Gompertz Analysis of Pneumonia and Influenza Death Rates by Age, United States, 1959-2006

Andrew Noymer, University of California, Irvine
Cécile Viboud, Fogarty International Center, NIH

Pneumonia and influenza incidence and death rates are often analyzed as time series. A variety of techniques are used therein, most notably Serfling regression to calculate excess mortality (or excess incidence) by establishing a baseline which is then subtracted from the observed data. In this paper we explore a different approach. Serfling regression and similar approaches are time domain techniques. We perform what might be called parametric age domain analysis, where we model he steepness of the increase, by age, in pneumonia and influenza death rates. Gompertz models fit an exponential curve to hazard rates by age. By performing this analysis for 47 years of available data, we get a picture of how the age profile of influenza mortality has changed in the United States over the last half-century, spanning the pre-vaccination era, the era of vaccine adoption, and the current era of reasonably high vaccination rates at old ages.

  See extended abstract

Presented in Session 122: Methodological Issues in Health and Mortality: Longitudinal studies