What “healthspan” is when you refuse to let it collapse onto mortality or disease. The wiki centres a health axis that includes function and healthspan, not only length of life (Rating Outcome Importance). This page holds the construct that operationalizes that axis as a measurable, person-centred functional outcome — so that “does this lever extend healthspan?” is a question with an answerable endpoint rather than a slogan.
The construct — two person-centred functional measures, not a biomarker
(Zheng et al., 2026) Healthspan is chronically underspecified — a recent review counted more than 100 definitions, and the most common (multimorbidity) is disease-centred. Zheng’s SR deliberately takes the alternative route: measures with «a person-centered focus on functioning rather than diagnoses».
- Intrinsic capacity (IC) — «the sum of an individual’s physical and mental abilities… proposed as an integrated measure of healthspan»; a WHO composite of 5 domains — cognition, locomotion, vitality, psychological, sensory — scored objectively. «Rather than a disease lens, IC focuses on the capacity of individuals to maintain good health».
- Quality of life (QoL) — a multidimensional indicator of self-perceived well-being; the subjective complement to IC’s objective tests (SF-36, EQ-5D, and others in the included trials).
- The construct claim: «a composite IC score would serve as a better indicator when assessing healthspan instead of single domains of IC (e.g., locomotion or cognitive domain alone)» — the whole is held to carry more than any one domain.
Why the composite matters for the outcome menu (the anti-streetlight point)
(inferred from Zheng et al., 2026) This is the outcome-menu architecture made concrete. A composite person-centred function score is an anti-streetlight device on the outcome side: it resists redefining “health” as whatever single, well-lit endpoint a trial happened to measure (a lab marker, a single cognitive test, a disease event). It is the positive-capacity mirror of Frailty — frailty measures reduced reserve / deficit accumulation; IC measures the remaining integrated ability — the two are complementary lenses on the same aging-function axis, not duplicates (frailty is deficit-defined; IC is capacity-defined).
- But the same measures are the ones the field measures WORST. QoL and self-rated function are self-reported and heterogeneous across scales — exactly the outcomes CLAUDE.md flags for more honest uncertainty, not more confident advice. Zheng’s own caveat confirms it: «self-reported data tend to overestimate the effects due to reporting bias in RCTs», compounded by the inability to blind exercise/cognitive interventions and by a Hawthorne effect. So a composite-function benefit on self-report is discounted, not taken at face value. (Zheng et al., 2026)
What moves it — the hedged evidence (single 2026 SR, recency-discounted)
(Zheng et al., 2026) 15 RCTs, 4656 participants, no meta-analysis (too heterogeneous — qualitative synthesis only). The one signal that survives the heterogeneity is exercise:
- «there is some evidence that exercise could extend intrinsic capacity and quality of life, either aerobic or resistance training alone, or a combination of different types of exercise; but further research is required to evaluate the effect of other interventions» — the exercise signal is reported in 11 of 13 exercise/multidomain studies; the «further research» caveat attaches to the other (non-exercise) interventions, not to the exercise finding. But even the exercise effect is small: «the effect sizes were generally small, and therefore, the clinical significance of these differences is unclear».
- The multidomain effect is probably the exercise component. All 6 multidomain trials included exercise; «none of the included studies investigated cognitive training or nutrition alone… the protective effect of multidomain interventions on IC and QoL may be predominantly due to the exercise component». So this is not independent support for cognitive-training or nutrition levers.
- Everything else is insufficient-evidence, not no-effect. Caloric restriction (1 study: 25% CR over 2 yr, SF-36 QoL +6.45 [3.93, 8.98]) and NMN supplementation (1 study) each showed an effect but rest on n=1; omega-3 alone was null. «no conclusion could be drawn regarding other interventions.»
Confidence: low, and recency-guarded. A single 2026 SR in a nascent, fast-moving field (the authors note multiple IC-focused trials reporting within 2-3 years) carries a lower prior of being the settled truth, per the crowded-field recency rule. Treat the exercise -> IC/QoL direction as directionally supported, magnitude small and uncertain, not established. (inferred from Zheng et al., 2026)
Decision relevance and transportability
(inferred from Zheng et al., 2026)
- No new lever, a second outcome on a lever already ranked high. Exercise is already a big rock for mortality and function (Physical Activity Dose and Mortality, Muscle-Strengthening Activity and Mortality); IC/QoL is a further patient-important outcome it moves, not a free-standing new intervention. The RCT bundle evidence for the cognitive slice of IC is developed separately and is weak/unreplicated -> Multidomain Lifestyle Intervention and Cognitive Decline (do not read that page’s cognitive-surrogate nulls as evidence about the whole IC composite — different endpoint).
- Healthy-user selection bounds transport. «due to the nature of exercise-related interventions, many studies excluded individuals with contraindications to physical activity… participants in these trials had better baseline health conditions than their peers» — so the estimates come from a healthier-than-average older stratum, and evidence «may be generalized to older Caucasians and Asians from high-income or upper-middle income countries», not younger or lower-income populations. (Zheng et al., 2026)
Open (G-gaps)
- A magnitude the SR structurally cannot give — no pooled effect size for exercise -> IC/QoL (no
meta-analysis; heterogeneous scales).
G (needs aggregation). Zheng flags multiple IC-focused RCTs underway (results expected 2-3 yr) that «will allow further quantitative synthesis» — a future meta-analysis, not a source held or yet registered, so no[AWAITS]handle is set. - Component isolation — is any non-exercise multidomain component (cognitive training, nutrition) an independent IC/QoL lever? Untested (none studied alone).
- Does moving IC/QoL move hard outcomes? IC/QoL are the person-centred targets, but the causal link from a composite-function gain to survival or disability-free years is unproven here -> Surrogate Outcomes.