Risk factors for unsuccessful tuberculosis treatment outcome (failure, default and death) in public health institutions, Eastern Ethiopia

Introduction Unsuccessful TB treatment outcome is a serious public health concern. It is compelling to identify, and deal with factors determining unsuccessful treatment outcome. Therefore, study was aimed to determine pattern of unsuccessful TB treatment outcome and associated factors in eastern Ethiopia. Methods A case control study was used. Cases were records of TB patients registered as defaulter, dead and/or treatment failure where as controls were those cured or treatment complete. Multivariate logistic regression models were used to derive adjusted odds ratios (OR) at 95% CI to examine the relationship between the unsuccessful TB treatment outcome and patients’ characteristics. Results A total of 990 sample size (330 cases and 660 controls) were included. Among cases (n = 330), majority 212(64.2%) were because of death, 100(30.3%) defaulters and 18(5.5%) were treatment failure. Lack of contact person(OR = 1.37; 95% CI 1.14-2.9, P, .024), sputum smear negative treatment category at initiation of treatment (OR = 1.8; 95% CI 1.3-5.5,P, .028), smear positive sputum test result at 2nd month after initiation treatment (OR = 14; 95% CI 5.5-36, P,0.001) and HIV positive status (OR = 2.5; 95% CI 1.34-5.7, P, 0.01) were independently associated with increased risk of unsuccessful TB treatment outcome. Conclusion Death was the major cause of unsuccessful TB treatment outcome. TB patients do not have contact person, sputum smear negative treatment category at initiation of treatment, smear positive on 2nd month after treatment initiation and HIV positive were factors significantly associated unsuccessful treatment outcome. TB patients with sputum smear negative treatment category, HIV positive and smear positive on 2nd nd month of treatment initiation need strict follow up throughout DOTs period.


Introduction
Tuberculosis (TB) is a major cause of illness and death worldwide. It is one of the leading causes of morbidity and death in sub-Saharan African countries. Ethiopia ranks 3 rd among sub-Saharan African countries. The burden is exacerbated by the spread of HIV infection [1,2]. In Ethiopia free TB diagnosis and treatment is undertaking at 1,448 state-owned health service institutions and more than 230 private health facilities. While treatment is integrated into general health services and DOTS geographical coverage is 95%, TB remains a major health problem in Ethiopia. Cure rate of 67% remains well below the 85% rate of WHO recommendation [3]. A retrospective study from north Ethiopia also depicts the treatment success rate of tuberculosis patients was unsatisfactory [4]. An earlier study on the impact of DOTS in the Southern Ethiopia reported one in five TB patients still continued to result with unsuccessful treatment outcome. Default is one of the unsuccessful forms of TB treatment outcome and a serious problem in the TB program of Ethiopia. According to the retrospective study in rural hospital in South Ethiopia defaulting from treatment rate was 11.4%. Another study from Northern part of the Country also reviled among unsuccessful treatment outcome, 18.3% were defaulted followed by death and treatment failure account 10.1% and 0.2% respectively [4][5][6].
In effort to reach the global target of 85% treatment success, it is compelling to identify, describe, and deal with factors determining poor treatment outcome. Several reasons and risk factors for unsuccessful TB treatment outcomes have been reported from different countries [4,5,7,8]. However, up to the researches awareness it is not clear which factors are major contributors to the unsuccessful TB treatment outcome of TB patients in the eastern part of Ethiopia. Failure to treatment completion or cure is believed to be the main reason for difficulties in controlling a disease that is far from new. For these reasons, determination of the pattern of unsuccessful treatment outcome and factors that predicts the unsuccessful treatment outcomes helps to design the possible future of TB treatment and control in the community. Therefore, this study was aimed to describe pattern of unsuccessful TB treatment outcome and associated factors among health institutions providing DOTS in eastern Ethiopia.

Settings and study design
Intuitional based case control study design was conducted using the patients' record in 6 TB clinics providing DOTS in East Hararge of Oromia Region, Dire Dawa Administration and Harari regional State, Eastern Ethiopian. Cases were those registered as defaulter, death or treatment failure on the TB registration log book. For each case, 2 controls enrolled in the same week or one week later or earlier and declared as cured or treatment complete were selected. Data was collected from September 1 st to October 30, 2012.

Sample Size and Sampling Procedures
Sample size was calculated based on the assumption previous treatment defaulter as one of the important risk factor for unsuccessful treatment outcome, 5% level of significance and 80% power. In an earlier study, we found that the proportion of success full treatment outcome among started treatment after defaulter in southern Ethiopia is 46% where as 55% of patient return after default result with unsuccessful treatment outcome [7]. Based on the above assumption the final sample size was 330 cases and 660 controls with 95% certainty that return after default is a statistically significant risk factor for not to complete treatment successfully.
Records of TB patients from September 2007 to August 30, 2012 were reviewed. The total sample size was proportionally allocated for each institutions based on the total patient started TB treatment in the last five years before the data collection time. Additionally, the total sample size allocated for each institution was also allocated for each Ethiopian Fiscal year based on the total number of TB patient started treatment during each year. All the list of defaulter, dead or treatment failure patients was reviewed until the sample size was fulfilled. To improve the comparison of cases and controls for one case two controls that started the treatment in the same week in the same institution was included. If more than two controls started in the same week; the one with the closest recoded with the case was selected.

Data Collection
Data was collected using a check list prepared from TB registration logbook form patient records, which was developed by the Ministry of Health. They were filled out by health workers working in the TB Page number not for citation purposes 3 clinics. To ensure the quality of data, pre-test of data collection tools was done in one hospital and one health centre not included in the study. Training was given for data collectors and data collection process was supervised by investigators. Every check list was checked for its completeness during data collection.

Data Analysis and Processing
The collected quantitative data was coded by investigators and   Sputum test positive for AFB result at 2nd month after initiation of treatment was significantly high among unsuccessful treatment outcome compared to those successful treatment outcome (OR= 14.23 (5.52-36.46). This result is in line with a study from Yunnan, China which showed positive 2-month smear test result is one of the risk factors for unsuccessful treatment outcome [14,15]. Our study showed sputum smear negative pulmonary TB at the begging of treatment is a factor that predict unsuccessful TB treatment outcome (OR 1.83, 95% CI 1.3-5.51). The finding is in line with previous studies from Ethiopia and other countries [4,8]. This might because of probability of miss diagnosis of the patients which resulted with poor treatment response. This study also depicts lack of registered contact person was also the risk factor associated with unsuccessful treatment outcome (OR=1.37(1.14-2.91, p<0.024).

Socio-demographic Characteristics
There are some limitations to this study. First, the retrospective nature of the study is a methodological limitation. Second, we used only routine programme data; so that the Ethiopian health

Competing interests
The authors declare no competing interests.

Authors' contributions
Both TD and TA conceived of the study, and participated in its design and coordination. Also carried out the literature search and performed the statistical analysis. TD has taken part in the data collection, data interpretation and writing. Both authors have read and approved the final manuscript.

Acknowledgments
The authors thank the health institutions' staff involved for their contribution to data collection and management. The study would not have been possible without the financial support of Haramaya University.