Marriage may be the most obvious social relationship linked to health, with research showing that divorced people tend to be less healthy than those who stay together.
But a study presented at this year's meeting of the American Sociological Association (ASA) goes further than all of this, looking beyond spouses to the presence of family and friends in older people's lives.
The researchers asked nearly 3000 volunteers aged 57 to 85 to list up to five of their "closest confidants" - excluding spouses - and to detail those relationships. By a lot of measures, most of the participants appeared to have happy, full lives. Most were married, in good physical health and said they weren't very lonely. On average, they reported having three close family members or friends.
In tracking how long these participants lived, the researchers found that those with more family members in their network were less likely to die. Having a lot of close friends didn't seem to make a difference.
James Iveniuk, a post-doctoral researcher at the University of Toronto's Dalla Lana School of Public Health, said he and his co-author were surprised by this.
"Because you can choose your friends, you might, therefore, expect that relationships with friends would be more important for mortality, since you might be better able to customise your friend network to meet your specific needs," Iveniuk explained.
"But that account isn't supported by the data. It is the people who in some sense you cannot choose, and who also have little choice about choosing you, who seem to provide the greatest benefit to longevity."
For the study, Iveniuk teamed up with biostatistician L. Philip Schumm. The pair found that people who said they felt "extremely close" to those family members had about a 6 per cent risk of mortality within the next five years. That compared with a 14 per cent risk of mortality for those who said they did not feel very close to family members.
The researchers theorised that maybe family members have more authority to exercise control over some aspects of health and that contact with family members may be more satisfying in some respects.
Furthermore, "when compared to friends, the provision of support from family is not as conditional on the emotional content of the relationship," they wrote. "Family members may be stressful or burdensome, but individuals may still provide support."
Iveniuk and Schumm were also able to tease out some findings about marriage and found that being married, even if it's not a great marriage, still seemed to have positive effects on longevity. "The presence of a marital bond may be more important for longevity than certain aspects of the bond itself," they concluded.
The paper adds to research about how individuals who lack supportive social relationships may be at a higher risk of health issues. Some papers have even tried to quantify that risk - and have suggested that the impact could be as detrimental as smoking.