Knowing your epidemic and knowing your response maximising

Knowing your epidemic and knowing your response – maximising routinely collected data to measure and monitor HIV epidemics in sub-Saharan Africa Monitoring mortality in ART programmes Matthias Egger Institute of Social & Preventive Medicine (ISPM) University of Bern, Switzerland

Improvements in life expectancy across the globe Wandeler et al. Curr Op HIV/AIDS 2016

Life expectancy on ART compared to the general population Poster TUPEB 031, Abstract 4072

• International consortium established to address clinical and health services research questions • >500 sites in 7 regions, harmonized data collection • > 1 million adults and children • Funded by NIH • www. iedea. org 4

Retention on ART • Data from 105 cohorts in 41 countries from six Ie. DEA regions (Southern Africa, Central Africa, West Africa, Asia-Pacific, Caribbean & Central/South America, North America). • 301, 756 adults and children with HIV who enrolled in an HIV care program and started ART between 2009 -2014. 5

Linkage with vital registry • Of deaths recorded in patients files after 2003, 94. 0% were recorded by civil registration, with higher completeness in urban areas, older adults and females. • Of deaths recorded by civil registration only 35. 0% were recorded in patient files • When the information from the two systems was combined, an estimated 96. 2% of all deaths were recorded.

Field tracing of patients lost to follow-up

Zurcher et al submitted Meta-analysis and meta-regression analysis of tracing studies

Mortality in patients lost Zurcher et al submitted

Undocumented transfer Zurcher et al submitted

Treatment interruption Zurcher et al submitted

Correcting mortality for loss to follow-up Egger M et al. PLo. S Med. 2011; 8: e 1000390

Correction factors (c) • 168 correction factors (2 genders x 3 time periods x 4 age groups x 7 CD 4 groups) • 2 correction factors Anderegg et al submitted Gender Age (years) CD 4 count (cells/µL) ART period (months) Female Female Female Female Female Etc. 15 -24 15 -24 15 -24 15 -24 15 -24 <50 50 -99 100 -199 200 -249 250 -349 350 -499 >=500 0 -6 6 -12 >12 0 -6 6 -12 >12 0 -6 6 -12 Correction factors (No. of factors in set) South Africa (2) South Africa (168) 1. 64397 2. 189916 2. 189916 1. 64397 2. 189916 1. 814623 1. 697969 1. 975792 1. 414255 1. 622381 2. 072795 1. 556734 1. 702032 1. 964165 1. 680239 1. 612833 1. 967266 1. 311076 1. 61538 2. 695073 3. 695248 2. 213003 2. 800414 3. 79234 2. 299502 Kenya (168) 2. 03894 1. 740988 2. 696108 2. 048315 1. 40328 2. 102754 2. 040634 1. 701063 2. 634279 1. 91695 1. 333495 1. 616284 1. 993837 1. 442466 1. 959174 2. 353412 1. 302914 1. 579218 1. 96031 1. 969346

Anderegg et al submitted Sensitivity analyses based on worst and best case scenarios

Conclusions • Monitoring of mortality in ART programmes requires data on mortality in patients lost to follow-up and patients transferred out • ART programmes should trace patients lost to followup to ascertain their status and bring them back to care • National programmes should introduce unique IDs so that patients can be traced across facilities • MESH will develop refine approaches to correcting mortality and develop applications for use in the field

Thank you • Special thanks to Nina Anderegg, Leigh Johnson, Beth Zaniewski, Olivia Keiser and our funders

Reserve slides

Study-level determinants of mortality Variable Calendar year Study duration Setting Per 1 -year increase Rural Urban Population Children Adults *adjusted for all variables listed Crude Adjusted* OR (95% CI) 0. 84 (0. 74 -0. 96) a. OR (95% CI) 0. 86 (0. 78 -0. 95) 0. 82 (0. 74 -0. 92) 0. 84 (0. 77 -0. 92) 1 0. 51 (0. 25 -1. 01) 1 1. 51 (0. 63 -3. 62) 1 0. 59 (0. 36 -0. 98) 1 1. 47 (0. 78 -2. 78)

Sensitivity analyses South African correction factors Set of two Set of 168 Southern Africa 6/168 (3. 6%) 7/168 (4. 2%) Central Africa 4/168 (2. 4%) 5/168 (3. 0%) Latin America 5/168 (3. 0%) Asia Pacific 1/168 (0. 6%) East Africa 0/168 (0%) 0/186 (0%) West Africa 0/168 (0%) North America 0/168 (0%) 0/186 (0%) East African correction factors (168) 10/168 (6. 0%) 7/168 (4. 2%) 2/168 (1. 1%) 0/168 (0%) Anderegg et al. submitted

Methods for Correction of Mortality Estimates for Loss to Follow-Up after ART Initiation • Double-sampling • Frangakis CE, Rubin DB. Addressing an idiosyncrasy in estimating survival curves using double sampling in the presence of self-selected right censoring. Biometrics 2001; 57(2): 333 -342. • Geng EH et al. Sampling-based approach to determining outcomes of patients lost to follow-up in antiretroviral therapy scale-up programs in Africa. JAMA 300; 2008: 506– 507. • Multiple imputation • Brinkhof MW et al. Adjusting mortality for loss to follow-up: analysis of five ART programmes in sub-Saharan Africa. PLo. S One 2010, 5(11), e 14149.

Web calculator http: //www. iedea-sa. org/index. php? id=2726

Comparisons between methods