How often friends or family visited had the largest association of the five. People who were never visited had a 39% higher risk of dying over the follow-up than those visited daily; visits once every three months carried 11%. Below monthly the risk climbed; at monthly or more it flattened, which the authors read as a threshold. Living alone added 25%, no weekly group activity 13%, and often feeling lonely 6%.
How it was done
458,146 people in UK Biobank, followed for a median of 12.6 years and linked to death registers. Five measures of social connection were taken at the start: how often they could confide in someone, whether they often felt lonely, how often friends or family visited, whether they did a weekly group activity, and whether they lived alone. Each was tested against all-cause mortality with the others held constant.
What it means
The shape matters more than the numbers. This is not a dose curve where every extra visit buys a little more; it is a floor. Once contact was at least monthly, more of it made little further difference in this data. That is a more forgiving finding than it first sounds: the line is not daily, or weekly. It is once a month.
It also separates the parts of social connection that are usually lumped together. Feeling lonely carried the smallest association here; actually being visited carried the largest. The 2023 Nature Human Behaviour meta-analysis found the same ordering across 90 cohorts: isolation predicted mortality more strongly than loneliness did.
The combinations are the sharpest part of the paper. Never being visited while also living alone carried a 77% higher risk, against 19% for daily visits while living alone. Living alone is survivable; living alone and unvisited is what the association is really about.
