Predictors of poor treatment outcomes in multidrug-resistant tuberculosis patients: a retrospective cohort study.

OBJECTIVES: We aimed to determine the characteristics, treatment outcomes and risk factors for poor treatment outcomes among multidrug-resistant (MDR) tuberculosis (TB) patients in Khyber Pakhtunkhwa province, Pakistan.

METHODS: A retrospective cohort study including all patients with MDR-TB who sought care at the MDR-TB unit in Peshawar was conducted between January 2012 and April 2014. Patients were followed until an outcome of TB treatment was recorded as successful (cured or completed) or unsuccessful. Binary logistic regression was used to identify predictors of poor outcome, i.e. unsuccessful treatment outcomes.

RESULTS: Overall, 535 patients were included. The proportion of female subjects was relatively higher (n = 300, 56.1%) than male subjects. The mean (standard deviation) age of patients was 30.37 (14.09) years. Of 535 patients for whom treatment outcomes were available, 402 (75.1%) were cured, 4 (0.7%) completed therapy, 34 (6.4%) had disease that failed to respond to therapy, 93 (17.4%) died and two (0.4%) defaulted; in total, 129 (24.1%) had an unsuccessful outcome. We found three significant predictors of unsuccessful treatment during multivariate logistic regression: being married (odds ratio (OR) = 2.17, 95% confidence interval (CI) 1.01, 4.66), resistance to second-line drugs (OR = 2.61, 95% CI 1.61, 4.21) and presence of extensively drug-resistant TB (OR = 7.82, 95% CI 2.90, 21.07).

CONCLUSIONS: Approximately 75% of the treatment success rate set by the Global Plan to Stop TB was achieved. Resistance to second-line drugs and presence of extensively drug-resistant TB are the main risk factors for poor treatment outcomes.