Effect of community-wide isoniazid preventive therapy on tuberculosis among South African gold miners

“Thibelo TB”

Aurum Health Research

LSHTM

JHU

Gold mining companies

Dept. of Health, SA

Study outline

• To investigate the effect of community-wide TB preventive therapy (TBPT) in setting of high HIV prevalence (gold mines in South Africa)

• Community randomised controlled study community-wide TBPT vs. standard TB control (targeted TBPT for high risk groups)

• Community = mine shaft(s) and associated hostel

• Paired design

Study outline (2)

• Primary outcome: TB incidence measured 12 months after enrolment (over a 12 month period)

• Secondary outcomes – TB incidence among HIV infected individuals,

measured 12 months after enrolment (over a 12 month period)

– TB case notifications, over the 24 month period– TB culture prevalence at the end of the follow-up

period

Study month0-3

4-6 7-9 10-12

13-15

16-18

19-21

22-24

25-27

28-30

31-33

34-36

37-39

40-41

42-45

46-48

49-51

52-54

Recruitment of clusters

Prevalence survey

Primary outcom

e

measurem

ent period

• 9 months of TBPT• Measure primary outcome over 12 months, from 13-24

months after enrolment

Enrol clusters over 15 months

9 months of TBPT

Eligibility criteria for communities

• Expanded DOTS program which includes– Active TB case finding using chest x-ray– Standardised monitoring and reporting

• VCT– same-day confidential testing and counselling– Individuals found to be HIV-infected are offered

referral for HIV care

• Isoniazid and cotrimoxazole preventive therapy for HIV-infected individuals

Baseline data

• TB case notifications rates measured over the 6 month preparation phase

• Baseline Survey (at recruitment)– Sampling ~ include everyone on certain days ?– TB and silicosis prevalence (using latest chest x-ray)

measured at the time of recruitment from a random sample [n=1000 per cluster]

– Baseline HIV prevalence (using saliva) measured at the time of recruitment from a random sample [n=1000 per cluster]

Paired versus stratified design?

• Important confounders are– Baseline TB case notification rates– Prevalence of HIV infection– Prevalence of silicosis– Baseline TB prevalence– Trends in the proportion of the HIV-infected workforce

receiving ART over the study period

• Paired design– mining company and baseline TB case rates

Sample Size (1)

• Primary outcome: TB incidence over 12 months, measured amongst hostel dwellers– 90% power, type I error of 5%– K=0.25, paired design– Average community size of 2500 (80% live in hostels)– Factored in a potential effect of ART– Assumed a 60% reduction in the community wide PT

arm

Indicates 7 matched pairs

Sample Size (2) • Secondary outcomes:

– 80% power, type I error of 5%, K=0.25, paired design– Average community size of 2500– Factored in a potential effect of ART

• HIV-specific TB incidence ~ 60% reduction in the community wide PT arm (over 12 months)

• TB case notification rates ~ 50% reduction in the community wide PT arm (over 24 months)

• TB culture prevalence ~ 50% reduction in the community wide PT arm, based on community size of 750

Enrolment

• Intervention and control enrolment teams will work at one pair of clusters at a time

• Both arms– Consent to use data – Baseline survey [previous x-ray for TB radiological

prevalence and silicosis score; saliva sample for HIV testing]

• Intervention arm– Symptom screen

Intervention (1)

• TB screening to exclude active TB• using symptom questionnaire and new and previous

chest x-ray • Investigated further if new abnormality or symptoms

• All consenting participants offered 9 months of IPT

• Monthly visits to• Dispense IPT• Monitoring for toxicity and side effects

Intervention (2)

• IPT adherence enhancing measures– Patient education– Self-adherence– Treatment supporters– incentives?

• Monitoring of adherence to IPT– Questionnaire (monthly); pill count etc– Urine testing for INH

Measuring primary outcome

TB incidence measured over a 12 month period

• Human Resources– List of miners living at each hostel (cluster)– Redundancies and death information

• TB database– All TB diagnoses collected– Case definitions applied

Measuring secondary outcomes

TB incidence measured over a 12 month period, amongst HIV-infected

• As before • All TB diagnoses will be offered anonymous-

unlinked HIV testing• Use the HIV prevalence from baseline survey

Measuring secondary outcomes (2)

TB case notification rates over a 24 month period

• Human Resources (info. collected every 6 mths)– List of miners living at each hostel (cluster)– Redundancies and death information

• TB database– All TB diagnoses collected

• Also calculate TB case notification in four 6 month intervals

Statistical Analysis

Unadjusted Analysis• Point estimate (GM of the pairwise estimates)

– Log(RR)=(1/c)∑log(RRj )=(1/c) ∑log(r1j /r0j)

where r1j = TB incidence in cluster j, intervention arm and r0j = TB incidence in cluster j, control arm

– And 95% CI

• Paired t test, applied to the log(rates)

Statistical Analysis (2)

• Poisson regression model fitted to the individual data, including all a priori confounders and an indicator variable for matched pairs

• Calculate the observed (Oij) and fitted (Eij) numbers of TB events for each cluster

• Calculate the GM of Oij/Eij

Other Issues

• Document TBPT use in the control arm (though the “Wellness” clinics offering care for HIV-infected individuals

• Pilot study of quantiferon & TST to measure TB infection