Related Experiment Videos

Competing risk adjustment reduces overestimation of opportunistic infection rates in AIDS

Y Yan1, R D Moore, D R Hoover

  • 1Department of Surgery, Washington University School of Medicine, St. Louis, MO 63110, USA. yany@msnotes.wust.edu

Insights

Adjusting for other causes of death is crucial when estimating acquired immunodeficiency syndrome (AIDS) related illnesses. Ignoring competing risks leads to overestimating disease occurrence and potentially misleading comparisons.

Area of Science:

  • Epidemiology
  • Biostatistics

Background:

  • Estimating the cumulative incidence of acquired immunodeficiency syndrome (AIDS) related illnesses requires careful consideration of competing risks.
  • Other causes of death can significantly impact the observed incidence rates of specific AIDS-defining conditions.

Purpose of the Study:

  • To highlight the importance of adjusting incidence estimates for competing risks of mortality.
  • To compare unadjusted and adjusted incidence estimates for four key AIDS-related illnesses.

Main Methods:

  • Retrospective cohort study of patients at Johns Hopkins Hospital AIDS Service (1989-1995).
  • Calculated unadjusted and adjusted cumulative incidence estimates for pneumocystis pneumonia (PCP), mycobacterium avium complex (MAC), cytomegalovirus (CMV), and esophageal candidiasis.
  • Analyzed ratios of unadjusted to adjusted estimates and corresponding cumulative death rates.

Main Results:

  • Ratios of 4-year unadjusted to adjusted incidence for four AIDS illnesses ranged from 1.38 to 1.86 (cumulative death rates: 61%-69%).
  • For cytomegalovirus (CMV), ratios ranged from 1.5 to 2.33 across patient groups (cumulative death rates: 48%-78%).
  • Substantial overestimation of disease occurrence was observed when competing risks were ignored.

Conclusions:

  • Failure to account for competing risks of death leads to significant overestimation of AIDS-related disease incidence.
  • Adjusted estimates are essential for accurate assessment and comparison of disease occurrence in populations with high mortality rates.
  • Misleading conclusions regarding disease prevalence and risk can arise from unadjusted analyses.

Related Concept Videos