2. Southern Africa out of control Swaziland South Africa Namibia Botswana Zimbabwe Lesotho Zambia 19902007 400 800 1200

3. P. Godfrey-Faussett, H. Ayles, N. Beyers Southern Africa Map Source: Google Earth October 2007 Zambia Western Cape 0 30 Km 15 Zambia 0 400 Km 200 A community randomized trial of two interventions delivered to ~1,000,000 people while strengthening the existing health systems

4. Enhanced Case finding (ECF)  Community Mobilization and sputum collection  School intervention  Open Access at the clinic  Guiding Principles ▪Every person able to give sputum within 30 min walk ▪Sputum smear results within 48 hours

5. Household Intervention (HH)  Using a TB patient as the Gateway to a household at risk of TB and HIV  3 visits (0,2, completion TB treatment)  Group education TB/HIV  TB screening  HIV testing (group, couple, individual)  Counselling and referral for care

7. Actual Trial  Total Population 962,655  6 communities per arm  Primary endpoint:  Prevalence of TB ●Enhanced case finding (ECF) Vs no ECF ●Household Intervention (HH) Vs no HH  Secondary Endpoint :  Community level: TB transmission TB/HIV at the clinic: 257,698 Enhanced Case Finding: 148,090 Household: 257,729 ECF & Household: 299,138

8. Prevalence surveys  Random sample of adults living in each of 24 communities  Geographically based clustered sample used  Every adult equal chance of being picked  Respiratory sample collected from every consenting adult and cultured on MGIT

9. Household enumerated 48,395 Eligible Individuals 123,790 No consent from household 7,055 Individual Consent 90,601 QuestionnaireRespiratory sample Individuals not found or no consent 33,189 Household visited 55,450 HIV test, blood sugar Field work Flow

10. Laboratory methods  Sample receipt  Samples transported to 5 laboratories (furthest site 800 km)  Each sample inoculated on two manual MGIT tubes ●Reduce proportion of contaminated samples ●Increase yield  QA: Each batch of samples processed with one positive and negative control  Each laboratory working at maximum capacity (100 per day in each mini-lab, 200 per day in centralised lab)  Isolate identification  ZN stain  MPB 64 Ag test  16s rRNA gene sequencing done on samples that were either ZN stain positive or MPB64 positive

12. Evaluable Culture 64,430 M.tb (16S) 884 No TB 63,546 Both cultures Contaminated 9,461 Consented 90,601 Batch rejected due to QA failure 16,710 Laboratory Flow

13. Prevalence of TB  Zambian sites ▪Prevalence 542/100,000 adults (SD 263) ▪Community range 221-1,096/100,000  South African sites ▪Prevalence 2,319/100,000 adults (SD 487) ▪Community range 1489-3054/100,000

14. TB/HIV @ clinic 770/100,000 (ref) ECF 1,010/100,000 aRR 1.03 [0.71-1.50] Household 700/100,000 aRR 0.77 [0.53-1.13] ECF & Household 880/100,000 aRR 0.89 [0.61-1.29]

15. TB/HIV @ clinic ECF Household ECF & Household Prevalence 880/100,00 (Ref) Prevalence 780/100,000 Adj RR 0.82 (0.64-1.05) P=0.07

16. TB/HIV @ clinic ECF Household ECF & Household Prevalence 730/100,00 (Ref) Prevalence 940/100,000 Adj RR 1.09 (0.85-1.40) P=0.48

17. Transmission endpoint  Longitudinal design  Direct measure of incidence of tuberculous infection  Follow children TST negative at baseline and measure rate of TST conversion  Advantage over repeated cross sectional design in that cumulative incidence would be acquired throughout child’s life and not just for the duration of the interventions  Trial outcome, so favour specificity over sensitivity

18. 2 nd TST Survey results  12,075 children seen, assented, injected and read at 2 nd survey.  8809 negative at baseline.  Median follow-up period was 4 years (IQR: 3.5- 4.7).

19. TST conversion rate (per 100 person-years) against baseline prevalence of TB infection in 6-11 year olds

21. Summary  Community randomized trial conducted in 24 communities in Zambia and Western Cape SA  Great need for interventions to reduce TB and HIV  HH intervention reduced the prevalence of culture positive TB by 18%  HH intervention also associated with an important reduction in incidence of tuberculous infection in children  No effect seen from ECF intervention

22. Future Directions  ZAMSTAR has built a huge team involving researchers, communities and health systems  Platform for the PopART (HPTN 071) study  Builds on ZAMSTAR household counselling intervention to deliver universal coverage of combination prevention and test and treat

23. Lessons for impact evaluations  Heterogeneity of communities ▪Opportunity for interventions ▪Challenge for trial design ▪TB>HIV  Herd effects ▪TB and HIV intervention trials need to be huge ▪Need to understand transmission networks  HIV incidence similar to TB incidence rates ▪TB and HIV intervention trials need to be huge

24.  ZAMSTAR Team; Communities; Participants; Zambian Ministry of Health; Western Cape Provincial and South African National Department of Heath; CREATE; Bill and Melinda Gates Foundation (Grant No. 19790.01)