User login
I-MOVE: Inpatient Pre-Discharge Mobility Score As a Predictor of Post-Discharge Mortality
From the Mayo Clinic Center for Innovation (Dr. Romero-Brufau) Department of Medicine (Drs. Manning, Borrud, Keller, Kashiwagi, Huddleston, and Croghan) Department of Health Sciences Research (Mr. Cha), Mayo Clinic, Rochester, MN.
Abstract
- Objective: To determine whether a score of 8 or greater on the I-MOVE, a bedside instrument that evaluates the need for assistance in turning, sitting, standing, transferring from bed to a chair, and ambulating, predicts lower risk for 30-day readmission or mortality.
- Design: Retrospective cohort study of patients discharged from 2003 to 2011 from a referral hospital in Southeastern Minnesota. We used a convenience sample of 426 inpatients who had at least one documented calculation of the I-MOVE score performed as part of the clinical process during the study.
- Results: Overall 30-day mortality rate, readmission rate, and rate of the combined death/readmission outcome were 6.1% (26 patients), 15% (64 patients) and 19.7% (84 patients), respectively. After controlling for confounding variables, an I-MOVE score ≥ 8 was a significant predictive factor for 30-day mortality (OR = 0.136, P < 0.01) but not 30-day readmission (OR = 1.143, P = 0.62) or the combined outcome death/readmission (OR = 0.682, P = 0.13).
- Conclusion: The clinical information provided by a patient's I-MOVE score before discharge does not provide information about readmission risk but may offer incremental information about 30-day mortality risk.
Risk factors for hospital 30-day readmission have been studied by Hasan et al [1], van Walraven et al [2], Allaudeen et al [3], and more recently, Donze et al [4]. Risk factors found to be associated with readmission include race, length of stay, and number of hospitalizations in the last 12 months. Additionally, patients identified “feeling unprepared for discharge” and “difficulty performing activities of daily living” as top issues contributing to readmission. The Affordable Care Act established the Value-Based Purchasing (VBP) model for defined hospital illnesses such as acute myocardial infarction, heart failure, and community acquired pneumonia. This has focused more attention on post-discharge 30-day mortality and readmissions as publicly reported metrics that in part determine the Centers for Medicare and Medicaid Services care reimbursement rates [5].
In our hospital, over 400 inpatients have been evaluated since 2004 using the I-MOVE scoring system in the course of their usual care. I-MOVE was most commonly employed by geriatricians in the division of hospital internal medicine, who collectively endorsed the tool in their practice meetings, especially for elderly patients returning to home alone whose mobility independence was uncertain.
Although it was initially designed to help clinicians understand the mobility independence of a patient before discharge, it may provide incremental value discerning risk of 30-day readmission and/or death. We therefore hypothesized that an I-MOVE score of less than 8 (not being able to transfer from a bed to a chair without assistance) would be a significant predictor of 30-day readmission and/or death.
Methods
Study Design
We performed a retrospective cohort study using a convenience sample including the patients in which the I-MOVE score had been calculated as part of the clinical process of care.
Setting and Participants
Participants were any inpatients discharged from the general medicine unit at Mayo Clinic Rochester from January 2003 to May 2011 who had at least one documented calculation of the I-MOVE score performed as part of the clinical process. Patients in the general medicine unit are adults not requiring subspecialty cardiovascular or neurology, coronary care unit, surgical, psychiatry, or rehabilitation. Patients were excluded if there was missing key outcome information or if they died during the hospitalization. For patients with more than one I-MOVE assessment, only the one closest to discharge was used. Data were abstracted from the electronic medical records between July and August 2011.
Variables
Outcome variables were 30-day readmission, 30-day mortality, and the combined outcome of mortality or readmission. We used the last I-MOVE score as a dichotomous variable with a cut-off of 8, which corresponds to the ability to transfer from bed to a chair unaided, for predicting the 2 outcomes. Only readmissions to the study hospital were captured. Deaths were identified from the electronic medical record. Mayo Clinic patient records are updated monthly with external reports of confirmed, actuarial records of deaths reported from public databases.
To control for possible confounding variables, we included the following covariates: age, gender, race/ethnicity, dates of admission and discharge, insurance (Medicare, Medicaid, self-pay, or private), marital status (currently married/not currently married), length of hospital stay, emergent admission, number of hospital admissions in the last 12 months, number of visits to the emergency department in the last 6 months and Charlson Index. All variables were abstracted from the electronic medical record.
Sample
A search was performed in the electronic medical record to find clinical documents (admission notes, progress notes, and hospital summaries) that mentioned the term “I-MOVE.” Manual review of the records was performed to confirm inclusion criteria.
Statistical Analysis
Separate analyses were performed for the 2 outcomes considered. First, a univariate analysis was performed with all covariates for variable selection. Variables that were significantly predictive with P < 0.1 were included in the multivariate model. Variables included in the first run of the multivariate model were excluded from the final multivariate model if they were not independently significant with P < 0.05. The I-MOVE variable was then added to that model to check its predictive power beyond that of the included covariates.
Results
Patient Characteristics
For the final dataset of 426 patients, 30-day mortality rate, readmission rate, and rate of the combined death/readmission outcome were 6.1% (26 patients), 15% (64 patients) and 19.7% (84 patients), respectively. A total of 6 patients were readmitted and died within 30 days after the initial discharge. The number of patients that had an I-MOVE score greater than or equal to 8 was 232 (54.4%). Table 2 presents the mean, standard deviation, and median I-MOVE score by patient discharge destination. Patients discharged home had an average I-MOVE score of 11.98, versus 7.24 for patients discharge to a skilled nursing facility (P = 0.2).
Analysis
Table 3 presents the odds ratios and coefficient estimate of the models. In the univariate analysis, an I-MOVE score greater than or equal to 8 was significantly correlated with 30-day mortality (P < 0.001), and the combined outcome (P = 0.044) but not with 30-day readmission (P = 0.76). After controlling for confounding variables, I-MOVE greater than or equal to 8 was a significant predictor of 30-day mortality (P < 0.01) but not 30-day readmission (P = 0.75) 
Discussion
An I-MOVE score of less than 8 (inability to transfer from bed to a chair unassisted) is a statistically significant predictor of 30-day post-discharge mortality but not readmission or the combined outcome of death/readmission.
A recent review that evaluated published models that attempted to predict readmissions concluded that most current readmission risk prediction models designed for either comparative or clinical purposes perform poorly and that efforts are needed to improve their performance as use becomes more widespread [8]. Health care providers’ ability to predict which patients would be readmitted within 30 days was also shown by a recent study to be very poor, with C-statistics around 0.60 [9]. This inability of both experts and statistical methods to accurately predict readmissions may reflect some inherent randomness or unpredictability of readmissions, or the fact that a paradigm shift is still needed in the identification of the most important risk factors for readmissions. Along the same line, a recent evaluation of interventions aimed at reducing readmissions found that none of those identified in the literature managed to consistently reduce readmission rates long-term [10]. In addition, hospitals with greater adherence to recommended care processes did not achieve meaningfully better 30-day hospital readmission rates compared to those with lower levels of performance.
Conceptually, readmissions are an example of what is called “complexity science,” where many agents or factors (including the patient’s underlying illness, quality of care delivered, continuity and coordination of care, and resources available in the patient’s environment) and their interactions all play a role in the outcome [11,12]. Since I-MOVE primarily evaluates the physical capacity of the patient, and not any of the other variables that strongly affect readmission, it is perhaps not surprising that it did not predict readmission. It can be argued, on the other hand, that short term (30-day) mortality is more dependent on the patient’s physical and functional status [13] and so more likely to correlate with a measure such as I-MOVE. Inouye et al [13] found that pre-hospital, self-reported need for assistance in 7 basic “activities of daily living” (among which are transfers and ambulation) correlated with 90-day, and 2-year, post-hospital mortality.
The study has the advantage of a relatively large sample size, and the fact that the I-MOVE score was assessed before discharge eliminates the possibility of assessor bias. However, it has some limitations. We used a convenience sample, which may have introduced selection bias. Although we have no data on how providers selected patients for I-MOVE assessment, it would be reasonable to assume that patients were selected from among those whose activity level was, in terms of independence, doubtful or uncertain. That is, those who were not clearly vigorous (up and walking easily), nor clearly debilitated (in need of great assistance) may have been more likely to be assessed using I-MOVE. A more systematic selection of subjects might increase or decrease the predictive performance of the I-MOVE assessment. In addition, although we attempted to control for potential confounders, it is possible that additional confounders were left out of our analysis.
In summary, although the predictive performance of I-MOVE still needs to be confirmed by prospective studies with a comprehensive selection of subjects, the I-MOVE score at discharge appears to be associated with 30-day post-discharge mortality.
Acknowledgments: We thank the Department of Medicine’s clinical research office for their help in study design, data acquisition, and statistical analysis.
Corresponding author: Santiago Romero-Brufau, MD, Mayo Clinic Center for Innovation, 200 First St. SW, Rochester, MN 55905, [email protected].
Funding/support: This publication was supported by grant number UL1 TR000135 from the National Center for Advancing Translational Sciences (NCATS). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.
1. Hasan O, Meltzer DO, Shaykevich SA, et al. Hospital readmission in general medicine patients: a prediction model.
J Gen Intern Med 2010;25:211–9.
2. van Walraven C, Dhalla IA, Bell C, et al. Derivation and validation of an index to predict early death or unplanned readmission after discharge from hospital to the community. CMAJ 2010;182:551–7.
3. Allaudeen N, Vidyarthi A, Maselli J, Auerbach A. Redefining readmission risk factors for general medicine patients. J Hosp Med 2011;6:54–60.
4. Donze J, Aujesky D, Williams D, Schnipper JL. Potentially avoidable 30-day hospital readmissions in medical patients: derivation and validation of a prediction model. JAMA Intern Med 2013;173:632–8.
5. Kocher RP, Adashi EY. Hospital readmissions and the Affordable Care Act: paying for coordinated quality care. JAMA 2011;306:1794–5.
6. Manning DM, Keller AS, Frank DL. Home alone: assessing mobility independence before discharge. J Hosp Med 2009;4:252–4.
7. Cook DJ, Manning DM, Holland DE, et al. Patient engagement and reported outcomes in surgical recovery: effectiveness of an e-health platform. J Am Coll Surg 2013;217:648–55.
8. Kansagara D, Englander H, Salanitro A, et al. Risk prediction models for hospital readmission: a systematic review. JAMA 2011;306:1688–98.
9. Allaudeen N, Schnipper JL, Orav EJ, et al. Inability of providers to predict unplanned readmissions. J Gen Intern Med 2011;26:771–6.
10. Hansen LO, Young RS, Hinami K, et al. Interventions to reduce 30-day rehospitalization: a systematic review. Ann Intern Med 2011;155:520–8.
11. Marks E. Complexity science and the readmission dilemma. JAMA Intern Med 2013;173:629–31.
12. Lindquist LA, Baker DW. Understanding preventable hospital readmissions: masqueraders, markers, and true causal factors. J Hosp Med 2011;6:51–3.
13. Inouye SK, Peduzzi PN, Robison JT, et al. Importance of functional measures in predicting mortality among older hospitalized patients. JAMA 1998;279:1187–93.
From the Mayo Clinic Center for Innovation (Dr. Romero-Brufau) Department of Medicine (Drs. Manning, Borrud, Keller, Kashiwagi, Huddleston, and Croghan) Department of Health Sciences Research (Mr. Cha), Mayo Clinic, Rochester, MN.
Abstract
- Objective: To determine whether a score of 8 or greater on the I-MOVE, a bedside instrument that evaluates the need for assistance in turning, sitting, standing, transferring from bed to a chair, and ambulating, predicts lower risk for 30-day readmission or mortality.
- Design: Retrospective cohort study of patients discharged from 2003 to 2011 from a referral hospital in Southeastern Minnesota. We used a convenience sample of 426 inpatients who had at least one documented calculation of the I-MOVE score performed as part of the clinical process during the study.
- Results: Overall 30-day mortality rate, readmission rate, and rate of the combined death/readmission outcome were 6.1% (26 patients), 15% (64 patients) and 19.7% (84 patients), respectively. After controlling for confounding variables, an I-MOVE score ≥ 8 was a significant predictive factor for 30-day mortality (OR = 0.136, P < 0.01) but not 30-day readmission (OR = 1.143, P = 0.62) or the combined outcome death/readmission (OR = 0.682, P = 0.13).
- Conclusion: The clinical information provided by a patient's I-MOVE score before discharge does not provide information about readmission risk but may offer incremental information about 30-day mortality risk.
Risk factors for hospital 30-day readmission have been studied by Hasan et al [1], van Walraven et al [2], Allaudeen et al [3], and more recently, Donze et al [4]. Risk factors found to be associated with readmission include race, length of stay, and number of hospitalizations in the last 12 months. Additionally, patients identified “feeling unprepared for discharge” and “difficulty performing activities of daily living” as top issues contributing to readmission. The Affordable Care Act established the Value-Based Purchasing (VBP) model for defined hospital illnesses such as acute myocardial infarction, heart failure, and community acquired pneumonia. This has focused more attention on post-discharge 30-day mortality and readmissions as publicly reported metrics that in part determine the Centers for Medicare and Medicaid Services care reimbursement rates [5].
In our hospital, over 400 inpatients have been evaluated since 2004 using the I-MOVE scoring system in the course of their usual care. I-MOVE was most commonly employed by geriatricians in the division of hospital internal medicine, who collectively endorsed the tool in their practice meetings, especially for elderly patients returning to home alone whose mobility independence was uncertain.
Although it was initially designed to help clinicians understand the mobility independence of a patient before discharge, it may provide incremental value discerning risk of 30-day readmission and/or death. We therefore hypothesized that an I-MOVE score of less than 8 (not being able to transfer from a bed to a chair without assistance) would be a significant predictor of 30-day readmission and/or death.
Methods
Study Design
We performed a retrospective cohort study using a convenience sample including the patients in which the I-MOVE score had been calculated as part of the clinical process of care.
Setting and Participants
Participants were any inpatients discharged from the general medicine unit at Mayo Clinic Rochester from January 2003 to May 2011 who had at least one documented calculation of the I-MOVE score performed as part of the clinical process. Patients in the general medicine unit are adults not requiring subspecialty cardiovascular or neurology, coronary care unit, surgical, psychiatry, or rehabilitation. Patients were excluded if there was missing key outcome information or if they died during the hospitalization. For patients with more than one I-MOVE assessment, only the one closest to discharge was used. Data were abstracted from the electronic medical records between July and August 2011.
Variables
Outcome variables were 30-day readmission, 30-day mortality, and the combined outcome of mortality or readmission. We used the last I-MOVE score as a dichotomous variable with a cut-off of 8, which corresponds to the ability to transfer from bed to a chair unaided, for predicting the 2 outcomes. Only readmissions to the study hospital were captured. Deaths were identified from the electronic medical record. Mayo Clinic patient records are updated monthly with external reports of confirmed, actuarial records of deaths reported from public databases.
To control for possible confounding variables, we included the following covariates: age, gender, race/ethnicity, dates of admission and discharge, insurance (Medicare, Medicaid, self-pay, or private), marital status (currently married/not currently married), length of hospital stay, emergent admission, number of hospital admissions in the last 12 months, number of visits to the emergency department in the last 6 months and Charlson Index. All variables were abstracted from the electronic medical record.
Sample
A search was performed in the electronic medical record to find clinical documents (admission notes, progress notes, and hospital summaries) that mentioned the term “I-MOVE.” Manual review of the records was performed to confirm inclusion criteria.
Statistical Analysis
Separate analyses were performed for the 2 outcomes considered. First, a univariate analysis was performed with all covariates for variable selection. Variables that were significantly predictive with P < 0.1 were included in the multivariate model. Variables included in the first run of the multivariate model were excluded from the final multivariate model if they were not independently significant with P < 0.05. The I-MOVE variable was then added to that model to check its predictive power beyond that of the included covariates.
Results
Patient Characteristics
For the final dataset of 426 patients, 30-day mortality rate, readmission rate, and rate of the combined death/readmission outcome were 6.1% (26 patients), 15% (64 patients) and 19.7% (84 patients), respectively. A total of 6 patients were readmitted and died within 30 days after the initial discharge. The number of patients that had an I-MOVE score greater than or equal to 8 was 232 (54.4%). Table 2 presents the mean, standard deviation, and median I-MOVE score by patient discharge destination. Patients discharged home had an average I-MOVE score of 11.98, versus 7.24 for patients discharge to a skilled nursing facility (P = 0.2).
Analysis
Table 3 presents the odds ratios and coefficient estimate of the models. In the univariate analysis, an I-MOVE score greater than or equal to 8 was significantly correlated with 30-day mortality (P < 0.001), and the combined outcome (P = 0.044) but not with 30-day readmission (P = 0.76). After controlling for confounding variables, I-MOVE greater than or equal to 8 was a significant predictor of 30-day mortality (P < 0.01) but not 30-day readmission (P = 0.75)
Discussion
An I-MOVE score of less than 8 (inability to transfer from bed to a chair unassisted) is a statistically significant predictor of 30-day post-discharge mortality but not readmission or the combined outcome of death/readmission.
A recent review that evaluated published models that attempted to predict readmissions concluded that most current readmission risk prediction models designed for either comparative or clinical purposes perform poorly and that efforts are needed to improve their performance as use becomes more widespread [8]. Health care providers’ ability to predict which patients would be readmitted within 30 days was also shown by a recent study to be very poor, with C-statistics around 0.60 [9]. This inability of both experts and statistical methods to accurately predict readmissions may reflect some inherent randomness or unpredictability of readmissions, or the fact that a paradigm shift is still needed in the identification of the most important risk factors for readmissions. Along the same line, a recent evaluation of interventions aimed at reducing readmissions found that none of those identified in the literature managed to consistently reduce readmission rates long-term [10]. In addition, hospitals with greater adherence to recommended care processes did not achieve meaningfully better 30-day hospital readmission rates compared to those with lower levels of performance.
Conceptually, readmissions are an example of what is called “complexity science,” where many agents or factors (including the patient’s underlying illness, quality of care delivered, continuity and coordination of care, and resources available in the patient’s environment) and their interactions all play a role in the outcome [11,12]. Since I-MOVE primarily evaluates the physical capacity of the patient, and not any of the other variables that strongly affect readmission, it is perhaps not surprising that it did not predict readmission. It can be argued, on the other hand, that short term (30-day) mortality is more dependent on the patient’s physical and functional status [13] and so more likely to correlate with a measure such as I-MOVE. Inouye et al [13] found that pre-hospital, self-reported need for assistance in 7 basic “activities of daily living” (among which are transfers and ambulation) correlated with 90-day, and 2-year, post-hospital mortality.
The study has the advantage of a relatively large sample size, and the fact that the I-MOVE score was assessed before discharge eliminates the possibility of assessor bias. However, it has some limitations. We used a convenience sample, which may have introduced selection bias. Although we have no data on how providers selected patients for I-MOVE assessment, it would be reasonable to assume that patients were selected from among those whose activity level was, in terms of independence, doubtful or uncertain. That is, those who were not clearly vigorous (up and walking easily), nor clearly debilitated (in need of great assistance) may have been more likely to be assessed using I-MOVE. A more systematic selection of subjects might increase or decrease the predictive performance of the I-MOVE assessment. In addition, although we attempted to control for potential confounders, it is possible that additional confounders were left out of our analysis.
In summary, although the predictive performance of I-MOVE still needs to be confirmed by prospective studies with a comprehensive selection of subjects, the I-MOVE score at discharge appears to be associated with 30-day post-discharge mortality.
Acknowledgments: We thank the Department of Medicine’s clinical research office for their help in study design, data acquisition, and statistical analysis.
Corresponding author: Santiago Romero-Brufau, MD, Mayo Clinic Center for Innovation, 200 First St. SW, Rochester, MN 55905, [email protected].
Funding/support: This publication was supported by grant number UL1 TR000135 from the National Center for Advancing Translational Sciences (NCATS). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.
From the Mayo Clinic Center for Innovation (Dr. Romero-Brufau) Department of Medicine (Drs. Manning, Borrud, Keller, Kashiwagi, Huddleston, and Croghan) Department of Health Sciences Research (Mr. Cha), Mayo Clinic, Rochester, MN.
Abstract
- Objective: To determine whether a score of 8 or greater on the I-MOVE, a bedside instrument that evaluates the need for assistance in turning, sitting, standing, transferring from bed to a chair, and ambulating, predicts lower risk for 30-day readmission or mortality.
- Design: Retrospective cohort study of patients discharged from 2003 to 2011 from a referral hospital in Southeastern Minnesota. We used a convenience sample of 426 inpatients who had at least one documented calculation of the I-MOVE score performed as part of the clinical process during the study.
- Results: Overall 30-day mortality rate, readmission rate, and rate of the combined death/readmission outcome were 6.1% (26 patients), 15% (64 patients) and 19.7% (84 patients), respectively. After controlling for confounding variables, an I-MOVE score ≥ 8 was a significant predictive factor for 30-day mortality (OR = 0.136, P < 0.01) but not 30-day readmission (OR = 1.143, P = 0.62) or the combined outcome death/readmission (OR = 0.682, P = 0.13).
- Conclusion: The clinical information provided by a patient's I-MOVE score before discharge does not provide information about readmission risk but may offer incremental information about 30-day mortality risk.
Risk factors for hospital 30-day readmission have been studied by Hasan et al [1], van Walraven et al [2], Allaudeen et al [3], and more recently, Donze et al [4]. Risk factors found to be associated with readmission include race, length of stay, and number of hospitalizations in the last 12 months. Additionally, patients identified “feeling unprepared for discharge” and “difficulty performing activities of daily living” as top issues contributing to readmission. The Affordable Care Act established the Value-Based Purchasing (VBP) model for defined hospital illnesses such as acute myocardial infarction, heart failure, and community acquired pneumonia. This has focused more attention on post-discharge 30-day mortality and readmissions as publicly reported metrics that in part determine the Centers for Medicare and Medicaid Services care reimbursement rates [5].
In our hospital, over 400 inpatients have been evaluated since 2004 using the I-MOVE scoring system in the course of their usual care. I-MOVE was most commonly employed by geriatricians in the division of hospital internal medicine, who collectively endorsed the tool in their practice meetings, especially for elderly patients returning to home alone whose mobility independence was uncertain.
Although it was initially designed to help clinicians understand the mobility independence of a patient before discharge, it may provide incremental value discerning risk of 30-day readmission and/or death. We therefore hypothesized that an I-MOVE score of less than 8 (not being able to transfer from a bed to a chair without assistance) would be a significant predictor of 30-day readmission and/or death.
Methods
Study Design
We performed a retrospective cohort study using a convenience sample including the patients in which the I-MOVE score had been calculated as part of the clinical process of care.
Setting and Participants
Participants were any inpatients discharged from the general medicine unit at Mayo Clinic Rochester from January 2003 to May 2011 who had at least one documented calculation of the I-MOVE score performed as part of the clinical process. Patients in the general medicine unit are adults not requiring subspecialty cardiovascular or neurology, coronary care unit, surgical, psychiatry, or rehabilitation. Patients were excluded if there was missing key outcome information or if they died during the hospitalization. For patients with more than one I-MOVE assessment, only the one closest to discharge was used. Data were abstracted from the electronic medical records between July and August 2011.
Variables
Outcome variables were 30-day readmission, 30-day mortality, and the combined outcome of mortality or readmission. We used the last I-MOVE score as a dichotomous variable with a cut-off of 8, which corresponds to the ability to transfer from bed to a chair unaided, for predicting the 2 outcomes. Only readmissions to the study hospital were captured. Deaths were identified from the electronic medical record. Mayo Clinic patient records are updated monthly with external reports of confirmed, actuarial records of deaths reported from public databases.
To control for possible confounding variables, we included the following covariates: age, gender, race/ethnicity, dates of admission and discharge, insurance (Medicare, Medicaid, self-pay, or private), marital status (currently married/not currently married), length of hospital stay, emergent admission, number of hospital admissions in the last 12 months, number of visits to the emergency department in the last 6 months and Charlson Index. All variables were abstracted from the electronic medical record.
Sample
A search was performed in the electronic medical record to find clinical documents (admission notes, progress notes, and hospital summaries) that mentioned the term “I-MOVE.” Manual review of the records was performed to confirm inclusion criteria.
Statistical Analysis
Separate analyses were performed for the 2 outcomes considered. First, a univariate analysis was performed with all covariates for variable selection. Variables that were significantly predictive with P < 0.1 were included in the multivariate model. Variables included in the first run of the multivariate model were excluded from the final multivariate model if they were not independently significant with P < 0.05. The I-MOVE variable was then added to that model to check its predictive power beyond that of the included covariates.
Results
Patient Characteristics
For the final dataset of 426 patients, 30-day mortality rate, readmission rate, and rate of the combined death/readmission outcome were 6.1% (26 patients), 15% (64 patients) and 19.7% (84 patients), respectively. A total of 6 patients were readmitted and died within 30 days after the initial discharge. The number of patients that had an I-MOVE score greater than or equal to 8 was 232 (54.4%). Table 2 presents the mean, standard deviation, and median I-MOVE score by patient discharge destination. Patients discharged home had an average I-MOVE score of 11.98, versus 7.24 for patients discharge to a skilled nursing facility (P = 0.2).
Analysis
Table 3 presents the odds ratios and coefficient estimate of the models. In the univariate analysis, an I-MOVE score greater than or equal to 8 was significantly correlated with 30-day mortality (P < 0.001), and the combined outcome (P = 0.044) but not with 30-day readmission (P = 0.76). After controlling for confounding variables, I-MOVE greater than or equal to 8 was a significant predictor of 30-day mortality (P < 0.01) but not 30-day readmission (P = 0.75)
Discussion
An I-MOVE score of less than 8 (inability to transfer from bed to a chair unassisted) is a statistically significant predictor of 30-day post-discharge mortality but not readmission or the combined outcome of death/readmission.
A recent review that evaluated published models that attempted to predict readmissions concluded that most current readmission risk prediction models designed for either comparative or clinical purposes perform poorly and that efforts are needed to improve their performance as use becomes more widespread [8]. Health care providers’ ability to predict which patients would be readmitted within 30 days was also shown by a recent study to be very poor, with C-statistics around 0.60 [9]. This inability of both experts and statistical methods to accurately predict readmissions may reflect some inherent randomness or unpredictability of readmissions, or the fact that a paradigm shift is still needed in the identification of the most important risk factors for readmissions. Along the same line, a recent evaluation of interventions aimed at reducing readmissions found that none of those identified in the literature managed to consistently reduce readmission rates long-term [10]. In addition, hospitals with greater adherence to recommended care processes did not achieve meaningfully better 30-day hospital readmission rates compared to those with lower levels of performance.
Conceptually, readmissions are an example of what is called “complexity science,” where many agents or factors (including the patient’s underlying illness, quality of care delivered, continuity and coordination of care, and resources available in the patient’s environment) and their interactions all play a role in the outcome [11,12]. Since I-MOVE primarily evaluates the physical capacity of the patient, and not any of the other variables that strongly affect readmission, it is perhaps not surprising that it did not predict readmission. It can be argued, on the other hand, that short term (30-day) mortality is more dependent on the patient’s physical and functional status [13] and so more likely to correlate with a measure such as I-MOVE. Inouye et al [13] found that pre-hospital, self-reported need for assistance in 7 basic “activities of daily living” (among which are transfers and ambulation) correlated with 90-day, and 2-year, post-hospital mortality.
The study has the advantage of a relatively large sample size, and the fact that the I-MOVE score was assessed before discharge eliminates the possibility of assessor bias. However, it has some limitations. We used a convenience sample, which may have introduced selection bias. Although we have no data on how providers selected patients for I-MOVE assessment, it would be reasonable to assume that patients were selected from among those whose activity level was, in terms of independence, doubtful or uncertain. That is, those who were not clearly vigorous (up and walking easily), nor clearly debilitated (in need of great assistance) may have been more likely to be assessed using I-MOVE. A more systematic selection of subjects might increase or decrease the predictive performance of the I-MOVE assessment. In addition, although we attempted to control for potential confounders, it is possible that additional confounders were left out of our analysis.
In summary, although the predictive performance of I-MOVE still needs to be confirmed by prospective studies with a comprehensive selection of subjects, the I-MOVE score at discharge appears to be associated with 30-day post-discharge mortality.
Acknowledgments: We thank the Department of Medicine’s clinical research office for their help in study design, data acquisition, and statistical analysis.
Corresponding author: Santiago Romero-Brufau, MD, Mayo Clinic Center for Innovation, 200 First St. SW, Rochester, MN 55905, [email protected].
Funding/support: This publication was supported by grant number UL1 TR000135 from the National Center for Advancing Translational Sciences (NCATS). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.
1. Hasan O, Meltzer DO, Shaykevich SA, et al. Hospital readmission in general medicine patients: a prediction model.
J Gen Intern Med 2010;25:211–9.
2. van Walraven C, Dhalla IA, Bell C, et al. Derivation and validation of an index to predict early death or unplanned readmission after discharge from hospital to the community. CMAJ 2010;182:551–7.
3. Allaudeen N, Vidyarthi A, Maselli J, Auerbach A. Redefining readmission risk factors for general medicine patients. J Hosp Med 2011;6:54–60.
4. Donze J, Aujesky D, Williams D, Schnipper JL. Potentially avoidable 30-day hospital readmissions in medical patients: derivation and validation of a prediction model. JAMA Intern Med 2013;173:632–8.
5. Kocher RP, Adashi EY. Hospital readmissions and the Affordable Care Act: paying for coordinated quality care. JAMA 2011;306:1794–5.
6. Manning DM, Keller AS, Frank DL. Home alone: assessing mobility independence before discharge. J Hosp Med 2009;4:252–4.
7. Cook DJ, Manning DM, Holland DE, et al. Patient engagement and reported outcomes in surgical recovery: effectiveness of an e-health platform. J Am Coll Surg 2013;217:648–55.
8. Kansagara D, Englander H, Salanitro A, et al. Risk prediction models for hospital readmission: a systematic review. JAMA 2011;306:1688–98.
9. Allaudeen N, Schnipper JL, Orav EJ, et al. Inability of providers to predict unplanned readmissions. J Gen Intern Med 2011;26:771–6.
10. Hansen LO, Young RS, Hinami K, et al. Interventions to reduce 30-day rehospitalization: a systematic review. Ann Intern Med 2011;155:520–8.
11. Marks E. Complexity science and the readmission dilemma. JAMA Intern Med 2013;173:629–31.
12. Lindquist LA, Baker DW. Understanding preventable hospital readmissions: masqueraders, markers, and true causal factors. J Hosp Med 2011;6:51–3.
13. Inouye SK, Peduzzi PN, Robison JT, et al. Importance of functional measures in predicting mortality among older hospitalized patients. JAMA 1998;279:1187–93.
1. Hasan O, Meltzer DO, Shaykevich SA, et al. Hospital readmission in general medicine patients: a prediction model.
J Gen Intern Med 2010;25:211–9.
2. van Walraven C, Dhalla IA, Bell C, et al. Derivation and validation of an index to predict early death or unplanned readmission after discharge from hospital to the community. CMAJ 2010;182:551–7.
3. Allaudeen N, Vidyarthi A, Maselli J, Auerbach A. Redefining readmission risk factors for general medicine patients. J Hosp Med 2011;6:54–60.
4. Donze J, Aujesky D, Williams D, Schnipper JL. Potentially avoidable 30-day hospital readmissions in medical patients: derivation and validation of a prediction model. JAMA Intern Med 2013;173:632–8.
5. Kocher RP, Adashi EY. Hospital readmissions and the Affordable Care Act: paying for coordinated quality care. JAMA 2011;306:1794–5.
6. Manning DM, Keller AS, Frank DL. Home alone: assessing mobility independence before discharge. J Hosp Med 2009;4:252–4.
7. Cook DJ, Manning DM, Holland DE, et al. Patient engagement and reported outcomes in surgical recovery: effectiveness of an e-health platform. J Am Coll Surg 2013;217:648–55.
8. Kansagara D, Englander H, Salanitro A, et al. Risk prediction models for hospital readmission: a systematic review. JAMA 2011;306:1688–98.
9. Allaudeen N, Schnipper JL, Orav EJ, et al. Inability of providers to predict unplanned readmissions. J Gen Intern Med 2011;26:771–6.
10. Hansen LO, Young RS, Hinami K, et al. Interventions to reduce 30-day rehospitalization: a systematic review. Ann Intern Med 2011;155:520–8.
11. Marks E. Complexity science and the readmission dilemma. JAMA Intern Med 2013;173:629–31.
12. Lindquist LA, Baker DW. Understanding preventable hospital readmissions: masqueraders, markers, and true causal factors. J Hosp Med 2011;6:51–3.
13. Inouye SK, Peduzzi PN, Robison JT, et al. Importance of functional measures in predicting mortality among older hospitalized patients. JAMA 1998;279:1187–93.