Comparing cross-sectional and longitudinal approaches to tuberculosis patient cost surveys using Nepalese data

Abstract

The World Health Organization has supported the development of national tuberculosis (TB) patient cost surveys to quantify the socio-economic impact of TB in high-burden countries. However, methodological differences in the study design (e.g. cross-sectional vs longitudinal) can generate different estimates making the design and impact evaluation of socio-economic protection strategies difficult. The objective of the study was to compare the socio-economic impacts of TB estimated by applying cross-sectional or longitudinal data collections in Nepal. We analysed the data from a longitudinal costing survey (patients interviewed at three time points) conducted between April 2018 and October 2019. We calculated both mean and median costs from patients interviewed during the intensive (cross-sectional 1) and continuation (cross-sectional 2) phases of treatment. We then compared costs, the prevalence of catastrophic costs and the socio-economic impact of TB generated by each approach. There were significant differences in the costs and social impacts calculated by each approach. The median total cost (intensive plus continuation phases) was significantly higher for the longitudinal compared with cross-sectional 2 (US$119.42 vs 91.63, P < 0.001). The prevalence of food insecurity, social exclusion and patients feeling poorer or much poorer were all significantly higher by applying a longitudinal approach. In conclusion, the longitudinal design captured important aspects of costs and socio-economic impacts, which were missed by applying a cross-sectional approach. If a cross-sectional approach is applied due to resource constraints, our data suggest that the start of the continuation phase is the optimal timing for a single interview. Further research to optimize methodologies to report patient-incurred expenditure during TB diagnosis and treatment is needed.

Keywords: Tuberculosis costing; longitudinal vs cross-sectional survey; tuberculosis and Nepal.