Factors Associated with Unsuccessful Treatment of Tuberculosis in Arsi-Robe Hospital, Arsi Zone, Ethiopia: A Retrospective Study

Background: Tuberculosis remains a major global health problem and ranks along with the human immunodeciency virus (HIV) as a leading cause of mortality worldwide. The aim of this study was to investigate the treatment outcome of tuberculosis, and factors associated with treatment outcome of tuberculosis in TB patients enrolled in Arsi-Robe Hospital, Oromia regional state, South eastern Ethiopia between January 2013 to December 2017. Methods: An Institutional-based retrospective study was conducted in Arsi-Robe Hospital from 2013 to 2017 in study patients who had all forms of TB in DOTS clinic. The predictors of treatment outcomes were analyzed through bivariate and multivariable logistic regression analysis and a P-value < 0.05 were considered statistically signicant. Results: Out of the 257 registered TB patients, most of them were males (57.9%), from rural areas (62.6%) and in age of 15-24 category (39.3%). PTB-, PTB+ and EPTB were recorded in 48.2%, 32% and 19.8% of the patients, respectively. Among all cases, 8.6% had TB-HIV co-infection. Among all TB cases, 84.0% had successful treatment outcome. TB patients from urban areas (AOR: 3.34, 95% CI: 1.33 8.38, P = 0.01), with failure treatment (AOR: 6.66, 95% CI = 1.12- 39.57; P = 0.037) and HIV positive (AOR: 4.92, 95% CI = 1.38-17.51; P = 0.014) had higher odd of unsuccessful treatment outcome of tuberculosis. However, TB patients with PTB+ (AOR: 0.1470, 95% CI = 0.031-0.687; P = 0.015) and EPTB (AOR: 0.194, 95% CI = 0.054-0.688; P = 0.011) had signicantly lesser odd of unsuccessful treatment outcome. Conclusions: Being urban resident, treatment failure and HIV positive considerably challenge the treatment outcome of tuberculosis, but being PTB+ and EPTB were associated with higher treatment success rate of TB. Continuous follow-up of patients with unsuccessful treatment

A DOTs (Directly Observed Treatment Short Course) is a global strategy launched by WHO in 1994 for prevention and control Tuberculosis. The DOTs program principally employs a direct observation of treatment means where health professionals must watch the patient taking each dose to ensure correct treatment and notice any problem rapidly to take action [11]. Between 1995 and 2012, 56 million people were successfully treated for TB in countries that had adopted the DOTS/Stop TB Strategy which saved approximately 22 million lives [3].
In Ethiopia, the DOTs strategy was started in Arsi Zone and Bale zones in 1992. It has been subsequently scaled up and implemented at national level [7], and almost all public, private and non-governmental health facilities implement the strategy [12], but the case detection and treatment outcome vary across different regions of Ethiopia [13]. Retrospective cohort studies of treatment outcome all forms of conducted in Asella Teaching Hospital during 2004 to 2014 revealed a TSR of 82.9% [10].
Studies of treatment success rate of smear positive pulmonary TB of 25 districts of Arsi zone, central Ethiopia from September 2004 to 2011 revealed a treatment success rate that ranged from 69.3 to 92.5%, but, the TSR in Arsi-Robe district (82.4%) [14] was still below the WHO TSR.
Factors contributing to treatment outcome of TB are likely to vary depending on the local settings of populations. Being older age, living in rural areas, sputum smear negative treatment category at initiation of treatment, smear positive sputum test result at second month after initiation treatment, retreatment cases and HIV positive status were reported as predictors factors associated with unsuccessful TB treatment outcome [10,[13][14]. It was suggested that analysis of factors affecting treatment outcomes may help to improve performance of DOTS services and provide useful evidence for decision making in disease control programs [15]. However, no study has assessed the treatment outcome of TB in Arsi-Robe hospital, South eastern Ethiopia from January 2013 to December 2017. To the best of knowledge, this is the rst and most comprehensive study that speci cally addressed the treatment outcome of TB and associated risk factors under DOTS strategy in this public hospital.

Study Design and Setting
The study was conducted Arsi-Robe Hospital, located in Arsi-Robe town in Arsi Administrative Zone, Southeastern Ethiopia. The town is situated at a longitude, a latitude and elevation of 09 36' N, 39 08' E, and 2435 meters above sea level, respectively. Arsi-Robe town is located at 226 Km east of Addis Ababa and 101 Km from Arsi zonal town, Asella. The hospital has commenced its clinical services (including TB treatment in DOTs clinic) since March 2011. In the hospital, the DOTS clinic is operating under the National Tuberculosis and Leprosy Control Program (NTLCP) of Ethiopia, under which the diagnosis different types of TB was conducted using detection of acid fast bacilli in (AFB) under microscopy. Patients diagnosed with tuberculosis were treated using combination of four drugs (rifampicin, isoniazid, pyrazinamide and ethambutol (2RHZE) according to the Ethiopian national tuberculosis and leprosy control guideline (NTLCP guideline) [12]. The Hospital is serving more than half a million population of the town and the adjacent districts.
An institutional retrospective design was used to study the treatment outcomes of TB and associated risks in tuberculosis patients registered from January 2013 to December 2017 at Arsi-Robe Hospital located at Arsi-Robe town. All forms of TB cases diagnosed with smear, and/or radiography for all forms of TB which were registered in the TB clinic of the Hospital and who took anti-TB treatment except MDR-TB between January 2013 to December 2017. Registries in which treatment outcomes were missing and patients who transferred to other districts were excluded from the treatment outcome evaluation, as information on their treatment outcome was not available. Those cases that were under ongoing TB treatment during data recording by the researcher were also excluded. Variables such as demographic data (patient's age, sex, and residence), clinical history (type of TB, patient category during the start of treatment), and information including HIV status of the patient, year of treatment and treatment outcomes were recovered from patients' registration document using structured questionnaire by the trained nurse working in the Arsi-Robe hospital. The completed datasets were supervised by the principal investigator for completeness and consistency.
Case and Treatment Outcome De nitions ∘ ∘

Results
Patients' characteristics Case de nition and treatment outcome were de ned based on the standard de nitions of the National Tuberculosis and leprosy control program (NTLCP) guideline of Ethiopia for the diagnosis and treatment of TB case [12].

Cases de nitions
New case. A patient who never had treatment for TB, or has been on anti-TB treatment for less than four weeks in the past.
Relapse. A patient who declared cured or treatment completed of any form of TB in the past, but who reports back to the health service and is now found to be AFB smear-positive or culture positive.
Treatment failure. A patient who, while on treatment, is smear-positive at the end of the fth month or later, after commencing.
Default. A patient previously recorded as defaulted from treatment and returns to the health facility with smear-positive sputum.
Transfer-in. A patient who is transferred-in to continue treatment in a given treatment unit after starting treatment in another treatment unit for at least four weeks; Others. A patient who does not t in any of the above mentioned categories.

Treatment outcomes de nitions
Cured. A patient whose sputum smear or culture was positive at the beginning of the treatment but who was smear-or culture-negative in the last month of treatment and on at least one previous occasion.
Treatment completed. A patient who completed treatment but who does not have a negative sputum smear or culture result in the last month of treatment and on at least one previous occasion.
Treatment failure. A patient whose sputum smear or culture is positive at 5 months or later during treatment or patients found to harbor a multidrug-resistant (MDR) strain at any point of time during the treatment, whether they are smearnegative or -positive.
Defaulter. A patient who has been on treatment for at least four weeks and whose treatment was interrupted for eight or more consecutive weeks.
Died. A patient died from any cause during the course of treatment.
Transferred out: A patient who started treatment and has been transferred to another reporting unit and for whom the treatment outcome is not known at the time of evaluation of treatment results.
Successful treatment outcome. If TB patients were cured or treatment completed with resolution of symptoms.
Unsuccessful treatment outcome. If treatment of TB patients resulted in treatment failure, default, or death.
Treatment success rate (TSR). The sum of the percentages of cured and patients who completed treatment.

Data Analysis
Data were entered and analyzed using SPSS version 21 statistical software. Descriptive statistics were computed and employed to summarize socio-demographic data, clinical data and treatment outcome. Pearson Chi-square test was used to evaluate the association of independent factors to tuberculosis types and the difference was signi cant at P < 0.05. Bivariate and multivariable logistic regression was used to explore possible associations between dependent variable (treatment outcome) and independent variables. The independent variables were checked for the presence multicollinearity using Variance In ation Factor (VIF). The goodness of t of the employed model was evaluated using the Hosmer-Lemeshow test. The COR of independent variables whose 95% con dence interval (CI) excluding one were tted to multivariable logistic regression to calculate the adjusted odd ratio (AOR). In AOR, the 95% con dence interval (CI) of odds ratio (OR) excluding one was signi cantly associated with the corresponding dependent variable.

Page 5/19
A total of 257 TB patients were registered for DOTs program at Arsi-Robe Hospital, Oromia, between January 2013 and December 2017. Majority of them were males (57.9%) and rural residents (62.6%).

Discussions
This study was performed to analyze treatment outcome and factors associated with unsuccessful treatment outcome of the TB cases treated at DOTS clinic of Arsi-Robe Hospital, Ethiopia between January 2013 and December 2017 ( Table 1). Evaluation of the treatment outcome and associated factors for TB patients has the greatest importance in assessing the effectiveness of DOTS program in a country [16].
In this study, the frequency of male TB patients (57.9%) was higher than female (42.1%) in agreements with studies in Arsi zone, central Ethiopia (male: 54.4%; female 45.6%) [14]. Similar studies from Southern Ethiopia (61.3%) [13] reported higher frequency of TB patients in males than females. Globally, gender inequalities have been reported in TB case noti cation, and TB is more common in males than females [18]. It was suggested that higher proportion of males being exposed to the infection as a result of more social contact or engagement in professions associated with a higher risk for tuberculosis [19], or underutilization of the DOTS service by female patients as a result of various socioeconomic and cultural in uences [20] or might be due to the biological differences between male and female in vulnerability to development of active disease [21].
In our study, more TB patients were noted from rural areas (62.6%) than urban areas (37.4%). A study from southern Ethiopia also revealed higher frequency of TB patients from rural Ethiopia (66%) than that of urban areas (44%) [13]. This might be because of better awareness of tuberculosis and health facility in urban areas than rural areas. The frequency of TB infection was low in earlier ages (≤ 14 years) ( Table 1) The frequency of TB infection was also higher among new patients as compared other TB treatment categories (Table 1).
Similar result was also reported in other study in Debre Tabor [23]. This might indicate that disease was widespread in the society, and the retreatment cases in the society may be the sources for subsequent infection new infections. TB cases were predominantly PTB-(48.3%) followed by PTB+ (32%) and the least was EPTB (19.8%) (Table 1). Similarly, earlier studies from Bale Robe Hospital, south eastern Ethiopia (40.7%) [8] and from Brazil (65.2%) reported higher frequency of PTB-over PTB+ [24].
According to WHO [3], continuous tracing of TB treatment outcome, mainly in TB/HIV co-infected patients, is essential. The overall TB-HIV co-infection rate at Arsi-Robe Hospital was 8.6% (Tables 1 and 2). This was lower than studies in northeastern parts of Ethiopia (24.3%) [25], and Ethiopia as a whole (29.4%) [26] and Kenya (41.8 %) [27]. Studies indicated that the unavailability of HIV counseling and testing services or refusal of patients to be tested for HIV [16] or low prevalence of HIV in the study population [28] could contribute to lower TB-HIV co-infection. In this study, PTB-cases had higher number of TB/HIV co-infection than PTB + cases ( Table 2). This is the fact that TB cases with HIV are less likely to be smear-positive [29] and high in HIV prevalence countries, the number of patients with SNPT is increasing rapidly [30].
The percentage of cured patients in our study (31.5%) was higher than the rate in the Bale Robe Hospital (19.7%) [8]. More than half of the study population completed their treatment. It was higher than study in Gonder Teaching Hospital, North Ethiopia (46.4%) [9]. The Arsi zone was the zone that historically commenced the rst DOTs program [7]. Deployment of health extension workers all across the zones of Ethiopia that provide preventive and curative interventions against TB has been suggested for increase in the treatment outcome (cure and treatment completion) of TB [31].
The failure rate in this study (5.1%) was higher than in Asella Teaching Hospital, Ethiopia (0.2%) [10]. Treatment failure is a result of MDR-TB development in previously treated TB cases [32]. Treatment failure steadily decreased across the study period although the trend was not consistent through the years. The defaulting rate in our study (5.1%) was slightly lower than in Bale Robe Hospital (6%) [8]. The overall death rates of TB patients in this study (5.8%) was similar to that in Asella Hospital, central Ethiopia (6%) [10], but lower than in Arsi Zone for smear positive cases (7.4%) [14]. In our study, higher death rates occurred in TB patients who were female, urban residents, older, HIV-positive, and in PTB-( Table 2). In agreement with our study, previous studies also reported that older ages (44-54 years), smear-negative pulmonary TB [33] and HIV-positive [34] were more likely to die than their counter parts. High death rates have been reported for urban residents and in PTB- [23] in concordance with our study. According to Dangisso et al. [15], the higher proportion of deaths in PTB-cases than PTB + and EPTB cases might be due to diagnosis and treatment delays, as well as HIV infection among PTB-cases. The recorded higher death in TB patients of older ages might be due to increasing comorbidities as well as the general immunological deterioration with age [35][36].
TB patients who were HIV positive experienced more default (9.1%), death (18.2%) and treatment failure (13.6%) than HIV negative TB patients (Table 3). It was in agreement with study in Arsi Asella [10]. Studies indicated that TB/HIV co-infection has signi cant impact on treatment outcome [37][38]. HIV affected the performance of TB control programmes by increasing the number of TB cases and by compromising the treatment outcome.
The overall TSR of all forms of TB in Arsi-Robe Hospital in this study was 84.0% (Table 4). It was nearly similar with studies in Southwest Ethiopia (85.2%) [39], but lower than the global target (> 90%) [1], WHO target of Ethiopia for the new cases in the years of 2016 (90%) [1] and 2017 (96%) [22], and China (95.01%) [17]. But, it was higher than Asella Teaching hospital (81.7%) [10], Bale Robe Hospital (78.2%) [8], and Kenya (82.4%) [40]. The variation in the TSR with other studies could be due to differences in variation in DOTs performance in various study areas, socio-economic of the patients, geographic setting, sample size, study period and the TB clinic management (Table 4).
Higher treatment success was noted in male TB patients than female (Table 4). Similar ndings were also noted from Addis Ababa [33], and South region [13], but different from studies in Bale Robe Hospital [8], and Arsi Asella hospital [10]. The treatment success rate was high in younger TB patients, and decrease in older ages but the trend was not consistent. Studies indicated that TB patients at higher age might be under immune-compromised situations as a result of associated diseases, and general physiological deterioration with age thereby resulted in poor-treatment outcome [35,39].
In our study, PTB + had the highest treatment success rates (95.1%) as compared other TB types, and was higher than Arsi Zone (83.6%) [14]. The TSR in TB/HIV co-infection cases (59%) in our study was lower than Bale Robe Hospital (67.1%) [8], and Asella Teaching Hospital (79.8%) [10]. The treatment success in Arsi-Robe Hospital showed an increment during the rst two years (76.6% in 2013 and 86.8% in 2014) followed by an inconsistent increase and decrease in the following years ( In multivariable logistic regression, the odd of unsuccessful treatment outcome of TB was higher in TB patients residing urban areas than rural residents (Table 4). It was agreement with recent study in Debre Tabor, Northern Ethiopia [23] and Sidama Zone, Ethiopia [15], but against study in south Ethiopia [13]. A model based study from India indicated that average treatment success rate decrease with increase in the proportion of urban population and with increase higher degrees of urbanization [43]. Comorbidities (HIV, and diabetes mellitus), temporary interruption of treatment and other addiction related factors were suggested for poor treatment outcome of TB among urban residents [23], and di culties in ensuring quality of care and prevention of infection and transmission in densely populated areas with large disparities in sanitation and access to health care [43] have been suggested to explain the observed difference.
TB patients with failure treatment had high odd of signi cant unsuccessful treatment as compared to new patients (the reference patient category). This was in agreement with the study of Melesse and Zeleke (2018) [23] where new cases have lower odd poor treatment as compared to retreatment cases. It was found that previously treated TB cases (including failure) had a multidrug resistance TB (MDR) [11,44] and was associated with unsuccessful treatment outcomes [10,44].
In our study, TB patients with HIV co-infection had higher odd of unsuccessful treatment of TB as compared HIV negative TB patients. Several studies documented lower TSR of the HIV/TB co-infected cases [9][10]13]. HIV co-infection also increases the risk of latent TB reactivation 20-fold, is the most known risk factor for progression of M. tuberculosis infection to active disease [18,33]. The low treatment success in TB/HIV co-infected patients could be due to co-administration of ART along with anti-TB therapy which can lead to drug-drug interactions, overlapping drug toxicities and immune reconstitution syndrome [45].
Smear positive pulmonary TB patients (PTB+) had lesser odd of unsuccessful treatment of TB as compared to smear negative pulmonary TB and extra pulmonary TB patients. This is in agreement with studies in Ethiopian University Hospital [9], and Southern Ethiopia [13]. According to Biruk et al. [9], there is a monitoring of the treatment outcome and testing of the sputum in the 2nd, 5th and 7th months of treatment for PTB + but for PTB-and EPTB, monitoring is only in clinical conditions. HIV coinfection in PTB-and EPTB patients may decrease the treatment success and increase mortality [6]. Global reports also indicated an increase in TB treatment success from 2012 to 2016 [1,42]. The higher TSR of PTB + in our study could be attributed to improved diagnosis and good adherence to DOTs program in the Arsi-Robe Hospital.
This study was performed on retrospective basis and used already registered socio-economic and clinical data. As compared to WHO international target (85%) [11, 18], our study revealed an appreciable treatment success rate of TB (84%) in the study area.
Many variables that might help to relate with treatment outcome were missed because they were absent in data registry or log book of the hospital. Secondly, this study doesn't nd data on other potential factors such as sputum smear test result at second month after initiation treatment, duration, malnutrition status, drug resistance pattern of the TB patients. The data was also collected from a single hospital, though it gives treatment services to more than a half a million of people of the town and the surrounding district. The absence of these critical factors has impact on the conclusion of the results of our study and any interpretation of the results should consider the mentioned limitations, and based on the variables considered.

Conclusions
The overall TB treatment success rate for all registered patients was 84.0% which comprised of 31.5% of cure and 52.5% of completion of treatment. The unsuccessful treatment outcome rate was 16% constituted by 5.8% of death, 5.1% default and 5.1% treatment failure. In multivariate analysis, residence of the patients, TB patients' category, HIV status of the patients and TB types were the predictor variables for unfavorable treatment success of tuberculosis in Arsi-Robe Hospital, Ethiopia. TB/HIV coinfected patients were found to have lesser chance of successful treatment success. It is therefore, important to strengthen the antiretroviral treatment of HIV coverage in order to reduce the morbidity and mortality due to TB. The higher proportion of PTBover and PTB + and EPTB in our study was also unusual as compared to many studies; therefore diagnosis of TB types should be con rmed by rapid and reliable methods with higher sensitivity and speci city. It is also essential to enhance strategy of TB