CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

1 University of Minnesota, Minneapolis, MN, USA, 2 Hennepin County Medical Center, Minneapolis, MN, USA Background: Standard measures of retention in HIV care are all-or-nothing classifications: persons living with HIV (PLWH) are either “in care” or “out of care”. However, in reality, engagement in care is an evolving process, with periods of more or less frequent care encounters. The goal of this analysis was to classify PLWH into different “care trajectory” classes based on longitudinal patterns of care and to determine how patient outcomes differ across care classes. Methods: We conducted a retrospective analysis of all PLWH ≥18 years-old who received medical care at a public, hospital-based clinic in Minneapolis, MN between 2008-2013. We analyzed clinic visit and laboratory test data from 2008-2015 merged with surveillance records to account for HIV care at other clinics, out-of-state relocation, and mortality. Individual HIV care trajectories were constructed in six-month intervals starting from first observed care event until death, relocation, or the end of 2015. PLWH observed for less than 1 year were excluded from analysis. Latent class analysis was used to identify care trajectory classes, described by the probability of having a clinical encounter in each six-month interval. The number of care classes was chosen to maximize model fit. Patient outcomes included retention (≥1 care events in every six- month interval), mortality, and sustained viral suppression (all viral loads <50 copies/mL in last 12 months of care). Results: The study population (N=2,110) was best divided into 5 care trajectory classes: (1) consistent care; (2) less frequent care; (3) return to care after initial attrition; (4) moderate attrition; and (5) rapid attrition. Only Class 1 had a substantial (78.4%) level of retention using standard measures. PLWH in Class 1 were also the most likely to have sustained viral suppression in the last 12 months of care. Though retention was consistently low among the other care classes, there was substantial variation in viral suppression, from 63.2% in Class 2 to 24.2% in Class 5. Mortality was greatest for those in Class 1, but the total number of deaths over the study period was relatively small. Conclusion: Characterizing the longitudinal patterns of HIV care identified five intuitive care trajectories, including four distinct patterns of suboptimal retention with differing levels of viral suppression. Care trajectories could be used to prioritize re-engagement efforts. 1133 FROM PRISON’S GATE TO DEATH’S DOOR: SURVIVAL ANALYSIS OF RELEASED PRISONERS WITH HIV Kelsey B. Loeliger 1 , Jaimie P. Meyer 1 , Maria M. Ciarleglio 1 , Mayur M. Desai 1 , Colleen Gallagher 2 , Frederick Altice 1 1 Yale University, New Haven, CT, USA, 2 Connecticut Department of Correction, Wethersfield, CT, USA Background: Prisoners returning to the community have high mortality, but there is limited data on death among those living with HIV (PLH). This study evaluates causes of and risk factors for death among PLH after prison or jail release. Methods: We created a retrospective cohort of all PLH released from a Connecticut jail or prison in 2007–14 (n=1,350) by linking prison pharmacy and custody data with state death indices and HIV surveillance data. Cox proportional hazards regression identified predictors of death. Results: Overall, 184 (13.6%) died during a median of 5.2 (IQR 3.0-6.7) years follow-up after release (crude mortality rate of 28/1,000 person-years). Among PLH who died, median time to death was 3.0 years (IQR 1.5-4.5); 179 died in the community and 5 died during a later re-incarceration. Among the 175 deaths with available data, main primary causes included HIV (n=76, 43.4%), drug overdose (n=26, 14.9%), liver disease/failure or hepatitis C (n=16, 9.1%), heart disease (n=11, 6.3%), infection (n=9, 5.1%), accidental injury or homicide (n=9, 5.1%), non-AIDS related cancer (n=5, 2.9%), diabetes (n=4, 2.3%), and suicide (n=2, 1.1%). Independently significant (p<0.05) predictors of death

1131 HEALTHCARE PROVIDER TRUST LINKED TO LONG-TERM HIV VIRAL SUPPRESSION Kate G. Michel 1 , Cuiwei Wang 1 , Tracey Wilson 2 , Kathryn Anastos 3 , Mardge H. Cohen 4 , Ruth Greenblatt 5 , Margaret Fischl 6 , Igho Ofotokun 7 , Mirjam-Colette Kempf 8 , Adaora Adimora 9 , Joel Milam 10 , Joanne Michelle F. Ocampo 1 , Stephen J. Gange 11 , Michael Plankey 1 , Seble Kassaye 1 1 Georgetown University, Washington, DC, USA, 2 SUNY Downstate Medical Center, Brooklyn, NY, USA, 3 Montefiore Medical Center, Bronx, NY, USA, 4 Cook County Health & Hospitals System, Chicago, IL, USA, 5 University of California San Francisco, San Francisco, CA, USA, 6 University of Miami, Miami, FL, USA, 7 Emory University, Atlanta, GA, USA, 8 University of Alabama at Birmingham, Birmingham, AL, USA, 9 University of North Carolina Chapel Hill, Chapel Hill, NC, USA, 10 University of Southern California, Los Angeles, CA, USA, 11 Johns Hopkins Hospital, Baltimore, MD, USA Background: Trust in the healthcare system (HS) and health providers (HP) is linked to medication adherence; however, many studies do not link to biological data. We hypothesize that trust differs by HIV status and is associated with longitudinal patterns of viremia. Methods: A 2006 cross-sectional survey assessed the Healthcare System Distrust Scale (HSDS, 0=trust; 50=distrust), an adapted Patient-Physician Trust Scale (PPTS, 0=distrust; 25=trust), HIV medication distrust and demographics in 1049 HIV+ and 463 high-risk HIV-negative women from the Women’s Interagency HIV Study. This study identified HIV viral load trajectories in 2440 HIV+ women who contributed ≥ 4 semi-annual visits from 1994-2015. Viral suppression was defined by assay detection limits (<80 to <20 copies/mL). Group-based probability trajectory analyses categorized women based on longitudinal viral load patterns, and identified 3 groups: sustained viremia (SV; n=1,010), intermittent viremia (IV; n=719), and non-viremia (NV; n=711). Ordinal logistic regression models assessed trajectory group and HP/HS trust, controlling for demographics. Results: Most women were African American (60%), currently insured (89%) non-smokers (56%). HIV+ women were more trusting of HS (HSDS 12.6 vs. 13.8, p=0.02) and HP (PPTS 20.6 vs. 18.8, p<0.0001) compared to HIV- women. HIV+ women with NV had higher HP trust compared to SV women (PPTS: SV 19.9, IV 20.6, NV 21.4, p<0.0001); there was no difference in HS trust between viral trajectory groups. Compared to NV women, SV women were less likely to agree that HIV medicines help people live longer (86% vs. 94%, p<0.0002) or that HIV medicines prevent hospitalizations (73% vs. 87%, p<0.0001). Only 52% of SV women believed HIV medicines work as well for African American/Latina women compared to white women (IV 61%, NV 66%, p=0.0005). In ordinal logistic regression, groups with higher viremia were associated with HP distrust (OR 1.49; 95% CI 1.12, 1.98), HS distrust (OR 1.49; 95% CI 1.13, 1.98), African American race (OR 1.70; 95% CI 1.1.29, 2.24), current smoking (OR 2.42; 95% CI 1.84, 3.19), and current unemployment (OR 1.62; 95% CI 1.23, 2.14). Conclusion: HIV+ women express high levels of HS and HP trust. HIV-negative women’s low trust in HS, HP, and HIV medicines may have implications for PrEP use. Trust in HPs and HIV medicines is less common among SV women. Successful long-term HIV management depends on HP and HS trust; current data is needed on healthcare perceptions linked to HIV biological data. 1132 CHARACTERIZING HIV CARE TRAJECTORIES AND DIFFERENCES IN PATIENT OUTCOMES Eva A. Enns 1 , Cavan Reilly 1 , Karen Baker 2 , Keith Henry 2

Poster Abstracts

CROI 2018 437