CROI 2020 Abstract eBook

Abstract eBook

Poster Abstracts

Conclusion: UTT has not led to an increase in early care-seeking, with most patients still presenting with CD4<350. The impact of UTT on numbers starting ART has been limited by late presentation in South Africa.

1135 CD4 COUNT AT ART INITIATION IN THE UTT ERA: AN INTERRUPTED TIME SERIES ANALYSIS H. Manisha Yapa 1 , Hae-Young Kim 2 , Kathy Petoumenos 1 , Frank Post 3 , Awachana Jiamsakul 1 , Jan-Walter De Neve 4 , Frank Tanser 5 , Collins C. Iwuji 6 , Kathy Baisley 7 , Maryam Shahmanesh 8 , Deenan Pillay 8 , Mark J. Siedner 9 , Till Bärnighausen 4 , Jacob Bor 10 1 Kirby Institute, Sydney, NSW, Australia, 2 Africa Health Research Institute, Mtubatuba, South Africa, 3 King's College Hospital NHS Foundation Trust, London, UK, 4 Heidelberg University, Heidelberg, Germany, 5 Africa Health Research Institute, KwaZulu-Natal, South Africa, 6 Brighton and Sussex Medical School, Brighton, UK, 7 London School of Hygiene & Tropical Medicine, London, UK, 8 University College London, London, UK, 9 Harvard T.H. Chan School of Public Health, Boston, MA, USA, 10 Boston University, Boston, MA, USA Background: South Africa implemented universal test and treat (UTT) in September 2016 in an effort to encourage earlier initiation of antiretroviral therapy (ART). We conducted an interrupted time series (ITS) analysis to assess the impact of UTT on median CD4 count at ART initiation among adults attending public sector primary care services in rural South Africa. Methods: We analysed data from individuals >=16 years old initiating ART between 2014 and 2019 at 17 clinics in northern KwaZulu-Natal and registered on TIER.net, the national ART clinical database. Our outcome of interest was CD4 count at ART initiation, defined as the value closest to ART start date in a window from 6 months prior to 3 months after ART initiation. Our primary exposure of interest was calendar period, based on CD4 eligibility expansions: (i) =September 2016). We used a segmented linear regression model with a continuous time variable, binary exposure variables for each policy change, and time-by-policy interaction terms. To distinguish between short- and longer-term effects of eligibility expansions, we allowed a change in trend 12 months after policy rollout. We fitted separate regression models for men and women. Results: Between July 2014 and March 2019, 20,603 (54% under UTT) individuals (69% female) aged >=16 years commenced ART. Median age at ART initiation was 30 (interquartile range 25-38) years. CD4 counts within window were available for 74% individuals. In January 2015 median CD4 at ART initiation was 381 cells/µL among women and 282 cells/µL among men. After UTT implementation, there was an immediate increase in median CD4 at ART initiation of 123 cells/µL (95% CI 81.7 to 164.3, p<0.001) among women, and 98.3 cells/µL (95% CI 75.6 to 121.0, p<0.001) among men (Figure 1). However, there was a significant downward monthly trend in CD4 count at ART initiation in both women (-12.5 cells/µL, 95%CI -18.1 to -6.9, p<0.001) and men (-7.0 cells/ µL, 95%CI -11.2 to -2.7, p=0.002) for 12 months after UTT implementation. The trends stabilised thereafter (Figure 1). Conclusion: UTT led to an immediate boost in earlier initiation of ART in this rural community. However, the effect declined over time before stabilising. More efforts are needed to increase early ART initiation, particularly among men.

1136 USING SOCIAL NETWORKS TO REACH INDIVIDUALS WITH LOW CD4 AT HIGH RISK OF DEATH Lillian Brown 1 , Yiqun Chen 2 , Laura B. Balzer 3 , Gabriel Chamie 1 , James Ayieko 4 , Dalsone Kwarisiima 5 , Jane Kabami 5 , Norton Sang 4 , Edwin D. Charlebois 1 , James Peng 1 , Yusuf Mwinike 5 , Elizabeth A.Bukusi 4 , Moses R. Kamya 6 , Diane V. Havlir 1 , Maya L. Petersen 7 1 University of California San Francisco, San Francisco, CA, USA, 2 University of Washington, Seattle, WA, USA, 3 University of Massachusetts Amherst, Amherst, MA, USA, 4 Kenya Medical Research Institute, Nairobi, Kenya, 5 Infectious Diseases Research Collaboration, Kampala, Uganda, 6 Makerere University, Kampala, Uganda, 7 University of California Berkeley, Berkeley, CA, USA Background: HIV+ persons with low CD4 (<200 cells/mm 3 ) are at high risk of death without effective treatment. We evaluated the potential for a social network strategy based on outreach to social contacts of HIV+ persons in care to identify HIV+ individuals who had CD4<200 and were out of care. Methods: Adult (≥15 years) residents enumerated during a 2013-2014 census in 32 rural Kenyan and Ugandan communities in the SEARCH Study (NCT01864603) named social contacts in five domains: health, money, emotional support, food, and free time. Named contacts were matched to enumerated residents to build social networks among 150,395 adult residents; 135,484 (90%) were tested for HIV; 117,593 had at least one contact and were included in analyses. The target population was defined as HIV+ adults with CD4<200 and out of care. We evaluated strategies for reaching this target population based on outreach to 1st degree contacts of two index populations: 1) all HIV+ adults in care, 2) HIV+ adults in care with CD4<350. For each strategy we calculated coverage (% of the target population potentially identified), number needed to screen (NNS, # of persons outreached to per target individual identified), and ratio of coverage and NNS of each index population relative to the other. Clustering was quantified with an assortative mixing coefficient; p-values were based on randomly permuting node labels. Results Among 10,285 adults known to be HIV+ at baseline with at least one contact, 8,168 had a record of HIV care, of whom 1,904 had CD4<350; 394 HIV+ adults had CD<200 and were out of care. HIV+ persons in care had an average of 4.3 1st degree network members; HIV+ persons in care with CD4<350 had an average of 4.4 1st degree network members. An outreach strategy to 1st degree contacts of all HIV+ adults in care would have reached 40% of target persons (p<0.001) and required outreach to 52 contacts per target individual identified (p<0.001). Outreach to 1st degree contacts of HIV+ in care with CD4<350 would have reached 15% of target persons (p<0.001) and required 31 contacts per target individual identified (p<0.001)[Table]. The assortative mixing coefficient was 0.009 for persons with CD4<200 out of care, 0.02 for HIV+ in care with CD4<350, 0.10 for all HIV+ persons. Conclusion: HIV+ persons with low CD4 who are out of care are socially connected to HIV+ individuals engaged in care. An outreach strategy through the social networks of HIV+ persons in care may be an effective way to reach this high-risk population.

Poster Abstracts

CROI 2020 429