CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

losses, 15 (3%) mothers were documented as transferring facilities, and 11 (2%) mothers moved out of the country. Among the remaining 436 HEI, 192 (44%) had outcome data, including 143 (74%) who were engaged in care, 31 (16%) disengaged, and 18 (9%) who had died after delivery. Of 113 HEI with data on HIV testing, 75 (66%) completed HIV testing at 6 weeks and 53 (47%) at 6 months. Conclusion: Majorities of pregnant women and HEI classified as LTF under Option B+ were engaged in care. These findings highlight a need to obtain more accurate outcome measures via strengthened systems for capturing and utilizing HF data for pregnant women and HEI. 885LB OPTIMIZING EID: A CLUSTER-RANDOMIZED TRIAL OF THE HIV INFANT TRACKING SYSTEM IN KENYA Sarah Finocchario Kessler 1 , Catherine Wexler 1 , An-Lin Cheng 2 , Niaman Nazir 1 , Brad Gautney 3 , May Maloba 3 , Kathy Goggin 4 , Melinda Brown 1 , Elizabeth Muchoki 5 , Shadrack Babu 5 , Eric Muriithi 5 , Martin Ochieng 5 , Samoel Khamadi 5 , Matthew Sandbulte 1 , Raphael Lwembe 5 1 University of Kansas Medical Center, Kansas City, KS, USA, 2 University of Missouri– Kansas City, Kansas City, MO, USA, 3 Global Health Innovations, Dallas, TX, USA, 4 Children’s Mercy Hospital, Kansas City, MO, USA, 5 Kenya Medical Research Institute, Nairobi, Kenya Background: Outcomes for HIV-exposed infants (HEI) depend on the quality and efficiency of early infant diagnosis (EID) services. In Kenya, we evaluated the impact of the HITSystem (system-level intervention linking EID stakeholders via e-alerts for providers and text messages for mothers of HEI) on key EID outcomes. Methods: In this non-blinded, phased, cluster randomized controlled trial (NCT02072603), 6 hospitals matched on geographic region, resource level, and volume were randomized to receive the HITSystem (n=3) or standard of care (SOC; n=3). Eligible participants (HIV+ women >18 yrs with HEI <24 weeks) enrolled between February 2014 and December 2016 were followed to evaluate the primary outcome of complete EID care; defined as receipt of all key EID services through 18-months (HIV-uninfected infants) or through ART initiation (HIV+ infants) per Kenyan guidelines. The HITSystemwas hypothesized to improve retention, ART initiation, and results turnaround times (TAT) compared to SOC. Using a stepwise approach, we conducted separate multivariate logistic and Poisson regression analyses with intervention group, site volume, and significant covariates included as fixed effects in the models. Bonferroni corrections for multiple comparisons were applied. Results: Among 809 eligible HEI, data from 690 were analyzed (n=392 intervention, n=298 SOC); excluding 28 deaths and 91 documented transfers/ moved. Median age at enrollment was 6.0 weeks; 50%were male. Infants enrolled in HITSystemwere significantly more likely to receive complete EID services compared to controls (85.2% [.82-.89] vs. 61.02% [.55-.66], p<0.001), including the following: receipt of OI prophylaxis (99.7% vs 89.6%), PCR results returned to the hospital (100% vs 96.98%), mothers notified of test result (98.9% vs 89.3%), re-testing among HIV-uninfected infants at 9 months (96.8% vs 91.1%) and 18 months (84.7% vs 69.3%), and ART initiation for HIV+ infants (100% vs. 72.7%). Mean results TAT (24.6 vs 49.2 days, p=0.003) and mother notification (19.0 vs 29.8 days, p=0.003) were faster at intervention sites. Receipt of initial HIV test was similar and time to ART was faster at SOC sites (median 68 vs. 51 days, p=0.045). Conclusion: HITSystem significantly increased completion of EID services and reduced TAT for results and notification. Hindered by intervention settings that required multiple adherence counseling sessions prior to initiation, time to ART was faster in SOC sites.

886 TRENDS IN CAUSE-SPECIFIC MORTALITY ON THE HIV CARE CASCADE, SOUTHERN & EASTERN AFRICA Kathryn A. Risher 1 , Clara Calvert 1 , Basia Zaba 1 , Emma Slaymaker 1 , Kobus Herbst 2 , Jessica Nakiyingi-Miiro 3 , Amelia C. Crampin 1 , Mark Urassa 4 , Baltazar Mtenga 4 , Georges Reniers 1 1 London School of Hygiene & Tropical Medicine, London, UK, 2 Africa Health Research Institute, Mtubatuba, South Africa, 3 MRC/UVRI Research Unit on AIDS, Entebbe, Uganda, 4 National Institute for Medical Research, Kisesa HDSS, Mwanza, Tanzania, United Republic of Background: Population-based studies have found decreasing mortality among those living with HIV, but it is unclear at what level of care engagement these deaths take place, to what causes the deaths are attributable, or how these patterns change as antiretroviral therapy (ART) programs mature. Methods: We assess trends in cause-specific mortality along the cascade of HIV care in data from four population-based HIV surveillance sites in eastern and southern Africa. Deaths are assigned to a most likely cause using InSilicoVA following verbal autopsy. We conduct a competing risks analysis of four cause-of-death groups at three key transitions on the HIV care cascade (i) among seroconverters prior to HIV diagnosis, (ii) among those diagnosed prior to ART initiation and (iii) among those on ART. Cox proportional hazards models are used to estimate cause-specific hazard ratios (HR) for period of HIV seroconversion, diagnosis or ART initiation, controlling for age and sex, meta- analysed by site. Results: Following seroconversion, HIV diagnosis, and ART initiation (respectively), 5,435, 13,186 and 7,778 adults contribute 19,213, 29,051, and 34,799 person-years of follow-up, and 226, 908 and 1,197 deaths. Overall five-year mortality on each step of the cascade decreased over time (Figure). However, these decreases were not evenly distributed across causes of death. The cause-specific hazard of mortality due to HIV decreased in the later cohort among seroconverters not yet diagnosed with HIV (HR=0.36, 95%CI=0.18- 0.75), those diagnosed but not yet initiated on ART (HR=0.51, 95%CI=0.39- 0.66), and among those initiated on ART (HR=0.60, 95%CI=0.40-0.89). Mortality due to TB decreased over time among those who had initiated ART (HR=0.57, 95%CI=0.44-0.75) but not among seroconverters not yet diagnosed or those diagnosed but not yet on ART. There was no change in non-HIV/TB mortality with changing year of seroconversion, diagnosis or ART initiation. Men had increased cause-specific mortality for all causes of death both following diagnosis prior to ART initiation and following ART initiation (HR range=1.34- 4.48), but only for TB among seroconverters who had not yet been diagnosed (HR=3.43). Conclusion: In these population-based studies, the cause-specific mortality among those living with HIV is changing over time as ART becomes more widely available. The mortality gains at three stages of the HIV care cascade appear to be attributable largely to HIV (seroconversion and diagnosis) or HIV and TB (ART) improvements.

Poster Abstracts

CROI 2018 336