Considering birth cohort when evaluating endemic HIV in the United States, age-period-cohort findings from CMS beneficiaries, 1999-2020

Authors

• M Alodadi
• E Lyons
• A Che
• D Watson
• S Roy
• GJ Tawa
• F Porter
• SJ Haugabook
• E Ottinger
• US Mudunuri

Abstract

Introduction:
There is high agreement that HIV+ cases are ‘surviving’ to later ages. Counterfactually, age at HIV diagnosis has increased since peak AIDS death in 1995-1996. Cases observed later in life may not be survivors infected young, but cases who were infected later in life. An ‘at-scale’ case by case assessment of HIV from at scale, real-world data is lacking for the United States.

Methods:
This study collected a 100% sample of Medicare and Medicaid claims from 1999 through 2020. Every claim with a declared HIV diagnosis (B20, 042, Z21) and said HIV status was attributed to the individual referenced. Date of birth and date of death was integrated with HIV status for a CMS wide HIV+ case index. Age-period-cohort disambiguation was performed for attributed mortality and case observation by study year.

Results:
CMS data found 283,688,152 cases, of which 1,543,041 were HIV+. 48,598,568 decedents were identified, with 363,425 being HIV+. There is birth cohort variation in eligibility for mortality. These mortality seasons have ‘years old at observation’ eligibility requirements that some birth cohorts have yet to reach because they are ‘too young’ in the final study year.

Conclusions:
Applying population survival outcomes to cases too young to die of a given illness is a prime example of APC confounding. Cases born after 1980 could experience mortality later in life similar to cases born in 1960, given additional study time.

Scientific Focus Area: Social and Behavioral Sciences

This page was last updated on Monday, September 25, 2023