Causes of death and mortality risk factors among people living with HIV: recent changes and unique risks

This PhD thesis is comprised of three studies using data from three international cohorts of people with HIV: EuroSIDA, D:A:D, and RESPOND. The thesis characterizes mortality among people with HIV, with each project focusing on different aspects of this broad topic.

Project 1 showed striking patterns in the EuroSIDA cohort with respect to mortality from suicide, substance use, or violent/accidental death. Participants who had recently been diagnosed with AIDS were at much higher risk of mortality, even given the much-improved prognosis in the modern era of ART. There were also striking regional differences, with participants in Eastern Europe at higher risk of violent/accidental or substance use mortality, even after accounting for HIV transmission by injection drug use.

Project 2 provided an overarching update to patterns of mortality across two large cohort collaborations, D:A:D and RESPOND. Mortality rates in these cohorts declined over the period of follow-up, from 1999 through 2020, but were shown to have remained stable over the last ten years of follow-up. Analyses of
cause-specific mortality showed that this trend of declining mortality was true for some causes of death (AIDS, CVD, and liver-related), but not for others (non-AIDS-defining malignancies). Though improvements in immunologic and virologic status among people in these cohorts was certainly a contributing factor to these improvements in mortality, it did not fully explain the phenomenon.

Project 3 investigated yet another aspect of mortality among people with HIV, namely a safety signal reported in the literature from another international cohort collaboration linking all-cause mortality to use of raltegravir (RAL). We found an age-adjusted association between RAL and mortality which was nullified after accounting for baseline confounding. RAL was initially often used as salvage therapy among highly treatment experienced people with HIV. We also showed that, in this population, starting RAL was associated with several other mortality risk factors, including end-stage renal disease, CVD, and HCV. We
conclude from this that the apparent association between RAL use and mortality is likely driven by a combination of baseline confounding and channeling bias.

These three projects investigated different yet highly relevant questions and highlight different avenues for intervention to reduce mortality among people with HIV. These projects have highlighted positive trends, with declining mortality overall and among some non-AIDS causes of mortality and more strong evidence for the relative safety of an ART that is still a valuable option in special cases. On the other hand, though improvements in ART has allowed for lower mortality and more focus on treating comorbidities, these projects highlight the continued impact of AIDS diagnosis on death by suicide, persistent inequities between regions of Europe, and the causes of mortality which have not declined over time.