Over all, an increase in treatment success rate was observed over the studied years and a relatively lower success was recorded for HIV co-infected TB patients. Similar studies conducted on TB/HIV co-infected patients have reported lower successful treatment outcome among these patients [23, 24, 25]. The overall treatment success was 89.3% and increased from 87.0% to 92.8% between the years analysed (2009- 2010 and 2013-2014). This increase was in line with the Annual Performance Report of the Ministry of Health which was shown to increase from 84% in 2010 to 92% in 2014 . The possible reason might be due to improved adherence of patients to anti- tuberculosis treatment. The overall successful treatment outcome recorded in the present study was in agreement with previous studies conducted in Ethiopia and elsewhere [13, 18, 27-30]. Our result is higher the national pooled estimates (68.1%) calculated for six European countries . In this report, a cure rate of 69.1% was found in pulmonary positive patients. Difference in outcome could be due to variation in DOTS performance. Other reasons could be difference in duration of study period, sample size and in the study setting.
The proportion of HIV+/TB co-infection (18.8%) in the present study was higher than the national average (10.7%) , but was lower than (29.4 % and 34.7%) reports from the Ethiopia  and from Cameroon (35.6%) . In 2012 in Tigray the HIV+/TB co-infection was 11.2%, 10.7% in whole Ethiopia. In the Sentinel Surveillance Report, a HIV+/TB co-infection rate of 17.2% was seen in Tigray in 2013/2014, with the highest rate of 34.7% in Addis Ababa . The three health centers performed differently, with the highest success in Semen health center and the lowest in Mekelle health center (92.6% and 85.3%, respectively p<0.0001). The reason why this difference exists warrants further research. Possible explanations could be the fact that Mekelle health center has a larger catchment area as compared to the other centers and hence this might compromise the quality of the health care for TB patients.
New TB treatment cases were more likely to have successful treatment outcome. Our finding was in agreement with other studies conducted in Somalia , Nigeria  and other countries [36, 37], indicating that previous treatment history lowered chances of successful outcome. Furthermore, being HIV negative was significantly associated with successful treatment outcome. This finding supports other similar findings suggesting that HIV co-infected TB patients have significantly lower cure rates and lower treatment success rates compared to non-HIV infected counterparts [27, 38, 39]. Viral co-infection might increase the risk of anti-tuberculosis treatment-induced hepatotoxicity leading to frequent discontinuation of the first-line anti-tuberculosis drugs and hence lower treatment adherence and lower cure rates . Furthermore, TB/HIV co-infected patients might have multiple individual, disease specific and treatment related factors that can adversely affect their treatment outcomes.
In conclusion, findings from this study show good treatment outcome of patients which is in line with Ethiopian national TB program. Being new treatment case and HIV negative are shown to facilitate successful TB treatment outcomes. Early detection of TB with prompt initiation of effective treatment of anti-TB drugs alongside the scaling up of HIV prevention activities are crucial to increase successful treatment outcome among TB patients.
One of the major limitations of this study was the use of retrospective secondary data, which is limited to whatever documented in the TB registers of the TB clinic. Variables such as treatment adherence and other disease conditions which might affect outcome were not captured in the patient’s file. Furthermore, selected sites encompass small geographic region and hence patients from these facilities might have a different profile from patients residing in other parts of the country. Another limitation might emerge from the study design where we cannot make causal inferences regarding the efficacy of the investigated treatment and the internal validity might be low due to the lack of comparator group.