Survival Status and Predictors of Mortality among Multidrug-Resistant Tuberculosis Patients in Saint Peter's Specialized Hospital, Addis Ababa, Ethiopia

Background Multidrug-Resistant Tuberculosis (MDR-TB) is tuberculosis that is resistant to at least both rifampicin and isoniazid. The World Health Organization as reported in 2019 revealed that Ethiopia is among the 20 countries with the highest estimated numbers of incident MDR-TB cases. However, supporting evidence is limited in the study area after the Ethiopian national strategic plan for tuberculosis prevention and control is started. Objective To determine survival status and predictors of mortality among multidrug-resistant tuberculosis patients treated in Saint Peter's Specialized Hospital at Addis Ababa, Ethiopia, 2020. Methods An institutional retrospective cohort study was conducted using all MDR-TB patients who were enrolled in Saint. Peter's Specialized Hospital from January 01, 2015, to December 31, 2017. A pretested data extraction form that had 5 items for sociodemographic and 15 items for the measurement of clinical characteristics of 484 MDR-TB patients was used. STATA software version 14.2 was used for data cleaning and analysis. A variable that fitted in the bivariable Cox proportional hazard model at p value <0.25 was used in the final multivariable Cox proportional hazard model, and independent predictors of time to event were determined at a p value of 0.05. Result A total of 484 patients were followed up for 5,078 person-months. Among the total patients, nearly half, 238 (48.8%), were males. The median age of patients was 30 years (interquartile range (IQR), 24–39), and 56 (11.6%) were aged between 1 and 19 years. During the follow-up period, 315 (65.1%) patients were cured, 125 (25.8%) completed treatment, 24 (5%) died, and 20 (4.1%) were lost to follow-up. The overall cumulative probability survival of the patients at the end of treatment was 94.85% (95% confidence interval (CI): 92.38%–96.53%). The independent predictors of time to death were being anemic (AHR = 3.65; 95% CI: 1.36, 9.79), having clinical complication (AHR = 3; 95% CI: 1.2, 7.5), and being HIV infected (AHR = 5.8; 95% CI: 2.2, 15.7). Conclusions MDR-TB patients' survival rate was high in St Peter's Specialized Hospital. MDR-TB patients with anemia, HIV coinfection, and clinical complications had higher risk of mortality. So, prevention and controlling of anemia, HIV/AIDS, and clinical complications will reduce the mortality of MDR-TB patients.


Introduction
Multidrug-resistant tuberculosis (MDR-TB) is tuberculosis (TB) disease resistant to at least both rifampicin and isoniazid, the two most powerful anti-TB drugs. Detection of MDR-TB requires bacteriological confirmation of TB and testing for drug resistance using rapid molecular tests, culture methods, or sequencing technologies. Treatment requires a course of second-line drugs at least for 9 months and up to 20 months, supported by counseling and monitoring for adverse outcomes [1,2].
TB remains a major cause of ill health and is one of the top 10 causes of death worldwide. An estimated 10.0 million people fell ill with TB in 2018. Globally, there were 1.2 million TB deaths among HIV-negative people in 2018 and an additional 251000 deaths among HIV-positive people. Since 2007, TB has been the leading cause of death from a single infectious agent, ranking above HIV/AIDS. In 2018, there were an estimated 390,000 new cases of multidrugresistant TB [1].
An anti-TB drug resistance surveillance data estimated 4.1% of new and 19% of previously treated TB cases in the world to have either multidrug-resistant or rifampicin-resistant tuberculosis (MDR/RR-TB). In 2016, an estimated 600,000 new cases of MDR/RR-TB emerged globally and caused 240,000 deaths. About 6.2% of MDR-TB cases have additional drug-resistance, extensively drug-resistant TB (XDR-TB) [3].
World Health Organization reports in 2019 reveal that Ethiopia is among the 20 countries with the highest estimated numbers of incident MDR-TB cases and listed among 14 countries that are the highest for TB, TB/HIV, and MDR-TB cases [1]. A major barrier to progressing toward TB elimination in Ethiopia is the TB "case detection rate" of only 60 percent, meaning that an estimated 80,000 Ethiopians who developed TB in 2014 were never diagnosed or treated, leading to ongoing spread of TB to family members and communities. e gap in case detection rate is even worse for the more severe MDR-TB, where less than a quarter of an estimated 2,200 Ethiopian MDR-TB patients are identified each year [4].
Different factors have been associated with mortality of MDR-TB patients. Some studies show that clinical complication, the presence of any chronic disease including HIV [5][6][7], older age [7,8], therapeutic delay [5,7], smoking [5], body mass index (BMI) less than 18 [9], poor drug adherence [6], alcohol use [7], hypokalemia [8], low CD4 [9], extra-pulmonary TB [6], and living in rural areas [8] are risk factors for the mortality of MDR-TB patients. However, contradictory evidence revealed that BMI, HIV status, extra-pulmonary TB [8], and smoking [7] are not risk factors for mortality of MDR-TB patients. So, the purpose of this study was to assess survival status and identify predictors of mortality among MDR-TB patients in Saint Peter's Specialized Hospital, Addis Ababa, Ethiopia.

Study Design and
Period. An institution-based retrospective cohort study design was used to assess the survival status and predictors of mortality among MDR-TB patients treated in Saint Peter's Specialized Hospital, Addis Ababa, Ethiopia, in March, 2020.

Study Setting and Population.
e study was conducted in Addis Ababa, the capital city of Ethiopia. It is the largest city in Ethiopia, established in 1887 by emperor Menilik II. It has the status of both a city and a state. e city is divided into ten subcities. In 2017, it had a projected 3,435,028 million population of whom 1,809,577 were females and the rest 1,625,451 were males [10]. e study was conducted in Saint Peter's Specialized Hospital, the largest TB referral center in the country, in Addis Ababa. e hospital has a long history of TB management since it is a hospital where TB case management was initiated for the first time in the country in June 1961. e study population was MDR-TB patients who started MDR-TB treatment from January 1, 2015, to December 31, 2017, and then, by considering the lately registered patients, we followed them up to the end of their treatment follow-up based on records which take up to December 2019. Five hundred and nineteen MDR-TB patients were managed at inpatient and outpatient levels during the indexed period.
All completely registered documents that contain identified variables of treated patients at Saint Peter's Specialized Hospital during the index period were included in the study. All transfer-out patients' documents whose treatment outcome was not available were excluded. By excluding 28 transfer outpatients and 7 patients with incomplete medical records, all the remaining 484 MDR-TB patients managed form 2015-2017 were included in the study.

Data Collection Tool and Procedures.
A structured data extraction form was prepared in English language by reviewing previous similar works of literature and considering the availability of variables on medical registration books and treatment cards [5,8,[11][12][13]. ere were 5 items for sociodemographic and 15 items for the measurement of clinical characteristics of MDR-TB patients. e baseline weight of patients was taken as recorded on the patient card nearest to 0.5 kilograms. Baseline weight was categorized using the median weight of patients. Nothing was written about how the weight was measured. e hemoglobin level was recorded to the nearest 0.1 g/dl. e CD4 count of HIV/ AIDS patients was measured in number of cells per cubic millimeter. e person-time of patients was measured from the time of diagnosis of MDR-TB to event (death). A pretest was conducted on records of 25 MDR-TB patients that started treatment in 2014 at St. Peter's Specialized Hospital. en, drug adherence and height variables were excluded because more than half of the records missed these variables.
Data were extracted by one BSc nurse who has been working at St. Peter's Specialized Hospital and 2 general practitioners (GPs), new graduates from Debre Berhan University.

Operational Definition
(i) Death (event): a patient who died of any cause during the course of MDR-TB treatment [14] (ii) Censored: when the outcome of interest has not been observed for an individual; this includes cured completed, treatment failure, and defaulter [15] (iii) Cured: treatment completed according to national recommendation without evidence of failure and three or more consecutive cultures taken at least 30 days apart are negative after the intensive phase [14] (iv) Treatment completed: treatment completed according to national recommendation without evidence of failure but no record that three or more consecutive cultures taken at least 30 days apart are negative after the intensive phase [14] (v) Treatment success: sum of cured and completed treatment [14] (vi) Treatment failure: treatment terminated or need for permanent regimen change of at least two anti-TB drugs at the end of the intensive phase or later during treatment [14] (vii) Defaulter (lost follow-up): a patient whose treatment was interrupted for two consecutive months or more for any reason without medical approval [14,16] (viii) Time to death: time to occurrence of death measured from confirmed diagnosis of MDR-TB to event (death) (40) (ix) erapeutic delay: a confirmed MDR-TB patient who starts treatment after one month [5] (x) Previously NOT treated for TB: patients have never been treated for TB or have taken anti-TB drugs for less than a month [14] (xi) Previously treated case: a patient who admits having been treated for TB for one month or more [14] (xii) Clinical complications: disorders that occur during the course of MDR-TB treatment (xiii) Anemia: <13 g/dl for adult men, <11 g/dl for age 6 months to 5 years, and <12 g/dl for nonpregnant women and children aged 6-14 years were applied [17] 2.5. Data Quality Control. Orientation was given for half a day on the objectives and relevance of the study, confidentiality of information, and how to extract data on MDR-TB logbook and treatment cards. Daily supervision of the data collection process and checking the completeness and consistency collected data were carried out by the principal investigator.

Data Processing and Analysis.
First, data were checked for completeness and consistency before entering into computer. en, data were coded, entered into Epi-Data version 3.1 Software, and exported to STATA statistical software version 14.2 for data cleaning and analysis.
A Shapiro-Wilk test (p value <0.05) showed that age, baseline weight, hemoglobin, and CD4 counts were not normally distributed. e descriptive analysis such as frequency distribution, percentages, and measures of central tendency for continuous variables (median and interquartile range (IQR)) was used. e maximum variance inflation factor was 1.26 among variables in the final model. A Kaplan-Meier Survival curves were carried out to compare survival probability between groups using a log-rank test. e incidence of death with respect to person-time at risk was calculated. A variable that fitted in the bivariable Cox proportional hazard model at p value <0.25 was used in the final multivariable Cox proportional hazard model, and independent predictors of time to event were determined at a p value of <0.05.
e value of Harrell's C was 0.9638 for the assessment of model adequacy.
e proportionality assumptions to the Cox proportional hazard model were checked using the goodness-of-fit (GOF) test by Schoenfeld residual. All candidate covariates for multivariable Cox regression did not violate the proportionality assumptions.
e goodness of fit of the final model was checked by likelihood ratio. Results of the likelihood ratio test were a chi-square of 104.47 with a p value of <0.0001 suggesting that the model was good fitted.
Crude and adjusted hazard ratios with 95% CI were calculated to measure the magnitude of the association between the covariates and time to MDR-TB death. A p value of <0.05 was taken as a cutoff point to identify statistically significant variables.

Survival Time.
All 484 patients were followed up for a median of 247.5 days (IQR: 237 to 497.75 months) ranging from 2 days to 1340 days. e total person-days of follow-up was 158,337 days. During this period, 24 deaths were recorded, resulting in 1.5 deaths per 10,000 person-days. e median time to death was 57.5 days (IQR, 30-95 days), indicating 75% of deaths occurred during the first 95 days of the MDR-TB course of treatment. e cumulative survival rate of MDR-TB patients at the end of treatment was 94.9%. At six months of the MDT-TB course of treatment, the survival probability was 95.2%. e cumulative failure probability of patients was high in the first six months of follow-up (Table 4 and Figures 1 and 2). Table 5, with bivariable analysis, baseline weight, hemoglobin level, HIV status, previous history of treatment, clinical complication, and the presence of lung cavities were significantly associated with the time of death. However, in multivariable binary Cox regression, being anemic, HIV coinfection, and clinical complications were significantly associated with death during MDR-TB medication. ose who had clinical complications were more than 3 times at risk to have a death outcome than those who had no clinical complications (AHR � 2.55; 95% CI: 1.06, 6.10). HIV-infected patients were about 6-folds at risk to have a death outcome than noninfected MDR-TB patients (AHR � 5.7; 95% CI: 2.2, 14.51), and anemic patients were more than 4folds at risk than nonanemic patients (AHR � 4.26; 95% CI: 1.65, 11.02) ( Table 5).

Discussion
A retrospective cohort study was conducted among 484 MDR-TB patients in Saint Peter's Specialized Hospital, Addis Ababa, Ethiopia, to assess the survival status and predictors of mortality. A majority, 440 (90.9%) (65.1% cured and 25.8% completed), of the patients had successful treatment outcomes, which reflects Saint Peter's specialized Hospital has achieved the "Global Plan to END TB 2016-2020" target of ≥87% [18]. e survival rate of 94.85% of MDR-TB patients in St. Peter's Specialized Hospital has shown considerable improvement compared to the 78.95% report in the same hospital 7 years ago [6]. It is also higher than 88% in SNNPR [14] and 80% in Wuhan, China [16]. e difference might be due to the change in time that is related to increased patients' awareness of their health and improvement in patient care. e death rate was 1.5 per 10,000 person-days which is lower than the report of Wuhan, China, which was 1.79 per 10,000 person-days, SNNPR, Ethiopia, 1.92, and much lower than the incidence rate in this hospital before 7 years which was 4 per 10,000 person-days [5,12,19]. e possible reason for these findings is as follows: first, it might be because the setting is a TB-specialized hospital that has good infrastructure with highly trained and skilled healthcare workers and the site is a research center for tuberculosis that helps the hospital to identify gaps to intervene for good achievement. Second, chronic diseases get more emphasis that decreases the mortality of patients related to it.
Almost all deaths of patients, 23 (95.8%), occurred in the first 6 months of the MDR-TB treatment course which is consistent with a study report in Northwest Ethiopia [11]. e following would be the possible explanations: Firstly, a diagnosis of MDR-TB and being hospitalized for 6 months might create psychosocial problems that led to a poor prognosis. Secondly, more than one-third were from out of Addis Ababa, so the distance might have led to feeling like a stranger and it would be difficult to trace defaulters. Lastly, the majority (66.7%) of the patients who died were on ART, and taking many drugs might have affected their psychological wellness that led to poor prognosis. Hemoglobin level was one of the factors associated with mortality of MDR-TB patients in this study. Anemic patients were more than 4-folds at risk of death during MDR-TB medication than nonanemic patients. is result is consistent with a study from Northwestern Ethiopia, Tanzania, and Haiti that describes anemia was an associated factor to the death of patients [11,20,21].
Another factor associated with death of patients was HIV/AIDS coinfection. HIV-infected patients were 5.7 times more at risk to die while on MDR-TB treatment. is is in line with findings in the Amhara Region of Ethiopia, SNNPR, Ethiopia, South Africa, and Latvia [5,8,12,22,23]. HIV/AIDS might compromise their immunity and lead to poor prognosis of their health status.
ose who had clinical complications were 2.55 times at risk of death than those who had no clinical complications. Other studies have also shown that MDR-TB patients with clinical complications were associated with an increased risk Canadian Journal of Infectious Diseases and Medical Microbiology of death [5,19]. Clinical complications may adversely affect the prognosis of patients involving the worsening of MDR-TB and affect other organ systems.
is study had limitations of inability to control potentially important confounding variables such as drug adherence, economic status, nutritional status, and

Conclusions
e probability of survival during MDR-TB medication was high that indicates St. Peter's Specialized Hospital has achieved 2020 milestones for a reduction in tuberculosisrelated deaths planed by the World Health Organization. Almost all deaths of MDR-TB patients were during the first six months of treatment. Being anemic, HIV coinfected cases, and clinical complications were associated with mortality from MDR-TB.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Ethical Approval
Ethical clearance was obtained from the Institution Review Board of the Public Health Department, College of Health Science, Debre Berhan University. en, a formal letter of cooperation was written to the Director of St. Peter's Specialized Hospital and permission was obtained. e information regarding any specific personal identifiers such as name of the patients was not collected, and also, the confidentiality of any personal information was maintained throughout the study process.

Conflicts of Interest
e authors declare no conflicts of interest.