Outcomes in ART treatment programmes with and without access to routine viral load monitoring

1. Outcomes in ART treatment programmes with and without access to routine viral load monitoring Olivia Keiser on behalf of IeDEA Southern Africa okeiser@ispm.unibe.ch

2. Background • Debate on place of routine viral load monitoring in scale-up programmes • South Africa: routine VL monitoring part of national programme Malawi and Zambia: no routine viral load monitoring 2

3. Selection of patients/ programmes • Treatment naïve patients > 15 years, NNRTI-based ART regimen. All sites participate in IeDEA Southern Africasee: www.iedea-sa.org 3

4. Mortality 0.10 No viral load monitoring(Malawi and Zambia) Routine viral load monitoring(South Africa) Cumulative mortality 0.05 0.00 0 1 2 3 Years since HAART start

5. 1) Differences in patient characteristics ? 5

6. Patient characteristics at start of ART 6

7. Sites with and without (reference) VL monitoring Competing risk regression with loss to follow-up and start of second-line therapy as competing events *Adjusted for age, sex, CD4 cell count and clinical stage of disease 7

8. 2) Differences in loss to follow-up ? 8

9. Sites with and without (reference) VL monitoring • Sites with < 15% loss to follow-up 2 years after ART start Competing risk regression *Adjusted for age, sex, CD4 cell count and clinical stage of disease 9

10. 3) Differences in background mortality? 10

11. Expected non-HIV-related death rate per 100 pyrs * Based on South African rates 11

12. 4) Delayed/missed detection of treatment failure and switching in non-VL sites? 12

13. Multistate model Treatment failure - virologic (viral load sites) - immunologic (non viral load sites) Six months after ART start Switch Death Loss to follow-up Putter et al., Stat Medicine 2007

14. 1 0.75 Proportion of patients 0.5 0.25 0 From 6 months after ART start Non VL sites VL sites 0 200 400 600 800 0 200 400 600 800 Time (days) Time (days) Remaining on first-line ART Death Second-line therapy Loss to follow-up CD4 criteria for switching

15. 1 0.75 Proportion of patients 0.5 0.25 0 From failure Non VL sites VL sites 0 200 400 600 800 0 200 400 600 800 Time (days) Time (days) Death Second-line therapy Loss to follow-up CD4 criteria for switching

16. 1 0.75 Proportion of patients 0.5 0.25 0 From switch Non VL sites VL sites 0 200 400 600 0 200 400 600 800 800 Time (days) Time (days) Death Second-line therapy Loss to follow-up

17. 1 0.75 Proportion of patients 0.5 0.25 0 0 200 400 600 800 From switch VL sites (linkage with death registry) Non VL sites 0 200 400 600 800 Time (days) Time (days) Death Second-line therapy Loss to follow-up

18. Conclusions • Mortality rate higher in sites without routine viral load monitoring • Difference probably not explained by differences in- Patient characteristics- Loss to follow-up- Background mortality • Alternative explanations: • Delayed/missed switching- Diagnostic and treatment capacities (for OIs), other differences in patient management 18

19. Diagnostic capacities Not generally available or off site yes no 19

20. Limitations • Few sites and countries included • Mortality in patients lost to follow-up unknown in many sites • Linkage with death registry only possible for patients with South African ID number available • Estimates of HIV-free mortality might be inaccurate This is an observational study - Ideally one would need a randomized trial 20

21. Acknowledgement University of Bern: Matthias Egger, Thomas Gsponer, Janne Estill, Gilles Wandeler, Franziska Schöni-Affolter, Martin Brinkhof, Fritz Käser, Claire Graber Data centerCape Town: Andrew Boulle, Morna Cornell, Leigh Johnson,Nicola Maxwell Site investigators: Benjamin Chi, Matt Fox, Mhairi Maskew, Catherine Orrell, Hans Prozesky, Ralf Weigel, Andrew Westfall 21