Trends in Mycobacterium tuberculosis detection and rifampicin resistance using Xpert MTB/RIF assay at Woldia comprehensive specialized hospital, Northeastern Ethiopia: a 5-year retrospective study

Tuberculosis (TB) remains a major public health challenge, and drug-resistant TB complicates disease control. The Xpert ® MTB/RIF assay enables rapid detection of Mycobacterium tuberculosis (MTB) and rifampicin resistance (RR). To assess five-year temporal patterns in MTB detection and Xpert-detected RR among presumptive-TB testing records at Woldia Comprehensive Specialized Hospital (WCSH), Northeastern Ethiopia, from 2020 to 2024. A retrospective laboratory record review was conducted using routine Xpert MTB/RIF records from January 2020 to December 2024. Descriptive statistics and logistic regression assessed MTB detection and associated factors. Annual proportions and linear trends were evaluated using chi-square tests and continuous-year logistic regression. A case-mix-adjusted sensitivity analysis assessed temporal changes in MTB detection, and Firth-penalized logistic regression evaluated the robustness of associations with Xpert-detected RR. MTB was detected in 504/4,351 records (11.6%), with no significant annual difference ( p = 0.460) or linear trend ( p = 0.832). Among 498 MTB-detected records with determinate RR results, 28 (5.6%) had RR, with no significant annual difference ( p = 0.905) or linear trend ( p = 0.504). Continuous-year models showed no significant temporal association with MTB detection (OR = 1.007, 95% CI: 0.946–1.071) or RR (OR = 1.089, 95% CI: 0.848–1.399). After adjustment for age, sex, TB treatment history, specimen type, and referring facility, calendar year remained unassociated with MTB detection (AOR = 1.025, 95% CI: 0.957–1.097; p = 0.481). Previous TB treatment was associated with MTB detection (AOR = 13.99, 95% CI: 7.85–24.95) and RR (AOR = 49.72, 95% CI: 18.60–132.90), while male sex was associated with RR (AOR = 4.59, 95% CI: 1.32–16.00). HIV positivity was associated with MTB detection in univariable analysis (COR = 7.03, 95% CI: 1.99–24.77). No statistically significant temporal trend was detected in MTB detection or Xpert-detected RR. Previous TB treatment was associated with both outcomes, while male sex was associated with RR. These findings describe facility-level testing patterns rather than population-level disease burden or unique-patient risks. Sparse RR events and extensive missing HIV data warrant cautious interpretation. Not applicable.