← All metrics

Social connection as a mortality risk factor

Each row is one study of death risk from social connection. The bright line is what that study found. The band around it is the range the data still supports — it fades where the evidence thins. Left of the centre line is lower risk.

what the study foundstill plausibleno change
highest vs lowest social isolation
Naito et al. 2023
33% higher
range 26–41% higher
loneliness vs no loneliness (pooled, both genders)
Rico-Uribe et al. 2018
22% higher
range 10–35% higher
loneliness vs no loneliness (older adults, pooled)
Zhou et al. 2023
9% higher
range 6–12% higher

What it is

This calculator asks a three-level self-report of how often you have meaningful contact with friends or family (Weak / Average / Strong). That is closer to social isolation — contact frequency — than to loneliness or living-alone. It does not compute a social-network index.

Why it matters for longevity

The engine is constructed: Strong = 0.85, Average = 1.0, Weak = 1.25. No paper reports that three-level table. Wang 2023 (90 cohorts, 2,205,199 people) is the matching-construct split meta: isolation ACM 1.32 (1.26–1.39), I² 77.8%; loneliness 1.14 (1.08–1.20), I² 91.1% — two separate pools. Combined isolation+loneliness 1.18, smaller than isolation alone. Weak 1.25 is milder than Wang’s isolated-vs-not 1.32; 0.85 is not 1/1.32. Holt-Lunstad 2015 is three separate ORs (isolation 1.29, loneliness 1.26, living alone 1.32), not HRs, not a joint model. Naito 2023 isolation-only HR 1.33. Nakou 2025 older-adult isolation 1.35 / loneliness 1.14 / living-alone 1.21. Engine not moved.

How to improve it

  • Schedule regular contact with close friends/family
  • Join a group around a shared interest
  • Prioritize a few deep ties over many shallow ones

Evidence, by endpoint

EndpointGradeFinding
—METHOD
0.85 / 1.0 / 1.25 is constructed
Strong / Average / Weak contact frequency. Average = 1.0 is the typical. Weak 1.25 is milder than Wang’s isolation 1.32. 0.85 is not 1/1.32. Engine not moved.
All-cause mortalityC
Isolation 1.32, loneliness 1.14 — two pools (Wang 2023)
90-cohort ACM split: isolation 1.32 (1.26–1.39), loneliness 1.14 (1.08–1.20). Combined 1.18, not stacked. Holt-Lunstad’s 1.29 / 1.26 / 1.32 are ORs for three separate pools, not this engine.
All-cause mortalityC
1.35 in Older Adults, Still Not 1.25 (Nakou 2025)
86-study older-adult meta (65+): isolation ACM HR 1.35 (1.27–1.43); loneliness 1.14 (1.10–1.18); living alone 1.21 (1.13–1.30) — three separate constructs, not one model. Isolation’s 1.35 is the closest age-matched analogue to this slider’s Weak-contact penalty, still not the engine’s constructed 1.25.
All-cause mortalityC
The Original Three: 1.29 / 1.26 / 1.32 (Holt-Lunstad 2015)
Classic meta of 70 independent prospective studies (3,407,134 people, mean follow-up 7 years) pooled three separate constructs as odds ratios, not hazard ratios: isolation OR 1.29, loneliness OR 1.26, living alone OR 1.32 — three distinct models, not one joint estimate, comparable in size to well-established risk factors.
All-cause mortalityC
Loneliness Alone, HR 1.22 (Rico-Uribe 2018)
Loneliness-specific meta of 35 articles (77,220 people): pooled ACM HR was 1.22 (1.10–1.35) for both genders. Isolation was not assessed in this pooling, so it cannot be compared to Wang’s separate isolation estimate.
All-cause mortalityC
Older-Adult Loneliness Is Smaller, HR 1.09 (Zhou 2023)
Older-adult loneliness meta of 36 studies (128,927 people): pooled ACM HR was a modest 1.09 (1.06–1.12), though older men alone carried a higher HR 1.18 (1.04–1.33) — smaller than Nakou’s mixed-age older-adult loneliness pool of 1.14.

What argues against this

No source in this metric's set contradicts the direction of the association — every pooled estimate here says weaker social connection carries higher all-cause mortality. The strongest disagreement is on magnitude: Zhou et al. 2023's older-adult loneliness meta (36 studies, 128,927 people) found a much smaller pooled HR of 1.09 (1.06–1.12) than Nakou's mixed-age older-adult loneliness pool (1.14) or the engine's constructed Weak-tier penalty of 1.25 — suggesting that if 'weak social contact' resembles loneliness more than isolation, the engine's 1.25 may overstate the true risk.

Last reviewed 2 September 2026

Evidence

  1. Wang et al. (2023) — Nature Human Behaviour
    A systematic review and meta-analysis of 90 cohort studies of social isolation, loneliness and mortality
    View source
  2. Holt-Lunstad et al. (2015) — Perspectives on Psychological Science
    Loneliness and Social Isolation as Risk Factors for Mortality: A Meta-Analytic Review
    View source
  3. Naito et al. (2023) — PLOS ONE
    Social isolation as a risk factor for all-cause mortality: Systematic review and meta-analysis of cohort studies
    View source
  4. Nakou et al. (2025) — Aging Clinical and Experimental Research
    Loneliness, social isolation, and living alone: a comprehensive systematic review, meta-analysis, and meta-regression of mortality risks in older adults
    View source
  5. Rico-Uribe et al. (2018) — PLOS ONE
    Association of loneliness with all-cause mortality: A meta-analysis
    View source
  6. Zhou et al. (2023) — Psychiatry Research
    A meta-analysis of the association between loneliness and all-cause mortality in older adults
    View source
See how social connection fits your full score
See how it fits the 8-metric index in the calculator.
Open calculator

Educational content only — not medical advice, diagnosis, or treatment. Not a medical device; not FDA evaluated. Consult a qualified healthcare professional about your own health, and call emergency services for urgent symptoms. Full medical disclaimer →