CROI 2020 Abstract eBook

Abstract eBook

Poster Abstracts

mass index>25 kg/m 2 , alcohol/drug use disorders, and common comorbidities (see Table footnote). Results: The study included 39,000 PWH (with 697 cancers) and 387,767 uninfected adults (with 2,876 cancers). Any cancer increased mortality for PWH with an RD of 62.2 deaths per 1,000 py, and for uninfected persons with an RD of 45.5 deaths per 1,000 py. This difference by HIV status persisted with adjustment for confounders with an adjusted excess mortality rate for any cancer of 20.5 per 1,000 py (P<0.001) for PWH compared with uninfected persons. Excess mortality rates for PWH with cancer varied by cancer group (Table) with the lowest for ADCs (11.8) and the highest for NADCs (30.3), virus- unrelated NADCs (30.6), and HPV-related NADC (24.7). Conclusion: Even with access to comprehensive HIV and cancer care, PWH have excess mortality after cancer, especially NADCs. Additional research is needed to understand this disparity, including studies evaluating effectiveness and tolerability of cancer treatments in PWH.

stable over the past 5 years, from 1.0 to 1.1 deaths per 100 PLWH. Though some deaths among PLWH can be directly attributed to HIV, an increasing proportion of deaths are due to other factors, including the aging of PLWH. We compared a population based cohort of all PLWH 2016-2018 to decedents with HIV over those years, and conducted in depth investigations of causes of death, comorbidities, and social determinants of health for 2017 deaths. Methods: Data were collected by provider interviews, medical record abstractions, and analysis of the CDC’s HIV/AIDS surveillance system (NHSS). 268 deaths occurred among King County PLWH 2016-2018 relative to 7,922 PLWH. Of 98 deaths in 2017 82% had a local death certificate and of these 85% had a local medical record available for review; medical providers completed surveys for 56% of these. Results: One third (34%) of decedents had CD4 counts <200 relative to 5% of PLWH; 44% of decedents were 60+ years relative to 18% of PLWH; and 57% of decedents were diagnosed with HIV in 2000 or earlier relative to 33% of PLWH (Table). Decedents were roughly twice as likely to have a history of injection drug use. Of the 68 patients whose medical records were abstracted, 10 (15%) had causes of death related to HIV; half had an AIDS-defining Opportunistic Illnesses (OI; 7%). Non-AIDS cancers were associated with death for 26%, heart disease for 18%, self-harm for 12%, and liver disease for 6%. An additional 10 had an AIDS OI within a year of death (making 22% total). More than half, 65% had a mental health diagnoses (mostly depression/anxiety), and 86% had some treatment of their mental illness. One quarter had HCV and one quarter of these had been treated, all of whom had sustained viral response. Provider interviews suggest roughly 1/3 of decedents had some social isolation. Those experiencing stigma (24%) had 9-fold higher odds of an HIV-related death relative to decedents without known stigma. Conclusion: Expectedly, decedents were older, had been diagnosed with HIV longer, had lower CD4 counts, and were more likely to have used injection drugs than PLWH. The deaths of the majority of PLWH in King County are from non-HIV/AIDS related causes though AIDS-OIs contributed to 7% of deaths and were present for 22%. Data suggest stigma may be associated with HIV-related deaths but a larger study is needed to validate this finding.

871 GAINS AND REMAINING CHALLENGES IN MORTALITY AMONG INDIVIDUALS WITH HIV, 1999-2017 Monina Klevens 1 , Maria McKenna 1 1 Massachusetts Department of Public Health, Boston, MA, USA Background: Improvements in HIV care have resulted in people living with HIV reaching older ages. An increased risk of non-AIDS comorbid conditions may pre-date HIV infection, complicating efforts to close the gap in life expectancy compared to uninfected people. This study assessed trends in, and causes of, deaths in a state with high health insurance coverage. Methods: We analyzed records of deaths in Massachusetts from 1999-2017 excluding non-residents. Using ICD-9 and -10 codes, we dichotomized deaths as with HIV or AIDS (ICD-9: 42, 43, 44; ICD-10: B20) or without HIV. We aggregated causes of death into broader system groupings (e.g. circulatory, digestive, respiratory, etc.) using WHO and CDC standards. We calculated the difference in the mean age at death for specific comorbidities during the earliest three- year period (1999-2001) with the most recent period (2015-2017) to assess improvements in longevity among individuals with HIV infection. Results: There were 1,018,132 deaths in Massachusetts from 1999-2017; of these, 3,384 (0.3%) were among HIV infected individuals. The number of deaths among infected individuals declined from 1319 deaths in 1999-2003 to 565 deaths in 2013-2017; deaths among uninfected individuals increased from 274,625 to 275,744. Mean age of death increased from 42.5 years in 1999 to 60.0 years in 2017 among infected individuals but was unchanged among the uninfected (76.1 and 76.2 years, respectively). In both groups, diseases of the circulatory system ranked first; diseases of the respiratory system ranked third among infected individuals and second among uninfected. The second cause (31.2% of recent deaths) among infected individuals were infectious diseases; these ranked 9th (23,946 deaths or 8.7% of deaths) among uninfected individuals. Increases in age at death varied by cause of death and ranged from 5.2 to 17.7 years among infected individuals. Areas with the least improvement were hypertension (5.2 years), lower respiratory tract diseases (5.6 years), and diabetes (8.6 years). Areas with the most improvement were renal failure (17.7 years) and heart failure (16.9 years). Conclusion: A gap in longevity remains. 872 A DETAILED LOOK AT HIV MORTALITY IN KING COUNTY, 2016-2018 Andrea L.Martin 1 , Meena Ramchandani 1 , Susan E. Buskin 2 1 University of Washington, Seattle, WA, USA, 2 Public Health–Seattle & King County, Seattle, WA, USA Background: While mortality among people living with HIV (PLWH) has declined 43% over the past decade in King County, death rates have remained

Poster Abstracts

873 USING MULTISTATE MODELS TO DISENTANGLE MORTALITY & LOSS TO FOLLOW-UP IN HIV+ PATIENTS Nanina Anderegg 1, Jonas Hector 2 , Juan Burgos 2 , Laura Jefferys 2 , Michael A. Hobbins 3 , Jochen Ehmer 3 , Matthias Egger 1 1 Institute of Social and Preventive Medicine, Bern, Switzerland, 2 SolidarMed, Pemba, Mozambique, 3 SolidarMed, Luzern, Switzerland Background: Estimating mortality in HIV-positive patients starting antiretroviral therapy (ART) is challenging, as clinics often face substantial loss to follow-up (LTFU). Many studies ignore LTFU, leading to biased estimates. Others correct for LTFU, but conventional methods give pooled estimates, which makes it impossible to asses risk factors and mortality estimates separately for those LTFU and remaining in care. We examined the use of multistate models to overcome this problem, using data from rural northern Mozambique, where patients LTFU are routinely traced. Methods: We used clinical and tracing data from Ancuabe District, Mozambique. We used a multistate illness-death model without recovery to describe progression of patients from the initial state “on ART” through the intermediate state “LTFU” to the final absorbing state “Death”. We used Nelson-Aalen and Aalen-Johansen estimators to estimate crude cumulative transition hazards and probabilities, respectively. We fitted Cox proportional hazards models to examine associations between patient characteristics and transition hazards.

CROI 2020 326