Attrition and delays before treatment initiation among patients with MDR-TB in China (2006-13): Magnitude and risk factors.

Background: China's national tuberculosis programme does not have cohort wise information regarding attrition and delays in the multidrug resistant tuberculosis (MDR-TB) diagnosis and treatment pathway.

Objective: Under the Global Fund programmatic management of drug-resistant TB (2006-13), we assessed the attrition and delay in the pathway and the factors associated.

Methods: Cohort study involving secondary programme data. All patients identified as presumptive MDR-TB (defined as i) previously treated TB patients which included recurrent TB, return after loss to follow up, treatment after failure and ii) new TB patients that were non-converters at three months of treatment or in close contact with a known MDR-TB patient) during October 2006 to June 2013 were eligible for phenotypic drug susceptibility testing (DST). Pre-diagnosis attrition (presumptive MDR-TB not undergoing culture and DST) and pre-treatment attrition (confirmed MDR-TB patients not initiated on treatment) was calculated. Diagnosis delay was the time interval from DST eligibility to DST result, treatment initiation delay was fom DST result to treatment initiation and total delay was from DST eligbility to treatment initiation. Factors associated with attrition and delay were identified using log binomial regression and linear regression, respectively.

Results: Of 78 564 presumptive MDR-TB patients, 2 470 (3.1%) underwent pre-diagnosis attrition. Of 9 283 MDR-TB patients, 3 361 (36.2%) underwent pre-treatment attrition. Median(IQR) diagnosis delay was 84 (64, 114) days; treatment initation delay was 23(6,68) days and total delay was 117(77,187) days. Long diagnosis delay was an independent predictor of pre-treatment attrition in a dose response relationship. While pre-treatment attrition was less likely among presumptive criterion 'previously treated' and with increasing time period, it was more likey among elderly and in east and west region. While the diagnosis delay increased with time period, treatment initiation delay and total delay reduced with time period. Short diagnosis delay was associated with west region, smear negative patients and presumptive criterion 'treatment after lost to follow up'. Short treatment initiation delay was associatied with east and west regions while long treatment initiation delay was associated with elderly and presumptive criterion 'recurrent TB'. Total delay predictors were similar to treatment initiation delay. In addition, short total delay was associated with presumptive criterion 'treatment after failure'.

Conclusion: The diagnosis and treatment delay were long and the pre-treatment attrition was considerable high. Long diagnosis delay is likely to predict pre-treatment attrition.