Abstract: Objective  To explore the operational feasibility of using medication monitors to monitor treatment adherence in TB patients by comparing the medication monitors’ records with those of patients’ urine tests for traces of TB drugs.   Methods  Four hundred and thirty-two newly registered initial treatment TB patients were enrolled in succession in 9 counties of China MOH-Gates Foundation TB Control Project case management pilot sites in Heilongjiang, Hunan, Chongqing and Jiangsu provinces. The male-to-female ratio of the enrolled patients is 3.15:1, and average age is (45.98±17.36) years old. Among all the enrolled patients, 226 cases were in the intensive phase of treatment and 206 cases were in the continuous phase of treatment. All the patients were treated using the regimen 2H₃R₃Z₃E₃/4H₃R₃. An electronic medication monitor stored the patients’ drugs and kept a record of every time patients took their medicine out of the monitor. Investigators determined the urine test date for patients by using a table of random numbers at the patients’ 2nd or 4th month of treatment. Investigators then visited the patients’ house to collect patients’ urine and conducted a simple urine test to detect rifampicin to confirm whether patients had taken their medicine within the past 24 hours. Among the 319 patients tested, 116 cases were in the intensive phase of treatment and 153 cases were in the continuous phase, with an overall test proportion of 73.84% (319/432). Among the 113 patients not tested, 64.60% (73/113) were not at home on the test day, 20.35% (23/113) had changed their treatment regimen or quit due to adverse effects of drugs, and 8.85% (10/113) had taken their drugs less than 6 hours before the doctors’ visit and did not meet the criteria for the urine test, 4.42% (5/113) had taken their drugs more than 24 hours before the doctors’ visit and the doctors misclassified them as unqualified cases and did not conduct urine test, and 1.77%（2/113） refused to the testing. The operational feasibility of medication monitors in monitoring adherence to TB treatment was evaluated in terms of sensitivity and specificity by comparing 24h intake records from the monitor against rifampicin incidence in urine samples.  Results  Compared to the detection of rifampicin in urine, the sensitivity and specificity of 24h medication monitor records as the evidence of TB drug adherence were respectively 99.52% (208/209) (95% CI：97.30%-100.00%) and 95.45% (105/110) (95% CI：91.56%-99.35%). Meanwhile the sensitivity (Male: 99.38%(160/161)，95% CI：98.16%-100.00%；Female:100.00%(48/48)，95% CI：87.90%-100.00%）and specificity（Male：95.24%(80/84)，95% CI：90.68%-99.79%；Female: 96.15%(25/26)，95% CI:88.76%-100.00%）between gender groups and the sensitivity（Intensive phase:99.06%(105/106)，95% CI:97.21%-100.00%；Continuous phase：100.00%(103/103)，95% CI:93.00%-100.00%）and specificity（Intensive phase:93.33%(56/60)，95% CI：87.02%-99.65%；Continuous phase:98.00%(49/50)，95% CI：97.12%-100.00%）between treatment phases show no significant statistical difference (χ²=0.8，0.0，0.8，0.0，1.5；P>0.05).   Conclusion  There is a high correlation between medication monitors’ records and the TB treatment adherence by the patients; hence the records of medication monitors can be used as evidence of TB patient’s treatment adherence.

Key words: Tuberculosis, pulmonary/drug therapy, Medication adherence, Directly observed therapy, Feasibility studies