Estimating a Set of Mortality Risk Functions with Multiple Contributing Causes of Death

Epidemiology. 2020 Sep;31(5):704-712. doi: 10.1097/EDE.0000000000001225.

Abstract

Background: There are few methodologic examples of how multiple causes of death may be summarized in cause-specific mortality analyses to address limitations of attributing death to a single underlying cause. We propose a cause-of-death weighting approach to estimate the set of risk functions of specific causes of mortality using both underlying and contributing cause-of-death information.

Methods: We constructed weights according to a user-specified function. Using data from four southern US human immunodeficiency virus (HIV) clinics, we constructed a cause of death-weighted Aalen-Johansen estimator of the cumulative incidence function to estimate risks of five specific causes of mortality in the full sample and by injection drug use history.

Results: Among 7740 HIV-positive patients initiating antiretroviral therapy between 1999 and 2014, the 8-year risk of all-cause mortality was 17.5% [95% confidence interval (CI) = 16.5, 18.4]. The cause of death-weighted risk of HIV-related mortality was 6.7% (95% CI = 6.0, 7.3) and accounted for 39% (95% CI = 35, 42) of total mortality risk. This compared with 10.2% (95% CI = 9.2, 11.2) using only the underlying cause, in which case HIV-related deaths accounted for nearly 60% of total mortality risk. The proportion attributable to cardiovascular disease among those whose HIV risk factor was injection drug use was twice as high using cause-of-death weights compared with only the underlying cause (8%; 95% CI = 5, 11 vs 4%; 95% CI = 1, 6).

Conclusion: Using cause of death-weighted estimators to incorporate multiple causes of death may yield different conclusions regarding the importance of certain causes of mortality. See video abstract: http://links.lww.com/EDE/B706.

MeSH terms

  • Adult
  • Cause of Death*
  • Female
  • HIV Infections* / drug therapy
  • HIV Infections* / mortality
  • Homosexuality, Male / statistics & numerical data
  • Humans
  • Male
  • Middle Aged
  • Risk Factors
  • United States / epidemiology