CROI 2017 Abstract e-Book

Abstract eBook

Poster and Themed Discussion Abstracts

Methods: We evaluated all HIV positive, treatment eligible patients presenting for care in 20 public facilities in southwestern Uganda. We examine visit adherence (the fraction of appointments made within 7 days) using generalized estimating equations and the cumulative incidence of being 14 days late for an appointment using the Kaplan-Meier method. We treated the randomized behavioral intervention as an instrumental variable to assess the effect of same day ART initiation on retention. We assumed that the randomized intervention had no effect on retention other than through the timing of ART initiation, which permits causal inference even in the presence of unmeasured, post-randomization common causes of same day start and retention. Results: Among 12,024 patients in 20 facilities, 63% female with a median CD4 count at eligibility of 310/ul (IQR: 179 to 424), the behavioral intervention did not change overall retention as measured by visit adherence (84.4% in intervention vs. 83.8% in control, p=0.18), but reduced overall the cumulative incidence of being 14 days late for an appointment at 12 months (44.4% to 38.8%, difference = 5.6%, 95% CI: 3.1% to 8.0%, p < 0.0001). The instrumental variable analysis found that same day ART initiation increased visit adherence (83.6% to 85.4%, RD=1.8%; 95% CI: 0.3% to 3.2%, p=0.014) and decreased the cumulative incidence of being 14 days late for a visit at 12 months by 11% in patients who would adopt same day ART initiation. Conclusion: More rapid ART initiation through an intervention that offered providers new knowledge, technology and feedback on performance but left professional judgment intact did not diminish retention. Improving one step in the cascade can in some cases enhance, rather than undermine, subsequent steps. 1006 LOWER RETENTION IN CARE WHEN ART IS INITIATED WITH CD4 ≥ 500 CELLS/ΜL Isabelle A. Meyer-Andrieux 1 , Jim Todd 2 , Robert Keango 3 , Elisabeth Poulet 4 , Prince Nkhoma 3 , Elisabeth Szumilin 4 , David Maman 5 , Leon Salumu 4 1 MSF, Geneva, Switzerland, 2 London Sch of Hygiene & Trop Med, London, UK, 3 MSF, Lilongwe, Malawi, 4 MSF, Paris, France, 5 Epicentre, Cape Town, South Africa Background: NewWHO guidelines for Universal test and treat imply many will start antiretroviral therapy (ART) without symptoms. A critical question is whether patients who initiate ART at high CD4 achieve good retention in care while on ART in sub-Saharan Africa. In the MSF supported project of Chiradzulu, we assessed the association between high CD4 count at ART initiation, retention and survival. Methods: All adults who started ART between 2011 and 2015 in Chiradzulu district were included. Lost to follow up was defined as 9 months after a clinic visit without contact, being dead or transferred out. All analysis were stratified by gender. Multivariate Cox regression models were used to present the effect of CD4 at initiation on retention, adjusted for WHO staging, Body Mass Index, and age. Results: Of 21,705 patients aged ≥15 years who initiated ART between 2011-2015, 16,258 (10,021 females and 6,237 males) had a CD4 count at initiation. Median age at ART start was 38[IQR 32-45] for men and 33[IQR27-40] for women. Among men and women, 239(3.8%) and 1,393(13.8%) initiated ART at CD4>500 cells/µL, respectively. At end of follow- up, 12,757 (78.5%) patients were still in care, 2,538 (15.6%) were lost to follow-up and 661 (4.1%) were dead. Mortality rates were higher for men (2.8 (95%CI 2.5-3.0)) than for women (1.3(95%CI 1.2-1.4)). Survival estimates 48 months after initiation were similar between patients with baseline CD4≥500cells/µL and 350-499, for both women (98.3% vs 98.1%, p=0.95) and men (91.2% vs 92.0%, p=0.33). Retention in care 48 months after ART initiation was 75.4%, 82.1%, 81.9% and 72.1% among women with initial CD4 at ≥500, 350-499, 200-349 and <200 CD4 cells/µL (Log Rank Test, p<0.01) respectively. In the multivariate analysis, women with CD4≥500 cells/µL had a lower retention in care than those who initiated between 350-499 (aHR 0.72; 95%CI 0.60-0.88) and 200-349 cells/µL (aHR 0.82; 95%CI 0.71-0.95). Similar trends were observed for men but the differences were not statistically significant (≥500 vs 350-499 aHR 0.84; 95%CI 0.62-1.14). Conclusion: Initiating ART at CD4>500 cells/µL was associated with lower retention in care when compared to those who initiated ART between 350-499 CD4 cells/µL. Even if these results come mainly fromwomen on PMTCT B+, in the treat-all era, specific attention to those initiating at high CD4 is needed to ensure good retention. Figure:Kaplan-Meier retention curves stratified by gender and CD4 count at initiation, Chiradzulu, Malawi,2011-2015

Poster and Themed Discussion Abstracts

1007 ART INITIATION FOR NEWLY DIAGNOSED PLHIV: IMPLICATIONS FOR TEST & TREAT, SOUTH AFRICA Anna Larsen 1 , Mireille Cheyip 1 , Abraham Tesfay 2 , Peter Vranken 1 , Henry Fomundam 2 , Anthony Wutoh 2 , Getahun Aynalem 1 1 US CDC, Pretoria, South Africa, 2 Howard Univ, Pretoria, South Africa

Background: Despite a decade advancing South African HIV/AIDS treatment policy, 20% of people living with HIV (PLHIV) eligible for antiretroviral treatment (ART) remain uninitiated. South Africa maintains the highest number of PLHIV -18% of the global burden. In anticipation of “test and treat” in South Africa, this analysis describes ART initiation, timeliness, and determinants among newly diagnosed PLHIV. Methods: This analysis used routine data extracted from 35 purposively selected primary health clinics in three high HIV burden districts of South Africa, Gert Sebande, uThukela, City of Johannesburg, from June 1, 2014 to March 31, 2015. Kaplan-Meier survival curves estimated rate of ART initiation. We identified predictors of ART initiation rate and timely initiation (within 14 days of eligibility determination) using Cox proportional hazards and multivariable logistic regression models respectively in Stata 14.1. Results: Based on national guidelines, 6,869 patients were eligible for ART initiation. Most were women (69.65%) and half were under age 30 (50.24%). Under half of men and non-pregnant women were initiated on ART within 14 days (men: 39.78%, 95% confidence interval: 37.7 – 41.9; women: 40.1%, 38.4 – 42.0) with median time to ART initiation over 20 days (men: 22, range: 7-61; women: 23, range: 7-61) and under 70% initiated within 60 days (men: 69.9%, 67.9 – 71.8; women: 68.8%, 66.8 – 70.2). Pregnant women initiated at a faster rate (day of HIV diagnosis: 82.0%, 80.3 – 83.7; within 14 days: 87.6%, 86.1 – 89.0; within 60 days 91.7%, 90.4 – 92.8; median days to ART initiation: 1, IQR: 1-1). From cox proportional hazards and multivariable logistic regression models for men, TB co-infection versus no co-infection predicted lower (adjusted hazard ratio (aHR): 0.7, 0.6 – 0.9) and less timely ART initiation (adjusted odds ratio (aOR): 0.3, 0.2 – 0.4) (Table 1). ART initiation and timeliness varied significantly by district and facility location among both genders, with minimal variation by age, WHO stage, or CD4 count. Conclusion: Men and non-pregnant women newly diagnosed PLHIV eligible for ART in South Africa show suboptimal timeliness of ART initiation. If treatment initiation performance is not improved, test and treat implementation will be challenging among men and non-pregnant women. Test and treat programming should be modeled after the successful implementation practices used to initiate pregnant women on ART at antenatal care in South Africa.

CROI 2017 434