Going into communities to actively screen for tuberculosis (TB) can uncover a very large number of TB cases in settings with a high prevalence of HIV — and within a couple of years, reduces the community’s burden of TB, according to the DetecTB study conducted in Harare, Zimbabwe and presented as a late breaker at the 40th Union World Conference on Lung Health held in Cancún, Mexico in December.
The study, which provided six rounds of periodic intensified case finding (ICF) (one week every six months) to a community with about 100,000 adults, found that TB screening provided via a mobile van visit detected significantly more cases than door-to-door screening. However, during the course of this two and a half year study, 41% of all smear-positive TB within the community was diagnosed by the study’s two interventions.
Even more importantly, by conducting parallel surveys for TB prevalence in the same population at the start and conclusion of the study, the researchers demonstrated that providing periodic ICF had a population level impact — reducing the burden of undiagnosed culture positive TB in this crowded urban setting by 43% (and the burden of smear-positive TB by 44%) by the end of the study.
“The effect we saw was actually huge,” said the primary investigator, Dr Liz Corbett of the London School of Hygiene and Tropical Medicine. “And it’s this secondary outcome that makes our study results so exciting.”
The failure
Indeed, because of the magnitude of the effect (and the rigour of the study), these findings are likely to have a major public health impact — potentially changing the approach to diagnosing TB in high HIV burden settings where traditional passive case finding and DOTS (therapy administered by a well-run TB programme that provides adequate support for treatment adherence) have failed to control the TB epidemic.
Well-supported TB treatment cures the disease — once it is diagnosed — and can reduce onward transmission, but according to Dr Corbett, “we’ve got lots of prevalence data to support the fact that in poor communities, DOTS simply does not penetrate well enough to control undiagnosed TB to the rate you want.”
Part of the problem is that programmes rely on passive TB case finding, which essentially means waiting for people with symptoms of TB to present themselves to a clinic for diagnosis.
But as Professor Bertel Squires of the Liverpool School of Tropical Medicine reminded conference participants during the opening session, each visit to a health facility costs time and money that the poor may ill afford — and given the difficulty in diagnosing TB, as well as problems in the health and laboratory services, it may take repeated visits to access TB services.
“Also, the symptom profile of undiagnosed TB can be quite mild,” said Dr Corbett. “And as everywhere, patients just put off going to see the doctor.”
The best way to address this is through intensified case finding (ICF) within the community according to Dr Corbett.
“We know that there has been a huge epidemic of HIV-related TB presented to the facilities,” she said, “but the way to control that is to go into the communities because most of these patients have been recently infected in the community by casual contact with undiagnosed smear-positive TB. That’s what makes these cases such an important target: you’ll be aiming to cut those cases — which you can do tomorrow if you want to, there’s no time-delay involved in case finding — and theoretically within a short period of time, you will see a reduction in the number of new TB cases and deaths occurring in that community.”
But while many smaller studies have shown that ICF can detect TB cases among people with HIV or within communities, they have been unable to demonstrate population impact due to a lack of baseline prevalence data.
DetecTB
So at the start and end of the study, Dr Corbett and her colleagues in Harare carried out two household enumerations in a community with a population of about 110,000 adults around 41,000 households, which were divided into 46 clusters containing two to three thousand individuals each.
Each census was linked to a survey (involving a one in ten random sample of the population) to establish the prevalence of undiagnosed TB in the same community. The survey at the start of the study contained 10,000 adults; the second, after the last round of case finding, had 11,000 adults.
In the prevalence surveys, sputum samples were obtained from every survey participant, regardless of symptoms, and cultured for undiagnosed culture-positive TB.
The 46 clusters were randomised to be provided with case finding services for one week every six months via one of the two community-based strategies (door-to-door or mobile van screening), with the primary outcome looking at the rate of diagnosis per round, and the secondary outcome comparing the burden of undiagnosed TB before/after the study.
Dr Corbett pointed out that the study also involved some community mobilisation about TB, delivered by loudspeakers and leafleting, with simple public health messages: ‘TB can be infectious for months or years with mild symptoms — especially if HIV-negative — and undiagnosed TB puts families and friends at risk.’
“So in addition to directly averting smear-positive days by diagnosing people earlier, we will have perhaps had an impact on health seeking between rounds and on TB transmission,” said Dr Corbett. However, she pointed out that the cluster size of two to three thousand was sub-optimal to pick up effects on TB transmission.
The interventions were “not terribly labour intensive,” said Dr Corbett. Both case finding strategies relied on community workers — one team of six for door to door screening and one team of three for the mobile van.
Community workers looked for people with symptoms of TB by asking whether anyone in the household had had a cough for two weeks or more or was experiencing unintentional weight loss, drenching night sweats, or coughing up blood. Any symptomatic individual could submit sputum specimens that were sent for smear microscopy, with positive results reported back to participants.
Baseline characteristics among the arms were well matched with an HIV prevalence of 21 to 22%. The prevalence of culture-positive TB was approximately 65 per 1000 in both arms in the 2005/2006 survey. By the time of the second survey, there was a 12-13% increase in the population and number of households (as well as a slight decrease in HIV prevalence to 19%).
Results: Cumulative TB case rate
Going into the study, everyone had expected that the door-to-door strategy would be more effective, according to Dr Corbett. “But at every round, to our surprise, we found the mobile outperformed door-to-door — and this was a strong effect.”
Overall, the study reached about 10,177 participants, but the difference was not in participation in each arm but in the yield: with 4.7% of the participants diagnosed with smear-positive TB in the mobile arm, 2.9% in the door-to-door arm.
Comparing the cumulative rate for smear-positive TB between the two arms, the unadjusted rate ratio in favour of the mobile arm was 1.7 (95% confidence interval (CI) 1.3 – 2.3), and 1.5 in an analysis that adjusted for other factors (95% CI 1.1- 2.0, p = 0.009).
The difference in yield was most pronounced in the crowded high HIV prevalence clusters. “In the lower HIV prevalence, more middle-class neighbourhoods, we did not see a difference between the two arms,” said Dr Corbett. “So again, it shows the mobile does the job best in the very communities where you’re most interested in improving TB.”
Dr Corbett suggested a number of reasons why the mobile van may have performed better. For instance, the mobile van would set up shop in the neighbourhood, and subjects with symptoms would have to go to them — so it was a more active process for the participants. Also, the mobile van intervention was available for a few days instead of just briefly.
“If you stop to think how you react to a knock on the door, you may start to understand perhaps why the mobile performed better,” said Dr Corbett. “You may catch people in their homes at an inconvenient time or they haven’t had time to think about it, and you’re asking for an ‘on the spot decision’. You’re asking people to report symptoms in their household for people who might not be there - they might not feel empowered to do that.”
“If the mobile is there people can see other people going there, can get encouragement from their neighbours. You can even get a sort of ‘party atmosphere’,” she said.
However, there were some differences in the population reached by each intervention: for instance, the door-to-door participants tended to be slightly older. So there may also be a value in performing some door-to-door screening.
“With the door-to-door you might be reaching people who just would not present in any other way. Whereas the mobile you are reaching people who basically want to be diagnosed but haven’t quite got round to it or haven’t quite managed to complete it,” she said.
Dr Corbett noted that 70% of the TB patients and 78% of the TB suspects had not previously sought healthcare anywhere else. “So we getting people before they’ve done any other health seeking,” she said.
Results: population level impact
Again, the study diagnosed 41% of the smear-positive TB cases within the community during this time. However, if the interventions only speeded up diagnosis by a little, there was potentially a chance that they would have little effect on the burden of TB in the community.
“If people are going to be diagnosed tomorrow, you’ve just wasted money and gone to a lot of effort for nothing,” said Dr Corbett. “But if you’ve actually managed to reduce the burden of undiagnosed TB through an intervention, it is likely to have an impact of major public health significance.”
And indeed, the TB prevalence surveys before and after the study showed the interventions resulted in a marked reduction in undiagnosed TB in the community overall: with a 43% in the pre-set definition of culture-positive TB (confirmed by follow-up studies), a 44% reduction in smear-positive TB; a 38% reduction in all TB cases and a 44% reduction in culture positive isolates.
“And the reduction was to rates that are low for rates regionally, so we’ve really had an impact,” said Dr Corbett.
Although the numbers were too small to be statistically significant, a subgroup analysis suggested that the reduction in undiagnosed TB cases was most pronounced in women and the HIV-negative population. For instance, there was a 59% reduction in the burden of undiagnosed TB in the HIV negative population, but only a 25% reduction in people with HIV in the prevalence survey. “I think this is an important proviso here, maybe that the six-month screening strategy is not so good for HIV-positive TB,” said Dr Corbett.
But it is also important to note that the interventions relied on smear microscopy to diagnose TB, and smear microscopy often fails to detect culture positive TB in people coinfected with HIV as they are often smear negative.
In addition, people with HIV are at increased risk of contracting TB within health facilities that do not practice adequate TB infection control — and so may be less impacted by interventions within the community.
Even so, continuing reductions in the number of undiagnosed cases within the community should benefit everyone in the long run.
Indeed, Dr Corbett believes that the reduction in the burden of undiagnosed TB in the community, “is likely to correspond with a major impact on TB transmission rates.”
Implications and discussion
More research is needed to expand the evidence base for case finding, Corbett added. For instance, more work is needed to find out how to engage men in TB case finding, and to improve yields in people with HIV.
“But I would also like to stress that in all case finding interventions, if you don’t have prevalence data, you really can’t interpret your results,” said Dr Corbett.
Community case finding has been attempted elsewhere in the pre-DOTS era, but Dr Corbett noted that its public health impact had never been well-evaluated because of the difficulty in doing concurrent prevalence surveys. As a result, community-based screening has been discouraged up till now, partly due to concerns about sustainability and cost-effectiveness.
But these data may help change that.
“As far as the implications for global TB control, I think we really do have something here,” said Dr Corbett, adding that this ‘not very intensive intervention’ could help meet the Millennium Development Goals on TB. “If it is readily replicable, we would expect major declines in TB incidence within a few years if this could be kept up. And I would just like to stress that the cost-effectiveness of intervening against an infectious disease far exceeds what you would predict by counting cases.”
During the discussion, Dr Ken Castro from the US Centers for Disease Control noted that similar interventions had been done in the US, but proved to no longer be cost-effective as the TB incidence went down.
“I think we can use this approach to fast-track TB control for a short period, I wouldn’t see this as being indefinite,” Dr Corbett responded. “But I think this is a way that you can make quick, cheap, high-impact gains and then move on to something more suitable for the long term.”
Corbett L. Impact of periodic case finding for symptomatic smear-positive disease on community control of prevalent infectious tuberculosis. A community randomized trial of two delivery strategies in Harare, Zimbabwe. 40th Union World Conference on Lung Health, Cancún, Mexico, 2009.