There’s growing recognition of how much our first years of life affect us not just as children but into adulthood. Parents are routinely told how important it is for them to talk to their babies and toddlers, for instance; we know that deficits in young children’s vocabulary and cognitive development are hard to rectify down the road. Experiencing poverty early in life likewise has long-term implications for children’s chances of flourishing. A new study sheds greater light on another childhood experience with a lasting impact: parental divorce.

Sociologists Jason Thomas of Penn State and Robin Högnäs of the University of Louisville used longitudinal data on almost 15,000 adults born in 1958 in England, Scotland, and Wales to explore how the timing of parents’ divorce affects people’s health later in life. Their findings, published in Longitudinal and Life Course Studies, suggest that experiencing parental divorce before the age of seven has a greater long-term effect on health than experiencing the same thing later in childhood.

Thomas and Högnäs discovered that parental divorce in childhood was positively associated with worse self-reported health at age 50, “but the estimates are significantly different from zero only…during the youngest age interval” (birth to age seven). Parental divorce in this period was also linked to lower physical functioning and to the number of health problems reported at age 50, though results for the latter were not statistically significant.

The two scholars then analyzed which mediating variables might explain these associations, looking into “the family’s economic resources, parental investment, cognitive skills, emotional and psychological problems, health behaviours, and the intergenerational transmission of divorce.” (The survey they used did not offer information on parental conflict and several other potential mediators; nor did it provide details about the timing and number of family transitions people experienced, or allow for addressing the issue of selection bias.)

Socioeconomic status and health behaviors—smoking in particular—emerged as the most important factors, with emotional/behavioral problems coming in third. In other words, it appears that early parental divorce puts people at a slightly greater risk of poor health at age 50 primarily because it leads to a decline in families’ socioeconomic status, an elevated risk of behavioral problems in late childhood, and an increased propensity to smoke in adulthood. Thomas and Högnäs hypothesize that “smoking may be a behavior adopted by children to deal with psychological and emotional stresses associated with a parental divorce.” The theory seems intuitively plausible, though an earlier study that they cite fails to bear it out.

Whatever the explanation for the smoking/divorce link, the study offers one more piece of evidence that parental divorce can undermine children’s well-being decades after the papers are signed.