Hardy SE, Gill TM. Recovery From Disability Among Community-Dwelling Older Persons. JAMA. 2004;291(13):1596–1602. doi:10.1001/jama.291.13.1596
Author Affiliations: Department of Internal Medicine, Yale University School of Medicine, New Haven, Conn.
Context Previous studies have found that a sizeable minority of newly disabled
older persons recover independent function; however, long intervals between
assessments have led to difficulty in determining the true incidence and duration
of disability, and therefore in accurately characterizing the probability
and course of recovery.
Objectives To determine the rate of and time to recovery of independent function
in community-dwelling older persons who become newly disabled in their activities
of daily living (ADLs), to determine the duration of recovery, and to compare
the likelihood of recovery among pertinent subgroups of older persons.
Design, Setting, and Participants Prospective cohort study, with monthly assessments of ADL function,
for 754 initially nondisabled, community-dwelling persons aged 70 years or
older, performed in a small urban area from March 1998 to May 2003.
Main Outcome Measures Demographic features, chronic conditions, cognitive function, and physical
frailty were determined during comprehensive assessments at 18-month intervals.
Disability, defined as needing personal assistance with 1 or more key ADLs
(bathing, dressing, walking, and transferring), was assessed during monthly
Results A total of 420 participants (56%) experienced disability during a median
follow-up of 51 months. Of these participants, 399 (81%) recovered (ie, regained
independence in all 4 ADLs) within 12 months of their initial disability episode,
and a majority (57%) of these maintained independence for at least 6 months.
Among participants who experienced 3 or more consecutive months of disability,
a majority (60%) recovered, but only a third of these maintained independence
for at least 6 months. Persons who were cognitively impaired, physically frail,
or severely disabled (ie, in 3-4 ADLs) at onset were less likely to recover
than those who were cognitively intact, nonfrail, or mildly disabled, respectively.
Nonetheless, a majority of participants within each subgroup recovered.
Conclusions Newly disabled older persons recover independent ADL function at rates
far exceeding those that have been previously reported. Recovery from disability,
however, is often short-lasting, suggesting that additional efforts are warranted
to maintain independence in this high-risk group.
Disability in basic activities of daily living (ADLs) is common among
community-dwelling older persons, with prevalence rates ranging from 7% in
those aged 65 to 74 years to 24% in those aged 85 years or older.1 Although disability in older persons is often thought
to be progressive or permanent, previous research has shown that it is a dynamic
process, with individuals moving in and out of states of disability.2 Indeed, recovery rates as high as 28% have been demonstrated
in previous longitudinal studies of community-dwelling older persons that
have included assessment intervals of 12 to 24 months.1,3,4 More
recent evidence has demonstrated that assessment intervals longer than 3 to
6 months lead to incomplete ascertainment of disability and that this incomplete
ascertainment is largely due to recovery from disability.5 These
results suggest that recovery may be considerably more common than previous
studies have indicated.
To set realistic goals and plan for appropriate care, disabled older
persons, along with their families and clinicians, need accurate information
about the likelihood and time course of recovery. The objectives of this study
were to determine the rate of and time to recovery of independent function
in community-dwelling older persons who become newly disabled in their ADLs,
to determine the duration of recovery, and to compare the likelihood of recovery
among pertinent subgroups of older persons.
The study population was drawn from members of the Precipitating Events
Project (PEP), a longitudinal study of 754 community-dwelling persons, aged
70 years or older, who were nondisabled (ie, required no personal assistance)
in 4 key ADLs—bathing, dressing, walking inside the house, and transferring
from a chair. The assembly of the cohort, which took place between March 1998
and October 1999, is summarized in Figure
1 and has been described in detail elsewhere.6 Potential
participants were identified from a computerized list of 3157 age-eligible
members of a large health plan in greater New Haven, Conn. To minimize potential
selection effects, each member was assigned a unique number using a computerized
randomization program, and screening for eligibility and enrollment proceeded
sequentially. Eligibility was determined during a screening telephone interview
and was confirmed during an in-home assessment. Persons who were physically
frail, as denoted by a timed score of greater than 10 seconds on the rapid
gait test (ie, walking back and forth over a 10-foot course as quickly as
possible), were oversampled to ensure a sufficient number of participants
at increased risk for ADL disability.7,8 Slow
gait speed has repeatedly been shown to be the single best predictor of ADL
Potential participants were excluded if they had a life expectancy less
than 12 months, planned to move out of the New Haven area during the next
12 months, or were unable to speak English. Participants with significant
cognitive impairment (as defined below) were excluded only if they had no
available proxy.5 Only 4.6% of the 2753 health
plan members who were alive and could be contacted refused to complete a screening
telephone interview, and 75.2% of the 1002 eligible members agreed to participate
in the study. Persons who refused to participate did not differ significantly
from those who were enrolled in terms of age or sex. The study protocol was
approved by the Yale Human Investigation Committee, and all participants gave
verbal informed consent.
PEP participants underwent comprehensive in-home assessments at baseline,
18, and 36 months and had monthly telephone interviews for up to 53 months.
The comprehensive assessments were completed by trained research nurses using
standard instruments. In addition to gait speed, data were collected on demographic
characteristics9; self-reported, physician-diagnosed
chronic conditions, namely, hypertension, myocardial infarction, congestive
heart failure, stroke, diabetes, arthritis, hip fracture, chronic lung disease,
and cancer (other than minor skin cancers); and cognitive function.11
During monthly telephone interviews, participants were assessed for
disability in 4 key ADL tasks—bathing, dressing, walking, and transferring.
Interviewers used standard questions4,8 that
have been described in detail elsewhere.5 For
each ADL, participants were asked, "At the present time, do you need help
from another person to" perform the task? Those participants who needed help
with or were unable to complete 1 or more of the ADL tasks were considered
disabled. Participants were not asked about eating, toileting, or grooming
because the incidence of disability in these 3 ADLs is low among community-dwelling
older persons,7,8 and disability
in these ADLs is uncommon without concurrent disability in bathing, dressing,
walking, or transferring.7,8,12 Among
a subgroup of 91 participants who were interviewed twice within a 2-day period
by different interviewers, we found that the reliability of our disability
assessment was substantial,13 with κ
= 0.75 for disability in 1 or more of the 4 ADLs; κ was 1.0 for the
18 paired interviews that were completed independently by different interviewers
on the same day. A designated proxy, defined as a person who is cognitively
intact and who either lives with the participant or visits the participant
at least 3 days per week,5 completed the interviews
for participants who had significant cognitive impairment, defined as recall
of none of the 3 items on the short-term memory portion of the Folstein Mini-Mental
State Examination11 (MMSE) or a score of less
than 20 on the MMSE and recall of 1 or 2 of the 3 memory items.5 The
accuracy of these proxy reports was found to be excellent, with κ =
Follow-up interviews completed through May 2003 were included. One hundred
fifty-one participants (20%) died after a median follow-up of 30 months, and
31 (4.0%) dropped out of the study after a median follow-up of 21 months.
Data are otherwise available for 99.4% of the remaining 34 220 monthly
telephone interviews. Seven percent (2521/34 014) of the interviews were
completed by proxy. Our results did not change appreciably when interviews
with proxies were excluded. Of the 641 participants interviewed at 12 months,
73 (10.2%) were disabled in 1 or more ADLs, a rate that is consistent with
previous point estimates of disability among previously nondisabled community-living
persons aged 70 years or older.1,4
Participants who reported ADL disability during at least 1 month of
the follow-up period were considered to have experienced disability. Disability
at onset was defined as severe if it was reported in 3 or 4 ADLs in the initial
month,14 and as mild if it was reported in
1 or 2 ADLs. Recovery occurred in the first month during which a participant
reported no disability in any of the 4 key ADLs.
We compared the baseline characteristics of the PEP participants who
did and did not experience ADL disability using the t test
for continuous variables and the χ2 test for categorical variables.
Subsequent analyses included only participants who experienced disability.
We entered PEP participants into the analytic sample at the time of their
first episode of disability and followed them up until they recovered independence
(ie, reported no disability in any of the 4 key ADLs), died, or were lost
to follow-up. Participants who recovered independence were subsequently followed
up until they developed recurrent disability, died, or were lost to follow-up.
We did not adjust for the original sampling strategy since the analytic sample
represented a select subgroup of participants who had developed disability
at differing times over the course of 53 months. Instead, we evaluated the
probability of recovery by physical frailty as described below.
We calculated Kaplan-Meier estimates of recovery over time (presented
as recovery curves) for all participants in the analytic sample. We censored
participants who had not recovered 12 months after the initial onset of disability,
since few participants remained eligible to recover after 12 months. To address
the potential concern that brief episodes of disability (ie, those lasting
only 1 month) could represent measurement error or very transient conditions,
we also calculated Kaplan-Meier estimates of recovery over time for "persistent"
disability,5,15 defined as a new
disability that was present for at least 2 consecutive months, and for "chronic"
disability,15,16 defined as a
new disability that was present for at least 3 consecutive months. To simplify
our presentation and avoid violating the statistical assumption of independence,
only the first episodes of any, persistent, and chronic disability were included,
respectively, for each participant in the time-to-recovery analyses. Thus,
a participant whose first episode of disability lasted 1 month and whose second
episode lasted 4 months would have the first episode included in the analyses
of any disability and the second episode included in the analyses of persistent
and chronic disability. We defined zero-time (ie, the time at which a participant
becomes eligible to recover)17 as the first
month of disability for any disability, the second month of disability for
persistent disability, and the third month of disability for chronic disability.
For each type of disability, we present the percentage of participants who
recovered and, among those who recovered, the percentage of participants who
achieved, respectively, 2 or more consecutive months and 6 or more consecutive
months of independence. We also present the mean duration of recovery, defined
as the time from regaining independence to recurrent disability, death, or
loss to follow-up.
Finally, we compared the likelihood of recovery among pertinent subgroups
of older persons using Kaplan-Meier recovery curves and the log-rank test.
Subgroups were defined on the basis of age at onset of disability, sex, cognitive
function (as measured by the MMSE), physical frailty, and severity of disability
at onset, each assessed at the most recent comprehensive assessment. A composite
"worst case" subgroup, which included participants who were physically frail,
who had MMSE scores less than 28, and who had severe disability at onset,
was also created and compared with the remaining participants. While not exhaustive,
the aforementioned subgroups reflect the most pertinent demographic features
and prognostic factors for disability.18,19
All analyses were performed using SAS version 8,20 and
all P values are 2-tailed. P<.05
was considered significant.
Of the 754 PEP participants, 420 (52%) experienced at least 1 month
of ADL disability during a median follow-up of 51 months. The baseline characteristics
of the PEP participants who did and did not experience ADL disability are
presented in Table 1. Participants
who experienced disability were older, were more likely to be physically frail,
and had fewer years of education, lower cognitive function, and more chronic
conditions than participants who did not experience disability.
Of the newly disabled participants, 339 (81%) recovered independence
within 12 months. Only 3 participants (<1%) recovered after more than 12
consecutive months of disability, and each of these participants experienced
only a single month of subsequent independence. The majority of disability
episodes were brief, with 272 (65%) lasting only 1 or 2 months.
Of the 754 PEP participants, 283 (38%) experienced an episode of persistent
disability and 217 (29%) experienced an episode of chronic disability. Of
the participants who experienced persistent and chronic disability, respectively,
193 (68%) and 131 (60%) subsequently recovered independence within 12 months.
Figure 2 shows Kaplan-Meier
estimates for the probability of recovering independent function for any,
persistent, and chronic disability. For all 3 of these types of disability,
the majority of participants recovered, and the vast majority of participants
who recovered did so within 6 months.
For participants who recovered independent function, the mean (SD) duration
of recovery was 11.9 (13.0) months for any disability, 6.5 (8.5) months for
persistent disability, and 5.3 (7.3) months for chronic disability. Table 2 shows the proportions of participants
who maintained independence for 2 or more months and 6 or more months, respectively,
among participants with any, persistent, and chronic disability. For each
of these 3 types of disability, a majority of participants who recovered maintained
independence for 2 or more months. Whereas a majority of participants who
recovered from any disability maintained independence for 6 or more months,
only a minority of participants who recovered from persistent or chronic disability
maintained independence for 6 or more months.
Figure 3 shows the Kaplan-Meier
estimates for the probability of recovery among pertinent subgroups of participants
with any disability. While recovery differed little by age or sex, participants
who had at least mild cognitive impairment (ie, MMSE scores ≤27), who were
physically frail, and who had severe disability were less likely to recover
independent function than participants who were not frail, who were cognitively
intact, and who had mild disability, respectively. Nonetheless, for all subgroups,
a majority of participants recovered independence. A majority (62%) also recovered
independence among participants with the worst combination of prognostic factors,
including cognitive impairment, physical frailty, and severe disability. Similar
results by subgroup were found for persistent and chronic disability (data
not shown), except that recovery rates across subgroups were lower.
In the current study, which included monthly assessments of ADL function,
we found that the vast majority of newly disabled community-dwelling older
persons recovered independent function, usually within the first 6 months
after disability onset. For those who recovered, independent function was
sustained for at least 6 months among a majority of persons with disability
of any duration, but only among a minority of persons with disability lasting
2 or more months. Persons who were cognitively impaired, physically frail,
or severely disabled at onset were less likely to recover than those who were
cognitively intact, nonfrail, or mildly disabled, respectively, but a majority
of participants with any disability recovered within each subgroup.
Our rates of recovery from ADL disability are much higher than those
that have been reported in previous studies.1,3,4 Our
results are unlikely to be due to measurement error, as the reliability of
our disability assessment was high, and persons with persistent and chronic
disability also had high rates of recovery. Because we oversampled persons
with physical frailty, our rates of recovery may actually underestimate the
true rate in the general population of community-dwelling older persons. While
the point prevalence of disability in our population was comparable to rates
reported in previous studies of community-dwelling older persons,1,4 the frequency of our assessments enabled
us to ascertain brief episodes of disability that are disproportionately missed
in longitudinal studies with assessment intervals of 6 to 24 months,5 likely accounting for our higher recovery rates. In
fact, brief episodes of disability were very common in our study population,
with over half of the initial disability episodes lasting only 1 or 2 months.
While the clinical relevance of short-term disability has been questioned,21 we have recently demonstrated that disability lasting
only 1 or 2 months is strongly associated with the development of future disability
Many of our participants developed recurrent disability, as evidenced
by the large minority of persons with any disability who did not maintain
independence for more than 6 months. In a recently published report, we found
that more than half of the PEP participants who experienced any disability
during a 2-year period experienced multiple episodes.15 In
one of the few other studies with assessment intervals less than 6 months,
Verbrugge and colleagues2 found that functional
status among a sample of 165 older persons following hospitalization often
fluctuated substantially during the course of a year, with many persons neither
consistently improving nor worsening. These results, together with ours, demonstrate
that the disabling process among many older persons is complex, with multiple
and possibly interrelated disability episodes, even over relatively short
periods of time. While other studies have incorporated multiple transitions
between disabled and independent states in models of the disabling process,23,24 the effects of prior disability episodes
on recovery from future episodes have not been explicitly considered.
The dynamic nature of disability among our participants raises important
questions regarding much prior research on the recovery process. In studies
with assessment intervals of 6 or more months, many participants likely experienced
multiple transitions between states of disability and independence within
each assessment interval. Studies of ADL recovery after specific events (eg,
hospitalization, hip fracture, or stroke) that have included assessments of
functional status 6 or more months later may have assessed participants during
or after a subsequent (rather than the initial) episode of disability. For
example, one study of recovery after hip fracture, which included follow-up
assessments at 6 and 12 months, found that about 10% to 20% of participants
had recovered at 6 months, but had declined again at 12 months.25 Studies
such as ours, which include frequent assessments of functional status, may
allow the course of recovery to be characterized more accurately. Although
recovery rates were consistently high among our participants, the duration
of recovery varied widely. This variation suggests that there may be different
patterns of recovery with potentially different predictive factors. Further
research is needed to elucidate the different patterns of recovery and to
determine predictors of these patterns.
While our finding that cognitive function and severity of disability
are associated with the likelihood of recovery is consistent with previous
research,1,3 we found no difference
in recovery among age groups, in contrast to most prior studies.1,3,26 Because
these other studies had long assessment intervals, it is possible that the
oldest old were more likely than the younger old to have died or experienced
another distinct disability episode prior to the next follow-up interval.
In fact, Hansen and colleagues,27 who also
used a short assessment interval of 1 month, found no age effect on recovery
from disability after hospitalization.
Unlike other studies, we did not evaluate recovery after a single disease
process or injury, such as stroke or hip fracture, nor did we have information
on the etiology of disability, which may not be readily apparent in the absence
of a catastrophic event.28 Disability, like
delirium and other geriatric syndromes, is thought to result from the interaction
of predisposing factors and precipitating events.29- 32 Disability
may have either a rapid or gradual onset, and many episodes of disability
are not preceded by an acute illness or injury leading to hospitalization.32,33 Because the likelihood and course
of recovery may differ depending on the type of precipitating event (eg, a
surgical procedure vs an acute illness vs a stressful life event),34 further research is warranted to evaluate the effect
of specific precipitating events on the recovery process, particularly noncatastrophic
events that have received relatively little attention to date. Many older
persons, for example, report common symptoms such as pain, weakness, and fatigue
as the cause of prevalent disability,28 and
recent evidence indicates that events leading to restricted activity are independently
associated with decline in ADL function.35
While the high rates of recovery across multiple subgroups of older
persons indicate that the short-term prognosis for any individual episode
of ADL disability is quite good, the high rates of recurrent disability suggest
the need for a paradigm shift on how ADL disability is viewed clinically.
In addition to treating the individual episodes, clinicians might be advised
to manage disability in the context of the chronic disease model. Buchner
and Wagner30 have described a state of reduced
physiologic reserve associated with increased susceptibility to disability.
Prevention of functional decline and disability would include not only management
of acute episodes of disability and promotion of recovery, but also ongoing
evaluation and management of key risk factors for disability and use of preventive
interventions. The high likelihood of recurrent disability among older persons
suggests that those who have recently recovered from an episode of disability
are an important target population for preventive interventions. While some
interventions designed to prevent recurrent disability may be disease-specific,
eg, anticoagulation after embolic stroke, others may be broadly applicable
regardless of the specific precipitant of disability, eg, exercise-based programs.36,37 Further research is needed to determine
the causes of recurrent disability episodes and to elucidate the relationships
Several other aspects of our study deserve comment. First, because our
participants were members of a single health plan in a small urban area, our
results may not be generalizable to older persons in other settings. However,
our population did reflect the demographic characteristics of persons aged
65 years or older in New Haven County, which are comparable to the United
States as a whole, with the exception of race (New Haven County has a larger
proportion of non-Hispanic whites in this age group than the United States,
91% vs 84%).38 Furthermore, generalizability
depends not only on the characteristics of the study population, but also
on its stability over time.39 The high participation
and follow-up rates of our study both enhance the generalizability of our
findings. Second, we had no information on the possible use of restorative
interventions among our participants after the onset of disability. Finally,
our study focused on basic ADLs because they are essential for living independently.
However, transitions between independence and disability for instrumental
ADLs and mobility are likely to be as common, suggesting that studies evaluating
disability in these higher-level tasks over shorter intervals need to be undertaken.
In summary, recovery from disability in essential ADLs among community-dwelling
older persons is much more common than previous studies have indicated, but
is often transient. While most newly disabled older persons can be reassured
that they will regain independent function, those who recover are at high
risk for recurrent disability. Our results provide additional evidence that
disability is a recurrent rather than an enduring condition and suggest that
interventions to maintain independence after recovery are needed.