Summary
Freedman, Martin, and Schoeni (2002) conduct the first systematic review of late-life disability and functioning trends in the United States, synthesizing 16 articles from 8 national surveys rated on 10 methodological criteria. They find consensus declines in any-disability (−0.92% to −1.55%/yr) and instrumental activities of daily living (IADL) disability (−0.40% to −2.74%/yr) among the elderly in the 1980s–1990s, but conflicting evidence on activities of daily living (ADL) disability — the most severe and expensive form — where only the National Long Term Care Survey (NLTCS) showed declines while other surveys showed no change or increases. Educational disparities widened: declines were concentrated among those with 13+ years of education and absent for less-educated groups.
Key Claims
- Consensus on any-disability and IADL decline: All 3 good/fair surveys showed statistically significant declines in any-disability (−0.92% to −1.55%/yr); all 4 good/fair surveys agreed on IADL decline (−0.40% to −2.74%/yr). These findings cover the 1982–1999 window depending on the survey.
- ADL disability: conflicting evidence: Of 6 surveys assessing ADL disability, 4 were rated good or fair but produced estimates ranging from −1.38% to +1.53%/yr. Only the NLTCS found declines; the Supplements on Aging (Liao et al. 2001) found statistically significant increases (+1.53%/yr) and Crimmins and Saito (2000) found increases in mean ADL limitations for men. This is the most policy-relevant finding: ADL disability drives the majority of long-term care costs.
- Functional limitations declined: Fair-rated surveys consistently showed declines in physical task difficulties (lifting, climbing stairs, walking ¼ mile) among those aged 50+ and 65+ from 1984–1993, −0.75% to −1.62%/yr.
- Cognitive impairment: preliminary decline: The Asset and Health Dynamics of the Oldest Old study (AHEAD) showed severe cognitive impairment fell from 6.1% to 3.6% (1993–1998, −6.83%/yr); the National Mortality Followback Study (NMFS) showed declines of −1.72% to −3.28%/yr by sex and age. Both labeled preliminary pending replication.
- Sensory limitations: mixed: Vision difficulty declined significantly in the Survey of Income and Program Participation (SIPP) (−2.26% to −2.42%/yr); hearing limitations remained constant in Supplements on Aging (1984–1995).
- Educational disparities widened: National Health Interview Survey (NHIS) data show disability declined only for those with 13 or more years of education; those with 0–8, 9–11, or 12 years showed no significant decline. The difference was statistically significant (p=.01–.06 across groups). This widening was the only disability disparity to survive a formal statistical test for differential trends.
- Racial disparities: possible narrowing, untested: AHEAD showed larger cognitive decline for nonwhites than whites (1993–1998); SIPP showed larger functional limitation declines for blacks than whites/others (1984–1993) — but neither was tested statistically.
- Causal mechanisms unknown: Education and demographic composition shifts are correlated with improvement across surveys. Direct disease contributions (cardiovascular, arthritis) explain some of the functional limitation decline. No study had established causal mechanisms; role of assistive technology and physical environment was speculative.
- Survey quality heterogeneity: Only 2 of 8 surveys rated "good" (NHIS and NLTCS); most surveys lacked one or more key features (national coverage including institutionalized, identical repeat cross-sections, sample size to detect 1–2% change/year). This drives the conflicting ADL results; method quality, not genuine disagreement in health trends, likely explains much of the cross-survey variance.
- Policy implication: Because the most expensive form of disability (ADL) showed conflicting trends and causal mechanisms were unknown, projections for future long-term care demand based on single-study findings were unreliable. The wide range of estimates across surveys — and the dearth of statistical tests for disparities — limited actionable policy conclusions.
Concepts Introduced or Extended
- Compression of Morbidity — first quality-rated synthesis of the U.S. evidence; identified IADL/any-disability compression as consistent but ADL evidence as conflicting; highlighted educational disparities as a key moderator
- Disability-Free Life Expectancy — documents 1980s–1990s trend basis for compression inference; IADL/functional limitation declines underpin DFLE gains; identifies methodological threats (ADL question wording, institutional coverage) to valid trend estimation
Entities Mentioned
Quotes
"Several measures of old age disability and limitation have shown improvements in the last decade. Research into the causes of these improvements is needed to understand the implications for the future demand for medical care."
"Among the 3 surveys providing trend estimates for the prevalence of 'any disability'… all 3 showed statistically significant declines."
"[For ADL disability,] 4 [surveys] were rated as good or fair but offered conflicting evidence resulting in a wide range of estimates of the average annual percent change, ranging from −1.38% per year to 1.53% per year."
"Declines in any disability were significantly larger for those with more than a high school education compared with those with just a high school education or less."
My Take
This is a foundational systematic review — the first to synthesize and quality-rate the full evidence base on late-life disability trends as of 2002. Its key contribution is methodological: by evaluating surveys on 10 criteria and providing a summary table of results by outcome and survey quality, it identified the ADL conflict as a genuine evidentiary gap rather than a mere discrepancy. The educational disparity finding — that the aggregate disability decline masked a non-decline for less-educated groups — was a prescient warning, later confirmed by Schoeni et al. (2005) and embedded in the Crimmins/Beltrán-Sánchez (2010) disease-expansion framework. The finding that cognitive impairment declined is noteworthy; it preceded the later confirmation from better-powered studies and is not prominently featured in the subsequent compression literature. The main limitation of the paper is its pre-2003 data horizon: the compression evidence strengthened considerably with Stallard (2011), Cutler et al. (2013), and Chernew et al. (2016). The 2002 paper correctly identified where the evidence was weak and uncertain; the post-2013 literature resolved those uncertainties in favor of compression for ADL disability at age 65+.