The clustering of premature deaths in families.

Background: Infant deaths cluster in families, but beyond infancy, little is known about familial aggregation of premature deaths. We hypothesized that an individual's risk of premature death would be influenced by prior premature deaths of any age in family members. Methods: For all Danish residents registered in the Civil Registration System, 1968-2005, information was linked to the Causes of Death Register and the Danish Family Relation Database, yielding a cohort of 4,870,821 persons linked to one or more relatives (93 million persons-years and 73,278 deaths). We used log-linear Poisson regression to estimate mortality rate ratios (relative risks) for premature death (before age 40 years) in persons with a family history of premature death, compared with persons without such a history. Results: Persons with a family history of premature death were 46% more likely themselves to die prematurely than persons without such a family history (relative risk [RR] = 1.46 [95% confidence interval = 1.42-1.50]). Relative risks were higher for concordant age at death, a close kinship relation, and similar causes of death. As expected, certain natural causes clustered among first-degree relatives (RRs ranged from 1.81-618), but unnatural causes of death, such as nonsimultaneous motor vehicle injuries, other injuries, and suicides, also clustered (RRs = 1.80, 3.53, and 4.23, respectively). Previous family history of premature death in a first-degree relative also increased the risk of dying from another cause (overall RR = 2.08 in infancy and 1.33 between ages 1 and 39). Conclusion: Premature deaths cluster in families, for both similar and dissimilar causes.