Differential loss to follow-up rates among adult tuberculosis patients – Findings from the largest private sector tuberculosis clinic database in Myanmar

BACKGROUND: Population Services International (PSI) Myanmar’s social franchise network of general practitioners, known as Sun Quality Health Clinics (SQHC), provided tuberculosis (TB) diagnosis and treatment with Direct Observed Treatment Short course (DOTs) across Myanmar since 2004, with a total of 898 clinics across the country. People who sought TB treatment from these SQHC by themselves were regarded as walk-in patients. To augment TB case notification, PSI Myanmar developed two treatment seeking channels: Community Health Services Providers (CHSP) and Interpersonal Communicators (IPC). They actively sought people who were suspected to have TB and referred them to SQH clinics. In this study, we compared the loss to follow-up rates of TB patients across three treatment seeking channels; and investigated risk-factors for loss to follow-up. METHODS AND FINDINGS: A retrospective cohort design was applied using TB client records between 2012 and 2016. Outcome was defined as loss to follow-up in comparison to successful TB treatment (completed or cured). Multivariate Poisson regression was conducted to estimate incidence rate ratio of loss to follow-up. Of the 62,664 TB patients registered at the SQHC, around 10% each were actively screened by the CHSP and the IPC, and 78.9% were walk-in patients. Overall cumulative incidence for loss to follow-up rate was significantly higher in the IPC channel (14.2%, 95% CI 13.4–15.1%) than walk-in patients (8.9%, 95% CI 8.6–9.1%) and the CHSP channel (5.5%, 95% CI 5.0–6.1%) (p<0.001). The median time after which patients were lost to follow-up from treatment was 4.04 months. We found that patients with older age, male sex, patients residing in hilly region, unknown smear status, retreated cases, HIV co-infection, and unknown HIV status were risk-factors for loss to follow-up in the continuation phase of treatment; whereas patients with higher initial body weight, patients who received travel support and patients taking treatment in older providers were less likely to be lost to follow-up. CONCLUSIONS: Based on these findings, we recommend that implementation strategies for improving case notification and treatment seeking should carefully consider retention strategies in parallel, and the identified influencing factors for loss to follow-up should be taken account for such consideration.