PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
J Am Geriatr Soc. Author manuscript; available in PMC 2013 November 25.
Published in final edited form as:
PMCID: PMC3839864
NIHMSID: NIHMS525413

A Clinical Index to Stratify Hospitalized Older Adults According to Risk for New-Onset Disability

Kala M. Mehta, DSc, MPH,* Edgar Pierluissi, MD,* W. John Boscardin, PhD,* Katharine A. Kirby, MS,* Louise C. Walter, MD,* Mary-Margaret Chren, MD,§ Robert M. Palmer, MD, MPH,|| Steven R. Counsell, MD,#** and C. Seth Landefeld, MD*

Abstract

BACKGROUND

Many older adults who are independent prior to hospitalization develop a new disability by hospital discharge. Early risk stratification for new-onset disability may improve care. Thus, this study’s objective was to develop and validate a clinical index to determine, at admission, risk for new-onset disability among older, hospitalized adults at discharge.

DESIGN

Data analyses derived from two prospective studies.

SETTING

Two teaching hospitals in Ohio.

PARTICIPANTS

Eight hundred eighty-five patients aged 70 years and older were discharged from a general medical service at a tertiary care hospital (mean age 78, 59% female) and 753 patients discharged from a separate community teaching hospital (mean age 79, 63% female). All participants reported being independent in five activities of daily living (ADLs: bathing, dressing, transferring, toileting, and eating) 2 weeks before admission.

MEASUREMENTS

New-onset disability, defined as a new need for personal assistance in one or more ADLs at discharge in participants who were independent 2 weeks before hospital admission.

RESULTS

Seven independent risk factors known on admission were identified and weighted using logistic regression: age (80–89, 1 point; ≥90, 2 points); dependence in three or more instrumental ADLs at baseline (2 points); impaired mobility at baseline (unable to run, 1 point; unable to climb stairs, 2 points); dependence in ADLs at admission (2–3 ADLs, 1 point; 4–5 ADLs, 3 points); acute stroke or metastatic cancer (2 points); severe cognitive impairment (1 point); and albumin less than 3.0 g/dL (2 points). New-onset disability occurred in 6%, 13%, 18%, 34%, 35%, 45%, 50%, and 87% of participants with 0, 1, 2, 3, 4, 5, 6, and 7 or more points, respectively, in the derivation cohort (area under the receiver operating characteristic curve (AUC) =0.784), and in 8%, 10%, 27%, 38%, 44%, 45%, 58%, and 83%, respectively, in the validation cohort (AUC =0.784). The risk score also predicted (P<.001) disability severity, nursing home placement, and long-term survival.

CONCLUSION

This clinical index determines risk for new-onset disability in hospitalized older adults and may inform clinical care.

Keywords: hospitalization, prognosis, disability, activities of daily living

A cute illness and hospitalization often lead to disability in older adults,14 increasing the odds of disability 60 times and accounting for nearly half of all new cases of disability.3 New-onset disability is defined as a recent transition from independence to dependence on another person to perform basic activities of daily living (ADLs), such as bathing, dressing, transferring from a chair, or using a toilet. This new-onset disability at hospital discharge often heralds long-term institutionalization, repeated hospitalizations, and death.5,6 However, acute illness and hospitalization do not lead uniformly to functional decline, and disability after hospitalization and is sometimes preventable or reversible.3,79

The accurate stratification of hospitalized older adults according to their disability risk would allow the targeting of interventions to those at high risk. Although previous efforts have identified risk factors for functional decline10 and developed screening tools in various clinical populations,1115 none has focused specifically on new-onset disability (new ADL dependence in hospitalized older adults who were independent before hospitalization—those for whom preventive and restorative measures may have the greatest effect).

Therefore, a clinical index was developed and validated to predict new-onset disability at hospital discharge in adults aged 70 and older who were independent in all ADLs 2 weeks before hospitalization. Based on conceptual models of disablement16 and studies of its risk factors,13,17,18 it was hypothesized that predisposing factors that indicate proximity to the threshold of disability (e.g., poor mobility), the effect of acute illness on function at admission, and inhibitors of recovery (e.g., cognitive impairment, malnutrition, or very old age) would predict new-onset disability. Potential predictors that are easily determined on admission, including sociodemographic, clinical, and functional characteristics, were considered. The goal was to develop a prognostic index that can be used to inform clinical decisions.

METHODS

Participants

Participants were drawn from two prospective studies of an intervention to improve functional outcomes of hospitalized older adults in two teaching hospitals in Ohio.19,20 Participants aged 70 and older who had emergency admissions to the general medical services between 1993 and 1998 were eligible. Exclusion criteria were admission to an intensive care unit or oncology ward, elective admission, or length of stay less than 2 days. Of 11,475 patients meeting eligibility criteria, 3,163 were randomly selected and enrolled in the original studies. The demographic, clinical, and functional characteristics of study participants enrolled were similar to those of people not enrolled.20 The original prospective studies included 1,632 participants at one site and 1,531 at the second site. Participants were randomized at admission to an intervention unit or a usual care unit using a block design that was determined a priori. Computer-generated random numbers were put in opaque envelopes that the admitting desk clerk opened sequentially. Seventy-nine participants were not admitted to the unit of randomization. The intervention was a multicomponent intervention including a special environment, patient-centered care, discharge planning, and review of medical care to reduce functional disability. ADL decline from baseline to discharge did not differ in the intervention (30%) and usual care (34%) groups (P =.11).20 These results were similar when analyzed using intention-to-treat or per-protocol analysis. In addition, the intervention and usual care groups did not differ significantly in their admission characteristics or self-reported measures of function at discharge.19,20

Of the 3,163 participants in the original studies, 1,656 reported that they did not require personal assistance 2 weeks before admission in any of five ADLs: bathing, dressing, transferring from bed to chair, using a toilet, and eating. For 18 participants, ADL information was missing at hospital discharge, leaving an analytical sample of 1,638. The clinical index was developed in a community teaching hospital (n =885) and then validated in a university teaching hospital (n =753) to test its transportability and accuracy.

Data Collection and Measurements

Trained interviewers obtained data from participants and surrogates, and trained chart abstractors obtained data from medical records. Standardized interviews were completed at hospital admission and discharge. On admission, participants were asked about sociodemographic characteristics, their health, and ADL and instrumental ADL (IADL) function at two time points: 2 weeks before admission and at admission. On discharge, participants were asked about ADL and IADL function. IADLs were using the telephone, transportation, shopping, preparing meals, doing housework, taking medicines and handling money.21 Prior work has demonstrated the validity of self-report of function.22 Surrogates were interviewed if the participant had more than five errors on the Short Portable Mental Status Questionnaire23 (SPMSQ) or if the participant was too ill to communicate.

Outcomes

The outcome of main interest was disability at hospital discharge, which was defined as need for personal assistance in performing one or more ADLs as reported to a trained interviewer by the participant or the surrogate, who was generally the participant’s primary nurse. ADL function was measured using a modified version of the Katz index of five ADLs: bathing, dressing, transferring from bed to chair, using a toilet, and eating.24 Because too few participants died in the hospital to identify separate predictors of death, the primary outcome for this study was the composite outcome disability at hospital discharge or death in hospital. Secondary outcomes included catastrophic disability (defined as dependence in three or more ADLs), nursing home placement at discharge, and long-term survival through December 31, 2009, confirmed by review of the National Death Index and a search of Ohio state death records.

Predictor Variables

Sociodemographic, clinical, and functional characteristics were considered as potential predictors of disability as conceptualized in the disablement model.16 Sociodemographic characteristics included age, sex, marital status, self-identified race, educational level, and current living situation. Clinical characteristics included the reason for admission, comorbid illnesses (Charlson Comorbidity Index),25 and acute severity of illness as indicated by the acute physiology component of the Acute Physiology and Chronic Health Evaluation (APACHE) II;26 Charlson and APACHE scores were then divided into tertiles. For laboratory values, the University Hospitals of Cleveland clinical chemistry laboratory measured creatinine, hematocrit, and albumin. Creatinine and hematocrit were evaluated using tertiles and albumin using an a priori defined clinically relevant cut point. Severe cognitive impairment was defined according to medical record documentation of dementia, five or more errors on the SPMSQ, or inability to complete the SPMSQ.23 Depressive symptoms were evaluated using the 10-item modified version of the Center for Epidemiologic Study Depression Scale (CES-D) and evaluated using a cut point of four or more symptoms.27 Self-rated general health was reported as excellent, very good, good, fair, or poor.28

Potential predictors included ADL function on admission (but not 2 weeks before admission, when all participants were independent in ADLs) and instrumental ADL (IADL) function, walking, and gait, at two time points (2 weeks before admission and on admission). IADLs were assessed as 1 point per dependency in using the telephone, transportation, shopping, preparing meals, doing housework, taking medicines, and handling money.29 Dependence was defined as needing the help of another person. IADLs were then evaluated in three groups, participants who were independent in all IADLs, dependent in one or two IADLs, and dependent in three or more IADLs. Walking ability was determined using a hierarchical scale based on participant reports of inability to run a short distance, walk up stairs or a hill, walk a block or more, or none of these.30 Gait was classified as steady or unsteady according to participant reports.31

Missing Data

For most variables, data were missing for fewer than 2% of participants. To avoid dropping participants with missing data, a multiple imputation strategy was used incorporating the Imputation for Chained Equations (ICE) library and the multiply imputed missing (mim) library32 in Stata version 10 (Stata Corp., College Station, TX). Analyses were calculated twice—with imputation and by dropping cases with missing values. Because the results did not vary substantially, the multiple imputation results were reported as the primary findings.

Analytical Strategy

This study sought to develop separate models to predict new-onset disability and death in the derivation cohort and to test the models in the validation cohort. Because too few participants died to allow the development of separate models, a model to predict the composite outcome disability or death at discharge was constructed; this composite outcome is referred to as “new-onset disability.”33 A secondary analysis was also performed excluding participants who died that gave results similar to those of the primary analysis, which are reported.

Bivariate associations between each characteristic and the composite outcome of disability or death at hospital discharge were assessed using single-predictor logistic regression analysis in the derivation cohort. To build a risk index for new-onset disability, multivariate logistic regression analyses were conducted using the best subsets methodology, as recommended, to incorporate biological plausibility and clinical utility in the selection of prognostic models.34 The score selection option of Proc Logistic in SAS 9.2 (SAS Institute, Inc., Cary, NC) was used. Best subsets regression produces a number of models with an increasing set of predictors. In this instance, a range of models with 1 to 20 predictors was evaluated. Several models of similar predictive power, reflected in statistically indistinguishable score statistics between many pairs of models, were then compared. The clinical relevance and practicality of potential predictors were then considered to select the final model from this larger set. In contrast, stepwise selection methods, which are extremely common in the literature, calculate only a single model; the natural tendency is to accord special status to this single model even though it is nearly always statistically indistinguishable from an extraordinarily large number of alternative models and it is not guaranteed to be the optimal model according to any known statistical criterion. To construct a clinical index, points were assigned to each predictor by dividing the β coefficient in the logistic model by the lowest β coefficient and rounding to the nearest integer.35 A risk score was calculated for each participant by summing the points assigned for each risk factor. The frequency of new-onset disability for each risk score was determined.

The clinical index was tested at a different site because this form of prospective validation testing assesses the generalizability of the model, as well as its accuracy.36,37 The predictive validity of the logistic model and the clinical index were assessed by comparing the frequency of new-onset disability observed in the validation cohort with the predicted frequency using the Hosmer-Lemeshow test (calibration) and by calculating the area under the receiver operating characteristic curve (AUC) in the derivation and validation cohorts (discrimination).38

To provide further information about the index’s potential clinical significance,39 the relationship between risk score and three other outcomes at discharge were examined: the number of ADL dependencies that participants were dependent in, catastrophic disability, and nursing home placement. The index’s performance was then examined in specific subgroups defined according to age, sex, cognitive status, and ADL function at admission. To determine whether the clinical index had similar discrimination in the intervention and usual care groups, its performance in all participants in the intervention group was compared with its performance in all participants in the usual care group. Finally, the relationship between risk score and survival from hospital discharge through 2009 was tested.

RESULTS

Characteristics of Participants

The mean age ± standard deviation of participants in the derivation cohort (n =885) was 78 ± 6; 59% were women, and 89% were white (Table 1). On admission, medical conditions included chronic lung disease (27%), congestive heart failure (24%), diabetes mellitus (23%), and severe cognitive impairment (17%), and 346 participants (39%) were dependent in one or more ADLs. At discharge, 242 (27%) were dependent in one or more ADLs, and 633 (72%) were independent in all ADLs; 10 participants (1%) had died.

Table 1
Characteristics of Patients in the Derivation and Validation Cohorts

The mean age of participants in the validation cohort (n =753) was 79 ± 7; 63% were women, and 66% were white (Table 1). On admission, medical conditions included chronic lung disease (18%), congestive heart failure (22%), diabetes mellitus (14%), and severe cognitive impairment (19%), and 279 participants (37%) were dependent in one or more ADLs. At discharge, 231 (31%) were dependent in one or more ADLs and 507 (67%) were independent in all ADL; 15 participants (2%) had died.

Bivariable Associations

In the derivation cohort, new-onset disability was associated with three demographic characteristics (age, sex, and marital status), two measures of functional status 2 weeks before admission (mobility and IADL function), and several admission characteristics (e.g., ADL function, unsteady gait, specific clinical diagnoses (e.g., stroke, severe cognitive impairment, congestive heart failure, pneumonia, chronic lung disease, and metastatic cancer)), Charlson comorbidity score, APACHE score, and albumin level of less than 3.0 mg/dL (Table 2).

Table 2
Bivariable Associations Between Risk Factors and New–Onset Disability at Discharge in the Derivation Cohort

Multivariable Results

Seven risk factors were independently associated with new-onset disability in the logistic model identified using best subsets regression, including one demographic characteristic (age), two indicators of functional status 2 weeks before admission (mobility and IADL function), and four admission characteristics (dependency in ≥2 ADLs, diagnosis of acute stroke or metastatic cancer, severe cognitive impairment, and albumin <3.0 mg/dL) (Table 3). After controlling for these independent risk factors, other characteristics and diagnoses were not associated with new-onset disability. Thus some predictors associated with new-onset disability on bivariable analyses were not included in the clinical index: sex, marital status, gait unsteadiness, congestive heart failure, pneumonia, chronic lung disease, Charlson comorbidity score, and APACHE acute physiology score.

Table 3
Characteristics Associated with New–Onset Disability at Discharge in the Derivation Cohort (Multivariable Analyses)

Risk Scoring System

The points assigned to each of the seven independent risk factors are listed in Table 3. A risk score was calculated for each participant by summing the points of each risk factor present. For example, an 86-year old woman (1 point) who was independent in IADLs and ADLs 2 weeks before admission but was unable to climb stairs (2 points) and was dependent in two ADLs (1 point) on admission for treatment of urosepsis would have a risk score of 4 points.

According to quintiles of predicted risk, the frequency of new-onset disability ranged from 8% in the lowest-risk quintile to 66% in the highest-risk quintile in the derivation group and from 7% to 65% in the validation group (Table 4). The discrimination of the final model was similar in the derivation and validation cohorts (AUC =0.787 and 0.791, respectively). Risk scores ranged from 0 to 14 points in the derivation cohort (mean 2.8 ± 2.5) and were directly related to the frequency of new-onset disability in the derivation and validation cohorts. The risk scoring system had acceptable calibration and good discrimination in the derivation and validation cohorts (AUC =0.784 in each).

Table 4
Validation of the Clinical Index: New–Onset Disability at Discharge in Derivation and Validation Cohorts According to Risk Stratum

The risk score was associated with disability severity, as indicated by the number of ADL dependencies participants were dependent in at discharge (Figure 1, P<.001). In participants with 0, 1 to 2, 3 to 4, 5 to 6, and 7 or more points, catastrophic disability, defined as dependent in three or more ADLs, occurred in 0%, 4%, 19%, 37%, and 65%, respectively, and discharge to a nursing home occurred in 0%, 1%, 4%, 6%, and 15%, respectively. The risk score was also associated with long-term survival. In participants with 0, 1 to 2, 3 to 4, 5 to 6, and 7 or more points, 5-year survival was 71%, 52%, 36%, 28%, and 19%, respectively, and 10-year survival was 41%, 24%, 16%, 11%, and 5%, respectively.

Figure 1
Proportion of participants with different numbers of activity of daily living (ADL) dependencies at discharge, according to the risk score. The number of ADLs participants were dependent in is shown in different colors, according to the key. Catastrophic ...

The risk scoring system performed well in several subgroups of the development and validation cohort combined (Appendix 1) (AUC =0.786 for men, AUC =0.781 for women; AUC =0.760 for participants aged 70–79, AUC = 0.763 for participants aged ≥80; AUC =0.725 for participants with normal cognition, AUC =0.787 for participants with severe cognitive impairment; AUC =0.744 for participants independent in all ADLs at admission, AUC =0.764 for participants dependent in one or more ADLs; AUC =0.774 for participants in the intervention arm, AUC =0.796 for participants who received usual treatment in the original trial).

Appendix 1
Disability or Death at Discharge, According to Risk Score in Patient Subgroups

DISCUSSION

A clinical index that uses information on hospital admission to stratify previously independent older adults according to their risk for new-onset disability was developed. The index identified groups of people with frequencies of disability that ranged from less than 10% to more than 90%, with excellent discrimination and calibration in prospective validation testing. Moreover, higher scores also identified people who had greater disability, more-frequent nursing home placement, and higher 10-year mortality rates after discharge.

This clinical index includes seven independent risk factors that are consistent with the disablement model16 and are identifiable within 24 hours of hospital admission. Some risk factors, such as impaired mobility and dependence in IADLs 2 weeks before hospitalization, probably indicate loss of functional reserve, physical frailty, and proximity to the threshold of ADL disability before a person became acutely ill. Dependence in two or more ADLs on admission reflects the magnitude of the effect of the acute illness on function; dependence in four or five ADLs at admission was the greatest risk factor for disability at discharge. Other risk factors, including aged 80 and older, severe cognitive impairment, hypoalbuminemia, and a diagnosis of stroke or metastatic cancer, are factors that may slow recovery of independence in ADLs, indicating low resilience. The congruence of the risk factors with the conceptual framework supports the “biological plausibility” of the clinical index and increases confidence that it is likely to be robust in different settings.

These findings extend previous studies in several ways. First, the clinical index demonstrated better discrimination, calibration, and validation than earlier indices, with AUCs of 0.784 in derivation and validation testing, compared with AUCs of 0.64 to 0.73 for previous indices.11,12 This better performance probably resulted from a theoretically based conceptual framework, a focus on adults at risk for new-onset disability, and the large size of the cohorts.

Second, this clinical index predicts new-onset disability in hospitalized people who had been independent in ADLs at baseline; prior studies predicted worsening ADL function in heterogeneous groups, some disabled before hospitalization and some not.11,12 Focusing on new-onset disability in previously independent adults may help target preventive and rehabilitative interventions.40

Finally, the findings extend the understanding of risk factors for disability. Risk factors for disability have been extensively studied in community-dwelling populations41 but have been studied less often in relation to acute illness and hospitalization,10 even though hospitalization accounts for half of all new-onset disability.3 This study confirms the importance of age (≥80), cognitive impairment, difficulty with IADLs, and signs of malnutrition as risk factors for disability in hospitalized older people, and the findings quantify the independent prognostic significance of each risk factor. The findings also provide quantitative estimates of the importance of difficulty walking, stroke, and advanced cancer as prognostic factors in hospitalized older adults. Measures of acute severity of illness and of comorbid illness were not independent prognostic risk factors, consistent with prior reports.15

This study has several implications. First, the findings provide a rationale for routine assessment on admission of the risk factors for new-onset disability. The clinical index may facilitate early planning for posthospital care, which is central in interventions to improve transitions in care.18,42,43 For example, patients at low risk for new-onset disability can plan to return home with little additional support. High-risk patients may need personal assistance, new equipment, or even rehabilitation to return home; these patients may also be at higher risk for unplanned readmission. Risk stratification can also identify patients who may benefit most from interventions aiming to prevent or ameliorate disability.18,20,44,45 Although the current study found similar performance of the index in the Acute Care for Elders intervention and usual care group, future studies could evaluate interventions that promote mobility46 and improve nutrition47 for patients at risk for new-onset disability. Second, prehospital interventions in the community may enhance modifiable risk factors and thereby reduce new-onset disability. For example, programs that enhance mobility and nutrition could increase an individual’s distance from the ADL disability threshold and promote resilience, preventing new-onset disability. Third, the clinical index may also be useful in comparing outcomes achieved by different large-scale providers. If two hospitals had different rates of new-onset disability in their older patients, it would be important to determine whether this difference was attributable to differences in admission characteristics (as in this index).

Several methodological considerations support the validity of these findings, including the large size of the cohort, the measurement of potential prognostic factors in conceptual domains, systematic use of a valid measure of disability, near-complete ascertainment of outcomes, and the reproducible accuracy of the clinical index when tested in a second site.

This study is the first to describe a clinical index for new-onset disability, and it had important limitations. First, measures of function were based on participant and surrogate reports, although these reports have demonstrated validity48 and are strongly associated with health outcomes such as mortality, nursing home placement, and resource use.1,19,49 Second, although many potential risk factors were evaluated, it cannot be excluded that other potentially important risk factors such as delirium, physical activity, and specific medications may add to predictive ability if evaluated. Delirium has been shown to contribute to disability in previously independent older adults,13,50 but it was not systematically assessed in the current study. Third, participants were enrolled in the 1990s, and the accuracy of the index should be tested in current patients. Although the frequency and predictors of new-onset disability may change over time, these changes may be small, as supported by the small fall in the average length of hospital stay in the United States for adults aged 65 and older from 4.8 days in 1998 to 4.6 days in 2007.51 In addition, the topic of new-onset disability may be of particular interest now, because the future prospective “bundling” of payment for hospital and posthospital costs may necessitate ways to identify and intervene upon disability in the peri-hospitalization period as a way to reduce overall costs. Finally, the generalizability of the findings to surgical and intensive care patients has not been determined.

In summary, a clinical index predicted new-onset disability at discharge in hospitalized older adults. It uses seven admission risk factors, most of which indicate loss of functional reserve and resilience, to calculate a simple point scoring system. The index had good discrimination and calibration and generalized well in testing at a different site. These features suggest that this index may be clinically useful.

Acknowledgments

We sincerely thank Kenneth E. Covinsky, MD, MPH, for his thoughtful review of this manuscript. We also wish to thank Ms. Priya Kamat and Ms. Carolyn Peterson for their contributions to this manuscript.

This work was presented at the Society of General Internal Medicine meeting, Miami, Florida (May 15, 2009).

Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper.

Author Contributions: Dr. Mehta had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Drs. Landefeld, Chren, Palmer, and Counsell contributed to original acquisition of data. Drs. Pierluissi and Walter acquired data from the National Death Index. Drs. Mehta and Landefeld contributed to study design and concept, data analysis and interpretation of data, drafting of the manuscript, and study supervision. Drs. Pierluissi and Boscardin and Ms. Kirby contributed to the data analysis, interpretation of data, and drafting of the manuscript. All authors contributed to the critical revision of the manuscript for important intellectual content, helped conceptualize ideas, interpreted findings, and reviewed drafts of the manuscript.

Financial Disclosure: This work was supported by National Institute on Aging (NIA) Grant RO1 AG029233 and a grant from the S.D. Bechtel, Jr. Foundation (PI: Dr. Lande-feld). Dr. Mehta was supported by a Research Career Scientist Award from the NIA (K-01AG025444). She was an affiliate of the Center for Aging in Diverse Communities at the University of California at San Francisco, a part of the P30 AG 15272 350 Resource Centers for Minority Aging Research Program (NIA, National Institute of Nursing Research, and the National Center for Minority Health and Health Disparities). Dr. Pierluissi was supported by an administrative supplement to NIA Grant RO1 AG029233. Dr. Chren was supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health (K24 AR052667). Dr. Walter was supported by a grant from the National Cancer Institute (R01CA134425-01A2) and a Veterans Affairs Merit Review Grant (IIR 04-427). The Acute Care for Elders studies were supported by grants from the John A. Hartford Foundation (88277-3G and 2006-0108), the NIA (AG-10418-03), and the Summa Health System Foundation.

Sponsor’s Role: The funding organizations and sponsors had no role in the design or conduct of the study; collection, management, analysis, or interpretation of the data; or preparation, review, or approval of the manuscript.

References

1. Covinsky KE, Justice AC, Rosenthal G, et al. Measuring prognosis and case mix in hospitalized elders. The importance of functional status. J Gen Intern Med. 1997;12:203–208. [PMC free article] [PubMed]
2. Covinsky KE, Palmer RM, Fortinsky RH, et al. Loss of independence in activities of daily living in older adults hospitalized with medical illnesses: Increased vulnerability with age. J Am Geriatr Soc. 2003;51:451–458. [PubMed]
3. Gill TM, Allore HG, Holford TR, et al. Hospitalization, restricted activity, and the development of disability among older persons. JAMA. 2004;292:2115–2124. [PubMed]
4. Sager MA, Franke T, Inouye SK, et al. Functional outcomes of acute medical illness and hospitalization in older persons. Arch Intern Med. 1996;156:645–652. [PubMed]
5. Fortinsky RH, Covinsky KE, Palmer RM, et al. Effects of functional status changes before and during hospitalization on nursing home admission of older adults. J Gerontol A Biol Sci Med Sci. 1999;54A:M521–M526. [PubMed]
6. Jencks SF, Williams MV, Coleman EA. Rehospitalizations among patients in the Medicare fee-for-service program. N Engl J Med. 2009;360:1418–1428. [PubMed]
7. Mor V, Wilcox V, Rakowski W, et al. Functional transitions among the elderly: Patterns, predictors, and related hospital use. Am J Public Health. 1994;84:1274–1280. [PubMed]
8. Boyd CM, Landefeld CS, Counsell SR, et al. Recovery of activities of daily living in older adults after hospitalization for acute medical illness. J Am Geriatr Soc. 2008;56:2171–2179. [PMC free article] [PubMed]
9. Gill TM, Allore H, Holford TR, et al. The development of insidious disability in activities of daily living among community-living older persons. Am J Med. 2004;117:484–491. [PubMed]
10. McCusker J, Kakuma R, Abrahamowicz M. Predictors of functional decline in hospitalized elderly patients: A systematic review. J Gerontol A Biol Sci Med Sci. 2002;57A:M569–M577. [PubMed]
11. Hoogerduijn JG, Schuurmans MJ, Duijnstee MS, et al. A systematic review of predictors and screening instruments to identify older hospitalized patients at risk for functional decline. J Clin Nurs. 2007;16:46–57. [PubMed]
12. Sutton M, Grimmer-Somers K, Jeffries L. Screening tools to identify hospitalised elderly patients at risk of functional decline: A systematic review. Int J Clin Pract. 2008;62:1900–1909. [PubMed]
13. Inouye SK, Wagner DR, Acampora D, et al. A predictive index for functional decline in hospitalized elderly medical patients. J Gen Intern Med. 1993;8:645–652. [PubMed]
14. McCusker J, Bellavance F, Cardin S, et al. Detection of older people at increased risk of adverse health outcomes after an emergency visit: The ISAR screening tool. J Am Geriatr Soc. 1999;47:1229–1237. [PubMed]
15. Sager MA, Rudberg MA, Jalaluddin M, et al. Hospital Admission Risk Profile (HARP): Identifying older patients at risk for functional decline following acute medical illness and hospitalization. J Am Geriatr Soc. 1996;44:251–257. [PubMed]
16. Verbrugge LM, Jette AM. The disablement process. Soc Sci Med. 1994;38:1–14. [PubMed]
17. Gill TM, Williams CS, Richardson ED, et al. Impairments in physical performance and cognitive status as predisposing factors for functional dependence among nondisabled older persons. J Gerontol A Biol Sci Med Sci. 1996;51A:M283–M288. [PubMed]
18. Landefeld CS, Palmer RM, Kresevic DM, et al. A randomized trial of care in a hospital medical unit especially designed to improve the functional outcomes of acutely ill older patients. N Engl J Med. 1995;332:1338–1344. [PubMed]
19. Walter LC, Brand RJ, Counsell SR, et al. Development and validation of a prognostic index for 1-year mortality in older adults after hospitalization. JAMA. 2001;285:2987–2994. [PubMed]
20. Counsell SR, Holder CM, Liebenauer LL, et al. Effects of a multicomponent intervention on functional outcomes and process of care in hospitalized older patients: A randomized controlled trial of Acute Care for Elders (ACE) in a community hospital. J Am Geriatr Soc. 2000;48:1572–1581. [PubMed]
21. Lawton MP, Brody EM. Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontologist. 1969;9:179–186. [PubMed]
22. Covinsky K, Palmer RM, Counsell SR, et al. Retrospective reports of functional status in hospitalized elders: Evidence of validity. J Am Geriatr Soc. 2000;48:164–169. [PubMed]
23. Pfeiffer E. A short portable mental status questionnaire for the assessment of organic brain deficit in elderly patients. J Am Geriatr Soc. 1975;23:433–441. [PubMed]
24. Katz S, Ford AB, Moskowitz RW, et al. Studies of illness in the aged: The index of ADL: A standardized measure of biological and psychosocial function. JAMA. 1963;185:914–919. [PubMed]
25. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chronic Dis. 1987;40:373–383. [PubMed]
26. Knaus WA, Draper EA, Wagner DP, et al. APACHE II: A severity of disease classification system. Crit Care Med. 1985;13:818–829. [PubMed]
27. Irwin M, Artin KH, Oxman MN. Screening for depression in the older adult: Criterion validity of the 10-item Center for Epidemiological Studies Depression Scale (CES-D) Arch Intern Med. 1999;159:1701–1704. [PubMed]
28. Idler EL, Kasl SV. Self-ratings of health: Do they also predict change in functional ability? J Gerontol B Psychol Sci Soc Sci. 1995;50:S344–S353. [PubMed]
29. Lawton MP, Brody EM. Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontologist. 1969;9:179–186. [PubMed]
30. Stewart AL, Ware JE, Jr, Brook RH. Advances in the measurement of functional status: Construction of aggregate indexes. Med Care. 1981;19:473–488. [PubMed]
31. Lindenberger EC, Landefeld CS, Sands LP, et al. Unsteadiness reported by older hospitalized patients predicts functional decline. J Am Geriatr Soc. 2003;51:621–626. [PubMed]
32. Carlin JB, Carlin JB, Royston P. A new framework for managing and analyzing multiply imputed data in Stata. Stata J. 2008;8:49–67.
33. Diehr P, Patrick DL. Trajectories of health for older adults over time: Accounting fully for death. Ann Intern Med. 2003;139:416–420. [PubMed]
34. King J. Running a best-subsets logistic regression: An alternative to stepwise methods. J Educ Psychol Meas. 2003;63:392.
35. Sullivan LM, Massaro JM, D’Agostino RB., Sr Presentation of multivariate data for clinical use: The Framingham Study risk score functions. Stat Med. 2004;23:1631–1660. [PubMed]
36. Justice AC, Covinsky KE, Berlin JA. Assessing the generalizability of prognostic information. Ann Intern Med. 1999;130:515–524. [PubMed]
37. McGinn TG, Guyatt GH, Wyer PC, et al. Users’ guides to the medical literature: XXII: How to use articles about clinical decision rules. Evidence-Based Medicine Working Group. JAMA. 2000;284:79–84. [PubMed]
38. Harrell FE, Jr, Lee KL, Mark DB. Multivariable prognostic models: Issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med. 1996;15:361–387. [PubMed]
39. Redelmeier DA, Lustig AJ. Prognostic indices in clinical practice. JAMA. 2001;285:3024–3025. [PubMed]
40. Working Group on Functional Outcome Measures for Clinical Trials. Functional outcomes for clinical trials in frail older persons: Time to be moving. J Gerontol A Biol Sci Med Sci. 2008;63:160–164. [PMC free article] [PubMed]
41. Stuck AE, Walthert JM, Nikolaus T, et al. Risk factors for functional status decline in community-living elderly people: A systematic literature review. Soc Sci Med. 1999;48:445–469. [PubMed]
42. Coleman EA, Parry C, Chalmers S, et al. The care transitions intervention: Results of a randomized controlled trial. Arch Intern Med. 2006;166:1822–1828. [PubMed]
43. Naylor M, Brooten D, Jones R, et al. Comprehensive discharge planning for the hospitalized elderly. A randomized clinical trial. Ann Intern Med. 1994;120:999–1006. [PubMed]
44. Rubenstein LZ, Goodwin M, Hadley E, et al. Working group recommendations: Targeting criteria for geriatric evaluation and management research. J Am Geriatr Soc. 1991;39:37S–41S. [PubMed]
45. Sennour Y, Counsell SR, Jones J, et al. Development and implementation of a proactive geriatrics consultation model in collaboration with hospitalists. J Am Geriatr Soc. 2009;57:2139–2145. [PubMed]
46. de Morton NA, Keating JL, Jeffs K. Exercise for acutely hospitalised older medical patients. Cochrane Database Syst Rev. 2007:CD005955. [PubMed]
47. Milne AC, Potter J, Avenell A. Protein and energy supplementation in elderly people at risk from malnutrition. Cochrane Database Syst Rev. 2005:CD003288. [PubMed]
48. Covinsky KE, Palmer RM, Counsell SR, et al. Functional status before hospitalization in acutely ill older adults: Validity and clinical importance of retrospective reports. J Am Geriatr Soc. 2000;48:164–169. [PubMed]
49. Chuang KH, Covinsky KE, Sands LP, et al. Diagnosis-related group-adjusted hospital costs are higher in older medical patients with lower functional status. J Am Geriatr Soc. 2003;51:1729–1734. [PubMed]
50. Inouye SK, Rushing JT, Foreman MD, et al. Does delirium contribute to poor hospital outcomes? A three-site epidemiologic study. J Gen Intern Med. 1998;13:234–242. [PMC free article] [PubMed]
51. Levit K, Wier L, Stranges E, et al. HCUP Facts and Figures: Statistics on Hospital-based Care in the United States, 2007. Rockville, MD: Agency for Healthcare Research and Quality; 2009. ( http://www.hcup-us.ahrq.gov/reports.jsp)