Association of Functional Impairment in Middle Age With Hospitalization, Nursing Home Admission, and Death

Key Points

Question  When difficulty performing basic activities of daily living (“functional impairment”) develops in middle-aged adults, is it associated with an increased risk of adverse outcomes including hospitalization, nursing home admission, and death?

Findings  In this nationally representative matched cohort study of 5540 community-dwelling adults, those who developed functional impairment between ages 50 and 64 years were at increased risk of hospitalization, nursing home admission, and death compared with those without functional impairment.

Meaning  Even in relatively young people, functional impairment has important clinical implications.

Importance  Difficulty performing daily activities such as bathing and dressing (“functional impairment”) affects nearly 15% of middle-aged adults. Older adults who develop such difficulties, often because of frailty and other age-related conditions, are at increased risk of acute care use, nursing home admission, and death. However, it is unknown if functional impairments that develop among middle-aged people, which may have different antecedents, have similar prognostic significance.

Objective  To determine whether middle-aged individuals who develop functional impairment are at increased risk for hospitalization, nursing home admission, and death.

Design, Setting, and Participants  This matched cohort study analyzed longitudinal data from the Health and Retirement Study, a nationally representative prospective cohort study of US adults. The study population included 5540 adults aged 50 to 56 years who did not have functional impairment at study entry in 1992, 1998, or 2004. Participants were followed biennially through 2014. Individuals who developed functional impairment between 50 and 64 years were matched by age, sex, and survey wave with individuals without impairment as of that age and survey wave. Statistical analysis was conducted from March 15, 2017, to December 11, 2018.

Exposures  Impairment in activities of daily living (ADLs), defined as self-reported difficulty performing 1 or more ADLs, and impairment in instrumental ADLs (IADLs), defined similarly.

Main Outcomes and Measures  The 3 primary outcomes were time from the first episode of functional impairment (or matched survey wave, in controls) to hospitalization, nursing home admission, and death. Follow-up assessments occurred every 2 years until 2014. Competing risks survival analysis was used to assess the association of functional impairment with hospitalization and nursing home admission and Cox proportional hazards regression analysis was used to assess the association with death.

Results  Of the 5540 study participants (2739 women and 2801 men; median age, 53.7 years [interquartile range, 52.3-55.2 years]), 1097 (19.8%) developed ADL impairment between 50 and 64 years, and 857 (15.5%) developed IADL impairment. Individuals with ADL impairment had an increased risk of each adverse outcome compared with those without impairment, including hospitalization (subhazard ratio, 1.97; 95% CI, 1.77-2.19), nursing home admission (subhazard ratio, 2.62; 95% CI, 1.99-3.45), and death (hazard ratio, 2.06; 95% CI, 1.74-2.45). After multivariable adjustment, the risks of hospitalization (subhazard ratio, 1.54; 95% CI, 1.36-1.75) and nursing home admission (subhazard ratio, 1.73; 95% CI, 1.24-2.43) remained significantly higher among individuals with ADL impairment, but the risk of death was not statistically significant (hazard ratio, 1.06; 95% CI, 0.85-1.32). Individuals with IADL impairment had an increased risk of all 3 outcomes in adjusted and unadjusted analyses.

Conclusions and Relevance  Similar to older adults, middle-aged adults who develop functional impairment appear to be at increased risk for adverse outcomes. Even among relatively young people, functional impairment has important clinical implications.

The ability to perform basic activities of daily living (ADLs) such as bathing and dressing is central to older adults’ quality of life and health. When older people develop difficulty performing these activities, often called functional impairment, they experience poorer quality of life and higher rates of acute care use, nursing home admission, and death.^1-4 Thus, a major focus of care for older people is to prevent or slow the progression to functional impairment and its associated adverse outcomes.

Functional impairment is often seen as a problem affecting people 65 years or older and especially the oldest old (≥80 years). However, functional impairment is also common in middle-aged adults. Nearly 15% of adults aged 55 to 64 years have difficulty performing at least 1 ADL,⁵ a group that includes people with longstanding impairments that are congenital or developed in young adulthood, as well as people with impairments that are newly acquired in middle age. Previous research shows that individuals with longstanding disabilities experience poorer outcomes than individuals without such disabilities.^6,7 However, it is unclear if functional impairment that develops for the first time in middle age has the same prognostic importance that it does in older adults. Functional impairment in older adults results from aging-related conditions including chronic illness, frailty, and cognitive impairment.⁸ Functional impairment that develops in middle-aged people occurs earlier in the life course, before these conditions become highly prevalent. Therefore, some hypothesize that incident functional impairment in middle-aged people may be primarily a transient issue due to acute illness or injury, and thus lacks the prognostic implications of functional impairment that develops among older adults.⁹

Although few studies have tested this hypothesis, 1 study shows that functional impairment in middle age is associated more strongly with hospital readmission than are chronic conditions,¹⁰ while another study shows that functional impairment in middle age is associated with death.¹¹ However, these studies assessed prevalent functional impairment, and thus did not distinguish between longstanding impairments due to trauma or congenital conditions vs impairments that develop in middle age and may have different clinical implications.¹² In addition, these studies did not account for potential confounders of the association between function and adverse outcomes, including socioeconomic status and health-related behaviors such as smoking.⁸

Understanding the clinical implications of functional impairment among middle-aged adults is key to developing appropriate strategies to address these impairments. We used nationally representative, longitudinal data to determine if functional impairment that develops in middle age is associated with hospitalization, nursing home admission, and death, and to examine the strength of the association of function with these outcomes compared with other risk factors.

We analyzed longitudinal data from the Health and Retirement Study (HRS), an ongoing nationally representative panel survey examining changes in health and wealth among Americans older than 50 years.¹³ The first participants were enrolled in 1992 and additional participants are enrolled every 6 years so the sample remains representative of the population older than 50 years. Participants are interviewed every 2 years by telephone; face-to-face interviews are conducted for those unable to access a telephone or who are too ill to participate by telephone. The institutional review boards of the University of California, San Francisco, and the San Francisco Veterans Affairs Medical Center approved this study. Participants provide written informed consent for participation in the HRS; because we used deidentified HRS data for this study, participant consent was waived by the institutional review boards of the University of California, San Francisco, and the San Francisco Veterans Affairs Medical Center.

A total of 8430 participants were aged 50 to 56 years at enrollment in the 1992, 1998, or 2004 survey waves and were assessed for study eligibility. For the 1992 wave, we considered the baseline to be 1994, because measures of function collected in 1992 differed from later measures. We excluded 1314 participants who reported difficulty performing ADLs or instrumental ADLS (IADLs) at baseline, 326 unavailable for follow-up after baseline, and 124 with missing functional status data at baseline. We also excluded individuals for whom we could not determine whether an outcome occurred before or after the onset of functional impairment. In each biennial survey wave, participants reported their current function as well as any hospitalizations and nursing home admissions that occurred during the previous 2 years. Thus, if a participant reported both functional impairment and 1 of these events in a single wave, we could not determine which occurred first. We excluded 453 participants who reported a hospitalization or nursing home stay in the same wave or a wave preceding their first impairment, and 15 who were in a nursing home when they first reported impairment. We also excluded 477 participants who first reported impairment in the final wave (2014).

Of 5721 eligible individuals, we identified 1097 who developed functional impairment between 50 and 64 years. To create a comparison group of individuals without impairment before 65 years, we matched these individuals by age, sex, enrollment year, and wave when they reported impairment with individuals without impairment as of that age and wave. For controls who could be matched with multiple individuals who developed functional impairment, we first randomly selected an individual who developed functional impairment. We then used the wave when impairment developed in that individual as the baseline date for the control. This date served as the baseline for time-to-event analyses for both cases and controls. In total, we identified 4624 controls; we excluded 181 who could not be matched. This matching procedure created a cohort balanced by follow-up time and age at first impairment (or matched wave, among controls). We analyzed longitudinal data from these 5540 participants, collected at approximately 2-year intervals, through 2014.

We examined 2 primary factors associated with outcomes of hospitalization, nursing home admission, or death: the first episode of ADL impairment between 50 and 64 years, and the first episode of IADL impairment. The HRS uses self-report measures to assess the ability to perform ADLs and IADLs, which have been found to be valid and reliable.¹⁴ At each study wave, participants are asked if they have difficulty (yes or no) performing each of 6 ADLs (bathing, dressing, transferring, toileting, eating, walking across a room)¹⁵ and 5 IADLs (managing money, managing medications, shopping for groceries, preparing meals, making telephone calls).¹⁶ We defined ADL impairment as reporting difficulty performing 1 or more ADLs and defined IADL impairment similarly.

Functional impairment among older adults is dynamic and has complex trajectories.¹⁷ Although some people have persistent impairment after an initial episode, others improve, although those who improve are at high risk for recurrence.^18,19 To examine the association of ADL trajectories with outcomes, we defined a secondary 4-level categorical variable associated with outcomes. We categorized impairment as (1) no impairment between 50 and 64 years, (2) transient impairment, (3) persistent impairment, and (4) recurrent impairment (eAppendix 1 in the Supplement). We created a similar variable for IADL impairment.

We examined 3 outcomes: time to hospitalization, nursing home admission, and death. We used self-reported data to define hospitalization and nursing home admission, because participants were aged 50 to 56 years on enrollment and most lacked Medicare claims data. We defined hospitalization as an overnight hospitalization since the last assessment and nursing home admission as an overnight stay since the last assessment, including short stays and stays for long-term care. We determined date of death from the National Death Index and family interviews.

Sociodemographic characteristics were assessed at study entry, including self-reported age, sex, race/ethnicity, marital status, and educational level; other variables were assessed at the wave when functional impairment was first reported (or the matched wave for controls). Participants reported household income in the previous year; household net worth was calculated based on self-reported assets minus debts. Measures of health status included the following self-reported medical conditions: hypertension, stroke, diabetes, cardiac disease, chronic lung disease, cancer, and arthritis. Participants reported if they had visual impairment (fair or poor eyesight despite best correction) and hearing impairment (fair or poor hearing or use of a hearing aid). We assessed cognitive impairment using a modified version of the Telephone Interview for Cognitive Status (with scores ranging from 0 to 35; impairment was defined as a score <5),²⁰ we assessed depression using the 8-item Center for Epidemiologic Studies Depression Scale (with scores ranging from 0 to 8; depressive symptoms were defined as a score ≥3),^21,22 and we assessed body mass index as self-reported weight and height.

Measures of health-related behaviors included self-reported alcohol use,²³ smoking status, and physical activity; we defined infrequent activity as once weekly or less.²⁴ Measures of health care use and access included self-reported hospitalization 2 years before the wave when functional impairment was first reported and health insurance. Measures of the physical environment included self-reported fair or poor neighborhood safety.

Statistical analysis was conducted from March 15, 2017, to December 11, 2018. To compare characteristics between participants who developed functional impairment in middle age vs those who did not, we used the t test and the Rao-Scott χ² test. All analyses accounted for the complex HRS survey design. All P values were from 2-sided tests and results were deemed statistically significant at P < .05.

We used 2 approaches to evaluate each outcome: cumulative incidence during follow-up and time to event. To determine the cumulative incidence of each outcome among participants with or without functional impairment, we used a survival analysis framework; we used this same framework to analyze functional trajectories. To obtain weighted estimates of cumulative incidence, we used a nonparametric competing risks approach accounting for the risk of death²⁵; for the outcome of death, death was the main event with zero competing events. We defined baseline as the wave when a participant first reported functional impairment (or the matched wave in controls), and event time as the date of hospitalization, nursing home admission, or death. Because HRS assessments occur every 2 years, we could not observe the exact date of hospitalization or nursing home admission. We estimated event times to be halfway between the interview date when an event was first reported and the date of the previous assessment. Exact dates of death were available from the National Death Index. We censored participants who ended their observation period or were lost to follow-up.

To compare the risk of each outcome between participants with and participants without functional impairment, we used a survival analysis framework. We used competing risks survival analysis to compare time to hospitalization and nursing home admission between the 2 groups, with death as a competing risk,²⁵ and used Cox proportional hazards regression analysis to compare time to death. We first conducted bivariable analyses to examine the association between functional impairment and each outcome. We then conducted multivariable analyses adjusting for functional impairment and traditional risk factors, including sociodemographics, health status, and health-related behaviors. In multivariable analyses, we qualitatively compared the magnitude of association of functional impairment with each outcome with that of other risk factors. We imputed missing variables using the fully conditional specification method. For race/ethnicity, we used the discriminant function method; for other variables, we used the logistic regression method. We created 50 imputed data sets (eAppendix 2 in the Supplement). We retained all covariates in the final models. In sensitivity analyses, we used continuous variables for income, net worth, and body mass index, and a 9-level categorical variable for depression. We conducted analyses using Stata, version 15.1 (StataCorp) and SAS, version 9.4 (SAS Institute Inc).

Among 5540 participants, the median age was 53.7 years (interquartile range [IQR], 52.3-55.2 years), 2801 were men, 2739 were women, 4066 were white, and 1113 had less than a high school education (Table 1). A total of 1097 participants (19.8%) developed ADL impairment between 50 and 64 years, and 857 (15.5%) developed IADL impairment. Overall, 390 participants (7.0%) had transient ADL impairment, 411 (7.4%) had persistent ADL impairment, and 296 (5.3%) had recurrent ADL impairment; for IADL impairment, 345 (6.2%) had transient impairment, 291 (5.3%) had persistent impairment, and 221 (4.0%) had recurrent impairment. The baseline characteristics of individuals with ADL impairment differed from those without. Participants with impairment were more likely to be women, racial or ethnic minorities, unmarried, and to have lower socioeconomic status (Table 1). They also had poorer health status and were more likely to smoke, exercise infrequently, and lack health insurance.

The median follow-up times for adverse outcomes were 6.0 years for hospitalization (IQR, 3.0-10.0 years), 10.0 years for nursing home admission (IQR, 6.5-14.8 years), and 11.4 years for death (IQR, 7.4-15.6 years). The cumulative incidence of each adverse outcome during follow-up was higher among participants with ADL impairment vs those without. Among participants with ADL impairment, the cumulative incidence of hospitalization was 80.8% (95% CI, 76.8%-84.1%) vs 64.7% (95% CI, 62.3%-70.0%) among participants without impairment (Figure 1), the cumulative incidence of nursing home admission was 15.7% (95% CI, 12.1%-19.7%) among participants with ADL impairment vs 7.3% (95% CI, 6.0%-8.9%) among participants without impairment, and the cumulative incidence of death was 31.9% (95% CI, 27.7%-36.1%) among participants with ADL impairment vs 18.7% (95% CI, 16.9%-20.6%) among participants without impairment (Figure 2). Findings for participants with IADL impairment were similar (eFigures 1 and 2 in the Supplement).

Participants who developed ADL impairment in middle age had an increased risk of each outcome, including hospitalization (subhazard ratio [sHR], 1.97; 95% CI, 1.77-2.19), nursing home admission (sHR, 2.62; 95% CI, 1.99-3.45), and death (HR, 2.06; 95% CI, 1.74-2.45) (Table 2). After multivariable adjustment, the associations with hospitalization and nursing home admission were attenuated, but still statistically significant (sHR for hospitalization, 1.54; 95% CI, 1.36-1.75; sHR for nursing home admission, 1.73; 95% CI, 1.24-2.43). The association with death was not statistically significant after adjustment (HR, 1.06; 95% CI, 0.85-1.32). Participants who developed IADL impairment had an increased risk of all 3 outcomes in adjusted and unadjusted analyses (Table 2). In sensitivity analyses including continuous versions of variables, findings were similar (eTable 1 in the Supplement). Analyses of ADL trajectories showed a generally similar pattern of findings to the primary analyses (eAppendix 1 and eTable 2 in the Supplement).

The associations of ADL impairment with hospitalization and nursing home admission were as strong or stronger than traditional risk factors, including chronic illnesses, obesity, and health-related behaviors (eTables 3-5 in the Supplement). Findings for the association of IADL impairment with hospitalization and nursing home admission were generally similar to those for ADLs (eTables 6 and 7 in the Supplement). However, IADL impairment was significantly associated with death (adjusted HR, 1.39; 95% CI, 1.12-1.73), with an HR similar to the HRs for lower income and net worth, chronic conditions, and health-related behaviors (eTable 8 in the Supplement).

In this nationally representative study, middle-aged people who developed functional impairment were at increased risk for hospitalization, nursing home admission, and death. After multivariable adjustment, those with ADL impairment remained at significantly higher risk of hospitalization and nursing home admission, but the risk of death was similar to that of individuals without ADL impairment. Individuals with IADL impairment had an increased risk of all 3 outcomes in both adjusted and unadjusted analyses. For most outcomes, developing functional impairment in middle age was as strongly associated with adverse outcomes as traditional risk factors.

A large body of research conducted among older adults shows that functional impairment is strongly associated with hospitalization and nursing home admission, even after multivariable adjustment.^1-4 We found that functional impairment has similar associations in middle-aged adults. Adjusted HRs for the association of functional impairment with hospitalization and nursing home admission were 1.5 to 2.0, similar to those seen in older adults.^26,27 Furthermore, these HRs were similar in magnitude to traditional risk factors for the use of health services among middle-aged adults, including chronic illness.^28,29 Our findings are consistent with those of a previous study showing that functional impairment in middle age was as strongly associated with hospital readmission as were chronic conditions.¹⁰

Among older adults, functional impairment is strongly associated with death, even after multivariable adjustment.^3,30 In contrast, among middle-aged adults, functional impairment had mixed associations with death. Middle-aged people with ADL impairment were not at increased risk for death in adjusted analyses, while those with IADL impairment were. It is not clear why functional impairment is associated with death in older adults but not in middle-aged adults. However, our findings reveal differences in the characteristics of functional impairment by age that may influence associations with outcomes. The cumulative incidence of IADL impairment in our cohort was lower than that of ADL impairment, at 15.5% vs 19.8%, consistent with previous research among middle-aged people.³¹ In contrast, among older adults, a large body of research shows that IADL impairment is more prevalent than ADL impairment.^32,33 Among older adults, this pattern is thought to reflect a hierarchical disabling process in which cognitive impairment affects the ability to perform cognitively complex IADL tasks.^8,9 In this middle-aged cohort, the lower incidence of IADL impairment may reflect the low prevalence of cognitive impairment. These differing characteristics of functional impairment by age may lead to differing associations with outcomes.

Our findings have clinical implications. Although a large body of research shows function’s prognostic importance among older adults,^1-4 function is seldom routinely assessed in US primary care settings.^34-36 This disconnect highlights the need to change how we approach function in our patients, from a concept on the sidelines of medical care to a measure as central as chronic illnesses. Although efforts are under way to improve the assessment and management of functional impairment for older adults,^37,38 our results suggest that we should consider similar efforts for some middle-aged people. Screening all middle-aged adults for functional impairment is unlikely to be appropriate or cost-effective. However, screening people at high risk for functional impairment may help identify a population at risk for adverse outcomes who would benefit from targeted interventions.

We currently lack evidence-based interventions for middle-aged people who develop functional impairment. However, recent research suggests that, with the exception of cognitive impairment, risk factors for developing functional impairment in middle age are similar to those in older adults, including chronic medical conditions, depression, obesity, and low physical activity.¹² Moreover, as in older adults, functional impairment in middle age is multifactorial, with risk factors including health status and health-related behaviors. These similar risk profiles suggest that multifactorial interventions to address risk factors for functional impairment and its sequelae in older adults may hold promise for middle-aged adults,^39,40 although additional study is needed to determine how to adapt these interventions for this younger group.

This study has several limitations. Because we excluded individuals with baseline impairments, our findings are not generalizable to people with longstanding disabilities. We also excluded individuals without follow-up data. Because individuals with functional impairment may have difficulty completing follow-up interviews, our findings may underestimate impairment. However, the HRS interviews participants in person if they are unable to complete telephone interviews, mitigating this issue. We excluded individuals who reported hospitalization or nursing home admission in the same wave or a wave preceding their first impairment. Because hospitalization can precipitate functional impairment, our findings may underestimate the association between impairment and these outcomes. We measured function by self-report rather than objective measures, and could not include physical performance measures because they were assessed in only a small subset of participants. However, self-reported function is a key person-centered measure that is strongly associated with adverse outcomes among older adults.^1-4,41 Because function is assessed every 2 years, shorter periods of impairment may be missed, leading to underestimation of functional impairment.⁴² Because participants were younger than 65 years on enrollment, we assessed self-reported health care use rather than Medicare claims data. However, self-reported hospitalization is highly correlated with actual hospitalization as measured by claims and medical records, with relatively small reporting errors.^43-45 Last, while we used matching and multivariable regression to control for differences between people with and people without functional impairment, residual confounding may remain.

We found that functional impairment in middle-aged people is associated with an increased risk for hospitalization, nursing home admission, and death. Even after multivariable adjustment, ADL impairment and IADL impairment remain strongly associated with most of these outcomes, with associations of similar magnitude to traditional risk factors. Our findings suggest that even among relatively young people, functional impairment has important clinical implications. Interventions commonly used to improve and delay functional impairment among older adults may hold promise for improving function and reducing adverse outcomes among middle-aged adults with functional impairment.

Accepted for Publication: January 2, 2019.

Corresponding Author: Rebecca T. Brown, MD, MPH, Division of Geriatric Medicine, Perelman School of Medicine, University of Pennsylvania, 3615 Chestnut St, Ralston-Penn Center, Philadelphia, PA 19104 (rebecca.brown@pennmedicine.upenn.edu).

Published Online: April 8, 2019. doi:10.1001/jamainternmed.2019.0008

Author Contributions: Dr Brown and Ms Diaz-Ramirez had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Brown, Lee, Steinman.

Acquisition, analysis, or interpretation of data: Brown, Diaz-Ramirez, Boscardin, Williams.

Drafting of the manuscript: Brown.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Diaz-Ramirez, Boscardin.

Obtained funding: Brown.

Administrative, technical, or material support: Brown.

Supervision: Boscardin, Williams, Steinman.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was supported by grant KL2TR000143 from the National Center for Advancing Translational Sciences, National Institutes of Health, through the University of California, San Francisco–Clinical and Translational Sciences Institute (Dr Brown); grants K23AG045290 and K76AG057016 from the National Institute on Aging, National Institutes of Health (Dr Brown); a grant from the American Federation for Aging Research (Dr Brown); grant P30AG044281 from the National Institute on Aging, National Institutes of Health (all authors); grant K24AG049057 from the National Institute on Aging, National Institutes of Health (Dr Steinman); grant R01AG0478897 (Dr Lee) from the National Institute on Aging; and grant IIR 15-434 from the Veterans Affairs Health Services Research and Development (Dr Lee).

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The opinions expressed in this manuscript may not represent those of the Department of Veterans Affairs. Drs Brown, Lee, Williams, and Steinman are employees of the Department of Veterans Affairs. Dr Boscardin is a Statistical Editor of JAMA Internal Medicine, but he was not involved in any of the decisions regarding review of the manuscript or its acceptance.