Thibela TB: Community-wide IPT did not improve TB control in South Africa’s mines at the population level

This article is more than 12 years old. Click here for more recent articles on this topic

Giving a six- to nine-month course of isoniazid preventive therapy (IPT) at a ‘community-wide level’ – to everyone working at randomly selected South African gold mines – had no effect on TB incidence, TB prevalence or all-cause mortality in the population, when compared to a cluster of gold mines randomised to standard TB programme management, according to the ‘Thibela TB’ study, the largest IPT intervention study ever to be conducted.

The results, released during a late-breaker presentation on the final day of the 19th Conference on Retroviruses and Opportunistic Infections (CROI) in Seattle, are clearly disappointing given the extremely high burden of tuberculosis (TB) facing workers in the mines of South Africa, where there is also a high prevalence of HIV co-infection – and had investigators scrambling to provide an explanation, which may involve inadequate uptake and adherence in the study.

However, part of the explanation could have been uncovered in a post-hoc analysis, described in a subsequent presentation, which found that IPT did have an effect at the individual level – reducing TB while people were taking it – but this effect was lost very quickly once IPT was stopped; suggesting that levels of TB transmission in the mining communities may be higher than realised – and that a longer, potentially continuous course may be necessary to better control TB.


person years

In a study “100 person years of follow-up” could mean that information was collected on 100 people for one year, or on 50 people for two years each, or on ten people over ten years. In practice, each person’s duration of follow-up is likely to be different.

active TB

Active disease caused by Mycobacterium tuberculosis, as evidenced by a confirmatory culture, or, in the absence of culture, suggestive clinical symptoms.

control group

A group of participants in a trial who receive standard treatment, or no treatment at all, rather than the experimental treatment which is being tested. Also known as a control arm.

latent TB

A form of TB that is not active. Persons with latent TB are infected with M. tuberculosis but do not have any symptoms and they cannot spread TB infection to others. Only specific tests will tell if anyone has latent TB. Treatment for latent TB is recommended in people living with HIV. 


Material coughed up from the lungs, which can be examined to help with diagnosis and management of respiratory diseases.

“We are working to better understand the reasons for these results so that we can make recommendations to improve TB control in the mines,” said lead investigator Professor Gavin Churchyard, of the Aurum Institute in South Africa, and the London School of Hygiene and Tropical Medicine.


Isoniazid preventive therapy (IPT) is an effective means of preventing the development of active TB in people with latent TB infection. It is estimated that up to a third of the global population will have latent infection with the bacteria that cause TB. Latent TB is more likely to develop into active tuberculosis where an individual has HIV infection or other challenges to their immunity. In people with HIV, IPT could reduce the risk of developing active TB by between 33 and 62%.

Since the mid-1990s, there has been an ever-worsening TB epidemic in the gold mines of South Africa, with case notification rates rising from 1000 per 100,000 in the early 1990s to 5000/100,000 by 2001 (in other words, around 5% of mining employees were developing TB each year). There is a high prevalence of silicosis in the mines which increases the risk of TB. In addition, the incidence of TB was clearly being exacerbated by the growing HIV epidemic (with a prevalence growing to 30% by 2001). A report at the 3rd South African TB Conference in 2010 suggested that the TB incidence may now be as high as 7% per year in the mines.

Detecting the rise in TB incidence, the mines implemented and improved all the standard practices of TB control, such as direct observed therapy, targeted IPT (in other words giving IPT to HIV-positive people and contacts of active cases) and scaling-up HIV testing and ART provision. Despite these standard interventions, the TB epidemic continued to rage, virtually unchecked.

Prior to the HIV epidemic, a public health programme sought to reduce similarly high rates of TB in Alaska by providing community-wide IPT, and this approach was ripe for evaluation in southern Africa, particularly in the mining industry.

The Aurum Institute designed a community-wide trial of IPT in which a number of clusters, comprising all the employees in a mine and its hostel, were randomised either to the standard-of-care intervention for TB control or to a workforce screening programme which identified suspected active TB cases and offered IPT to all employees without active TB. At the request of the mineworkers' unions the study did not test participants for HIV; the lack of data on HIV status makes it more difficult to interpret the results of this study.

The intervention included community mobilisation offered to the entire workforce; TB screening (all potential participants were offered extensive screening for TB symptoms) and chest X-ray, and TB suspects then provided one sputum specimen for microscopy, culture, speciation and drug-sensitivity testing. People with suspected TB were referred for investigation and treatment. Those determined not to have TB were offered nine months of IPT and had monthly follow-up visits for dispensing and screening for TB symptoms and possible side-effects.

Fifteen clusters were included in the study, eight in the intervention arm and seven in the control arm.

Uptake in the first four intervention clusters was moderate, averaging 58% after 16 months; in the second four clusters uptake reached 78%. The latter level of uptake was described by Professor Churchyard as the “best achievable”.


The study collected data in a number of ways: there were baseline and post-study surveys, all cases of TB were of course recorded. But the primary outcome measure was for the 12-month period after IPT was given to the last person entering the study – since the study was trying to determine whether having given IPT for a period would have long-ranging effects on the community. The secondary endpoint was TB prevalence, as determined by sputum culture in a representative sample of participants from the clusters at the end of the study.

The entire study included a total population of 75,596 people, 41,387 of whom were in the clusters randomised to IPT versus 37,209 in the control clusters.  Notably, in the intervention cluster, only 65% (27,126) consented to take part in the intervention and only 23,644 (87.2%) of those who consented actually started IPT – some couldn’t start because of contraindications, or because they were TB suspects etc. So already, only a little more than half actually consented and began taking the intervention.

A comparison of TB incidence at intervention and control sites showed no significant difference nine months after the end of the intervention (3.04 cases per 100 person-years versus 2.96 in the control arm), an adjusted incidence rate ratio of 1.00 (95% confidence interval 0.77-1.31).

TB prevalence was assessed in a sample of employees at each site (7050 at the intervention sites and 5557 at the control sites). Again, no significant difference was found (2.34 vs 2.14%, adjusted rate ratio 0.95, 95% CI 0.83-1.10).

IPT’s effect on individual risk of TB

But did IPT work for the individuals taking it? To find out, Dr Katherine Fielding, also of the London School of Hygiene and Tropical Medicine, conducted an analysis comparing the TB incidence risk among the subset of participants who were involved in the baseline survey portion of the study, comparing those who took IPT versus those who didn’t (in the clusters not randomised to take IPT). For those on IPT, the analysis of time at risk of TB was divided between 0-9 months, when they should have been on treatment, and 9-18 months, and post-18 months.

Once those who did not consent to IPT and people who were ineligible for other reasons were excluded, there were 6263 people in the control arm, and 4646 people in the IPT arm, with baseline characteristics fairly matched and not dissimilar from the study overall.

Over the entire follow-up period, there were 577 TB episodes over 22,939 person-years of follow-up: 382 TB episodes were observed in the control arm over 13,776 person-years for a TB incidence of 2.77/100 person-years; while 175 TB episodes were seen in the IPT arm over 9163 person-years for a TB incidence of 1.91/100 person-years.

Effect of IPT during and following the IPT period



Rate per 100 person-years

Unadjusted incidence risk ratio (IRR) (95% Confidence Interval)

Adjusted IRR

(95% Confidence Interval

0-9 month

IPT period



0.38 (0.19-0.75)

0.37 (0.19-0.72)





9-18 months



0.97 (0.57-1.65)

0.97 (0.57-1.65)





> 18 months



0.83 (0.54-1.28)

0.79 (0.54-1.17)





“During IPT treatment period TB incidence was reduced by two thirds among those who started IPT but the effect was not durable,” concluded Dr Fielding.


Indeed, the lack of durable effect beyond the on-treatment period could partly explain why IPT failed to have a population-level effect in Thibela – suggesting either that treatment (and its uptake and duration) were not long enough to clear latent TB, or that new infections are being transmitted at a much higher rate than previously thought in the mines. Either way, a more potent regimen; a longer or perhaps continuous course of IPT may be necessary. Notably however, this had not been the case in those earlier population-wide studies in Alaska. Though of course, that was before HIV.

But speaking of HIV, the fact that so few men in the South African mines acknowledge having it, strains credulity – particularly bearing in mind a median age of participants (45) and the likelihood that older men are more likely to be infected with HIV than younger men, and the prevailing idea that it is while men are away to work, and at the mines, that so many become infected. That this study did not include HIV screening and treatment (or at least accurate figures about HIV prevalence) could mask whatever effects HIV co-infection and HIV treatment could be having in this population.

Dr Kevin de Cock of CDC-Kenya also raised an important question about the possible impact of silicosis on the results in post-presentation discussion, prompting Prof. Churchyard to respond: “There's little evidence that there's been any improvement in dust control over the past decade, and the risk of TB in people with radiological silicosis is similar to what we observe in people with HIV, and when you have them combined the risk is multiplied, and so we really do need to address the dust control issue in the mines.”

How might IPT perform in a setting where dust control was more rigorous, where IPT uptake was higher and where antiretroviral therapy was more commonly used?

Prof. Churchyard said there are plans underway do some mathematical modelling studies with their data, to perhaps see if transmission rates are higher, or if continuous IPT or alternative regimens may perform better, before embarking on any other study in this population.

In the meantime, these findings can only reinforce the urgent need to address all aspects of the TB epidemic in South Africa's mines and to improve workplace programmes.


Churchyard G et al. Community-wide Isoniazid Preventive Therapy Does Not Improve TB Control among Gold Miners: The Thibela TB Study, South Africa. 19th Conference on Retroviruses and Opportunistic Infections, Seattle, abstract 150aLB, 2012.

Fielding K Individual-level Effect of Isoniazid Preventive Therapy on Risk of TB: The Thibela TB Study. 19th Conference on Retroviruses and Opportunistic Infections, Seattle, abstract 150bLB, 2012. 

A webcast of the session, TB and vaccine preventable diseases, is available through the official conference website.