Effects of GeneXpert MTB/RIF Testing and GxAlert eHealth Platforms on MDR-TB Diagnosis and Linkage to Care in Sofala Province, Mozambique

Thesis (Master's)--University of Washington, 2016-06

Introduction: Mozambique ranks as one of the 22 highest tuberculosis (TB) burden countries in the world but has one of the lowest TB case detection rates of only 39%. To improve Multi-Drug Resistant TB (MDR-TB) case detection rates and linkage to care, we installed three GeneXpert MTB/RIF machines in Sofala Province starting in 2012 and installed an eHealth platform called GxAlert in 2014. Methods: We conducted a retrospective clinical review using routine clinical data from GeneXpert MTB/RIF assays (Xpert), and the TB treatment and laboratory registries from 16 health facilities from March 2012-September 2015. Using multi-level logistic regression analyses, we conducted pre-post and time-series analyses of GxAlert to see if an innovative SMS notification system to key Ministry of Health (MoH) personnel of rifampin resistant (RR) patients would improve linkage to care. Results: A total of 22,543 Xpert tests were recorded and 17,024 (75.5%) did not detect TB, 2,841 (12.6%) detected TB without rifampin resistance (RR), 2,256 (10%) were invalid, had no result, or had errors, 249 (8.3%) patients were diagnosed with RR-TB, and 173 (0.76%) detected TB and were RR indeterminate. 139 (55.8%) RR-TB study subjects were started on MDR-TB treatment. Across the 46-month period there was an increase in 0.66 (95% Confidence Interval [CI]: 0.32, 0.99; p<0.001) MDR patients diagnosed every 6 months, and 0.21 (CI: 0.07, 0.35; p=.003) patients started on MDR-TB treatment. Using a simple pre-post analysis and a complex interrupted time-series analysis, there was no statistically significant change in the proportion of MDR-TB patients linked to care pre- versus post- GxAlert implementation (95% CI Confidence Interval [CI]: 0.75, 2.0; p-value 0.41), (Odds Ratio [OR]: 0.83; CI: 0.30, 2.3); p=0.72), respectfully. Conclusion: The introduction of three Xpert machines in Sofala Province led to an increase in the total number of RR-TB patients being diagnosed and started on treatment over a 4 year period from 2012-2015. Although the GxAlert platform did not show significant differences in linkage to care, it led to other positive benefits including rapidly assessing Xpert data using any computer with internet access, monitoring the performance of Xpert machines and cartridge capacity in real time, creation of summary reports, and formulating lists of patients with a positive test for MTB.