Healthy Life Expectancy

Definition

Healthy life expectancy (HALE) is the average number of years a person can expect to live in full health — free from the burden of disease or injury. It adjusts period life expectancy (LE) downward by the time spent in states of less-than-full health, weighted by severity. HALE was developed within the Global Burden of Disease (GBD) framework and is the standard metric for cross-national comparisons of health, not merely survival.

$\text{HALE} = \text{LE} - \text{YLD (weighted by disability weights)}$

Because years lived with disability (YLDs) are subtracted from LE, HALE reflects both mortality and morbidity simultaneously. A population with improving mortality but stagnant or worsening morbidity will show HALE growing more slowly than LE — or even declining while LE rises.

Key Ideas

HALE is a more complete measure of population health than life expectancy alone, because LE measures only survival while HALE measures healthy survival.
The gap between LE and HALE — call it the "disability gap" — measures the average years spent in less-than-full health. A widening gap indicates morbidity expansion (see Morbidity-Mortality Distinction).
A country or region can simultaneously show improving LE and deteriorating HALE if morbidity rises faster than mortality falls.
HALE rankings diverge from LE rankings: the US HALE rank declined faster than its LE rank over 1990–2021, confirming that morbidity worsened relative to peers even beyond what mortality trends reveal.

Calculation

HALE is calculated using age-specific mortality rates (which determine LE) and age-specific YLDs per capita (from GBD's disease burden estimates). The Sullivan method is commonly used: for each age group, the proportion of expected years lived in less-than-full health is estimated from YLD rates, then summed across the life table.

In GBD 2021:

HALE = Sullivan-method estimate using GBD YLD rates and standard GBD life tables
Produces estimates by sex, age, location, and year
95% uncertainty intervals propagated from YLD and mortality uncertainty

US HALE: National Levels (GBD 2021)

Year	All-sex HALE	Male HALE	Female HALE
1990	64.8 years	—	—
2010	66.7 years	—	—
2019	66.2 years	—	—
2021	64.4 years	63.2	65.7

US HALE peaked around 2010, then stagnated and declined by 2021 — driven by the drug use disorder surge and COVID-19. The LE/HALE gap widened from $\approx 10.8$ years (1990) to $\approx 12.7$ years (2021), indicating morbidity expansion over the study period.

HALE Rank Decline: 1990–2021

	1990 rank	2021 rank	Change
Males	42nd of 204	69th	↓ 27
Females	32nd of 204	76th	↓ 44

HALE rank declined faster than LE rank (males: LE fell $11$ places, HALE fell $27$ ; females: LE fell $28$ , HALE fell $44$ ). This differential confirms the morbidity expansion interpretation: the US fell further behind peers in healthy survival than in raw survival.

State-Level HALE Extremes (2021)

Location	HALE (2021)	Global rank (males)
Hawaii (best US state)	68.3 years	39th
National average	64.4 years	69th
West Virginia (worst US state)	59.5 years	141st

West Virginia's HALE of $59.5$ years means the average WV male can expect fewer healthy life-years than men in $140$ other countries, including many middle-income countries. The Hawaii–West Virginia HALE gap of $8.8$ years within a single country is larger than the LE gap between many pairs of high-income and upper-middle-income countries.

HALE vs. Disability-Free Life Expectancy (DFLE)

HALE and DFLE are related but methodologically distinct concepts often used in different literatures:

HALE (GBD): uses continuous disability weights ( $0$ – $1$ scale) for each health state; a year spent with moderate low back pain is counted as a partial year of health loss, not a full year of disability. Produced for all countries and US states.
DFLE (Cutler, Ghosh, and Landrum 2013; Stallard 2011; Chernew et al. 2016; US elderly literature): uses a binary or multi-state disability indicator (activities of daily living [ADL]/instrumental activities of daily living [IADL] limitations) from survey data; years are counted as "with disability" if a functional limitation threshold is met. Stallard (2011) uses the National Long-Term Care Survey (NLTCS) chronic disability definition ( $\geq 1$ ADL or $\geq 1$ IADL for $\geq 90$ days) in a seven-state multi-state Markov model (non-disabled, IADL-only, $1$ – $2$ ADL, $3$ – $4$ ADL, $5$ – $6$ ADL, institutionalized, dead) to estimate expected years in each state from age 65 — the actuarial-demography version of DFLE that decomposes into incidence, duration, and intensity components. Cutler, Ghosh, and Landrum (2013) and Chernew et al. (2016) both use the Medicare Current Beneficiary Survey (MCBS) linked to death records; Cutler et al. document strong compression 1992–2005 (DFLE $+1.6$ yrs, disabled LE $-0.9$ yrs at age 65) and identify disease-disability decoupling as the primary mechanism via Oaxaca decomposition; Chernew et al. extend through 2008 (total LE $+1.3$ yr, DFLE $+1.8$ yr, disabled LE $-0.5$ yr) and add disease-level attribution: time-until-death structural shift ( $-4.9$ pp, the largest single component), cardiovascular disease (CVD) ( $-2.5$ pp, primarily decoupling: ischemic heart disease (IHD) disability coefficient $0.11 \to 0.06$ ), vision ( $-1.7$ pp, purely prevalence: $38.4\% \to 24.7\%$ ). Treatment attribution via IMPACT model assigns $51\%$ of IHD mortality reduction to treatment (statins $\approx 48\text{k}$ lives, angiotensin-converting enzyme (ACE)+beta-blockers $\approx 43\text{k}$ ), yielding $+0.26$ DFLE years. Cataract surgery diffusion ( $20\% \to 33\%$ of beneficiaries) explains $27\%$ of vision-related disability decline $= +0.08$ DFLE years. All are confined to age $65+$ . See Compression of Morbidity.

These measure different things. HALE captures a broader and graded burden; DFLE captures a threshold-crossing. Chernew et al. (2016) found DFLE at age 65 improved 1992–2008 (disability compression for the elderly); GBD HALE shows no equivalent improvement at working ages. The two findings are not contradictory — they measure different age groups, different disability concepts, and different time periods. See Morbidity-Mortality Distinction.

Multimorbidity and the HALE Gap

Multimorbidity — simultaneous co-occurrence of $2+$ chronic diseases — is the primary proximate mechanism by which aging populations widen the gap between LE and HALE. Marengoni et al. (2011) showed that $55$ – $98\%$ of persons $60+$ have multimorbidity, and that it consistently impairs physical functioning (quality of life [QoL], disability, functional decline) even when its mortality impact is uncertain. Every year lived in a multimorbid state contributes YLDs that the Sullivan method subtracts from LE to produce HALE. As the Demographic Transition delivers larger cohorts of very old adults into Phase 3c, multimorbidity's contribution to the HALE-LE gap will grow.

Why It Matters

Program design: Disability Insurance (DI) and Supplemental Security Income (SSI) are designed to address inability to work — a concept closer to HALE than to LE. Stagnant or declining working-age HALE is a more direct signal of pressure on disability rolls than rising LE.
Retirement age policy: Arguments for raising Social Security retirement ages based on rising LE implicitly assume HALE is rising proportionally. GBD data show HALE stagnating at working ages ( $20$ – $64$ ) even as LE rose — making age-indexing proposals implicitly regressive for the bottom of the income distribution. See Income-Mortality Gradient.
Cross-national benchmarking: The US HALE rank of $69$ th/ $76$ th (male/female) is the appropriate benchmark for assessing whether US health policy is meeting the standard of peer countries — not raw LE rank, which is more favorable.

Open Questions

Working-age ( $20$ – $64$ ) HALE is less well-measured than elderly HALE in the US because the most detailed survey data (MCBS, Health and Retirement Study [HRS]) focus on older ages.
The relationship between GBD HALE and Social Security Administration (SSA)-defined work capacity for borderline DI applicants is not directly estimable from existing data.