Married status and lifespan

https://pubmed.ncbi.nlm.nih.gov/22975777/ p.15
if a man in the Netherlands has a wife after 65, then he lives until 81, and if not, only until 77
four years difference
How much can depend on the wife
Though it is not adjusted for many things such as alcohol consumption, smoking, healthy eating, following the recommendations of doctors, etc.

It is clear why married status might slightly help even for a highly motivated life extensionist: S.O. can call an ambulance, take care of illness, etc (also even highly motivational people sometimes lose their motivation for a while).
On the other hand, S.O. can be a source of chronic stress.
Probably much of influence of S.O. existence on lifespan comes from its influence on bad habits or from reverse causality from marriage selection (more healthy / wealthy people are more likely to get married). Perhaps also the effect is different for introverts and extroverts.

Most of papers (and especially meta-analyses) on married status influence on lifespan are not really interesting. Let's show that on one example:
https://pubmed.ncbi.nlm.nih.gov/32161813/
meta-analysis with 2 mln deaths. For men never married / married RR = 1.67 (1.52 - 1.83) (supp.fig.1c)
RR = 1.67 (1.52 - 1.83) it's about -5 years (minus 4 - 5.6)
mostly due to higher mortality of CVD -- for men, never marr/marr RR = 2.06 (1.50 - 2.82) (supp.fig.5d)
cancer mortality for men doesn't depend on married status: never married / married RR = 1.03 (0.89 - 1.18) (supp.fig.3d)
But this meta-analysis is of low quality. Let's open one of the studies with the highest weight:
https://pubmed.ncbi.nlm.nih.gov/23482379/ (Frisch, Danemark, 6.5mln adults) 21% of total weight
we see in tables 2, 4 -- they don't adjust on many important things.
While it's obvious that we need to adjust for healthy eating, physical activity, alcohol etc etc
So perhaps never married may have worse diet habits, take less care of their health, drink more alcohol, etc.
Let's open another study of their meta-analysis with a big weight:
https://pubmed.ncbi.nlm.nih.gov/16905719/ (NHIS) 19% of total weight
Again, no adjustments in tables for healthy eating, physical activity, alcohol etc
Also for 65+ years, RR = 1.097 (0.704 - 1.353). RR is higher for younger people (who don't really die often).
https://pubmed.ncbi.nlm.nih.gov/9545122/ (Nilsson) 18% of total weight
data taken from table 4, where authors adjust only for age.
https://pubmed.ncbi.nlm.nih.gov/17484786/ (Ikeda 2007) 9% of total weight
Table 2 -- only 74 people died in single group of all causes. Too few. Though RR=1.85 CI 1.46 - 2.34
Adjusted for smoking, alcohol, physical activity. Not adjusted for healthy eating. Not too bad.
https://pubmed.ncbi.nlm.nih.gov/19501442/ (Molloy) 9% of total weight
data taken from table 2 -- but there there is adjustment only for age
https://pubmed.ncbi.nlm.nih.gov/10854957/ (Johnson) 10% of total weight
for age 65+, RR = 1.10 (1.02 - 1.18). Not enough adjustments.