The effect of engaging unpaid informal providers on case detection and treatment initiation rates for TB and HIV in rural Malawi (Triage Plus): A cluster randomised health system intervention trial

Abstract

Background: The poor face barriers in accessing services for tuberculosis (TB) and Human Immuno-deficiency Virus (HIV) disease. A cluster randomised trial was conducted to investigate the effectiveness of engaging unpaid informal providers (IPs) to promote access in a rural district. The intervention consisted of training unpaid IPs in TB and HIV disease recognition, sputum specimen collection, appropriate referrals, and raising community awareness.

Methods: In total, six clusters were defined in the study areas. Through a pair-matched cluster randomization process, three clusters (average cluster population = 200,714) were allocated to receive the intervention in the Early arm. Eleven months later the intervention was rolled out to the remaining three clusters (average cluster population = 209,564)-the Delayed arm. Treatment initiation rates for TB and Anti-Retroviral Therapy (ART) were the primary outcome measures. Secondary outcome measures included testing rates for TB and HIV. We report the results of the comparisons between the Early and Delayed arms over the 23 month trial period. Data were obtained from patient registers. Poisson regression models with robust standard errors were used to express the effectiveness of the intervention as incidence rate ratios (IRR).

Results: The Early and Delayed clusters were well matched in terms of baseline monthly mean counts and incidence rate ratios for TB and ART treatment initiation. However there were fewer testing and treatment initiation facilities in the Early clusters (TB treatment n = 2, TB testing n = 7, ART initiation n = 3, HIV testing n = 20) than in the Delayed clusters (TB treatment n = 4, TB testing n = 9, ART initiation n = 6, HIV testing n = 18). Overall there were more HIV testing and treatment centres than TB testing and treatment centres. The IRR was 1.18 (95% CI: 0.903-1.533; p = 0.112) for TB treatment initiation and 1.347 (CI:1.00-1.694; p = 0.049) for ART initiation in the first 12 months and the IRR were 0.552 (95% CI:0.397-0.767; p<0.001) and 0.924 (95% CI: 0.369-2.309, p = 0.863) for TB and ART treatment initiations respectively for the last 11 months. The IRR were 1.152 (95% CI:1.009-1.359, p = 0.003) and 1.61 (95% CI:1.385-1.869, p<0.001) for TB and HIV testing uptake respectively in the first 12 months. The IRR was 0.659 (95% CI:0.441-0.983; p = 0.023) for TB testing uptake for the last 11 months.

Conclusions: We conclude that engagement of unpaid IPs increased TB and HIV testing rates and also increased ART initiation. However, for these providers to be effective in promoting TB treatment initiation, numbers of sites offering TB testing and treatment initiation in rural areas should be increased.

Trial registration: ClinicalTrials.gov NCT02127983.

Conflict of interest statement

Competing Interests: SBS, BF and RT are employed by the Liverpool School of Tropical Medicine (LSTM) which is a registered charitable, UK-based research and teaching institution. Its mission is to save lives in resource poor countries through research, education and capacity strengthening. LSTM, in collaboration with REACH Trust, Malawi, applied to LHL for funds to carry out the work described in this manuscript. RM works for LHL International which is a Norwegian Non-Governmental Organisation (NGO) that receives funding from NORAD and the ATLAS Alliance (an umbrella organisation consisting of Norwegian NGO's working with people with disabilities and the fight against TB). LHL International is a member of the ATLAS Alliance. Funding from the ATLAS Alliance comes from individual Norwegian citizens who have given non-earmarked donations. LHL International, using funding as described above, is the funder of the Triage-plus project and therefore also of this article. While RM had no decision on whether or not the article was to be written and published, he had both technical input and financial management responsibility for funding of the project, in addition to being involved in the actual writing and work linked to being an author of this article. GB, LS, IM and HB are employed by the Research for Equity and Community Health Trust (REACH Trust) which is a registered charitable, local NGO based in Malawi. It's mission is to address the inequity in access to health services through research. REACH Trust in collaboration with LSTM, applied to LHL for funds to carry out the work described in this manuscript. This does not alter our adherence to PLOS ONE policies on sharing data and materials.

Figures

Fig 1. CONSORT participant flowchart.

Fig 2. Map of Lilongwe District showing the distinction between urban and rural areas and the rural clusters allocated to the Early (pink) and Delayed (brown) arms of the trial.

Fig 3. Triage Plus study design: Phased, pair-matched, parallel cluster design used to randomise study clusters to Early and Delayed arms.

The blue colour represents the baseline period and the dark pink represents when the intervention package was implemented in the respective arms and the light pink represents study period when no intervention was implemented in the Delayed arm.

Fig 4. Monthly and cumulative TB testing rates per 10,000 adults: Results in individuals aged 12 years and above over the 23 months of the study are shown.

The line graphs represent the cumulative TB testing rates. The green triangle line graph indicates the Early Intervention arm and the purple cubed line graph indicates the Delayed Intervention arm. The solid bar graphs represent the monthly TB testing rates per 1,000 adults for each month over the 23 months of the study. The blue bars represent the Early Intervention Arm while the grey bars represent the Delayed Intervention Arm.

Fig 5. Monthly and cumulative TB treatment initiations rates per 10,000 adults: Results in individuals aged 12 years and above over the 23 months of the study are shown.

The line graphs represent the cumulative TB treatment initiation rates. The green triangle line graph indicates the Early Intervention arm and the purple cubed line graph indicates the Delayed Intervention arm. The solid bar graphs represent the monthly TB treatment initiation rates per 1,000 adults for each month over the 23 months of the study. The blue bars represent the Early Intervention Arm while the grey bars represent the Delayed Intervention Arm.

Fig 6. Monthly and cumulative HIV testing rates per 1,000 adults: Results in individuals aged 12 years and above over the first 11 months of the study before the scale up of the Delayed intervention arm are shown.

The line graphs represent the cumulative HIV testing rates. The green triangle line graph indicates the Early Intervention arm and the purple cubed line graph indicates the Delayed Intervention arm. The solid bar graphs represent the monthly HIV testing rates per 1,000 adults for every month over the 11 months of the study. The blue bars represent the Early Intervention Arm while the grey bars represent the Delayed Intervention Arm.

Fig 7. Monthly and cumulative ART treatment initiations rates per 10,000 adults: Results in individuals aged 12 years and above over the 23 months of the study before the scale up of the Delayed intervention arm are shown.

The line graphs represent the cumulative ART initiation rates. The green triangle line graph indicates the Early Intervention arm and the purple cubed line graph indicates the Delayed Intervention arm. The solid bar graphs represent the monthly ART initiation rates per 1,000 adults for every month over the 23 months of the study. The blue bars represent the Early Intervention Arm while the grey bars represent the Delayed Intervention Arm.