TB diagnosis usually starts with asking someone whether they have any of the classic symptoms of TB — coughing, fever, weight loss and drenching night sweats. Only people who have one or more of these symptoms are usually offered confirmatory molecular tests.

Evidence is mounting that, as with Covid-19, people can have active TB disease and be infectious without having any symptoms. More than half of the people diagnosed with TB in South Africa’s first national TB prevalence survey would have been missed had asymptomatic people not been offered additional tests as part of the survey. 

One response to the shortcomings of symptom screening is to offer people considered to be at high risk of TB a molecular test irrespective of whether they have any symptoms.

Last year, Spotlight reported on a cluster randomised study that found such a targeted universal testing (TUT) approach increased TB detection by 17%. Since then, TUT has been embraced by the Department of Health and is being rolled out in some districts. Early signs are that it is working well.

Another approach is to use digital X-ray technology to screen for TB. (It is classified as screening since an X-ray suggestive of TB has to be followed by a molecular test to confirm the diagnosis.) Here, too, there is compelling evidence. Of the 234 people found to have TB in the prevalence survey, about 58% had abnormal X-rays without any TB symptoms.

The World Health Organization (WHO) endorses the use of digital X-ray screening for TB. There is also compelling evidence that computers are now as good as humans at interpreting X-ray images.

Boosting TB detection in the real world

Spotlight reported last year that digital X-ray pilot projects were to be conducted in six of South Africa’s 52 districts. The pilots are coordinated by the Department of Health and supported by the Global Fund to Fight AIDS, Tuberculosis and Malaria and the United States Agency for International Development (USAID). These donors have contracted several non-profit organisations to run the pilots in the six districts. Then, in April, Spotlight reported that early signs from the pilots were promising. An independent assessment of the pilots has now been completed and was presented at the recent South African TB Conference held in Durban. A written report of the assessment hasn’t yet been made public.

Dr Lucy Connell, TB programme head at the NGO Right to Care, conducted the independent assessment of the Global Fund-supported parts of the pilot. She said that in each of the six districts with clinic-based digital chest X-ray (DCXR) containers, case finding had dramatically increased since 2019 – three-fold in some instances.

“Early results show that augmenting TB symptom screening with DCXR screening helps to detect more people with TB,” she said.

In the first phase of the project — 1 September 2020 to 30 June 2022 — 114,126 people visited the DCXR containers for TB screening (symptom screening followed by X-ray screening); 96% of these people were screened with DCXR.  

“In the first phase of this project, 15 semi-movable containers equipped with DCXR machines were placed at 24 high-volume clinics in six districts across South Africa,” she said. The districts were: Cape Town with three containers; Ehlanzeni with two; Ekurhuleni with two; eThekwini with two; Johannesburg with three; and OR Tambo, also with three.

“These districts were selected because all of them had very large gaps between the number of people estimated to be living with TB and the number that were being diagnosed and treated,” Connell said.

In clinics in the OR Tambo district that hosted DCXR units, the number of people testing positive for TB went from two out of every 1,000 clinic visitors in 2019 (before the use of DCXR TB screening) to six out of every 1,000 clinic visitors in the period when DCXR-augmented TB symptom screening took place. This occurred even though the capacity of the DCXR units is limited to about 80 people a day, far short of the number of clinic attendees.

Similarly, in the Ehlanzeni district, one out of every 1,000 clinic visitors was diagnosed in 2019. It was more than three times that when the clinics were provided DCXR TB screening. These are, however, correlations and other factors besides X-ray screening may also have contributed to the increase.

At least 28% would have been missed

Connell was upbeat about the role of DCXR and its diagnostic accuracy. 

“[It] is now the WHO-preferred tool for TB screening among groups at high risk for TB and in countries with the resources to do this,” she said.

“Furthermore, the new guidelines recommend the use of computer-aided detection (CAD) for the interpretation of X-rays. In addition to detecting more TB among symptomatic patients, DCXR TB screening has also detected TB in asymptomatic people: 28% of all people diagnosed with TB in this phase of the project were asymptomatic and would have been missed if the clinic was relying on symptom screening alone.”

She said the use of DCXR TB screening alone would find more people with TB than TB symptom screening alone, and the use of DCXR alone finds 84% of all cases diagnosed with both screening methods combined.

“The return on investment on the DCXR lies in not only reducing the prevalence of TB but significantly reducing the costs that families incur if a member of the family has TB,” she said. “It impoverishes families.”

Visit Daily Maverick’s home page for more news, analysis and investigations

Where X-rays fit in

Professor Salome Charalambous, group chief scientific officer at the Aurum Institute, questioned the benefits X-ray screening would offer in places where TUT was already provided, such as in some correctional facilities.

“We are not sure in this scenario whether there will be an advantage to adding DCXR since all high-risk individuals will already be tested. The biggest issue is the capital outlay that is quite substantial to be able to deliver DCXR in prisons,” she said.

“There would need to be a purchase of X-ray equipment for each facility or a mobile unit that moves from facility to facility and there would need to be about nine to 10 such units. As far as I know, they have not been routinely implemented. As the technology becomes easier to use and cheaper, it may become more feasible for them to be used routinely in prisons,” she says.

Although Charalambous described the technology as useful, she questioned how the technology could be used most effectively. 

“How to best use the few machines that we have in-country belonging to the Department of Health is the problem. We also need to work with the regulators to reduce the restrictions that are currently in place with regard to the use of X-ray to make it easier and more feasible to implement. X-rays offer a great opportunity to increase diagnosis but we still don’t know how best to utilise them,” she said.

She added that they were working with the government to see if they could quantify how much they could increase diagnosis in clinics to see if the costs would be justified.

“We are also looking at a strategy to see how digital X-rays can be used to find TB in communities as we know that TB affects men disproportionately and men do not generally visit health facilities so we need to find strategies that will identify men with TB,” said Charalambous. Most X-ray screening so far in South Africa has been clinic based.

Low sputum collection rate

One weakness highlighted in the independent assessment was a very low sputum collection rate. According to Connell, only a third of all people eligible for a laboratory TB test had a sputum sample submitted. Sputum samples are needed to run the standard molecular TB test needed to confirm a diagnosis after someone has an X-ray suggestive of TB. 

“Of all people eligible for TB testing, the sputum collection rate was highest (67%) for those who screened positive on both symptom screen and DCXR screen, lower for those who had no TB symptoms but had chest X-ray changes suggestive of TB, and lowest for those who had a positive symptom screen but a normal chest X-ray,” Connell said.

The low sputum collection rate resulted in missing people who had TB, Connell said, and it undermined the evaluation of the true impact of DCXR screening.

“This finding clearly shows that a new shiny technology that has promise is only as good as the performance at every other step of the diagnostic cascade. More effort needs to be directed at ensuring every person with presumed TB is assisted to produce a sputum sample for testing, or the investment in such technology will not bear fruit.

“This can be done with dedicated sputum collection staff who have access to multiple nebulisers that they can use to help patients secrete and cough up mucus from the lungs,” said Connell. She added that there was also a need to continue research to find alternative specimens for TB testing that were easier to collect, such as tongue swabs.

Phase two

The second phase of the project to pilot DCXR for TB screening is under way. In addition to the use of the DCXR containers in clinics, a fleet of seven mobile vans is now travelling daily to different community locations, said Connell.

“A further 15 DCXR vans will be active by the end of 2022. This phase is supported by GIS mapping, which is being used to identify ‘TB hotspots’ — areas with high numbers of people being diagnosed with TB and areas with many structural drivers for TB transmission.

“Mobile vans are placed in high-traffic areas in these hotspots where all people are invited to come for screening through community mobilisation and information dissemination,” she said.

Connell said the capacity to use DCXR for community TB screening was critical because the SA National TB Prevalence Survey tells us that 60% of all people with classic TB symptoms had not attended a clinic to seek care for these symptoms. The government had already incorporated new screening and laboratory testing approaches into the TB Recovery Plan, and into the recently released TB screening and testing Standard Operating Procedure (SOP).

“For example, a major change is that this SOP recommends that certain groups of people with a higher risk for TB should be tested routinely for TB at regular intervals, irrespective of TB symptoms.

“People who have been treated and completed TB treatment in the past two years, irrespective of TB symptoms, must be tested annually for two years, and each time they report TB symptoms on symptom screening,” said Connell.

“People living with HIV must be tested annually, and each time they report TB symptoms. People who have been in close contact with a person diagnosed with TB or TB treatment in the past year irrespective of TB symptoms, and at each new contact thereafter.” DM/MC

This article was produced by Spotlight — health journalism in the public interest.