The objective of this prospective study of the risks of treatment failure in patients with drug-susceptible pulmonary tuberculosis (PTB) was to provide reference data to help develop a disease control strategy. Participants were recruited in eight provinces of China from October 2008 to December 2010. A total of 1447 patients with drug-susceptible PTB and older than 15 years of age were enrolled. Demographic characteristics, bacteriological test results, and patient outcome, i.e., cure or treatment failure were recorded and compared using the chi-square or Fisher’s exact tests. Multivariate logistic regression was used to identify factors associated with risk of treatment failure. Of the 1447 patients who were enrolled, 1349 patients (93.2%) were successfully treated and 98 (6.8%) failed treatment. Failure was significantly associated with age 65 years [odds ratio (OR)=2.522, 95% confidence interval (CI): (1.097-5.801)], retreatment [OR=2.365, 95% CI: (1.276-4.381)], missed medicine [OR=1.836, 95% CI: (1.020-3.306)], treatment not observed [OR=1.879 95% CI: (1.105-3.195)], and positive culture result after the first [OR=1.971, 95% CI: (1.080-3.597)] and second month [OR=4.659, 95% CI: (2.590-8.382)]. The risk factors associated with treatment failure were age 65 years, retreatment, missed medication, treatment not observed, and positive culture at the end of month 1 or month 2. These risk factors should be monitored during treatment and interventions carried out to reduce or prevent treatment failure and optimize treatment success.

Tuberculosis (TB) continues to be a significant public health problem in the world. According to the World Health Organization, China was third among 22 high-TB burden countries, with an estimated 930, 000 pulmonary tuberculosis (PTB) patients in 2015^[1]. The global TB control target is to end the TB epidemic by 2035, which represents a 95% reduction in TB deaths, and a 90% reduction in incidence rate compared with 2015. The basic strategy to detect and cure all PTB patients and end the TB epidemic is patient centered. Patients with tuberculosis require retreatment if they are not cured in order to avoid being a source of infection or developing drug-resistant TB, which is a serious public health challenge that makes TB control more difficult^[2]. We designed this prospective study to identify the factors associated with risk of treatment failure and to provide evidence to help develop TB control strategies.

Study Setting and Population The study was conducted from October 2008 to December 2010 in eight provinces, Tianjin, Hebei, Henan, Shanghai, Chongqing, Yunnan, Guangxi, and Guangdong, situated in eastern, central, and western China. The study was integrated with the routine procedures of the national TB control program of each province, including sputum smears, chemotherapy, and treatment management. Sputum culture and drug susceptibility testing (DST) were added during the conduct of the study.

All study participants satisfied the following inclusion criteria: 1) ≥15 years of age; 2) provided informed consent; 3) PTB culture-positive at initial treatment and retreatment; 4) free of nontuberculous mycobacterium infection; and 5) infected with a strains susceptible to all four first-line anti-TB drugs (isoniazid, rifampicin, streptomycin, and ethambutol). A total of 1447 participants receiving initial treatment or retreatment of drug-susceptible PTB were enrolled. Treatment was successful in 93.2% (1349/1447) of the participants and failed in 6.8% (98/1447). Sputum smears, sputum cultures, and DST were performed during months 1, 2, and 5 of anti-TB treatment.

Study Definitions The definitions used in this study were those recommended in the guidelines for implementing tuberculosis control programs in China^[3]. Initial treatment is that given to 1) a patient who has never been treated for TB, or 2) a patient who has taken anti-TB drugs for less than 1 month. Retreatment is that given to 1) a patient who has taken anti-TB drugs for 1 month or more, or 2) a patient who failed initial treatment or has relapsed. Cured patients are those who completed prescriptive treatment and had two negative sputum smears during treatment and at least one after completing treatment. Patients who complete treatment are sputum smear-positive at the beginning, smear-negative at the end of the intensive-treatment phase, have not more than one positive smear during the continuing phase of treatment, and remain sputum smear-negative in the last month of treatment. Successful treatment describes cured patients and those who completed treatment; Treatment failure describes smear-positive patient who is still positive at 5 months or smear-negative patients who become smear-positive during treatment.

Treatment Regimens Anti-TB treatment followed the guidelines of the tuberculosis control program in China^[3]. The initial treatment and retreatment regimens were standard short-course chemotherapy protocols. Initial PTB patients received 2 months ofisoniazid (H), rifampin (R), and pyrazinamide (Z) plus 4 months of H and R (2H₃R₃Z₃E₃/4H₃R₃). Retreatment PTB patients received streptomycin (S), H, R, Z, and E during the intensive phase plus 6 months of H, R, and E (2H₃R₃Z₃ S₃E₃/6H₃R₃ E₃).

Ethical Statement This study was approved by the Institutional Ethics Review Committee of Beijing Chest Hospital, Capital Medical University, Beijing Tuberculosis and Thoracic Tumor Research Institute, and the eight provincial TB Control and Prevention Centers (TB hospitals).

Statistical Analysis The participant characteristics were analyzed descriptively. Chi-square or Fisher’s exact tests were used to compare demographic characteristics, bacteriological test results, and management outcomes in the treatment success and treatment failure groups. Variables found to be significant in univariate analysis were subsequently analyzed by multivariate logistic regression with odds ratios (ORs) and their 95% confidence intervals (CIs) to identify the factors that remained statistically significant in the final model. Data analysis was performed using SAS version 9.1 (SAS institute Inc., Cory, NC, USA), and P-values ＜0.05 were considered as statistically significant.

Demographic Characteristics of Study Participants Of 2142 confirmed smear-positive PTB patients evaluated during the study period, 67 were infected with nontuberallous mycobacteria (NTM), 54 were sputum-culture negative, 383 had drug-resistant TB, 36 died, 39 were default cases, 44 transferred out, and 72 others refused treatment. The remaining 1447 participants with drug-susceptible TB included 1349 treatment success cases and 98 treatment failure cases. DST confirmed that all cases were treatable with the four first-line anti-TB drugs; all 1447 were enrolled and received standard anti-TB treatment. Table 1 includes the demographic characteristics of the 1447 study participants.

Table 1

Table 1 Demographic Characteristics of the 1447 Study Participants Demographic Characteristics n % Sex Male 1039 71.80 Female 408 28.20 Age 15-24 289 19.97 25-44 547 37.80 45-64 423 29.23 65- 188 12.99 Profession Farmers and migrant workers 729 50.38 Employees 231 15.96 Students 84 5.81 others 403 27.85 Educational attainment Illiteracy 125 8.64 Primary school 345 23.84 Junior middle school and high school 818 56.53 College degree and above 159 10.99 Census register The local population registered 1044 72.15 Not local population registered 403 27.85 Marriage Unmarried 388 26.81 Married 990 68.42 Others 69 4.77 Treatment type Initial treatment 1273 87.98 Retreatment 174 12.02

Table 1 Demographic Characteristics of the 1447 Study Participants

Risk Factors Associated with Treatment Failure The treatment failure rate was 6.77% (98/1447). The chi-square and Fisher’s exact tests found significant differences (P＜0.5) in sex, age, profession, educational attainment, treatment type, missed medication, directly observed treatment (DOT) for patients, and sputum culture at the end of months 1^st and 2^nd in the treatment success and treatment failure groups (Table 2). Treatment failure was more frequent in men (7.80%) than in women (4.17%). By age, the failure rate was highest in those 65 years of age (11.70%). By profession, the rate was highest in ‘employees’ (9.52%); and by educational attainment, was highest in those who completed primary school (9.57%). Treatment failure occurred more often with retreatment (12.64%) than in initial treatment (5.97%), in participants missing medicine (12.00%), and in those who were culture-positive at the end of month 1^st (13.08%) or month 2^nd (16.34%) compared with culture-negative at the end of month 1 (3.28%) or month 2 (4.37%).

Table 2

Table 2 Univariate Analysis of Risk Factors Associated with Treatment Failure Variables Treatment Success (n=1349) Treatment Failure (n=98) χ2 P n % n % Sex Male 958 71.02 81 82.65 6.112 0.013* Female 391 28.98 17 17.35 Age 15-24 282 20.90 7 7.14 32.884 ＜0.001* 25-44 524 38.84 23 23.47 45-64 377 27.95 46 46.94 65- 166 12.31 22 22.45 Profession Farmers and migrant workers 676 50.11 53 54.08 Fisher exact test 0.020* Employees 209 15.49 22 22.45 Students 84 6.23 0 0.00 Others 380 28.17 23 23.47 Educational attainments Illiteracy 114 8.45 11 11.23 Fisher exact test 0.009* Primary school 312 23.13 33 33.67 Junior middle school and high school 767 56.86 51 52.04 College degree and above 156 11.56 3 3.06 Treatment type Initial treatment 1197 88.73 76 77.55 17.751 0.001* Retreatment 152 11.27 22 22.45 Missed medicine No 1107 82.06 65 66.33 14.694 ＜0.001* Yes 242 17.94 33 33.67 DOT for patients Doctor 758 56.19 36 36.73 Fisher exact test 0.001* Family 528 39.14 55 56.12 Nobody 15 1.11 3 3.06 Others 48 3.56 4 4.08 Sputum-culture of the end of 1st month Negative 797 68.94 27 33.3 43.163 ＜0.001* Positive 359 31.06 54 66.7 Sputum-culture of the end of 2nd month Negative 1072 89.33 49 66.2 35.249 ＜0.001* Positive 128 10.67 25 33.8 Note. Statistically significant.

Table 2 Univariate Analysis of Risk Factors Associated with Treatment Failure

Table 3

Table 3 Multivariate Analysis of Risk Factors Associated with Treatment Failure Variables OR (95% CI) P Age 15-25 25-44 0.719 (0.262-1.977) 0.523 45-64 1.815 (0.717-4.593) 0.208 65- 2.522 (1.097-5.801) 0.029 Treatment type Initial treatment Retreatment 2.365 (1.276-4.381) 0.006 Missed medicine No Yes 1.836 (1.020-3.306) 0.043 DOT for patients Doctor Family 3.923 (0.350-44.011) 0.268 Nobody 1.879 (1.105-3.195) 0.020 Others 0.938 (0.110-8.028) 0.954 Sputum-culture of the end of 1st month Negative Positive 1.971 (1.080-3.597) 0.027 Sputum-culture of the end of 2nd month Negative Positive 4.659 (2.590-8.382) ＜0.001 Note. Statistically significant.

Table 3 Multivariate Analysis of Risk Factors Associated with Treatment Failure

Multivariate logistic regression analysis found that six variables included in the univariate analysis remained independently associated with treatment failure. These were ≥65 years of age (OR, 2.522; 95% CI: 1.097-5.801), retreatment (OR, 2.365; 95% CI:1.276-4.381), missed medicine (OR, 1.836; CI: 1.020-3.306), treatment not observed (OR, 1.879; 95% CI: 1.105-3.195), and positive culture at the end of month 1 (OR, 1.971; 95% CI: 1.080-3.597), or month 2 (OR, 4.659; 95% CI: 2.590-8.382).

Previous studies have reported TB treatment failure rates ranging from 0.2% to 18.6%^[4-11]. In this study of drug-susceptible PTB patients in China, the treatment failure rate was 6.8% (98/1447). Failed patients easily develop drug-resistant or chronic TB, and thus become an important infection source that makes TB control even more difficult. Therefore, it is very important to recognize and act on factors that increase the risk of treatment failure.

We found that the risk of treatment failure was 2.522 times greater in patients ≥65 years of age than in those younger than 25 years of age, which is consistent with other reports^[12-16]. The increased risk may result from age-related factors such as long onset time, atypical symptoms, a weakened immune system, complications of disease, or poor adherence to treatment. Specific interventions should be directed at the elderly to promote early detection, diagnosis, and treatment. Overall supervision and management should stress, individualized treatment when necessary and as indicated by specific patient conditions in order to improve the cure rate and reduce the failure rate.

The result of sputum bacteria testing at the end of 2^nd month was an important index of the effectiveness of intensive phase treatment^[17]. Recent studies^[18-23] reported that non-conversion of sputum bacteria test results predicted treatment failure and relapse. In our study, nonconversion of sputum culture at the end of 1^st month increased the risk of treatment failure by 1.971 times compared with conversion. The risk of failure was increased 4.659 times in patients who had not become sputum-culture negative at the end of 2^nd month. The significance of nonconversion of sputum culture results at the end of 2^nd month is in line with recent studies^[24-30].Monitoring of sputum bacteria is thus a key element in the process of TB treatment, especially during the first two months. Regular sputum testing should be carried out every month, and DST should be performed if the results of bacteriological evaluation remain positive. The treatment regimen can be adjusted depending on the DST results.

Poor adherence to anti-TB regimens is known to result in treatment failure^[31-32]. If patients feel better at a time when their treatment is not being observed, they may decide to discontinue their medication, which predicts impending treatment failure. Our study revealed that missed medicine and lack of treatment observation were independently associated with risk of failure. We should adopt appropriate measures to improve treatment adherence and thus avoid increased disease severity. There is evidence that counseling and regular communication between patients and health-care stuff is effective in improving patient adherence to treatment^[33-34]. We should improve the health education of patients and their family members to raise awareness of the dangers of irregular medication. Implementing strict observation of drug taking, and intensified medicine management with enhanced tracking to prevent missed and intermittent medication doses would further improve adherence, ensure a full treatment course, and thus reduce failure rates.

Retreatment significantly affects treatment outcome, and need for PTB retreatment has been related to not using standard regimens for initial treatment^[35-38], resulting in increased incidence of drug-resistance^[39]. However, there is still hope to cure PTB by treatment using standard, first-line drugs. In this study we found that risk of treatment failure was 2.365 times greater in retreatment patients than in those receiving initial treatment. Retreatment has also been associated with development drug-resistant TB^{[13-14, 40-41]}. Thus, improved success of initial treatment is very important, and the choice of standard first-line drugs or second-line drugs for retreatment of patients can be individualized by DST.

In conclusion, the treatment failure rate (6.8%) of drug-susceptible PTB in this study was relatively high. By identifying factors that increased the risk of treatment failure, our findings provide evidence to help develop public health policy designed to improve treatment effectiveness. First, monitoring of sputum testing during intensive phase treatment should be strengthened and DST of positive results should be routinely used to adjust treatment regimens. Second, all PTB patients who need retreatment should be evaluated by DST to guide choice of the treatment regimen. Third, patient treatment and management needs improvement to reduce the occurrence of missed doses of medicine. New management tools such as electronic medication monitors might be used to improve patient adherence and reduce reliance on direct observation. Comprehensive TB control strategies are needed for effective reduction or prevention of treatment failure.

Study Limitations This study has several limitations. First, because it was based on data collected in questionnaires it may be subject to recall bias. Second, some factors that may influence treatment failure, including comorbidities such as chronic obstructive pulmonary disease, chronic hepatitis, and HIV were not considered in our study, and thus could not be assessed.

Acknowledgements

We would like to express our deep thanks to the participants for their cooperation in this study.