An Early-Life Conditions Scale, and Applications to Suicide and All-Cause Mortality

Michael S. Hollingshaus, University of Utah

The goal of this paper is to construct an early-life conditions scale that can be utilized to study later-life mortality and morbidity in historical demographic databases. We identified a sample of 536,850 individuals born between 1849 and 1972 in the Utah Population Database. Variables include advanced maternal and paternal age, large sibship size, early maternal and paternal death, high infant mortality rate in county of birth, late birth order, sibling died as an infant, sibling died as a child, low initial SES, being a twin or other multiple-birth sibling, out-of-wedlock birth, and short birth interval. Using a quarter sample of 134,213, we performed exploratory factor analysis. We present results from parallel analysis that suggest the variables consist of four factors. Factor patterns are also presented. After confirmatory factor analysis on a separate subsample, the final scale will be used to predict all-cause mortality and suicide hazards with Cox regression models.