Testing the Influenza-Tuberculosis Selective Mortality Hypothesis with Union Army Data

Andrew Noymer, University of California, Berkeley

In 2000, I postulated that there was selective mortality in the great 1918 influenza pandemic (Noymer & Garenne, in PDR). This hypothesis --- that tuberculosis (TB) enhances risk to death from influenza --- is potentially important for mortality studies generally, and may add new information to the frailty modeling approaches that have been predominant in mortality studies since the 1979 watershed methodological papers of Keyfitz & Littman; and Vaupel, Manton, & Stallard. Elaboration with historical microdata is a next step. Fogel's Union Army data, which by now are well-known to historical demographers, fits the bill. I use UA data to examine, using Cox regressions (etc), if lifetime TB exposure influences death due to influenza. The hypothesis predicts: Yes. The data agree. The risk ratio is large and is statistically significant. This informs current debates in demographic methods, and tells an interesting story about life and death in the UA cohorts.