CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

900 SUBSTANTIAL MORTALITY AND LOSS PRIOR TO TREATMENT IN ART- ELIGIBLE PATIENTS IN ZAMBIA Aaloke Mody 1 , Izukanji Sikazwe 2 , Theodora Savory 2 , Kombatende Sikombe 2 , Paul Somwe 2 , Mwanza wa Mwanza 2 , Carolyn Bolton Moore 2 , Nancy Padian 3 , Charles B. Holmes 4 , Elvin Geng 1 1 University of California San Francisco, San Francisco, CA, USA, 2 Centre for Infectious Disease Research in Zambia, Lusaka, Zambia, 3 University of California Berkeley, Berkeley, CA, USA, 4 Johns Hopkins University, Baltimore, MD, USA Background: An effective cascade in the treat-all era must include timely and complete ART initiation. Most epidemiologic studies begin with ART initiation, and thus neglect failures to initiate, loss to follow-up (LTFU), and deaths prior to treatment, thereby failing to identify important opportunities for improvement. Methods: We analyzed a population-representative sample of newly- enrolling, ART-eligible adults (CD4≤500 cells/μl or pregnant/breastfeeding women) at 64 clinics in Zambia between April 1, 2014 and July 31, 2015. We used data from electronic medical records supplemented by active tracing of a random sample of lost (>90 days late for last visit) patients to ascertain outcomes (e.g., mortality, care status, reasons for LTFU, virologic suppression [DBS HIV RNA≤1000 copies/ml]). We used a weighted multistate analysis to estimate the prevalence and incidence of six retention states over time: 1) In Care prior to ART 2) In Care on ART 3) LTFU prior to ART 4) LTFU on ART 5) Died prior to ART 6) Died on ART. We estimated overall virologic suppression and identified predictors of mortality prior to initiation using a modified Poisson regression with robust variances. Results: 23,229 patients (58% female; median age 34y [IQR 29-41], median CD4 234 cells/μl [IQR 120-365]) were eligible for ART at enrollment. At one year, 87.4% of patients had been initiated on ART (95% CI 86.0-88.8%), 70.5%were still in care and on ART (CI 69.8-71.1%), 22.3% (CI 21.5-23.2%) were LTFU (9.2% prior to vs. 13.1% after ART), and 7.0% (CI 6.5-7.5%) had died (3.2% prior to vs. 3.8% after ART). 11.8% of patients were LTFU after only one visit (CI 11.3-12.4%). Among those who died, younger age (aRR 1.23 per 10y decrease, p<0.01), lower CD4 (aRR 1.19 per 50 cell/μl decrease, p<0.01), and attending a clinic with higher proportions of follow-up scheduled at 1m (aRR 1.16 per 10% increase, p=0.025) were associated with dying prior to, as opposed to after, ART initiation. Among those LTFU, patients lost prior to ART were more likely to be out of care rather than silent transfers (77.8% prior to vs. 38.2% after ART, p<0.01). In Lusaka province, virologic suppression amongst alive patients was 80.0% in those started on ART (CI 64.1-93.6%) and 56.9% overall (CI 45.7-68.3%). Conclusion: There is substantial loss and mortality even prior to treatment initiation among ART-eligible patients. Efforts to ensure engagement prior to treatment and rapid ART initiation are needed to optimize the population-level effects of treat-all.

901 LONG-TERM MEASURES OF HIV EXPOSURE AND MORTALITY IN THE HIV OUTPATIENT STUDY Frank J. Palella 1 , Carl Armon 2 , Stephen R. Cole 3 , Rachel Hart 2 , Nabil Rayeed 2 , Ellen Tedaldi 4 , Bienvenido Yangco 5 , Richard Novak 6 , Linda Battalora 7 , Stacey Purinton 2 , Kate Buchacz 8 1 Northwestern University, Chicago, IL, USA, 2 Cerner Corp, Kansas City, MO, USA, 3 University of North Carolina Chapel Hill, Chapel Hill, NC, USA, 4 Temple University, Philadelphia, PA, USA, 5 Infectious Disease Research Institute, Tampa, FL, USA, 6 University of Illinois at Chicago, Chicago, IL, USA, 7 Colorado School of Mines, Golden, CO, USA, 8 CDC, Atlanta, GA, USA Background: Elucidating associations between long-term HIV viral exposure (VE) measures and mortality can inform the management and monitoring of ART-treated patients. Methods: We analyzed medical records of HIV Outpatient Study (HOPS) participants seen at 12 US HIV clinics who initiated ART between March 1995 and June 2015, were followed for ≥ 6 months, had ART prescribed ≥ 75% of time, and had ≥ 2 plasma viral load (VL) and ≥ 1 CD4+ cell count (CD4) during observation. We evaluated all-cause mortality from 6 months after ART start until 30 June 2016. VE was quantified using two time-updated variables: viremia copy-years (VCY) and percent of person-time (pPT) spent above 200 or 50 copies/mL. We fit Cox models to estimate associations between VE and mortality. Results: The 1,645 patients had median age at ART start (baseline, BL) of 38 years (interquartile range [IQR]: 31-45), 78%were male, 37% non-Hispanic black, 14% Hispanic/Latino, 57%were men who have sex with men (MSM), 6% had injection drug use (IDU) risk, and 31% had public health insurance. Median BL CD4 was 293 (IQR 140-460). Patients contributed 10,453 person years [py], with median 14 VLs (IQR: 7-24) per patient. Median pPT > 200 or > 50 were 10% (IQR: 1%-47%) and 26% (IQR: 6%-72%), respectively, and median VCY was 3.0 log 10 (IQR: 2.3-4.2). There were 115 deaths; among decedents median pPT > 200 or > 50 were 44% (IQR: 12%-92%) and 82% (IQR: 26%-100%) respectively, and median VCY was 3.7 log 10 (IQR: 2.7-4.9). In Cox models, mortality was associated with BL age (Hazard Ratio [HR] 1.7, 95% confidence interval [CI]: 1.4-2.0, per 10 years), IDU risk (HR 3.8, CI: 2.2-6.5), heterosexual HIV risk (HR 1.7, CI: 1.1-2.7), and BL CD4 (HR 0.8, CI: 0.8-0.9 per 100 cells/mm 3 ). When pPT > 200 or > 50 was added to the model, HR for death was 1.22 (CI: 1.15-1.28) and 1.19 (CI: 1.12-1.25) for each 10% increase in pPT above VE threshold value, respectively. Each log 10 increase in VCY was associated with 70% greater mortality risk (HR=1.70, CI: 1.45-1.99). When VCY and pPT > 50 were both added to the model, each remained significant, but when VCY and pPT > 200 were both added, only pPT > 200 remained independently predictive of death. Conclusion: Each measure of long-term VE was independently associated with mortality. Combined VCY and pPT > 50 captured more independent information about mortality than combined VCY and pPT > 200, suggesting pPT > 50 is a better indicator of viral danger in ART-treated patients.

Poster Abstracts

CROI 2018 343