The intervention consisted of training non-paid informal healthcare providers (such as store-keepers) in TB and HIV disease recognition, sputum specimen collection, referral to the public health system, and raising community awareness. Front line public health personnel and community leaders were sensitised to support the intervention.

A phased, matched, parallel cluster design was used to randomise three clusters (average population size per cluster = 200,714) to the Early intervention arm (received the intervention early in the first 12 months) and an equal number to the Delayed intervention arm (average population size per cluster = 209,564) which received the intervention after one year. Data for impact evaluation were obtained from routine patient registers in all the health facilities and patients were blindly allocated to the respective clusters based on residential address. Treatment initiation rates (expressed as incidence rate ratios) for TB and Anti Retroviral Therapy (ART) over the 12 months period were the primary outcome measures for each of the studied conditions. Poisson regression models with robust standard errors were used to assess the effectiveness of the intervention.

Inclusion Criteria:

• People accessing informal providers with possible TB or HIV

Exclusion Criteria:

• children
• people already with a diagnosis of TB or HIV