Time to death and its predictors among adult with drug-resistance tuberculosis patients, in Eastern and East-Central Ethiopia, 2012-2018: A retrospective cohort study

Background Drug-resistant tuberculosis (DRTB) is becoming a global public health problem in developing country including Ethiopia. It poses a greater challenge to the tuberculosis control program. Tuberculosis drugs namely; rifampicin and isoniazid, were the two most effective anti-tuberculosis drugs for which the agent become resistant. Understanding the survival time and the predictors of DRTB patients would be helpful to policy-makers and health practitioners in Ethiopia. However, there is a limited previous study on the aspect. Therefore, this study aimed to estimate the survival time and predictors of adult DRTB patients, in Eastern and East-Central Ethiopia. Methods A retrospective follow-up study was conducted in the Eastern and East-Central part of Ethiopia among adult drug resistance-tuberculosis patients from 1st September 2012 to 30th August 2017. The checklist was used to retrieve information among a total of 362 drug-resistant tuberculosis patients. Kaplan Meier curve method was used to estimate the median survival time with its interquartile range and risks. Multivariable Cox proportional regression modelling was used to investigating predictors of survival time. Hazard ratio with 95% CI was used to report the findings of regression modelling.

Conclusion To summarize, the survival time of patients with drug-resistant tuberculosis was low. DRTB related death was higher among patients with weight loss, people with extra-pulmonary, HIV co-infection, comorbidity, and history of relapse. Background Drug-resistant tuberculosis (DRTB) became a global public health problem. It is an emerging, especially in developing country posing a greater challenge for the national tuberculosis control program. It is a type of tuberculosis (TB) which is resistant for a drug like rifampicin and isoniazid, the two most powerful anti -TB first-line drugs [1][2][3][4]. Now a day, the low and middle-income countries are significantly suffering from DRTB and its complications. Moreover, in a country where HIV infection is epidemic, the occurrence of DRTB reported as increasing. One-third of the world population is infected by TB and an estimated 1.5 million death occurred each year [ 5,6]. The highest DRTB mortality rate was reported from Africa and Southeast Asia [ 7,8]. Accordingly, Ethiopia is among the 30 DRTB high-burden countries, with an estimated 5800 cases reported in 2016 with the prevalence rate of 2.7% among new cases and 14% among previously treated patients [ 8,9]. Half of the DRTB treatment has a poor outcome as compared with non-drug resistance TB (10%). Poor treatment response is what lengthens treatment since people will continue to receive treatment if they do not culture-convert [ 10].
HIV infection, undernutrition and limited access to a well-equipped health facilities were identified as common factors extensively worsen the treatment outcome DRTB [ 11].
Even with, the notable sound effects of directly observed therapy strategies (DOTS) implemented to reduce, control and also increase the positive treatment outcome of tuberculosis the rate of DRTB become increased and prone to be the leading problem of Ethiopia [22][23][24]. Lack of social support and poor patient Cohort recognized a significant factor for the increased rate of MDR-TB. These patients might also have a chance to transmit the disease in case of bacterial reversion and leads to poor prognosis [ 25].
Therefore, this study aimed to assess the time to death and its predictors among adult  The required sample was determined using Stata 14 version statistical software based on the assumption adjusted hazard ratio (AHR) for clinical complication, 1.9, event 82 and the probability of event 0.29 [ 21] and adding 10% for the incomplete record. Based on these assumptions we found a sample of 377. Therefore, 377 DRTB patients were included in this study, which 167 were from Hailemariam hospital, 194 from Dil Chora Referral Hospital and 16 from Amir Nur health center. After excluding incomplete patient medical records, the total number of DRTB patient records in the three health facilities was 362.
Since it was manageable, all records were included in the study. Cured: is defined as patients without evidence of failure and three or more consecutive cultures taken at least 30 days apart are negative after the intensive phase.
Completed treatment: A patient who completed the anti-TB regime for 18 months or over 12 months.
Poor treatment outcome: is defined as unsuccessful treatment leading to death, TB relapse, loss to follow up treatment, or fail to complete treatment regimen or treatment interruption. Death: is defined as a patient who died during the treatment respective of the cause (i.e., death after the end of DRTB treatment course is not included or considered by this study).
Failed treatment: is defined as a smear-positive patient who remained smear-positive by the end of the intensive phase, or bacteriological reversion in the continuation phase after conversion to negative.
Censored: when the outcome of interest has not been observed for an individual during the treatment course period. DRTB who cured, treatment completed, lost to follow up, transfer out with unknown treatment results and those on treatment were treated as censored.
Data extraction checklist and course of action An English version data extraction checklist was developed from related literature and patient registration book. The checklist contains characteristics like socio-demographic data, clinical characteristics, HIV co-infection, co-morbidities, nutritional status, site of involvement, radiological findings. Patient medical records: including registration and monitoring records were used as a source of data.

Data quality control
To maintain the quality of data the following activities were conducted: the data extraction checklist was pretested on other but similar population. Moreover, a one-day training was given for the data collectors (nurses with first-degree holder working in TB clinics) focusing on the data extraction producers, checklist, and objectives of the study.
Furthermore, close and routine supervision was done by the investigators of the study. Addition, the authors and an experienced data clerk entered and cleaned the data before the analysis.

Data Processing and analysis
The collected data were entered to EpiData version 4.2 and further analyses were done using stata version 14 statistical packages. Survival curves were compared between different exposure groups using a log-rank test (Chi-square test with 1df under H o ).
Survival trend over the follow-up time was computed using the Kaplan Meier (KM) method and covariates significant in Bivariable analysis at p<0.25 were fitted to multiple Cox regression model to show the independent predictor of death. After fitting the Cox proportional hazard regression model, the adequacy of the fitted model to the survival data was checked using Cox-Snell residuals with the Nelson-Aalen cumulative hazard graph. An assumption to Cox-proportional hazard model that is the hazards remaining proportionately constant was checked using graphical representation (Log-log plot) and Schoenfeld residuals. In addition to this, time varies covariates were checked and all of the covariates were constant overtime. Hazard ratio with its 95% CI and the p-value was used to measure the association of dependent and independent variables.   were significantly shorter compared to their counterparts ( Figure 4).

Treatment Outcome
Concerning about treatment outcome, this study found that about half, 180 (49.7%) of DRTB patients were completed and cured (Table 3). In addition, after fitting the Cox proportional model, the adequacy of the model was checked by Nelsen Aalen cumulative hazard with Cox-Snell residuals (supplementary file 1). person-days [ 19].
In another way, our finding is higher than the study conducted in southern Ethiopia which was 2 patients per 10,000 person-day [ 27]. This difference might be due to the difference in the sample size variation. In this study, a large number of patients register in five years were followed compared to the previous studies conducted in southern Ethiopia.
In this study, both the extended and restricted mean were used to determine survival status. Therefore the overall mean survival probability at the end of the follow-up period was 73.79% which lends support to a study conducted at St. Peter hospital (78.95%) [ 19].
In another way, this finding on mean survival probability is lower than the study conducted in southern Ethiopia (88%) [ 27]. The possible explanation for this might be due to the difference in the follow-up periods. That means as the follow-up period becomes longer, the probability of the event becomes decreased.
This study also identified a predictor of mortality among DRTB patients. In this study, DRTB who experience weight loss during treatment was 9 (AHR: 9, 95%CI: 2.0-20.5) times more risk to die as a compared with counterpart. This finding goes in good agreement with the study reported by southern Ethiopia, Bulgaria and Lima, Peru [ 14,27,28]. The possible explanation for this may be, DRTB participants experience severe gastrointestinal intolerance (nausea, vomiting and gastritis) and drug toxicities during treatment that causes malnutrition and this may reduce the survival probability of the participants.
This is scientifically supported that anti-TB drugs have serious adverse effects including nausea, vomiting and electrolyte disturbance which leads to poor prognosis [ 29].
The current study also looked at predictors of mortality among DRTB patients of included  [ 14,15,[30][31][32][33]. The high mortality rate of DRTB patients among HIV positive participants might be due to the synergistic effects of the two co-infection HIV and DRTB.
Patients receive treatment for a minimum of two years. The patients might experience serious drug adverse effects and toxicities due to the high burden of pills. As a result, patients could experience non-adherence of the treatment and end-up with poor outcome.
In turn, the diagnosis of DRTB in HIV positive patients is more difficult as may be confused with other pulmonary or systemic infection and the rate of smear negativity is also high in HIV patients. This can result in misdiagnosis or delay in diagnosis and leads to higher morbidity and mortality [ 3,34]. Diabetes is a common comorbidity in DRTB patients, and also similar in this study. Diabetes can worsen the clinical course of DRTB and DRTB can also worsen the glycaemia control of diabetes. Anti-TB drugs and diabetic medications may have overlapping toxicities like peripheral neuropathy which leads to death [ 31,32].
This study also found that previously TB treated DRTB patients, had a history of treatment relapse and re-treatment failure were significant predictors of death. This is also agreed with study conducted in Estonia [ 18,33,35,36]. The possible reason for this might be Mycobacterium tuberculosis acquires further resistance to anti-TB drugs by spontaneous mutations and changed to the most potent form of bacteria which is called Beijing Lineage. Unless appropriate drugs are administered, this could lead to unfavorable treatment outcome. This is also scientifically supported that Beijing Lineage Mycobacterium strain of bacteria overproduces Triglycerides which leads to lethal disease [ 17,37].
In addition, the site of DRTB involvement also found to be predictors for mortality.
Pulmonary with extra-pulmonary DRTB was found to be a predictor of DRTB patients mortality that proves report in Lima Peru [ 16].

Limitation of the Study
This study included all deaths recorded during the follow-up period. The cause of death was not recorded as the data were extracted from medical records. Several sociodemographic variables namely family size, employment, educational status, and income, and substance abuse, radiologic were missed due to the incompleteness of the records.

Conclusions And Recommendations
To summarize, the current study found that survival time of patients with drug-resistant tuberculosis was unacceptably low. Then permission letter was taken from the three Hospitals' administrative bodies. Personal identifiers that were recorded on medical records were not used.

Consent for publication
"Not applicable" Availability of data and materials "The data set would not be shared to anyone in order to protect the participants' identities"

Competing interests
The authors declared that there was no conflict of interest. The overall survival status of DRTB in Eastern and East-Central Ethiopia, from September 2012 to 30th August 2017