Offering TB screening as part of a home-based HIV testing
intervention has the potential to identify numerous TB cases that
would otherwise have gone undiagnosed, a
report from a large community-based study in Zambia shows. The findings, from
the PopART study, were reported by Comfort Phiri of Zambart at the Conference
on Retroviruses and Opportunistic Infections (CROI 2016) in Boston on Thursday.
The high prevalence of tuberculosis among people living with
HIV makes it desirable to improve TB screening alongside HIV testing. However,
unlike HIV testing, which can be carried out with rapid antibody tests in the
community or even in the home, screening for TB requires several stages, since
a point-of-care test is not yet available.
The World Health Organization recommends that TB screening
should start with a symptom screen, which has a high likelihood of picking up
people with symptomatic TB. But, because of the non-specific nature of the
symptoms, laboratory confirmation is needed in order to exclude people who are
not infected with tuberculosis before treatment can be provided.
TB screening is hampered by the need to send samples away
for testing, and by the challenge of getting people to come back to the health
care facility to receive their results and then start treatment. Studies from
numerous countries have recorded significant losses at this stage, which result
in a high burden of untreated TB and ongoing transmission in the community.
The PopART study is a community-randomised study testing
three different approaches to the provision of testing and treatment and their
impact on HIV incidence at population level in Zambia and South Africa. The
study includes a community HIV testing and prevention package delivered by Community HIV care
Provider (CHiPs). Preliminary results from the study were presented in a
separate session at the conference this week (see
The PopART study carried out door-to-door offers of HIV
testing, and also conducted a symptom screen for TB in eight communities in
Zambia. People in the intervention communities who reported either weight loss of more
than 1.5kg in the past month, cough for ≥ 2 weeks, and/or night sweats, or who
lived in a household where another member was on TB treatment were invited to
give two sputum samples for TB testing by smear microscopy or Xpert MTB/RIF. The
sample was then transported to the health facility by the CHiPs who carried out
the symptom screen, and the individual was asked to go to the clinic for
further evaluation. Positive TB test results were taken back to the individual’s
home by the CHiPS.
A total of 212,819 residents in 102,511 households consented to take part in the study (an uptake of 84% – approximately two-thirds of those who did
not take part were away from home, indicating the challenge that labour
mobility poses to community-based delivery of treatment and prevention in
southern Africa). Of those who gave consent to take part in the study, 98.4% consented to being screened
for TB. Only 1.2% had TB symptoms (2,583). The most common symptom resulting in referral
was cough (1884 people) and of those subsequently diagnosed with TB, 91% would
have been diagnosed on the basis of the presence of cough alone, without the
need for confirmation by other symptoms. Only 15 people diagnosed with TB did
not have a cough lasting two weeks or more.
Of the 2583 people with TB symptoms at screening, 82% gave a sputum sample. Producing sputum can be difficult for people with symptomatic TB, but the proportion of missing samples at this stage emphasises the importance of both clinic follow-up and of the need for a point-of-care test – or alternative sampling methods – that can be used in community settings to aid active case finding.
Approximately three-quarters are known to have received the
results of a TB test (1918 people), of which 167 (8.7%) were diagnosed with TB,
representing a TB notification rate of 417 cases per 100,000 people. One
hundred and thirty three started TB treatment (79.6%). The remainder are still
being followed up in order to start treatment.
People living with HIV were disproportionately represented among
those with symptoms suggestive of TB. They comprised 12.5% of those screened
but 41% of those with symptoms and 55% of those diagnosed with TB, indicating
the enormous benefit of integrating a TB active case finding element into the
The study investigators are currently unable to explain why the
proportion of people with TB symptoms in the total study population was so low.
One possibility is that lay health workers may not have sufficient knowledge of
TB symptoms to probe further, but Comfort Phiri said that lay health workers received
training in how to ask the questions. Another possible reason for the low TB
symptom rate is that 9.5% of all household members were away from home at the
time of the screening visit, and just over three-quarters of these missing
household members were men, who might be expected to have a lower likelihood of
seeking medical attention for any symptoms.
Regardless of these questions, the study findings show that large-scale screening using symptom checks and community health workers can be integrated into a large household HIV testing programme, underlining the synergy between global efforts to expand the coverage of HIV testing and treatment and efforts to improve TB diagnosis and treatment.