High mortality among patients with tuberculosis accessing primary care facilities: secondary analysis from an open-label cluster-randomised trial.

Background: Tuberculosis (TB) mortality remains persistently high, despite global TB control efforts. The aim of this study was to assess if a quality improvement (QI) intervention reduced deaths in TB patients accessing primary healthcare (PHC) services.

Methods: In this pre specified secondary analysis of a cluster-randomized controlled study conducted in 2016-2018 in South Africa (Clinicaltrials.gov, NCT02654613), we compared 18-month case-fatality rates among newly diagnosed TB patients irrespective of HIV status randomized to clinics receiving the QI intervention and standard of care (SOC) [(eight clusters and 20 clinics per arm)]. Statistical inferences used a t-test from a two-stage approach recommended for cluster-randomized trials with fewer than 15 clusters per arm.

Findings: Among the 5817 newly diagnosed TB patients enrolled (intervention = 3473; control = 2344), 562 died by 18-months [case-fatality rate (CFR) = 9·7%]. Ninety percent of the deaths (506/562) occurred within six months of TB treatment initiation. Quality improvement intervention arm clinics compared to control arm clinics did not demonstrate a significant difference in TB CFR. Case-fatality rates were 9·5% [95% Confidence Interval (CI): 6·9-12·9] and 11·3% (95% CI: 8·7-14·7) [adjusted rate ratio (aRR), 0·9 (95% CI: 0·6-1·2)] in the intervention and control arms, respectively. In people living with HIV/AIDS (PLWHA) CFR in the intervention and control arms: were 10·8% (95% CI: 7·8-14·7) and 14·4% (95% CI: 9·3-22·4) in those on antiretroviral therapy (ART) and 18·6 (95% CI: 9·1-38·0) and 33·0 (95% CI: 16·2-67·3), in those with no ART data respectively. In the intervention and control arms CFR in HIV-TB coinfected patients was 6·5 (95% CI: 3·6-11·6) and 11·5 (95% CI: 6·5-20·0) in those on ART with viral loads <200 copies/ml and 22·4 (95% CI: 16·7-30·2) and 19·7 (95% CI: 11·3-34·5) in those with no viral load data as they commenced ART within 12 months before initiating TB treatment, respectively.

Interpretation: The quality improvement intervention did not significantly reduce mortality. We observed that TB CFR was higher among PLWHA not on ART and HIV-TB coinfected patients.

Funding: Research reported in this publication was supported by South African Medical Research Council (SAMRC), and UK Government's Newton Fund through United Kingdom Medical Research Council (UKMRC).