Data for this historical cohort study were extracted from the documentation of government health care settings in Addis Ababa between December 2009 and March 2010. Addis Ababa is the capital city of Ethiopia. In 2007, Addis Ababa had an estimated population of 2.74 million.18 Administratively, the city is divided into 10 Sub-cities which are in turn divided into 99 kebeles (the smallest government administrative units). The health service delivering institutions in the city comprise 41 hospitals, 169 higher clinics, 146 private medium clinics, and 31 health centers. With the objective of increasing accessibility DOTS program is currently implemented in all health centers in the city.19

Health service delivery institution-based historical cohort study was conducted in Addis Ababa between December 2009 and March 2010. The objective was to investigate treatment outcomes among TB patients registered in DOTS program. Study participants were all TB patients registered from July 2004 to June 2009 in three randomly selected health centers’ DOTS clinic in Addis Ababa. The required information was extracted from TB patients who received care at Teklehaymanot, Selam and Kolfe Heath Centers.

The registration documents of the DOTS clinic in each heath center contains information about TB patients’ age, sex, address, weight, type of TB, Acid-Fast Bacilli (AFB) smear result at baseline, 2nd, 5th and 7th month after treatment initiation, treatment regimen used to treat each patient, date at which treatment was started, date at which treatment was stopped, and treatment outcome. According to the standard definitions of the National Tuberculosis and Leprosy control Program guideline of Ethiopia (NLCP),7 the following definitions were used for treatment outcome: (a) cured if patients have finished treatment with negative bacteriological result at the end of treatment, (b) treatment completed if patients have finished treatment, but without bacteriological result at the end of treatment, (c) treatment failure if a pulmonary TB patient was smear-positive at five month follow-up despite correct intake of medication, (d) defaulter if the patient interrupted treatment for two consecutive months or more than two months after registration, (e) died if the patient died from any cause during the course of treatment, (f) transferred out if treatment result is unknown due to transfer out to another health facility, and (g) successfully treated if patients were declared “cured” and “completed” treatment as per the protocol.

Three types of TB were considered in this study. The first type was smear-positive pulmonary TB (PTB+) and it was identified if a patient had at least two initial sputum smear examinations positive for AFB by direct microscopy, or one initial smear-positive examination for AFB by direct microscopy and a positive culture, or a patient has one initial smear-positive examination for AFB by direct microscope and radiographic abnormalities consistent with active TB as determined by a clinician. The second type was smear-negative pulmonary TB (PTB−) and it was characterized by a patient having (1) symptoms suggestive of TB with at least three initial smear-negative examinations for AFB by direct microscopy and no response to a course of broad-spectrum antibiotics; (2) three smear-negative examinations by direct microscopy, and radiological abnormalities consistent with pulmonary tuberculosis, and decision by a clinician to treat with a full course of anti-tuberculosis; or (3) a diagnosis based on a positive culture for Mycobacterium tuberculosis after three initial smear-negative examinations by direct microscopy. The third type was extra-pulmonary TB (EPTB). In this case TB occurs in organs other than the lungs, proven by one positive-culture from specimens of an extra-pulmonary site or histo-pathological evidence from a biopsy, or TB based on strong clinical evidence consistent with active EPTB and the decision by a physician to treat with a full course of anti-TB therapy.

Data were extracted from the registers of the selected health centers using a structured data sheet specially designed for this study. Data extraction was conducted by nurse/health officers working at the TB clinic of the selected health centers. Before embarking on the data collection process, all data collectors attended a one-day training provided by the principal investigator on how to fill the structured data collection sheet. To ensure data quality, the following measures were taken: (a) one-day training was given for data collectors before the start of data collection, (b) the overall activities of data extraction were monitored by the principal investigator, and there was strict supervision during data collection, (c) all completed datasets were examined by the principal investigator for completeness during data collection, and (d) from the data extracted from each health center, 5% of the sample was randomly selected and validated against the registration book by the principal investigator.

The main outcome of interest was treatment success of all TB patients. Study participants were categorized as having successful treatment if their record showed that they were cured or they had completed treatment. Otherwise, they were categorized as treatment not successful (i.e. the record showed that the patient was either treatment failure, defaulter, died, or transferred out). The pre-specified independent variables for treatment success were patients’ age, sex, type of TB (smear-positive pulmonary TB, smear-negative pulmonary TB, extra-pulmonary TB), patient category at the start of treatment (Category I: new sputum smear-positive, seriously ill new sputum smear-negative, new EPTB, and others; Category II: relapse, failure and return after interruption; Category III: new sputum smear-negative, not seriously ill new EPTB, not seriously ill; Category IV: chronic TB patients defined as smear-positive pulmonary TB patients who had previously received a supervised re-treatment regimen. Transfer-in: TB patients referred in from other health service delivery institutions) and AFB smear-positive pulmonary TB patients result at selected follow-up time points (i.e. at baseline, 2 month, 5 month, and 7 months).

The quantitative data extracted from the registration book of patients registered in DOTS program were checked for completeness and consistency by the principal investigator. Data entry and descriptive analysis were carried out using SPSS version 16.0 for Windows (SPSS, Chicago, IL, USA). STATA version 11 was used for simple and multivariate modeling. Descriptive statistical methods were used to generate frequencies for categorical variables and to summarize frequencies using graphical methods. Negative binomial logistic regression analysis was used to investigate the effect of selected risk factors on treatment success. Although logistic regression is the commonly used method to model binary outcome variable, odds ratio obtained from logistic regression is more likely to overestimate the true relative risk. Hence, we have reported incidence rate ratio and its corresponding 95% confidence interval, which was generated from negative binomial logistic regression. In this study any specific variable was not hypothesized as the main risk factor for treatment success. Hence, patient age, sex, treatment center, year of treatment initiation and baseline smear result of the patient were all considered in univariate and multivariate negative binomial logistic regression. Moreover, the effect of treatment category of the patient while starting treatment and sputum smear result at the 2nd, 5th, and 7th month of treatment on treatment success were investigated in bivariate negative binomial logistic regression. Results were reported as being statistically significant if p-value was less than 5%.

Ethical clearance for the conduct of this study was obtained from the Institutional Research Ethics Review Board of Aklilu Lemma Institute of Pathobiology, College of Health Sciences, Addis Ababa University (ALIPB AAU) and Addis Ababa City Administration Health Bureau with registration number of AAHB/3088/451. After this approval, the three Sub-cities health offices approved the study. In order to ensure confidentiality of the information, names or identification numbers of TB patients were not included in the data sheet.

A total of 6450 (100%) registered TB patients were included in this study, with 3017 (46.8%) patients being male. These patients had a mean, standard deviation and median age of 30.1, 13.7 and 28.0 years, respectively. In total, 1652 (25.6%) were pulmonary positive, 2187 (33.9%) were pulmonary negative, and 2611 (40.5%) were extra-pulmonary TB patients. In terms of treatment categories 5736 (88.9%) were new and 160 (2.5%) were relapse TB patients (Table 1).

Of the 1641 (25.4%) pulmonary positive TB patients, 1376 (83.9%) had AFB staining laboratory examination at the 2nd month of treatment. From these 1376 (100%) patients 49 (0.8%) were AFB positive. In addition 1193 (72.7%) of pulmonary positive TB patients had AFB staining examination at the 5th month with 24 (0.4%) being positive. At the 7th month 19 (0.3%) were AFB staining positive out of 1052 (63.6%) pulmonary positive TB patients.

A detailed summary of treatment outcomes of the study participants is presented in Table 2. A total of 6450 tuberculosis patients were analyzed: 1167 (18%) cured, 4164 (64.6%) completed treatment, 2.36 (3.7%) died, 26 (0.4%) were treatment failure, 328 (5.1) defaulted, and 351 (5.4) transferred out. The rates of treatment completion and death were 810 (65.9) and 51 (4.1) in July 2004–June 2005 to 622 (69.7) and 37 (4.1) in July 2008–June 2009. In the same line, defaulting rate decreased drastically from 107 (8.7%) in July 2004–June 2005 to 17 (1.9%) in July 2008–June 2009.

Between July 2004 and June 2009 the mean treatment success of all registered TB patients was 82.7% without significant gender effect (84.4% among males and 81.4% among females; p-value>0.05). The treatment success was 81.4% among pulmonary positive patients, 83.3% among pulmonary negative patients, and 83.1% among extra pulmonary TB patients.

The proportions of smear-positive pulmonary TB (PTB+) patients were less than 30% per year. The proportion of smear-negative pulmonary TB and extra pulmonary TB increased from 2007 onwards whereas smear-positive pulmonary TG had a slight decrement. The trend of treatment success across the years for registered TB patients from July 2004 to June 2009 showed a rise in treatment success rates from 2004 onwards. Again, unsatisfactory TB treatment outcomes (died, failed, and defaulted) across 2004–2009 showed a rise in the rate of death from 2008 onwards and reduction of defaulting rate.

Results from negative binomial logistic regression taking treatment success as an outcome of interest is summarized in Table 3.

After adjusting for pre-specified selected risk factors patients were less likely to have treatment success if they were treated in Selam health center (adjusted IRR=0.76; 95%CI: (0.69–0.83) or Kolfe health center (adjusted IRR=0.88; 95%CI: (0.82–0.95) compared to patients treated in Teklehaymanot health center. Although the statistical significance was marginal patients registered into DOTS program between July 2004 and June 2005 were more likely to have treatment success compared to patients registered into the DOTS program between July 2005 and June 2006 (Table 3).