Count marriage’s health benefit in, but discount it as observational data

Among older adults, married people’s probability of dying in the same period is about 12% lower than non-married people’s, and never-married people’s is about 24% higher than married people’s. But that is only two things occurring together; you cannot tell whether marriage caused it. These figures are mostly from Europe and North America; for China they are only an order-of-magnitude reference.

Cost

No money and no time. Just remember one point: these figures say married people have lower mortality in the same period — two things occurring together; you cannot tell whether marriage made you live longer.

Benefit

A meta-analysis of older adults pooled multiple studies and recalculated them, covering 53 comparisons and more than 250000 people. All were cohort studies — follow-up only, with no randomized comparison groups. Married versus non-married all-cause mortality RR 0.88, 95% CI 0.85–0.91 (RR is the risk ratio; CI is the credible range), about 12% lower probability of dying in the same period. In the same analysis, widowed 1.11, divorced or separated 1.16, never married 1.11 — about 11%, 16%, and 11% higher respectively. No difference by sex. The most conservative estimate was RR 0.94, only about 6% lower. Another meta-analysis compared never-married versus married, using 95 papers and more than 500000000 people. Result HR 1.24 (1.19–1.30), about 24% higher probability of dying in the same period. And the older the age, the smaller the gap. A further meta-analysis split by sex, 21 studies and more than 7890000 people: for both sexes, non-married status co-occurred with total mortality, cancer death, and cardiovascular death. On the male side the gaps for cardiovascular death and total mortality were larger. Never-married men versus never-married women had 31% more stroke deaths and 9% more all-cause deaths.

Notes

The controversy is that people who marry may already be selected — for example, healthier people may be more likely to marry. That is the author’s inference from study design; none of the three meta-analyses measured how large selection is. All three used observational data from existing populations — follow-up only, no randomized comparison. The authors themselves noted publication bias or method limits. Roelfs also found that the more rigorous the study, the smaller the gap calculated. These figures are mainly from European and North American samples; for China they are only an order-of-magnitude reference.

Sources · 3

1. Manzoli L, et al. (2007). Marital status and mortality in the elderly: a systematic review and meta-analysis. Soc Sci Med 64:77–94. doi.org/10.1016/j.socscimed.2006.08.031
2. Roelfs DJ, et al. (2011). The rising relative risk of mortality for singles: meta-analysis and meta-regression. Am J Epidemiol 174(4):379–389. doi.org/10.1093/aje/kwr111
3. Wang Y, et al. (2020). Sex differences in the association between marital status and the risk of cardiovascular, cancer, and all-cause mortality: a systematic review and meta-analysis of 7,881,040 individuals. Glob Health Res Policy 5:4. doi.org/10.1186/s41256-020-00133-8