CROI 2018 Abstract eBook

Abstract eBook

Poster Abstracts

Rwanda’s national HIV program, one of the first cohorts receiving third-line ART in sub-Saharan Africa. Methods: We assessed outcomes on third-line therapy for a period of 5 years. Virological failure was defined as 2 VL >1,000 copies/mL separated by at least three months and appropriate adherence counseling. We calculated the proportion of patients achieving VL suppression after 3rd-line initiation.VL suppression were defined as < 1000 copies/mL. Patients were censored at date of death, date of stopping ART, or June 2016. We reported treatment-limiting adverse and drug resistance mutations on third line ART. Results: Of 7,625 patients on 2nd line ART in Rwanda, 38 with treatment failure on PIs based regimen, started 3rd line ART. one patient died (2,6%) and 37 were retained over a median follow-up time of 30 months (IQR 12-38), median age was 42 years (IQR 26-50) and 40.5%were male.Median CD4 cell count at start of 3rd-line was 165 cells/µL (IQR 70-329) and median VL was 57,000 copies / mL (IQR 16,200-151,000). Genotyping was performed in 83.8% of participants prior to 3rd-line initiation; across all drugs, 96.5%, 63.4% and 75.8% had two or more mutations on NRTI, NNRTI and protease inhibitors, respectively.Overall, the majority of patients were susceptible to raltegravir and darunavir/ritonavir (Table1). Twenty-five of 28 self-reported good adherence to medication; 71.4 % achieved viral suppression < 1000 copies/mL; 8.8% of participants experiencing adverse events on ART. Conclusion: Our report is one of the few evaluations of 3rd-line ART program in SSA. Almost 3 in 4 patients achieved viral suppression.Our findings demonstrate the feasibility of providing third-line ART in a routine programme setting and indicate a need for strong surveillance to achieve better clinical outcomes.

non-pregnant. Deaths were similar (19 in UG, 15 in SA; p=0.21). In UG, median adherence was 89% (IQR 74-97) and viral suppression was 85%, and did not differ among cohorts (p>0.10; Figure). In SA, median adherence was higher in early/non-pregnant vs early/pregnant or late/non-pregnant (72%, 33%, 47%, respectively, p<0.001), with similar trends in viral suppression (91%, 59%, 73%; p<0.001). Adherence was higher with increasing year of age (0.5 percentage points [pp] in UG and SA), employment (10.9pp in UG, 9.2pp in SA), and marriage (12.3pp in SA). Adherence was lower with heavy alcohol (-20.5pp in UG), depression (-20.5pp in SA), and food insecurity (-6.5pp in SA). Non-significant predictors were sex, education, physical well-being, prior knowledge of HIV, stigma/disclosure, and other medications. Adherence was lower in months 0-6 vs months 6-12 (median difference -5.6pp in UG, -13pp in SA; p<0.001). Conclusion: ART adherence with early initiation is as high or higher than with late initiation, suggesting current universal access policies may indeed lead toward an AIDS-free generation. However, challenges remain for some, including pregnant women and those with late ART initiation in South Africa. Significant contextual differences highlight need for differentiated care with attention to alcohol use, depression, food security, youth, marital status, and employment.

Poster Abstracts

513 ADHERENCE TO COMBINATION ANTIRETROVIRAL THERAPY (ART) IN SUB-SAHARAN AFRICA Judith J. Lok 1 , Elisavet Syriopoulou 2 , Dustin J. Rabideau 1 , Beverly Musick 3 , Jeffrey N. Martin 4 , Kara K. Wools-Kaloustian 3 , Ronald Bosch 1 , Ann Mwangi 5 , Constantin T. Yiannoutsos 3 1 Harvard University, Cambridge, MA, USA, 2 University of Leicester, Leicester, UK, 3 Indiana University, Indianapolis, IN, USA, 4 University of California San Francisco, San Francisco, CA, USA, 5 Moi University, Eldoret, Kenya Background: Accurate estimation of adherence to ART is crucial for assessing the effectiveness of HIV treatment programs. Routine estimates of adherence to ART are based on crude percentages among patients in HIV care in the treatment program. These crude percentages significantly overestimate program effectiveness, because they do not consider data from patients who are deceased or no longer in care. Dead patients obviously can’t have perfect adherence, but they could perhaps have been alive with perfect adherence in a different treatment program. Or, patients who died because of poor adherence in another treatment program could perhaps have been alive with perfect adherence in this treatment program. We should analyze death and adherence jointly, and optimize the number of patients with the most favorable prognosis: alive with perfect adherence. Methods: We examined data from 25,260 HIV-positive patients from the East Africa IeDEA Consortium. Adherence was self-reported. We considered additional information obtained from a subset of patients who were lost to program but were traced later (the “outreach sample”). We used Inverse Probability of Censoring Weighting (IPCW) adapted to Missing Not At Random data to adjust for patient characteristics predicting loss to program. In the absence of treatment information in the outreach sample, we assumed (rather conservatively in this setting) that the probability of being on ART among those alive and lost to programwas half the probability of being on ART among similar patients still in the treatment program. Results: 9190 of 25,260 patients (36%) were male; median age 37 years (IQR 31-44); median CD4 count at ART initiation 112 cells/μl (IQR 49-180). 7974 (32%) were lost to program through 36 months after starting ART. Of these, 1053 (13%) were traced. The crude one-year estimate of on-ART and perfect adherence was 94%, which is substantially higher than the adjusted estimate of 74%, which considers the patients who left the program or died. The adjustment was larger for later times.

512 ADHERENCE IN EARLY VERSUS LATE ART INITIATION IN SUB-SAHARAN AFRICA Jessica E. Haberer 1 , Bosco M. Bwana 2 , Catherine Orrell 3 , Stephen Asiimwe 4 , Gideon Amanyire 5 , Nicholas Musinguzi 2 , Mark J. Siedner 1 , Kathleen Bell 1 , Annet Kembabazi 2 , Stephen Mugisha 4 , Victoria Kibirige 2 , Anna Cross 3 , Nicola Kelly 3 , David R. Bangsberg 1 1 Massachusetts General Hospital, Boston, MA, USA, 2 Mbarara University of Science and Technology, Mbarara, Uganda, 3 Desmond Tutu HIV Foundation, Cape Town, South Africa, 4 Kabwohe Clinical Research Center, Kabwohe, Uganda, 5 Makerere University Joint AIDS Program, Kampala, Uganda Background: Universal ART access policies and aspirations for an AIDS-free generation depend on high ART adherence with early stage HIV infection; however, adherence may be difficult in the absence of illness and associated support. Methods: We prospectively observed 3 cohorts initiating ART in routine care in southwestern Uganda (UG) and Cape Town, South Africa (SA): early (CD4≥350 cells/μl) initiation for men and non-pregnant women; early initiation for pregnant women; and late (CD4‹200 cells/μl) initiation for men and non-pregnant women. Socio-behavioral questionnaires and viral load were performed at 0, 6 and 12 months. Adherence was monitored in real-time (Wisepill). Loss to follow-up was treated as non-adherence/viremia and death as viremia; data were censored at disenrollment. Predictors of adherence were assessed by multivariable linear regression; sites were analyzed separately given socio-economic/cultural differences. Changes over time were assessed by fixed effects regression. Results: Of 904 individuals enrolled, data were available for 868 (96%): 322 (37%) early/non-pregnant, 198 (23%) early/pregnant, and 348 (40%) late/

CROI 2018 186