomized‐controlled studies. Unfortunately, the majority of antibiotic trials did not use superficial wound cultures, did not report mean number of isolates, or did not provide microbiological data in sufficient detail to calculate concordance ratesand therefore, did not meet eligibility criteria. Randomized‐controlled trials were a minority among our included articles; the majority of study designs were retrospective cohorts and case‐controlled studies.

Despite these limitations, we were able to conclude that superficial wound culture provides mediocre sensitivity (49%) and specificity (62%). The positive LR of 1.1 is unhelpful in decision making, having a CI that includes 1. Interestingly, the negative LR of 0.67 could be somewhat helpful in medical decision making, modifying the pretest probability and assisting in ruling out a deeper bacterial infection. Although, according to Fagan's nomogram, a negative LR of 0.67 has only a mild effect on pretest odds.22

The bacterial bioburden assessed by the number of isolates obtained by culture method serves as a proxy for reliability of culture results14, 23 by suggesting that fewer organisms isolated from deep tissue or bone samples reflects a less contaminated specimen. Our assessment of the bioburden found that the median number of isolates was slightly higher in surface cultures than deeper cultures, though not to a significant degree (P = 0.75). This indicates that the degree of contamination in superficial cultures was neither significantly worse nor better than deep cultures.

We attempted to define a role for surface wound cultures; however, we found that these did not show any greater utility than deep cultures for identifying the microbiologic etiology of diabetic wound infections. While the negative LR provides some quantitative verification of the common clinical practice that a negative culture argues against infection, the finding is not especially robust.

Although for this meta‐analysis we grouped all organisms in the same way, we recognize that the sensitivity and specificity may differ according to various subclasses of bacteria. Interpretations of culture results also vary (eg, Gram positive vs. negative; aerobic vs. anaerobic); practitioners will not interpret superficial cultures of coagulase‐negative Staphylococcus in the same way as Pseudomonas. However, this study seeks to establish a reasonable starting point for the medical decision‐making process by providing quantitative values in an area with previously conflicting data. We anticipate that as laboratory techniques improve and research into superficial wounds continues, greater sensitivity of superficial wound cultures will result.

Ultimately, physicians use culture data to target therapy in an effort to use the least toxic and most effective antimicrobial agent possible to successfully treat infections. Clinical outcomes were not described in all included articles; in those that did, the endpoints were too dissimilar for meaningful comparison. Limiting our review to studies reporting treatment outcomes would have resulted in too few included studies. Thus, we were unable able to assess whether superficial wound cultures were associated with improved patient‐oriented outcomes in this meta‐analysis.

There is a significant paucity of trials evaluating the accurate concordance of superficial swabs to deep tissue culture. The current data shows poor sensitivity and specificity of superficial culture methods. The presumption that deeper cultures (such as a bone biopsy) should result in a less contaminated sample and more targeted culture results was also not borne out in our review. When presented with a patient with a wound infection, physicians mentally supply a pretest (or a pretreatment) probability as to the microbiologic etiology of the infection. Careful history will, of course, be critical in identifying extenuating circumstance or unusual exposures. From our meta‐analysis, we cannot recommend the routine use of superficial wound cultures to guide initial antibiotic therapy as this may result in poor resource utilization.5 While clinical outcomes from the use of routine superficial cultures are unclear, we suggest greater use of local antibiograms and methicillin‐resistant Staphylococcus aureus (MRSA) prevalence data to determine resistance patterns and guide the selection of empiric therapies.

A 6‐year‐old girl with a history of bilateral myringotomies, tonsillectomy, and adenoidectomy 6 and 8 months prior, presented with forehead swelling. Eight days prior, she developed right ear pain, sore throat and fever followed by eye pain and headache for which she was evaluated and diagnosed with viral illness. On the day of presentation she awoke with forehead swelling, persistent headache, and recurrent fever.

On exam she was afebrile. Central forehead swelling was noted without overlying erythema or fluctulence (Figure 1). Neurologic exam was normal. Noncontrast computed tomography (CT) scan of the head showed pan sinusitis with an extra‐axial fluid collection in the left frontal region (Figure 2).

Figure 1

Figure 2

Vancomycin, ceftriaxone, and metronidazole were started empirically. She underwent bilateral maxillary antrostomy, total ethmoidectomy, and burr‐hole evacuation of the epidural abscess. Operative specimen cultures grew out Group A streptococci, after‐which antibiotic therapy was narrowed down to ampicillin/sulbactam alone.

Pott's puffy tumor, or osteomyelitis of the frontal bone with subperiosteal abscess formation, is rare in children less than 7 years of age and usually the result of a delay in diagnosis or inadequate treatment of rhinosinusitis.1 Risk factors include frontal sinusitis, head trauma, and less commonly, cocaine use, dental infection, or delayed neurosurgical or sinus surgery complications.2, 3 Fever, vomiting, forehead tenderness, and headache are the most common complaints, though seizure and focal neurologic findings have been described.2 CT scanning is the imaging modality of choice. Most commonly cultured organisms include Streptococci, Haemophilus influenzae, Bacteroides, and less commonly, Staphylococcus aureus.2, 4 Treatment includes empiric broad‐spectrum antibiotics that penetrate the blood‐brain barrier with early surgical drainage of the abscess and debridement of the osteomyelitic bone.4

A 6‐year‐old girl with a history of bilateral myringotomies, tonsillectomy, and adenoidectomy 6 and 8 months prior, presented with forehead swelling. Eight days prior, she developed right ear pain, sore throat and fever followed by eye pain and headache for which she was evaluated and diagnosed with viral illness. On the day of presentation she awoke with forehead swelling, persistent headache, and recurrent fever.

On exam she was afebrile. Central forehead swelling was noted without overlying erythema or fluctulence (Figure 1). Neurologic exam was normal. Noncontrast computed tomography (CT) scan of the head showed pan sinusitis with an extra‐axial fluid collection in the left frontal region (Figure 2).

Figure 1

Figure 2

Vancomycin, ceftriaxone, and metronidazole were started empirically. She underwent bilateral maxillary antrostomy, total ethmoidectomy, and burr‐hole evacuation of the epidural abscess. Operative specimen cultures grew out Group A streptococci, after‐which antibiotic therapy was narrowed down to ampicillin/sulbactam alone.

Pott's puffy tumor, or osteomyelitis of the frontal bone with subperiosteal abscess formation, is rare in children less than 7 years of age and usually the result of a delay in diagnosis or inadequate treatment of rhinosinusitis.1 Risk factors include frontal sinusitis, head trauma, and less commonly, cocaine use, dental infection, or delayed neurosurgical or sinus surgery complications.2, 3 Fever, vomiting, forehead tenderness, and headache are the most common complaints, though seizure and focal neurologic findings have been described.2 CT scanning is the imaging modality of choice. Most commonly cultured organisms include Streptococci, Haemophilus influenzae, Bacteroides, and less commonly, Staphylococcus aureus.2, 4 Treatment includes empiric broad‐spectrum antibiotics that penetrate the blood‐brain barrier with early surgical drainage of the abscess and debridement of the osteomyelitic bone.4

A 6‐year‐old girl with a history of bilateral myringotomies, tonsillectomy, and adenoidectomy 6 and 8 months prior, presented with forehead swelling. Eight days prior, she developed right ear pain, sore throat and fever followed by eye pain and headache for which she was evaluated and diagnosed with viral illness. On the day of presentation she awoke with forehead swelling, persistent headache, and recurrent fever.

On exam she was afebrile. Central forehead swelling was noted without overlying erythema or fluctulence (Figure 1). Neurologic exam was normal. Noncontrast computed tomography (CT) scan of the head showed pan sinusitis with an extra‐axial fluid collection in the left frontal region (Figure 2).

Figure 1

Figure 2

Vancomycin, ceftriaxone, and metronidazole were started empirically. She underwent bilateral maxillary antrostomy, total ethmoidectomy, and burr‐hole evacuation of the epidural abscess. Operative specimen cultures grew out Group A streptococci, after‐which antibiotic therapy was narrowed down to ampicillin/sulbactam alone.

Pott's puffy tumor, or osteomyelitis of the frontal bone with subperiosteal abscess formation, is rare in children less than 7 years of age and usually the result of a delay in diagnosis or inadequate treatment of rhinosinusitis.1 Risk factors include frontal sinusitis, head trauma, and less commonly, cocaine use, dental infection, or delayed neurosurgical or sinus surgery complications.2, 3 Fever, vomiting, forehead tenderness, and headache are the most common complaints, though seizure and focal neurologic findings have been described.2 CT scanning is the imaging modality of choice. Most commonly cultured organisms include Streptococci, Haemophilus influenzae, Bacteroides, and less commonly, Staphylococcus aureus.2, 4 Treatment includes empiric broad‐spectrum antibiotics that penetrate the blood‐brain barrier with early surgical drainage of the abscess and debridement of the osteomyelitic bone.4

Hospital readmissions are common, costly in both economic and human terms, and often preventable. This perfect storm of attributes has placed hospital readmissions at the center of discourse among payers, providers, and policy makers, which is leading to innovations in care delivery. Evolving efforts to enhance discharge communication, to improve care coordination and accountability, and to meaningfully involve primary care, show promise of reducing readmissions.14 These pockets of success demonstrate that improving care transitions can increase quality of care while decreasing costs.

Within the hospital, it is now clear that the discharge process typically requires the same intensity of effort as admission. The hospitalist guides an interdisciplinary team, including nurses, pharmacists, case managers, and social workers, through a checklist of discharge tasks. Some tasks require substantial hospitalist involvement and expertise, such as medication reconciliationa detail‐oriented, time‐consuming process. Other tasks can be accomplished by team members, overseen by the hospitalist, such as scheduling timely follow‐up appointments, coordinating outpatient services, and assembling educational materials. Taken as a whole, significant time and effort must be devoted by the inpatient team to address the complex landscape of a patient's medical and psychosocial needs.

That of course, is only half of the equation, as patient care must be transferred to an equally invested outpatient team led by a primary care provider (PCP). Several influential medical societies have endorsed the medical home (a multidisciplinary care team led by a PCP) as the primary agent to coordinate patient care across settings.5 Indeed, promptly reconnecting with their PCP and primary care team after discharge can have profound meaning for patients, who may otherwise be unsupported with their postdischarge clinical needs. In this issue of the Journal of Hospital Medicine, 4 important articles provide evidence in support of an outpatient partner to actively assume patient care responsibility after hospital discharge.

van Walraven et al.6 conducted an elegant study to evaluate the impact of postdischarge PCP visits on readmissions. Following more than 5000 patients for nearly 6 months, they demonstrated that increased PCP follow‐up was significantly and independently associated with a decreased risk of hospital readmission. This confirms the positive impact that a primary care connection can have on postdischarge care. This study also highlights some challenges: 18% of the original cohort were excluded from the final analyses because they had only 1 or no PCP visit in the 6 months following discharge, indicating inadequate postdischarge follow‐up for a substantial sub‐group. Misky et al.7 similarly established that patients with timely PCP follow‐up (within one month of discharge) were 10 times less likely to be readmitted for the same condition as their index admission. These are also encouraging findings for those patients with PCP follow‐up. Yet among patients in their study, PCP follow‐up was even less common, with only 49% of patients having appointments within one month. Future studies should consider how more intensive outreach strategies might engage difficult‐to‐reach patients and communities.

PCP follow‐up may be beneficial because discharged patients often have ongoing issues that need to be addressed. Arora et al.8 surveyed inner city patients and their PCPs 2 weeks after hospital discharge to assess whether patients experienced any problems in the postdischarge period, and whether PCPs were aware of their patients' hospitalization. Nearly half of all patients recounted 1 or more postdischarge problems. The likelihood of reporting such a problem was twice as common among those patients whose PCP was unaware of their hospitalization. Again, this is strong validation of the importance of PCP involvement in posthospital care, but equally concerning is their finding that fully 3 in 10 PCPs were unaware of their patient's hospitalization.

Finally, Mitchell et al.9 further refine our understanding of risk factors for readmission. In an ethnically diverse inner city population, they screened 738 inpatients for depression. Among the 238 (32%) patients who screened positive, there was a marked 73% increase in hospital utilization (emergency department [ED] visits and readmissions) within 30 days of discharge.9 This confirms previous research that depression is a risk factor for rehospitalization.10 Depression, however, is amenable to treatment and receiving care through one's primary care practice can potentially mitigate how depression negatively affects patients' medication adherence, self‐care behavior, and ultimately readmission rates.

Collectively, these 4 articles give us reason to experience both despair and hope. It is discouraging that large numbers of patients do not have timely encounters with their PCP after discharge, confirming previous findings,11 and that too often PCPs are unaware of their patients' hospitalizations. In any other industry this sort of inefficiency and poor customer service would put a company out of business; that it persists in medicine is embarrassing. Because our medical system is not sufficiently incentivized by quality outcomes, such poor practices continue to be tolerated, and our patients suffer the consequences.

But we are also shown a way forward, with accumulating evidence complimenting existing studies that a primary care connection can improve the quality of postdischarge care and decrease readmissions.1214 Recognizing this central role of primary care, however, forces us to acknowledge the diminishing availability of primary care nationwide; in many inner city and rural locations, accessible primary care is largely nonexistent. This shortage must be corrected to attain needed access and to advance health care reform.

Postdischarge PCP involvement is particularly essential with shorter hospital stays, as patients will predictably have complex postdischarge needs as they complete their recuperation at home. Indeed, Arora et al.8 indicate that posthospitalization problems may be more the rule than the exception, and that specific types of problems can be foreseen. Most commonly reported were challenges obtaining follow‐up appointments, difficulties managing or obtaining medications, feeling unprepared for discharge, having unanswered questions, or needing an urgent reevaluation. While well organized predischarge efforts help to prepare patients, even the most perfect discharge process cannot anticipate all possible pitfalls. Fortunately, most postdischarge problems can be effectively handled by those who often know the patient best, the patient's primary care team. The outpatient team is ideally situated to assist patients with the logistics of accessing the care system, to provide ongoing education, and to help with such basic needs as transportation and social support. It makes sense that such personalized outreach can prevent small problems from blossoming into more serious issues that might ultimately require rehospitalization.

Upon discharge, patients and families are also often expected to assume new self‐care responsibilities, to implement new dietary restrictions, to use new medications, and to monitor and respond to new and evolving symptoms. Gaining the knowledge, the confidence, and the experience to adopt new behaviors is critical to successful postdischarge self‐care. Adult learning theory informs us that education is an ongoing process. Mastering new material occurs with repetition, over time, and under different circumstances. An ancient Chinese proverb encapsulates the key aspects of learning:

• I hear and I forget;

• I see and I remember;

• I do and I understand.

Thus, a single didactic discharge session in the hospital is unlikely to provide patients with sufficient depth of understanding that one attains through experiential learning. Hospital‐based discharge teaching is further compromised in the setting of patient fatigue, anxiety and illness: it is not surprising that patients comprehend and retain only 50% of the medical information discussed with their physicians.15

While teach back has been effectively used to ensure that information is registered when initially presented, it does not ensure that information has been internalized in such a way that it can be utilized hours or days later. Instead, it is the active engagement with the primary care team that provides opportunities for ongoing learning, personalized to the educational needs of each patient. With nurses, dieticians, pharmacists, and medical educators playing a central guiding role, patients can receive appropriately tailored instruction, as well as opportunities to practically apply their new knowledge.

The evolving partnership among hospitals, hospitalists, and primary care holds great promise to reduce avoidable readmissions, and the Patient Protection and Affordable Care Act of 2010 will provide financial incentives to support this partnership. Health care reform aims to adjust hospital payments based on rates of preventable Medicare readmissions; bundling payments (paying for episodes of illness based on outcomes) and accountable care organizations (ACOs) will ideally foster seamless care coordination.16 Nurturing and developing this collaboration between the inpatient and outpatient care teams will be essential as we seek to provide patients the safest transition possible from the hospital to home.

Hospital readmissions are common, costly in both economic and human terms, and often preventable. This perfect storm of attributes has placed hospital readmissions at the center of discourse among payers, providers, and policy makers, which is leading to innovations in care delivery. Evolving efforts to enhance discharge communication, to improve care coordination and accountability, and to meaningfully involve primary care, show promise of reducing readmissions.14 These pockets of success demonstrate that improving care transitions can increase quality of care while decreasing costs.

Within the hospital, it is now clear that the discharge process typically requires the same intensity of effort as admission. The hospitalist guides an interdisciplinary team, including nurses, pharmacists, case managers, and social workers, through a checklist of discharge tasks. Some tasks require substantial hospitalist involvement and expertise, such as medication reconciliationa detail‐oriented, time‐consuming process. Other tasks can be accomplished by team members, overseen by the hospitalist, such as scheduling timely follow‐up appointments, coordinating outpatient services, and assembling educational materials. Taken as a whole, significant time and effort must be devoted by the inpatient team to address the complex landscape of a patient's medical and psychosocial needs.

That of course, is only half of the equation, as patient care must be transferred to an equally invested outpatient team led by a primary care provider (PCP). Several influential medical societies have endorsed the medical home (a multidisciplinary care team led by a PCP) as the primary agent to coordinate patient care across settings.5 Indeed, promptly reconnecting with their PCP and primary care team after discharge can have profound meaning for patients, who may otherwise be unsupported with their postdischarge clinical needs. In this issue of the Journal of Hospital Medicine, 4 important articles provide evidence in support of an outpatient partner to actively assume patient care responsibility after hospital discharge.

van Walraven et al.6 conducted an elegant study to evaluate the impact of postdischarge PCP visits on readmissions. Following more than 5000 patients for nearly 6 months, they demonstrated that increased PCP follow‐up was significantly and independently associated with a decreased risk of hospital readmission. This confirms the positive impact that a primary care connection can have on postdischarge care. This study also highlights some challenges: 18% of the original cohort were excluded from the final analyses because they had only 1 or no PCP visit in the 6 months following discharge, indicating inadequate postdischarge follow‐up for a substantial sub‐group. Misky et al.7 similarly established that patients with timely PCP follow‐up (within one month of discharge) were 10 times less likely to be readmitted for the same condition as their index admission. These are also encouraging findings for those patients with PCP follow‐up. Yet among patients in their study, PCP follow‐up was even less common, with only 49% of patients having appointments within one month. Future studies should consider how more intensive outreach strategies might engage difficult‐to‐reach patients and communities.

PCP follow‐up may be beneficial because discharged patients often have ongoing issues that need to be addressed. Arora et al.8 surveyed inner city patients and their PCPs 2 weeks after hospital discharge to assess whether patients experienced any problems in the postdischarge period, and whether PCPs were aware of their patients' hospitalization. Nearly half of all patients recounted 1 or more postdischarge problems. The likelihood of reporting such a problem was twice as common among those patients whose PCP was unaware of their hospitalization. Again, this is strong validation of the importance of PCP involvement in posthospital care, but equally concerning is their finding that fully 3 in 10 PCPs were unaware of their patient's hospitalization.

Finally, Mitchell et al.9 further refine our understanding of risk factors for readmission. In an ethnically diverse inner city population, they screened 738 inpatients for depression. Among the 238 (32%) patients who screened positive, there was a marked 73% increase in hospital utilization (emergency department [ED] visits and readmissions) within 30 days of discharge.9 This confirms previous research that depression is a risk factor for rehospitalization.10 Depression, however, is amenable to treatment and receiving care through one's primary care practice can potentially mitigate how depression negatively affects patients' medication adherence, self‐care behavior, and ultimately readmission rates.

Collectively, these 4 articles give us reason to experience both despair and hope. It is discouraging that large numbers of patients do not have timely encounters with their PCP after discharge, confirming previous findings,11 and that too often PCPs are unaware of their patients' hospitalizations. In any other industry this sort of inefficiency and poor customer service would put a company out of business; that it persists in medicine is embarrassing. Because our medical system is not sufficiently incentivized by quality outcomes, such poor practices continue to be tolerated, and our patients suffer the consequences.

But we are also shown a way forward, with accumulating evidence complimenting existing studies that a primary care connection can improve the quality of postdischarge care and decrease readmissions.1214 Recognizing this central role of primary care, however, forces us to acknowledge the diminishing availability of primary care nationwide; in many inner city and rural locations, accessible primary care is largely nonexistent. This shortage must be corrected to attain needed access and to advance health care reform.

Postdischarge PCP involvement is particularly essential with shorter hospital stays, as patients will predictably have complex postdischarge needs as they complete their recuperation at home. Indeed, Arora et al.8 indicate that posthospitalization problems may be more the rule than the exception, and that specific types of problems can be foreseen. Most commonly reported were challenges obtaining follow‐up appointments, difficulties managing or obtaining medications, feeling unprepared for discharge, having unanswered questions, or needing an urgent reevaluation. While well organized predischarge efforts help to prepare patients, even the most perfect discharge process cannot anticipate all possible pitfalls. Fortunately, most postdischarge problems can be effectively handled by those who often know the patient best, the patient's primary care team. The outpatient team is ideally situated to assist patients with the logistics of accessing the care system, to provide ongoing education, and to help with such basic needs as transportation and social support. It makes sense that such personalized outreach can prevent small problems from blossoming into more serious issues that might ultimately require rehospitalization.

Upon discharge, patients and families are also often expected to assume new self‐care responsibilities, to implement new dietary restrictions, to use new medications, and to monitor and respond to new and evolving symptoms. Gaining the knowledge, the confidence, and the experience to adopt new behaviors is critical to successful postdischarge self‐care. Adult learning theory informs us that education is an ongoing process. Mastering new material occurs with repetition, over time, and under different circumstances. An ancient Chinese proverb encapsulates the key aspects of learning:

• I hear and I forget;

• I see and I remember;

• I do and I understand.

Thus, a single didactic discharge session in the hospital is unlikely to provide patients with sufficient depth of understanding that one attains through experiential learning. Hospital‐based discharge teaching is further compromised in the setting of patient fatigue, anxiety and illness: it is not surprising that patients comprehend and retain only 50% of the medical information discussed with their physicians.15

While teach back has been effectively used to ensure that information is registered when initially presented, it does not ensure that information has been internalized in such a way that it can be utilized hours or days later. Instead, it is the active engagement with the primary care team that provides opportunities for ongoing learning, personalized to the educational needs of each patient. With nurses, dieticians, pharmacists, and medical educators playing a central guiding role, patients can receive appropriately tailored instruction, as well as opportunities to practically apply their new knowledge.

The evolving partnership among hospitals, hospitalists, and primary care holds great promise to reduce avoidable readmissions, and the Patient Protection and Affordable Care Act of 2010 will provide financial incentives to support this partnership. Health care reform aims to adjust hospital payments based on rates of preventable Medicare readmissions; bundling payments (paying for episodes of illness based on outcomes) and accountable care organizations (ACOs) will ideally foster seamless care coordination.16 Nurturing and developing this collaboration between the inpatient and outpatient care teams will be essential as we seek to provide patients the safest transition possible from the hospital to home.

Hospital readmissions are common, costly in both economic and human terms, and often preventable. This perfect storm of attributes has placed hospital readmissions at the center of discourse among payers, providers, and policy makers, which is leading to innovations in care delivery. Evolving efforts to enhance discharge communication, to improve care coordination and accountability, and to meaningfully involve primary care, show promise of reducing readmissions.14 These pockets of success demonstrate that improving care transitions can increase quality of care while decreasing costs.

Within the hospital, it is now clear that the discharge process typically requires the same intensity of effort as admission. The hospitalist guides an interdisciplinary team, including nurses, pharmacists, case managers, and social workers, through a checklist of discharge tasks. Some tasks require substantial hospitalist involvement and expertise, such as medication reconciliationa detail‐oriented, time‐consuming process. Other tasks can be accomplished by team members, overseen by the hospitalist, such as scheduling timely follow‐up appointments, coordinating outpatient services, and assembling educational materials. Taken as a whole, significant time and effort must be devoted by the inpatient team to address the complex landscape of a patient's medical and psychosocial needs.

That of course, is only half of the equation, as patient care must be transferred to an equally invested outpatient team led by a primary care provider (PCP). Several influential medical societies have endorsed the medical home (a multidisciplinary care team led by a PCP) as the primary agent to coordinate patient care across settings.5 Indeed, promptly reconnecting with their PCP and primary care team after discharge can have profound meaning for patients, who may otherwise be unsupported with their postdischarge clinical needs. In this issue of the Journal of Hospital Medicine, 4 important articles provide evidence in support of an outpatient partner to actively assume patient care responsibility after hospital discharge.

van Walraven et al.6 conducted an elegant study to evaluate the impact of postdischarge PCP visits on readmissions. Following more than 5000 patients for nearly 6 months, they demonstrated that increased PCP follow‐up was significantly and independently associated with a decreased risk of hospital readmission. This confirms the positive impact that a primary care connection can have on postdischarge care. This study also highlights some challenges: 18% of the original cohort were excluded from the final analyses because they had only 1 or no PCP visit in the 6 months following discharge, indicating inadequate postdischarge follow‐up for a substantial sub‐group. Misky et al.7 similarly established that patients with timely PCP follow‐up (within one month of discharge) were 10 times less likely to be readmitted for the same condition as their index admission. These are also encouraging findings for those patients with PCP follow‐up. Yet among patients in their study, PCP follow‐up was even less common, with only 49% of patients having appointments within one month. Future studies should consider how more intensive outreach strategies might engage difficult‐to‐reach patients and communities.

PCP follow‐up may be beneficial because discharged patients often have ongoing issues that need to be addressed. Arora et al.8 surveyed inner city patients and their PCPs 2 weeks after hospital discharge to assess whether patients experienced any problems in the postdischarge period, and whether PCPs were aware of their patients' hospitalization. Nearly half of all patients recounted 1 or more postdischarge problems. The likelihood of reporting such a problem was twice as common among those patients whose PCP was unaware of their hospitalization. Again, this is strong validation of the importance of PCP involvement in posthospital care, but equally concerning is their finding that fully 3 in 10 PCPs were unaware of their patient's hospitalization.

Finally, Mitchell et al.9 further refine our understanding of risk factors for readmission. In an ethnically diverse inner city population, they screened 738 inpatients for depression. Among the 238 (32%) patients who screened positive, there was a marked 73% increase in hospital utilization (emergency department [ED] visits and readmissions) within 30 days of discharge.9 This confirms previous research that depression is a risk factor for rehospitalization.10 Depression, however, is amenable to treatment and receiving care through one's primary care practice can potentially mitigate how depression negatively affects patients' medication adherence, self‐care behavior, and ultimately readmission rates.

Collectively, these 4 articles give us reason to experience both despair and hope. It is discouraging that large numbers of patients do not have timely encounters with their PCP after discharge, confirming previous findings,11 and that too often PCPs are unaware of their patients' hospitalizations. In any other industry this sort of inefficiency and poor customer service would put a company out of business; that it persists in medicine is embarrassing. Because our medical system is not sufficiently incentivized by quality outcomes, such poor practices continue to be tolerated, and our patients suffer the consequences.

But we are also shown a way forward, with accumulating evidence complimenting existing studies that a primary care connection can improve the quality of postdischarge care and decrease readmissions.1214 Recognizing this central role of primary care, however, forces us to acknowledge the diminishing availability of primary care nationwide; in many inner city and rural locations, accessible primary care is largely nonexistent. This shortage must be corrected to attain needed access and to advance health care reform.

Postdischarge PCP involvement is particularly essential with shorter hospital stays, as patients will predictably have complex postdischarge needs as they complete their recuperation at home. Indeed, Arora et al.8 indicate that posthospitalization problems may be more the rule than the exception, and that specific types of problems can be foreseen. Most commonly reported were challenges obtaining follow‐up appointments, difficulties managing or obtaining medications, feeling unprepared for discharge, having unanswered questions, or needing an urgent reevaluation. While well organized predischarge efforts help to prepare patients, even the most perfect discharge process cannot anticipate all possible pitfalls. Fortunately, most postdischarge problems can be effectively handled by those who often know the patient best, the patient's primary care team. The outpatient team is ideally situated to assist patients with the logistics of accessing the care system, to provide ongoing education, and to help with such basic needs as transportation and social support. It makes sense that such personalized outreach can prevent small problems from blossoming into more serious issues that might ultimately require rehospitalization.

Upon discharge, patients and families are also often expected to assume new self‐care responsibilities, to implement new dietary restrictions, to use new medications, and to monitor and respond to new and evolving symptoms. Gaining the knowledge, the confidence, and the experience to adopt new behaviors is critical to successful postdischarge self‐care. Adult learning theory informs us that education is an ongoing process. Mastering new material occurs with repetition, over time, and under different circumstances. An ancient Chinese proverb encapsulates the key aspects of learning:

• I hear and I forget;

• I see and I remember;

• I do and I understand.

Thus, a single didactic discharge session in the hospital is unlikely to provide patients with sufficient depth of understanding that one attains through experiential learning. Hospital‐based discharge teaching is further compromised in the setting of patient fatigue, anxiety and illness: it is not surprising that patients comprehend and retain only 50% of the medical information discussed with their physicians.15

While teach back has been effectively used to ensure that information is registered when initially presented, it does not ensure that information has been internalized in such a way that it can be utilized hours or days later. Instead, it is the active engagement with the primary care team that provides opportunities for ongoing learning, personalized to the educational needs of each patient. With nurses, dieticians, pharmacists, and medical educators playing a central guiding role, patients can receive appropriately tailored instruction, as well as opportunities to practically apply their new knowledge.

The evolving partnership among hospitals, hospitalists, and primary care holds great promise to reduce avoidable readmissions, and the Patient Protection and Affordable Care Act of 2010 will provide financial incentives to support this partnership. Health care reform aims to adjust hospital payments based on rates of preventable Medicare readmissions; bundling payments (paying for episodes of illness based on outcomes) and accountable care organizations (ACOs) will ideally foster seamless care coordination.16 Nurturing and developing this collaboration between the inpatient and outpatient care teams will be essential as we seek to provide patients the safest transition possible from the hospital to home.

A 25 year‐old hitherto healthy female Asian immigrant presented with gradually worsening painful neck swelling for 5 weeks and a 40‐pound weight loss. Neck examination showed tender, firm, matted and fixed lymph nodes in left and right cervical chains. Contrast computed tomography of the neck confirmed extensive necrotic lymphadenopathy. Excisional lymph node biopsy revealed granulomatous lymphadenitis with microabscesses. Staining for acid‐fast bacilli was positive; tissue culture confirmed the diagnosis of Mycobacterium tuberculosis. She was started on rifampin 600 mg, isoniazid 300 mg, pyrazinamide 1000 mg, and ethambutol 800 mg; ethambutol was discontinued when the mycobacteria proved susceptible to isoniazid. Pyrazinamide was discontinued after 2 months and isoniazid and rifampin continued to complete 6 months of therapy.0, 0

Tuberculous lymphadenitisscrofula when occurring in the cervical regionis a common cause of extrapulmonary tuberculosis, especially in developing countries. In developed countries, tuberculous lymphadenitis occurs mainly in immigrants but can also arise in travelers to endemic areas and immunodeficient persons.1 Painless lymphadenopathy of the superficial lymph nodes is typical. Fine needle aspiration with microscopy, culture and polymerase chain reaction (PCR) may be used as an initial diagnostic tool.2 However, if the results are negative despite high clinical suspicion, excisional biopsy of the lymph node may be necessary. Various nontuberculous mycobacteria also cause adenitis, so definitive speciation is needed to determine optimal therapy.3

Recommended initial treatment for M. tuberculosis includes isoniazid, rifampin, pyrazinamide and ethambutol or streptomycin, based on local resistance patterns. The fourth drug is withdrawn when the isolate is found susceptible to the first 3 drugs; isoniazid and rifampin are used for a total of 6 months of treatment, while pyrazinamide is withdrawn after 2 months.4 Twice‐weekly directly observed therapy (DOT) can be as effective as daily therapy supervised weekly.5 DOT maximizes compliance and reduces treatment failure.

Figure 1

Figure 2

A 25 year‐old hitherto healthy female Asian immigrant presented with gradually worsening painful neck swelling for 5 weeks and a 40‐pound weight loss. Neck examination showed tender, firm, matted and fixed lymph nodes in left and right cervical chains. Contrast computed tomography of the neck confirmed extensive necrotic lymphadenopathy. Excisional lymph node biopsy revealed granulomatous lymphadenitis with microabscesses. Staining for acid‐fast bacilli was positive; tissue culture confirmed the diagnosis of Mycobacterium tuberculosis. She was started on rifampin 600 mg, isoniazid 300 mg, pyrazinamide 1000 mg, and ethambutol 800 mg; ethambutol was discontinued when the mycobacteria proved susceptible to isoniazid. Pyrazinamide was discontinued after 2 months and isoniazid and rifampin continued to complete 6 months of therapy.0, 0

Tuberculous lymphadenitisscrofula when occurring in the cervical regionis a common cause of extrapulmonary tuberculosis, especially in developing countries. In developed countries, tuberculous lymphadenitis occurs mainly in immigrants but can also arise in travelers to endemic areas and immunodeficient persons.1 Painless lymphadenopathy of the superficial lymph nodes is typical. Fine needle aspiration with microscopy, culture and polymerase chain reaction (PCR) may be used as an initial diagnostic tool.2 However, if the results are negative despite high clinical suspicion, excisional biopsy of the lymph node may be necessary. Various nontuberculous mycobacteria also cause adenitis, so definitive speciation is needed to determine optimal therapy.3

Recommended initial treatment for M. tuberculosis includes isoniazid, rifampin, pyrazinamide and ethambutol or streptomycin, based on local resistance patterns. The fourth drug is withdrawn when the isolate is found susceptible to the first 3 drugs; isoniazid and rifampin are used for a total of 6 months of treatment, while pyrazinamide is withdrawn after 2 months.4 Twice‐weekly directly observed therapy (DOT) can be as effective as daily therapy supervised weekly.5 DOT maximizes compliance and reduces treatment failure.

Figure 1

Figure 2

A 25 year‐old hitherto healthy female Asian immigrant presented with gradually worsening painful neck swelling for 5 weeks and a 40‐pound weight loss. Neck examination showed tender, firm, matted and fixed lymph nodes in left and right cervical chains. Contrast computed tomography of the neck confirmed extensive necrotic lymphadenopathy. Excisional lymph node biopsy revealed granulomatous lymphadenitis with microabscesses. Staining for acid‐fast bacilli was positive; tissue culture confirmed the diagnosis of Mycobacterium tuberculosis. She was started on rifampin 600 mg, isoniazid 300 mg, pyrazinamide 1000 mg, and ethambutol 800 mg; ethambutol was discontinued when the mycobacteria proved susceptible to isoniazid. Pyrazinamide was discontinued after 2 months and isoniazid and rifampin continued to complete 6 months of therapy.0, 0

Tuberculous lymphadenitisscrofula when occurring in the cervical regionis a common cause of extrapulmonary tuberculosis, especially in developing countries. In developed countries, tuberculous lymphadenitis occurs mainly in immigrants but can also arise in travelers to endemic areas and immunodeficient persons.1 Painless lymphadenopathy of the superficial lymph nodes is typical. Fine needle aspiration with microscopy, culture and polymerase chain reaction (PCR) may be used as an initial diagnostic tool.2 However, if the results are negative despite high clinical suspicion, excisional biopsy of the lymph node may be necessary. Various nontuberculous mycobacteria also cause adenitis, so definitive speciation is needed to determine optimal therapy.3

Recommended initial treatment for M. tuberculosis includes isoniazid, rifampin, pyrazinamide and ethambutol or streptomycin, based on local resistance patterns. The fourth drug is withdrawn when the isolate is found susceptible to the first 3 drugs; isoniazid and rifampin are used for a total of 6 months of treatment, while pyrazinamide is withdrawn after 2 months.4 Twice‐weekly directly observed therapy (DOT) can be as effective as daily therapy supervised weekly.5 DOT maximizes compliance and reduces treatment failure.

Figure 1

Figure 2

As resident physicians, we find ourselves at the forefronts of medicine. While tackling the complexities of clinical medicine, all too soon we become entrenched in the bureaucracy of the medical system. We have come to accept realism and practicality over idealism. Shackled by the inefficiencies of a medical system in dire need of reform, we trek forward in hopes of a new future.

I remember the day quite clearly. My 30‐hour shift was nearing an end. It had not been a rough night, but covering both telemetry and coronary care unit services does take a toll on one's physical and mental well‐being. We were ready to discharge Mr. H, an indigent man recovering from a heart attack. I entered his room. He was all dressed. I explained all the important details and handed him a prescription. These medications are very important. Here is a taxi voucher to the pharmacy to get these medications.

He smiled, Thank you very much. You guys saved my life and I really appreciate it.

You are very welcome, I replied. In addition to the medications, it is very important that you follow up with your doctor within 1 to 2 weeks. Do you have any questions?

No problem. You guys saved my life. Thank you so much!

His eyes glistened with tears as he stared right through me, not hearing anything I had said. Mr. H, could you please wait a few more minutes? I will help you schedule a clinic visit.

I quickly located his clinic number at the county hospital. After 3 attempts, I reached a live person. I explained my situation: I am a doctor at a local hospital trying to help a patient setup a clinic appointment. She rattled off a slurry of information, clearly from a script, and left me with 5 different numbers to choose from (the main clinic line, an alternative number, a third number for the urgent advice nurse, one for new patients, and another for subspecialists). The main line was busy, and despite 3 additional attempts, this avenue ended in a recording advising me to leave a message. Unfortunately, my patient lived in an single resident occupancy (SRO) and did not have a phone, and thus leaving a message would not be helpful. I tried another number and the operator informed me that Mr. H was not in the system, and therefore, he would transfer me to the new patients department. The phone rang 5 times before the voicemail message answered, instructing me to leave a detailed message with my contact information. I hung up, and called the original operator back.

I'm sorry. Only the new patient department can schedule appointments for new patients. Here is the direct number. They must be busy. You can try again.

After several attempts, someone answered. My system shows that Mr. H has been assigned to the 7th Street Clinic. He's been seen there before, so you need to contact the main clinic line. I can only make appointments for new patients. She transferred me back to the first operator.

Oh yes, it looks like Mr. H is assigned to 7th Street Clinic. Please hold while I try to locate the next available appointment. He returns 3 minutes later. The next appointment I have is 4 months from now9:30 _AM or 3:30 _PM?

Is there anyway to schedule an earlier appointment? Mr. H has just recovered from a heart attack and needs to be seen sooner.

I'm sorry. This is the earliest appointment available. If he needs urgent care, he can go to the urgent care clinic Monday through Friday 7:30 _AM to 6:00 _PM. I persisted and he gave me the clinic's direct line. Sometimes, they may have earlier openings.

I called the 7th Street Clinic. The receptionist informed me that Mr. H was not in the system and she could not help me. I would need to speak with the new patient department. I just spoke with that department, and they were not able to schedule an appointment because their records show he has been seen at 7th Street Clinic in the past.

I'm sorry. He is not in our system. I can't schedule any appointments for him.

Again I went through all the previous numbers and spoke with the same 4 or 5 people, transferring me back and forth through this ridiculous maze, dodging voicemails and busy tones. One hour after I first began this endeavor, I finally succeeded in securing an appointment for Mr. H within 2 weeks as a new patient at a new clinic.

Why had this task been so difficult? Mr. H is the type of individual most at risk of falling through the cracks. Why is it that these individuals, homeless and indigent patients that lack social support, suffering already from countless barriers to health care access and resources because of their socioeconomic status continue to face such a horridly complex system of inefficiency and bureaucracy when trying to make a simple clinic appointment? How difficult and frustrating it was for me to accomplish this taskhow could I expect my patient living in an SRO without a phone to succeed?

Identifying barriers to health care access is the first step in addressing these issues. Previous studies have demonstrated that the majority of barriers to adequate follow‐up after a hospital visit occur among minority groups.17 Lower socioeconomic status is often associated with financial limitations from inability to take time off from work. Among immigrant cohorts, language and cultural barriers also play an important role in affecting follow‐up care. In addition to suboptimal follow‐up care, these barriers often lead to increased patient morbidity and increased rates of hospital readmission. For example, studies have reported that certain minority groups were more likely to receive no pain medication after bone fractures and were less likely to receive adequate analgesia for cancer‐related pain.1, 2, 7 In an attempt to address these issues, several studies have reported on multidisciplinary discharge planning interventions.810 One program in particular involved emergency departments providing free medication, transportation vouchers to and from the patient's primary care clinic, and telephone reminders to schedule follow‐up appointments.10 The implementation of these programs translated into improved primary care follow up, decreased hospital readmission rates, and decreased costs.

It is clear that our current health care system is wrought with inefficiencies that pose significant barriers to access by certain cohorts. The fact that minority groups and cohorts of the lowest socioeconomic status suffer most from these obstacles is concerning. Studies have shown that comprehensive programs to address these barriers including greater access to language interpreters and implementing a multidisciplinary approach to discharge planning improve patient outcomes. As we move forward in the ever‐evolving US medical system, there needs to be greater emphasis on preventative care. Education and resources to improve access to primary care physicians through identification of barriers and developing programs to address these issues is only the first step.

As resident physicians, we find ourselves at the forefronts of medicine. While tackling the complexities of clinical medicine, all too soon we become entrenched in the bureaucracy of the medical system. We have come to accept realism and practicality over idealism. Shackled by the inefficiencies of a medical system in dire need of reform, we trek forward in hopes of a new future.

I remember the day quite clearly. My 30‐hour shift was nearing an end. It had not been a rough night, but covering both telemetry and coronary care unit services does take a toll on one's physical and mental well‐being. We were ready to discharge Mr. H, an indigent man recovering from a heart attack. I entered his room. He was all dressed. I explained all the important details and handed him a prescription. These medications are very important. Here is a taxi voucher to the pharmacy to get these medications.

He smiled, Thank you very much. You guys saved my life and I really appreciate it.

You are very welcome, I replied. In addition to the medications, it is very important that you follow up with your doctor within 1 to 2 weeks. Do you have any questions?

No problem. You guys saved my life. Thank you so much!

His eyes glistened with tears as he stared right through me, not hearing anything I had said. Mr. H, could you please wait a few more minutes? I will help you schedule a clinic visit.

I quickly located his clinic number at the county hospital. After 3 attempts, I reached a live person. I explained my situation: I am a doctor at a local hospital trying to help a patient setup a clinic appointment. She rattled off a slurry of information, clearly from a script, and left me with 5 different numbers to choose from (the main clinic line, an alternative number, a third number for the urgent advice nurse, one for new patients, and another for subspecialists). The main line was busy, and despite 3 additional attempts, this avenue ended in a recording advising me to leave a message. Unfortunately, my patient lived in an single resident occupancy (SRO) and did not have a phone, and thus leaving a message would not be helpful. I tried another number and the operator informed me that Mr. H was not in the system, and therefore, he would transfer me to the new patients department. The phone rang 5 times before the voicemail message answered, instructing me to leave a detailed message with my contact information. I hung up, and called the original operator back.

I'm sorry. Only the new patient department can schedule appointments for new patients. Here is the direct number. They must be busy. You can try again.

After several attempts, someone answered. My system shows that Mr. H has been assigned to the 7th Street Clinic. He's been seen there before, so you need to contact the main clinic line. I can only make appointments for new patients. She transferred me back to the first operator.

Oh yes, it looks like Mr. H is assigned to 7th Street Clinic. Please hold while I try to locate the next available appointment. He returns 3 minutes later. The next appointment I have is 4 months from now9:30 _AM or 3:30 _PM?

Is there anyway to schedule an earlier appointment? Mr. H has just recovered from a heart attack and needs to be seen sooner.

I'm sorry. This is the earliest appointment available. If he needs urgent care, he can go to the urgent care clinic Monday through Friday 7:30 _AM to 6:00 _PM. I persisted and he gave me the clinic's direct line. Sometimes, they may have earlier openings.

I called the 7th Street Clinic. The receptionist informed me that Mr. H was not in the system and she could not help me. I would need to speak with the new patient department. I just spoke with that department, and they were not able to schedule an appointment because their records show he has been seen at 7th Street Clinic in the past.

I'm sorry. He is not in our system. I can't schedule any appointments for him.

Again I went through all the previous numbers and spoke with the same 4 or 5 people, transferring me back and forth through this ridiculous maze, dodging voicemails and busy tones. One hour after I first began this endeavor, I finally succeeded in securing an appointment for Mr. H within 2 weeks as a new patient at a new clinic.

Why had this task been so difficult? Mr. H is the type of individual most at risk of falling through the cracks. Why is it that these individuals, homeless and indigent patients that lack social support, suffering already from countless barriers to health care access and resources because of their socioeconomic status continue to face such a horridly complex system of inefficiency and bureaucracy when trying to make a simple clinic appointment? How difficult and frustrating it was for me to accomplish this taskhow could I expect my patient living in an SRO without a phone to succeed?

Identifying barriers to health care access is the first step in addressing these issues. Previous studies have demonstrated that the majority of barriers to adequate follow‐up after a hospital visit occur among minority groups.17 Lower socioeconomic status is often associated with financial limitations from inability to take time off from work. Among immigrant cohorts, language and cultural barriers also play an important role in affecting follow‐up care. In addition to suboptimal follow‐up care, these barriers often lead to increased patient morbidity and increased rates of hospital readmission. For example, studies have reported that certain minority groups were more likely to receive no pain medication after bone fractures and were less likely to receive adequate analgesia for cancer‐related pain.1, 2, 7 In an attempt to address these issues, several studies have reported on multidisciplinary discharge planning interventions.810 One program in particular involved emergency departments providing free medication, transportation vouchers to and from the patient's primary care clinic, and telephone reminders to schedule follow‐up appointments.10 The implementation of these programs translated into improved primary care follow up, decreased hospital readmission rates, and decreased costs.

It is clear that our current health care system is wrought with inefficiencies that pose significant barriers to access by certain cohorts. The fact that minority groups and cohorts of the lowest socioeconomic status suffer most from these obstacles is concerning. Studies have shown that comprehensive programs to address these barriers including greater access to language interpreters and implementing a multidisciplinary approach to discharge planning improve patient outcomes. As we move forward in the ever‐evolving US medical system, there needs to be greater emphasis on preventative care. Education and resources to improve access to primary care physicians through identification of barriers and developing programs to address these issues is only the first step.

As resident physicians, we find ourselves at the forefronts of medicine. While tackling the complexities of clinical medicine, all too soon we become entrenched in the bureaucracy of the medical system. We have come to accept realism and practicality over idealism. Shackled by the inefficiencies of a medical system in dire need of reform, we trek forward in hopes of a new future.

I remember the day quite clearly. My 30‐hour shift was nearing an end. It had not been a rough night, but covering both telemetry and coronary care unit services does take a toll on one's physical and mental well‐being. We were ready to discharge Mr. H, an indigent man recovering from a heart attack. I entered his room. He was all dressed. I explained all the important details and handed him a prescription. These medications are very important. Here is a taxi voucher to the pharmacy to get these medications.

He smiled, Thank you very much. You guys saved my life and I really appreciate it.

You are very welcome, I replied. In addition to the medications, it is very important that you follow up with your doctor within 1 to 2 weeks. Do you have any questions?

No problem. You guys saved my life. Thank you so much!

His eyes glistened with tears as he stared right through me, not hearing anything I had said. Mr. H, could you please wait a few more minutes? I will help you schedule a clinic visit.

I quickly located his clinic number at the county hospital. After 3 attempts, I reached a live person. I explained my situation: I am a doctor at a local hospital trying to help a patient setup a clinic appointment. She rattled off a slurry of information, clearly from a script, and left me with 5 different numbers to choose from (the main clinic line, an alternative number, a third number for the urgent advice nurse, one for new patients, and another for subspecialists). The main line was busy, and despite 3 additional attempts, this avenue ended in a recording advising me to leave a message. Unfortunately, my patient lived in an single resident occupancy (SRO) and did not have a phone, and thus leaving a message would not be helpful. I tried another number and the operator informed me that Mr. H was not in the system, and therefore, he would transfer me to the new patients department. The phone rang 5 times before the voicemail message answered, instructing me to leave a detailed message with my contact information. I hung up, and called the original operator back.

I'm sorry. Only the new patient department can schedule appointments for new patients. Here is the direct number. They must be busy. You can try again.

After several attempts, someone answered. My system shows that Mr. H has been assigned to the 7th Street Clinic. He's been seen there before, so you need to contact the main clinic line. I can only make appointments for new patients. She transferred me back to the first operator.

Oh yes, it looks like Mr. H is assigned to 7th Street Clinic. Please hold while I try to locate the next available appointment. He returns 3 minutes later. The next appointment I have is 4 months from now9:30 _AM or 3:30 _PM?

Is there anyway to schedule an earlier appointment? Mr. H has just recovered from a heart attack and needs to be seen sooner.

I'm sorry. This is the earliest appointment available. If he needs urgent care, he can go to the urgent care clinic Monday through Friday 7:30 _AM to 6:00 _PM. I persisted and he gave me the clinic's direct line. Sometimes, they may have earlier openings.

I called the 7th Street Clinic. The receptionist informed me that Mr. H was not in the system and she could not help me. I would need to speak with the new patient department. I just spoke with that department, and they were not able to schedule an appointment because their records show he has been seen at 7th Street Clinic in the past.

I'm sorry. He is not in our system. I can't schedule any appointments for him.

Again I went through all the previous numbers and spoke with the same 4 or 5 people, transferring me back and forth through this ridiculous maze, dodging voicemails and busy tones. One hour after I first began this endeavor, I finally succeeded in securing an appointment for Mr. H within 2 weeks as a new patient at a new clinic.

Why had this task been so difficult? Mr. H is the type of individual most at risk of falling through the cracks. Why is it that these individuals, homeless and indigent patients that lack social support, suffering already from countless barriers to health care access and resources because of their socioeconomic status continue to face such a horridly complex system of inefficiency and bureaucracy when trying to make a simple clinic appointment? How difficult and frustrating it was for me to accomplish this taskhow could I expect my patient living in an SRO without a phone to succeed?

Identifying barriers to health care access is the first step in addressing these issues. Previous studies have demonstrated that the majority of barriers to adequate follow‐up after a hospital visit occur among minority groups.17 Lower socioeconomic status is often associated with financial limitations from inability to take time off from work. Among immigrant cohorts, language and cultural barriers also play an important role in affecting follow‐up care. In addition to suboptimal follow‐up care, these barriers often lead to increased patient morbidity and increased rates of hospital readmission. For example, studies have reported that certain minority groups were more likely to receive no pain medication after bone fractures and were less likely to receive adequate analgesia for cancer‐related pain.1, 2, 7 In an attempt to address these issues, several studies have reported on multidisciplinary discharge planning interventions.810 One program in particular involved emergency departments providing free medication, transportation vouchers to and from the patient's primary care clinic, and telephone reminders to schedule follow‐up appointments.10 The implementation of these programs translated into improved primary care follow up, decreased hospital readmission rates, and decreased costs.

It is clear that our current health care system is wrought with inefficiencies that pose significant barriers to access by certain cohorts. The fact that minority groups and cohorts of the lowest socioeconomic status suffer most from these obstacles is concerning. Studies have shown that comprehensive programs to address these barriers including greater access to language interpreters and implementing a multidisciplinary approach to discharge planning improve patient outcomes. As we move forward in the ever‐evolving US medical system, there needs to be greater emphasis on preventative care. Education and resources to improve access to primary care physicians through identification of barriers and developing programs to address these issues is only the first step.

If you wish to receive credit for this activity, which begins on the next page, please refer to the website: www.blackwellpublishing.com/cme.

Accreditation and Designation Statement

Blackwell Futura Media Services designates this educational activity for a 1 AMA PRA Category 1 Credit. Physicians should only claim credit commensurate with the extent of their participation in the activity.

Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians.

Educational Objectives

Continuous participation in the Journal of Hospital Medicine CME program will enable learners to be better able to:

• Interpret clinical guidelines and their applications for higher quality and more efficient care for all hospitalized patients.

• Describe the standard of care for common illnesses and conditions treated in the hospital; such as pneumonia, COPD exacerbation, acute coronary syndrome, HF exacerbation, glycemic control, venous thromboembolic disease, stroke, etc.

• Discuss evidence‐based recommendations involving transitions of care, including the hospital discharge process.

• Gain insights into the roles of hospitalists as medical educators, researchers, medical ethicists, palliative care providers, and hospital‐based geriatricians.

• Incorporate best practices for hospitalist administration, including quality improvement, patient safety, practice management, leadership, and demonstrating hospitalist value.

• Identify evidence‐based best practices and trends for both adult and pediatric hospital medicine.

Instructions on Receiving Credit

For information on applicability and acceptance of continuing medical education credit for this activity, please consult your professional licensing board.

This activity is designed to be completed within the time designated on the title page; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period that is noted on the title page.

Follow these steps to earn credit:

• Log on to www.blackwellpublishing.com/cme.

• Read the target audience, learning objectives, and author disclosures.

• Read the article in print or online format.

• Reflect on the article.

• Access the CME Exam, and choose the best answer to each question.

• Complete the required evaluation component of the activity.

If you wish to receive credit for this activity, which begins on the next page, please refer to the website: www.blackwellpublishing.com/cme.

Accreditation and Designation Statement

Blackwell Futura Media Services designates this educational activity for a 1 AMA PRA Category 1 Credit. Physicians should only claim credit commensurate with the extent of their participation in the activity.

Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians.

Educational Objectives

Continuous participation in the Journal of Hospital Medicine CME program will enable learners to be better able to:

• Interpret clinical guidelines and their applications for higher quality and more efficient care for all hospitalized patients.

• Describe the standard of care for common illnesses and conditions treated in the hospital; such as pneumonia, COPD exacerbation, acute coronary syndrome, HF exacerbation, glycemic control, venous thromboembolic disease, stroke, etc.

• Discuss evidence‐based recommendations involving transitions of care, including the hospital discharge process.

• Gain insights into the roles of hospitalists as medical educators, researchers, medical ethicists, palliative care providers, and hospital‐based geriatricians.

• Incorporate best practices for hospitalist administration, including quality improvement, patient safety, practice management, leadership, and demonstrating hospitalist value.

• Identify evidence‐based best practices and trends for both adult and pediatric hospital medicine.

Instructions on Receiving Credit

For information on applicability and acceptance of continuing medical education credit for this activity, please consult your professional licensing board.

This activity is designed to be completed within the time designated on the title page; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period that is noted on the title page.

Follow these steps to earn credit:

• Log on to www.blackwellpublishing.com/cme.

• Read the target audience, learning objectives, and author disclosures.

• Read the article in print or online format.

• Reflect on the article.

• Access the CME Exam, and choose the best answer to each question.

• Complete the required evaluation component of the activity.

If you wish to receive credit for this activity, which begins on the next page, please refer to the website: www.blackwellpublishing.com/cme.

Accreditation and Designation Statement

Blackwell Futura Media Services designates this educational activity for a 1 AMA PRA Category 1 Credit. Physicians should only claim credit commensurate with the extent of their participation in the activity.

Blackwell Futura Media Services is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians.

Educational Objectives

Continuous participation in the Journal of Hospital Medicine CME program will enable learners to be better able to:

• Interpret clinical guidelines and their applications for higher quality and more efficient care for all hospitalized patients.

• Describe the standard of care for common illnesses and conditions treated in the hospital; such as pneumonia, COPD exacerbation, acute coronary syndrome, HF exacerbation, glycemic control, venous thromboembolic disease, stroke, etc.

• Discuss evidence‐based recommendations involving transitions of care, including the hospital discharge process.

• Gain insights into the roles of hospitalists as medical educators, researchers, medical ethicists, palliative care providers, and hospital‐based geriatricians.

• Incorporate best practices for hospitalist administration, including quality improvement, patient safety, practice management, leadership, and demonstrating hospitalist value.

• Identify evidence‐based best practices and trends for both adult and pediatric hospital medicine.

Instructions on Receiving Credit

For information on applicability and acceptance of continuing medical education credit for this activity, please consult your professional licensing board.

This activity is designed to be completed within the time designated on the title page; physicians should claim only those credits that reflect the time actually spent in the activity. To successfully earn credit, participants must complete the activity during the valid credit period that is noted on the title page.

Follow these steps to earn credit:

• Log on to www.blackwellpublishing.com/cme.

• Read the target audience, learning objectives, and author disclosures.

• Read the article in print or online format.

• Reflect on the article.

• Access the CME Exam, and choose the best answer to each question.

• Complete the required evaluation component of the activity.

A 46‐year‐old man with a history of heavy marijuana use for 30 years presented to the emergency room with 1 month of progressively worsening distal extremity pain and numbness that had developed into necrosis at the tips of his fingers (Figure 1A, arrows). The patient did not have other systemic symptoms. Transthoracic echocardiogram was normal. An evaluation for hypercoagulable disorders, blood cultures, C‐ANCA, cryoglobulins, Hepatitis C serologies, and human immunodeficiency virus (HIV) serologies were all negative. Erythrocyte sedimentation rate was within normal limits. An arteriogram of the hands showed distal subsegmental obliteration of digital arteries (Figure 1B, arrows), which was consistent with thromboangiitis obliterans (Buerger's Disease). While the patient rarely consumed tobacco, he reported smoking up to 10 marijuana cigarettes daily and a diagnosis of cannabis arteritis was made. The patient was encouraged to discontinue smoking and at follow‐up he had marked improvement in symptoms with abstinence from marijuana.

Figure 1

A 46‐year‐old man with a history of heavy marijuana use for 30 years presented to the emergency room with 1 month of progressively worsening distal extremity pain and numbness that had developed into necrosis at the tips of his fingers (Figure 1A, arrows). The patient did not have other systemic symptoms. Transthoracic echocardiogram was normal. An evaluation for hypercoagulable disorders, blood cultures, C‐ANCA, cryoglobulins, Hepatitis C serologies, and human immunodeficiency virus (HIV) serologies were all negative. Erythrocyte sedimentation rate was within normal limits. An arteriogram of the hands showed distal subsegmental obliteration of digital arteries (Figure 1B, arrows), which was consistent with thromboangiitis obliterans (Buerger's Disease). While the patient rarely consumed tobacco, he reported smoking up to 10 marijuana cigarettes daily and a diagnosis of cannabis arteritis was made. The patient was encouraged to discontinue smoking and at follow‐up he had marked improvement in symptoms with abstinence from marijuana.

Figure 1

A 46‐year‐old man with a history of heavy marijuana use for 30 years presented to the emergency room with 1 month of progressively worsening distal extremity pain and numbness that had developed into necrosis at the tips of his fingers (Figure 1A, arrows). The patient did not have other systemic symptoms. Transthoracic echocardiogram was normal. An evaluation for hypercoagulable disorders, blood cultures, C‐ANCA, cryoglobulins, Hepatitis C serologies, and human immunodeficiency virus (HIV) serologies were all negative. Erythrocyte sedimentation rate was within normal limits. An arteriogram of the hands showed distal subsegmental obliteration of digital arteries (Figure 1B, arrows), which was consistent with thromboangiitis obliterans (Buerger's Disease). While the patient rarely consumed tobacco, he reported smoking up to 10 marijuana cigarettes daily and a diagnosis of cannabis arteritis was made. The patient was encouraged to discontinue smoking and at follow‐up he had marked improvement in symptoms with abstinence from marijuana.

Figure 1

Hospitalists are common in North America.1, 2 Hospitalists have been associated with a range of beneficial outcomes including decreased length of stay.3, 4 A primary concern of the hospitalist model is its potential detrimental effect on continuity of care5 partly because patients are often not seen by their hospitalists after discharge.

Continuity of care6 is primarily composed of provider continuity (an ongoing relationship between a patient and a particular provider over time) and information continuity (availability of data from prior events for subsequent patient encounters).6 The association between continuity of care and patient outcomes has been quantified in many studies.720 However, the relationship of continuity and outcomes is especially relevant after discharge from the hospital since this is a time when patients have a high risk of poor patient outcomes21 and poor provider22 and information continuity.2325

The association between continuity and outcomes after hospital discharge has been directly quantified in 2 studies. One found that patients seen by a physician who treated them in the hospital had a significant adjusted relative risk reduction in 30‐day death or readmission of 5% and 3%, respectively.22 The other study found that patients discharged from a general medicine ward were less likely to be readmitted if they were seen by physicians who had access to their discharge summary.23 However, neither of these studies concurrently measured the influence of provider and information continuity on patient outcomes.

Determining whether and how continuity of care influences patient outcomes after hospital discharge is essential to improve health care in an evidence‐based fashion. In addition, the influence that hospital physician follow‐up has on patient outcomes can best be determined by measuring provider and information continuity in patients after hospital discharge. This study sought to measure the independent association of several provider and information continuity measures on death or urgent readmission after hospital discharge.

Methods

Results

Between October 2002 and July 2006, we enrolled 5035 patients from 11 hospitals (Figure 1). Of the 5035 patients, 274 (5.4%) had no follow up interview with study personnel. A total of 885 (17.6%) had fewer than 2 post discharge physician visits and were not included in the continuity analyses. This left 3876 patients for this analysis (77.0% of the original cohort), of which 3727 had complete follow up (96.1% of the study cohort). A total of 531 patients (10.6% of the original cohort) had incomplete follow‐up because: 342 (6.8%) were lost to follow‐up; 172 (3.4%) refused participation; and 24 (0.5%) were transferred into a nursing home during the first month of observation.

Figure 1

Figure 2

The 3876 study patients are described in Table 1. Overall, these people had a mean age of 62 and most commonly had no physical limitations. Almost a third of patients had been admitted to the hospital in the previous 6 months. A total of 7.6% of patients had no regular prehospital physician while 5.8% had more than one regular prehospital physician. Patients were evenly split between acute and elective admissions and 12% had a complication during their admission. They were discharged after a median of 4 days on a median of 4 medications.

Patients were observed in the study for a median of 175 days (interquartile range [IQR] 175‐178). During this time they had a median of 4 physician visits (IQR 3‐6). The first postdischarge physician visit occurred a median of 10 days (IQR 6‐18) after discharge from hospital.

Discussion

This large, prospective cohort study measured the independent association of several provider and information continuity measures with important outcomes in patients discharged from hospital. After adjusting for potential confounders, we found that increased continuity with physicians who regularly cared for the patient prior to the admission was significantly and independently associated with a decreased risk of urgent readmission. Our data suggest that continuity with the hospital physician did not independently influence the risk of patient death or urgent readmission after discharge.

Although hospital physician continuity did not significantly change patient outcomes, we found that follow‐up with a physician who regularly treated the patient prior to their admission was associated with a significantly decreased risk of urgent readmission. This could reflect the important role that a patient's regular physician plays in their health care. Other studies have shown a positive association between continuity with a regular physician and improved outcomes including decreased emergency room utilization7, 8 and decreased hospitalization.10, 11

We were somewhat disappointed that information continuity was not independently associated with improved patient outcomes. Information continuity is likely more amenable to modification than is provider continuity. Of course, our study findings do not mean that information continuity does not improve patient outcomes, as in other studies.23, 33 Instead, our results could reflect that we solely measured the availability of information to physicians. Future studies that measure the quality, relevance, and actual utilization of patient information will be better able to discern the influence of information continuity on patient outcomes.

We believe that our study was methodologically strong and unique. We captured both provider and information continuity in a large group of representative patients using a broad range of measures that captured continuity's diverse components including both provider and information continuity. The continuity measures were expressed and properly analyzed as time‐dependent variables in a multivariate model.34 Our analysis controlled for important potential confounders. Our follow‐up and data collection was rigorous with 96.1% of our study group having complete follow‐up. Finally, the analysis used multiple imputation to appropriately handle missing data in the one incomplete variable (post‐discharge information continuity).3537

Several limitations of our study should be kept in mind. We are uncertain how our results might generalize to patients discharged from obstetrical or psychiatric services or people in other health systems. Our analysis had to exclude patients with less than two physician visits after discharge since this was the minimum required to calculate postdischarge physician and information continuity. Data collection for postdischarge information continuity was incomplete with data missing for 19.0% of all 15 401 visits in the original cohort.38 However, a response rate of 81.0% is very good39 when compared to other survey‐based studies40 and we accounted for the missing data using multiple imputation methods. The primary outcomes of our studytime to death or urgent readmissionmay be relatively insensitive to modification of quality of care, which is presumably improved by increased continuity.41 For example, Clarke found that the majority of readmissions in all patient groups were unavoidable with 94% of medical readmissions 1 month postdischarge judged to be unavoidable.42 Future studies regarding the effects of continuity could focus on its association with other outcomes that are more reflective of quality of care such as the risk of adverse events or medical error.21 Such outcomes would presumably be more sensitive to improved quality of care from increased continuity.

We believe that our study's major limitation was its inability to establish a causal association between continuity and patient outcomes. Our finding that increased consultant continuity was associated with an increased risk of poor outcomes highlights this concern. Presumably, patient follow‐up with a hospital consultant indicates a disease status with a high risk of bad patient outcomesa risk that is not entirely accounted for by the covariates used in this study. If we accept that unresolved confounding explains this association, the same could also apply to the association between preadmission physician continuity and improved outcomes. Perhaps patients who are doing well after discharge from hospital are able to return to their regular physician. Our analysis would therefore identify an association between increased preadmission physician continuity and improved patient outcomes. Analyses could also incorporate more discriminative measures of severity of hospital illness, such as those developed by Escobar et al.43 Since patients may experience health events after their discharge from hospital that could influence outcomes, recording these and expressing them in the study model as time‐dependent covariates will be important. Finally, similar to the classic study by Wasson et al.44 in 1984, a proper randomized trial that measures the effect of a continuity‐building intervention on both continuity of care and patient outcomes would help determine how continuity influences outcomes.

In conclusion, after discharge from hospital, increased continuity with physicians who routinely care for the patient is significantly and independently associated with a decreased risk of urgent readmission. Continuity with the hospital physician after discharge did not independently influence the risk of patient death or urgent readmission in our study. Further research is required to determine the causal association between preadmission physician continuity and improved outcomes. Until that time, clinicians should strive to optimize continuity with physicians their patients have seen prior to the hospitalization.

Hospitalists are common in North America.1, 2 Hospitalists have been associated with a range of beneficial outcomes including decreased length of stay.3, 4 A primary concern of the hospitalist model is its potential detrimental effect on continuity of care5 partly because patients are often not seen by their hospitalists after discharge.

Continuity of care6 is primarily composed of provider continuity (an ongoing relationship between a patient and a particular provider over time) and information continuity (availability of data from prior events for subsequent patient encounters).6 The association between continuity of care and patient outcomes has been quantified in many studies.720 However, the relationship of continuity and outcomes is especially relevant after discharge from the hospital since this is a time when patients have a high risk of poor patient outcomes21 and poor provider22 and information continuity.2325

The association between continuity and outcomes after hospital discharge has been directly quantified in 2 studies. One found that patients seen by a physician who treated them in the hospital had a significant adjusted relative risk reduction in 30‐day death or readmission of 5% and 3%, respectively.22 The other study found that patients discharged from a general medicine ward were less likely to be readmitted if they were seen by physicians who had access to their discharge summary.23 However, neither of these studies concurrently measured the influence of provider and information continuity on patient outcomes.

Determining whether and how continuity of care influences patient outcomes after hospital discharge is essential to improve health care in an evidence‐based fashion. In addition, the influence that hospital physician follow‐up has on patient outcomes can best be determined by measuring provider and information continuity in patients after hospital discharge. This study sought to measure the independent association of several provider and information continuity measures on death or urgent readmission after hospital discharge.

Methods

Results

Between October 2002 and July 2006, we enrolled 5035 patients from 11 hospitals (Figure 1). Of the 5035 patients, 274 (5.4%) had no follow up interview with study personnel. A total of 885 (17.6%) had fewer than 2 post discharge physician visits and were not included in the continuity analyses. This left 3876 patients for this analysis (77.0% of the original cohort), of which 3727 had complete follow up (96.1% of the study cohort). A total of 531 patients (10.6% of the original cohort) had incomplete follow‐up because: 342 (6.8%) were lost to follow‐up; 172 (3.4%) refused participation; and 24 (0.5%) were transferred into a nursing home during the first month of observation.

Figure 1

Figure 2

The 3876 study patients are described in Table 1. Overall, these people had a mean age of 62 and most commonly had no physical limitations. Almost a third of patients had been admitted to the hospital in the previous 6 months. A total of 7.6% of patients had no regular prehospital physician while 5.8% had more than one regular prehospital physician. Patients were evenly split between acute and elective admissions and 12% had a complication during their admission. They were discharged after a median of 4 days on a median of 4 medications.

Patients were observed in the study for a median of 175 days (interquartile range [IQR] 175‐178). During this time they had a median of 4 physician visits (IQR 3‐6). The first postdischarge physician visit occurred a median of 10 days (IQR 6‐18) after discharge from hospital.

Discussion

This large, prospective cohort study measured the independent association of several provider and information continuity measures with important outcomes in patients discharged from hospital. After adjusting for potential confounders, we found that increased continuity with physicians who regularly cared for the patient prior to the admission was significantly and independently associated with a decreased risk of urgent readmission. Our data suggest that continuity with the hospital physician did not independently influence the risk of patient death or urgent readmission after discharge.

Although hospital physician continuity did not significantly change patient outcomes, we found that follow‐up with a physician who regularly treated the patient prior to their admission was associated with a significantly decreased risk of urgent readmission. This could reflect the important role that a patient's regular physician plays in their health care. Other studies have shown a positive association between continuity with a regular physician and improved outcomes including decreased emergency room utilization7, 8 and decreased hospitalization.10, 11

We were somewhat disappointed that information continuity was not independently associated with improved patient outcomes. Information continuity is likely more amenable to modification than is provider continuity. Of course, our study findings do not mean that information continuity does not improve patient outcomes, as in other studies.23, 33 Instead, our results could reflect that we solely measured the availability of information to physicians. Future studies that measure the quality, relevance, and actual utilization of patient information will be better able to discern the influence of information continuity on patient outcomes.

We believe that our study was methodologically strong and unique. We captured both provider and information continuity in a large group of representative patients using a broad range of measures that captured continuity's diverse components including both provider and information continuity. The continuity measures were expressed and properly analyzed as time‐dependent variables in a multivariate model.34 Our analysis controlled for important potential confounders. Our follow‐up and data collection was rigorous with 96.1% of our study group having complete follow‐up. Finally, the analysis used multiple imputation to appropriately handle missing data in the one incomplete variable (post‐discharge information continuity).3537

Several limitations of our study should be kept in mind. We are uncertain how our results might generalize to patients discharged from obstetrical or psychiatric services or people in other health systems. Our analysis had to exclude patients with less than two physician visits after discharge since this was the minimum required to calculate postdischarge physician and information continuity. Data collection for postdischarge information continuity was incomplete with data missing for 19.0% of all 15 401 visits in the original cohort.38 However, a response rate of 81.0% is very good39 when compared to other survey‐based studies40 and we accounted for the missing data using multiple imputation methods. The primary outcomes of our studytime to death or urgent readmissionmay be relatively insensitive to modification of quality of care, which is presumably improved by increased continuity.41 For example, Clarke found that the majority of readmissions in all patient groups were unavoidable with 94% of medical readmissions 1 month postdischarge judged to be unavoidable.42 Future studies regarding the effects of continuity could focus on its association with other outcomes that are more reflective of quality of care such as the risk of adverse events or medical error.21 Such outcomes would presumably be more sensitive to improved quality of care from increased continuity.

We believe that our study's major limitation was its inability to establish a causal association between continuity and patient outcomes. Our finding that increased consultant continuity was associated with an increased risk of poor outcomes highlights this concern. Presumably, patient follow‐up with a hospital consultant indicates a disease status with a high risk of bad patient outcomesa risk that is not entirely accounted for by the covariates used in this study. If we accept that unresolved confounding explains this association, the same could also apply to the association between preadmission physician continuity and improved outcomes. Perhaps patients who are doing well after discharge from hospital are able to return to their regular physician. Our analysis would therefore identify an association between increased preadmission physician continuity and improved patient outcomes. Analyses could also incorporate more discriminative measures of severity of hospital illness, such as those developed by Escobar et al.43 Since patients may experience health events after their discharge from hospital that could influence outcomes, recording these and expressing them in the study model as time‐dependent covariates will be important. Finally, similar to the classic study by Wasson et al.44 in 1984, a proper randomized trial that measures the effect of a continuity‐building intervention on both continuity of care and patient outcomes would help determine how continuity influences outcomes.

In conclusion, after discharge from hospital, increased continuity with physicians who routinely care for the patient is significantly and independently associated with a decreased risk of urgent readmission. Continuity with the hospital physician after discharge did not independently influence the risk of patient death or urgent readmission in our study. Further research is required to determine the causal association between preadmission physician continuity and improved outcomes. Until that time, clinicians should strive to optimize continuity with physicians their patients have seen prior to the hospitalization.

Hospitalists are common in North America.1, 2 Hospitalists have been associated with a range of beneficial outcomes including decreased length of stay.3, 4 A primary concern of the hospitalist model is its potential detrimental effect on continuity of care5 partly because patients are often not seen by their hospitalists after discharge.

Continuity of care6 is primarily composed of provider continuity (an ongoing relationship between a patient and a particular provider over time) and information continuity (availability of data from prior events for subsequent patient encounters).6 The association between continuity of care and patient outcomes has been quantified in many studies.720 However, the relationship of continuity and outcomes is especially relevant after discharge from the hospital since this is a time when patients have a high risk of poor patient outcomes21 and poor provider22 and information continuity.2325

The association between continuity and outcomes after hospital discharge has been directly quantified in 2 studies. One found that patients seen by a physician who treated them in the hospital had a significant adjusted relative risk reduction in 30‐day death or readmission of 5% and 3%, respectively.22 The other study found that patients discharged from a general medicine ward were less likely to be readmitted if they were seen by physicians who had access to their discharge summary.23 However, neither of these studies concurrently measured the influence of provider and information continuity on patient outcomes.

Determining whether and how continuity of care influences patient outcomes after hospital discharge is essential to improve health care in an evidence‐based fashion. In addition, the influence that hospital physician follow‐up has on patient outcomes can best be determined by measuring provider and information continuity in patients after hospital discharge. This study sought to measure the independent association of several provider and information continuity measures on death or urgent readmission after hospital discharge.

Methods

Results

Between October 2002 and July 2006, we enrolled 5035 patients from 11 hospitals (Figure 1). Of the 5035 patients, 274 (5.4%) had no follow up interview with study personnel. A total of 885 (17.6%) had fewer than 2 post discharge physician visits and were not included in the continuity analyses. This left 3876 patients for this analysis (77.0% of the original cohort), of which 3727 had complete follow up (96.1% of the study cohort). A total of 531 patients (10.6% of the original cohort) had incomplete follow‐up because: 342 (6.8%) were lost to follow‐up; 172 (3.4%) refused participation; and 24 (0.5%) were transferred into a nursing home during the first month of observation.

Figure 1

Figure 2

The 3876 study patients are described in Table 1. Overall, these people had a mean age of 62 and most commonly had no physical limitations. Almost a third of patients had been admitted to the hospital in the previous 6 months. A total of 7.6% of patients had no regular prehospital physician while 5.8% had more than one regular prehospital physician. Patients were evenly split between acute and elective admissions and 12% had a complication during their admission. They were discharged after a median of 4 days on a median of 4 medications.

Patients were observed in the study for a median of 175 days (interquartile range [IQR] 175‐178). During this time they had a median of 4 physician visits (IQR 3‐6). The first postdischarge physician visit occurred a median of 10 days (IQR 6‐18) after discharge from hospital.

Discussion

This large, prospective cohort study measured the independent association of several provider and information continuity measures with important outcomes in patients discharged from hospital. After adjusting for potential confounders, we found that increased continuity with physicians who regularly cared for the patient prior to the admission was significantly and independently associated with a decreased risk of urgent readmission. Our data suggest that continuity with the hospital physician did not independently influence the risk of patient death or urgent readmission after discharge.

Although hospital physician continuity did not significantly change patient outcomes, we found that follow‐up with a physician who regularly treated the patient prior to their admission was associated with a significantly decreased risk of urgent readmission. This could reflect the important role that a patient's regular physician plays in their health care. Other studies have shown a positive association between continuity with a regular physician and improved outcomes including decreased emergency room utilization7, 8 and decreased hospitalization.10, 11

We were somewhat disappointed that information continuity was not independently associated with improved patient outcomes. Information continuity is likely more amenable to modification than is provider continuity. Of course, our study findings do not mean that information continuity does not improve patient outcomes, as in other studies.23, 33 Instead, our results could reflect that we solely measured the availability of information to physicians. Future studies that measure the quality, relevance, and actual utilization of patient information will be better able to discern the influence of information continuity on patient outcomes.

We believe that our study was methodologically strong and unique. We captured both provider and information continuity in a large group of representative patients using a broad range of measures that captured continuity's diverse components including both provider and information continuity. The continuity measures were expressed and properly analyzed as time‐dependent variables in a multivariate model.34 Our analysis controlled for important potential confounders. Our follow‐up and data collection was rigorous with 96.1% of our study group having complete follow‐up. Finally, the analysis used multiple imputation to appropriately handle missing data in the one incomplete variable (post‐discharge information continuity).3537

Several limitations of our study should be kept in mind. We are uncertain how our results might generalize to patients discharged from obstetrical or psychiatric services or people in other health systems. Our analysis had to exclude patients with less than two physician visits after discharge since this was the minimum required to calculate postdischarge physician and information continuity. Data collection for postdischarge information continuity was incomplete with data missing for 19.0% of all 15 401 visits in the original cohort.38 However, a response rate of 81.0% is very good39 when compared to other survey‐based studies40 and we accounted for the missing data using multiple imputation methods. The primary outcomes of our studytime to death or urgent readmissionmay be relatively insensitive to modification of quality of care, which is presumably improved by increased continuity.41 For example, Clarke found that the majority of readmissions in all patient groups were unavoidable with 94% of medical readmissions 1 month postdischarge judged to be unavoidable.42 Future studies regarding the effects of continuity could focus on its association with other outcomes that are more reflective of quality of care such as the risk of adverse events or medical error.21 Such outcomes would presumably be more sensitive to improved quality of care from increased continuity.

We believe that our study's major limitation was its inability to establish a causal association between continuity and patient outcomes. Our finding that increased consultant continuity was associated with an increased risk of poor outcomes highlights this concern. Presumably, patient follow‐up with a hospital consultant indicates a disease status with a high risk of bad patient outcomesa risk that is not entirely accounted for by the covariates used in this study. If we accept that unresolved confounding explains this association, the same could also apply to the association between preadmission physician continuity and improved outcomes. Perhaps patients who are doing well after discharge from hospital are able to return to their regular physician. Our analysis would therefore identify an association between increased preadmission physician continuity and improved patient outcomes. Analyses could also incorporate more discriminative measures of severity of hospital illness, such as those developed by Escobar et al.43 Since patients may experience health events after their discharge from hospital that could influence outcomes, recording these and expressing them in the study model as time‐dependent covariates will be important. Finally, similar to the classic study by Wasson et al.44 in 1984, a proper randomized trial that measures the effect of a continuity‐building intervention on both continuity of care and patient outcomes would help determine how continuity influences outcomes.

In conclusion, after discharge from hospital, increased continuity with physicians who routinely care for the patient is significantly and independently associated with a decreased risk of urgent readmission. Continuity with the hospital physician after discharge did not independently influence the risk of patient death or urgent readmission in our study. Further research is required to determine the causal association between preadmission physician continuity and improved outcomes. Until that time, clinicians should strive to optimize continuity with physicians their patients have seen prior to the hospitalization.