Factors influencing length of hospital stay during the intensive phase of multidrug resistant tuberculosis treatment at Amhara Regional State hospitals, Ethiopia: retrospective follow up study

Background: Tuberculosis (TB) generally considered as an ambulatory disease. However, hospitalization remains an important component for isolation and medical stabilization of patients. Hence, this study aimed to identify factors influencing the length of hospital stay during the intensive phase of multidrug resistant tuberculosis treatment at Amhara Regional State hospitals, Ethiopia: retrospective follow up study. Methods: An institution based retrospective follow up study was conducted at three hospitals, namely the University of Gondar comprehensive specialized, Borumeda and Debremarkos referral from September 2010 to December 2016 (n=465). Data were extracted from hospital admission/discharge logbooks and individual patient medical charts. Logistic regression was used to identify factors associated with longer hospital stays during the intensive phase of multidrug resistant tuberculosis treatment. Result: Most patients (92.5%) had a pulmonary form of multidrug resistant tuberculosis and a quarter of them HIV co-infections. The median length of hospital stay was 61 (interquartile range 34 to 101) days. The pulmonary form of tuberculosis (Adjusted odds ratio [AOR], 3.20, 95% confidence interval [CI]; 1.28 to 7.96), treated at the University of Gondar (AOR= 2.11, 95%CI; 1.02 to 4.41) and Borumeda Hospital (AOR= 3.59, 95%CI; 1.67 to 7.72), functional status of ambulatory (AOR=2.25, 95% CI; 1.19 to 4.27) and bedridden (AOR= 3.39, 95%CI; 1.57 to 7.35), and reported adverse drug reactions (AOR=2.54, 95%CI; 1.60 to 4.02) were significant predictors of extended hospital stays. Conclusion: The study revealed that longer hospital stay and significant differences were observed among hospitals. Decreased functional status at admission, pulmonary form of tuberculosis and reported adverse drug reactions were determinants of longer hospital stays. This underscores the importance of early case detection and prompt treatment of adverse effects.


Introduction
Tuberculosis is the leading cause of mortality and morbidity and an increased concern of global health [1][2][3]. The emergence of drug resistant strains of Mycobacterium tuberculosis further complicated tuberculosis treatment and control efforts worldwide [4].
According to the 2016 World Health Organization (WHO) report, there were about 600,000 multi-drug resistant tuberculosis (MDRTB) cases and 250,000 MDRTB related deaths each year globally [5]. Thirty high burden countries carry more than 85% of the world's drug resistant tuberculosis (DRTB) cases [5]. Ethiopia is the third high burden country in Africa with an estimated 2100 cases annually [6].
Despite the recent short regimen approval for drug resistance tuberculosis treatment, the standard treatment took two or more years [7]. The intensive phase, the first 8-12 months, is the period in which patients suffer from critical disease conditions and drug side effects. Moreover, patient infectiousness, safety, tolerability, and adverse effects of second line anti-TB drug led to more frequent hospitalizations during this phase of the treatment [8,9].
The length of hospital stay (LOS) is one of the health care system metrics for measuring the duration of hospitalization. The LOS reflects disease severity, health care efficiency, resource consumption, and health care facility policy for patient admission and discharge [10]. Globally, the reported median length of hospital stay was 90 days in high burden countries [11]. African countries are characterized by a high burden of drug resistance tuberculosis and inefficient and inadequate health care facilities for the treatment of the disease. Findings from South Africa showed a LOS median of 144 days [12] and those Nigeria 135 days [7]. Patient functional status, co-morbidities, extensive lung damage, and adverse drug reactions were determinants of the length of hospital stay in the course of MDRTB treatment [2,[12][13][14][15][16].
Concerning health facility availability and efficiency, WHO recommends a conditional ambulatory model of care in the standard treatment of drug resistant tuberculosis [17,18]. Although Ethiopia is one of the high MDRTB burden countries, there have been only limited health facilities providing MDRTB treatments with scarce evidences on LOS and its determinants during treatment.
Therefore, this study aimed to determine the length of hospital stay and its predictors during the intensive phase of MDRTB treatment. The study could be of paramount importance to clinicians and hospital administrators for a more efficient planning of tuberculosis treatment programs and resource allocations.

Study design, area and period
An institution-based retrospective study was conducted at the University of Gondar comprehensive specialized and Borumeda and Debremarkos referral hospitals from September 2016 to December 2016. In Amhara region, there are nine drug resistant tuberculosis treatment initiating centers. Three hospitals, namely the University of Gondar comprehensive specialized and Borumeda and Debremarkos referral hospitals were selected out of the nine treatment initiating centers. The hospitals were selected because they were located in the main cities of the region and covered more than 85% of the services. Besides, the hospitals had large and organized MDRTB treatment data for a long period of time. The remaining six centers included recently were located in the districts with a major purpose of supporting the three main TICs mentioned above as referrals for outpatient follow ups and enhancing the accessibility of services. All treatment centers used the same standardized multi-drug resistant tuberculosis regimen with no individualized regimen because resistance patterns were not understood for all ant-TB drugs.

Population and sample
Patients who were admitted and discharged from the selected hospitals during the intensive phase of MDRTB treatment were the study population. The ingle population proportion with the assumption of 50% longer hospital stay, 5% level of precision, and 10% non-response rate was used to determine the final sample of 422. A total of 490 multidrug resistant tuberculosis patients were enrolled for DRTB treatment in the three hospitals from September 2010 to December 2016, and 465 patients who fulfilled the inclusion criteria took part. Twenty five patients were excluded owing to incomplete data, death during treatment and transfer before completing the intensive phase.

Data collection procedures
Data available on patient records were examined and the appropriate extraction format was prepared in English. Six data collectors and supervisors (nurses and health officers) were recruited. A two day training was given on research objectives and on how to review documents as per the data extraction format before the process. Prior to data collection, records were identified by their medical registration numbers. The trained collectors reviewed and extracted data from patient charts and hospital admission/discharge logbooks using the check lists.

Variables of the study
The dependent variable was length of hospital stay in days from the date of admission to discharge. LOS refers to the duration of stay in days from the date of admission to the date of discharge under the intensive phase of MDRTB treatment with zero days of stay for a patient with less than 24 hours of ward stay. When the patient stays admitted for more than a median cut off point, LOS ≥61 days, whereas socio-demographic characteristics (sex, age, residence ,housing condition, educational status, marital status), behavioral factors (smoking, alcohol use, khat chewing), and clinical characteristics (HIV co-infection, registration group, form of TB, type of resistance, chronic diseases, clinical complications, radiological findings, treatment delay, base line BMI, and functional status were the independent variables. Multidrug resistant tuberculosis is defined as tuberculosis that is resistant to the first-line drugs isoniazid and rifampicin, or when an individual is resistant only to rifampicin and treated as multi-drug resistant. A previously treated case was defined as a patient who was treated for TB for one month or more. A patient who had less than <18.5 kg/m 2 body mass index was classified as underweight, whereas a patient who had ≥18.5kg/m 2 body mass index was classified as normal BMI. Treatment initiating centers (TIC) are health facilities selected by the TB program to provide patient care and treatment services from the time of DRTB diagnosis and throughout the course of treatment with SLDs.

Data analysis
Data were entered in to EPI info version 7 and analyzed using Stata version 14 (StataCorp. Descriptive statistics, such as percentages and medians with interquartile rage (IQR) were used to summarize categorical and continuous variables, respectively.
Based on their LOS, patients were dichotomized using the median value, <61 days (0) vs. 61 or more (1). Logistic regression was used to identify factors associated with longer (61 or more days of hospital stay) during the intensive phase of MDRTB treatment. Odds ratio (OR) with 95% confidence intervals (CI) was computed to assess the associations between socio-demographic and clinical factors and the LOS.

Baseline socio-demographic characteristics
A total of 465 patients were included in the final analysis. Most of the patients (61.1%) had follow ups at the University of Gondar comprehensive specialized hospital, followed by Borumeda (28.2%) and the rest at Debremarkos.
More than half (58%) of the patients were male with the median age at initiation of treatment of 28(IQR, 22 to 38 years), and 60.8% of them were aged between 15 and 34 years. Of the participants, 43.2% and 34.4% were married and single, respectively; 58% had some primary and above educational status, while rural dwellers constituted 52% of the respondents. As far as substance use was concerned, 18.9%, 12.9%, and 8.6% drunk alcohol, smoked cigarettes, and chewed khat, respectively (Table 1).

Clinical characteristics
The pulmonary form of MDRTB accounted for 92.5% while the rest were extra pulmonary forms. One-fourth (25.4%) of the MDRTB patients 94% of whom were on ART had HIV coinfections. One or more medical co-morbidities reported, involved 9.2% of the participants of whom 2.58% had diabetes mellitus. One or more radiological abnormalities were seen in 72% of the patients. The most common radiological findings included 42.2% cavitation, 28.6% infiltration, and 24.5% chronic changes, like fibrosis. Out of the total patients, 72% had one and above adverse drug reactions with gastro-intestinal upset, (81.5%) and electrolyte disturbance (33.6%), the most common side effects ( Table 2).

Tuberculosis diagnosis and treatment characteristics
Most patients (89.5%) had one or more previous TB treatment history, one patient for a maximum of seven times. Line probe assay (LPA) (45.6%) and Gene Xpert (46.7%) were the most commonly used diagnostic methods for confirmation of drug resistance TB. In  Table 3).

Length of in-hospital stay (LOS)
The median length of hospital stay during the intensive phase of MDRTB treatment was 61 length of hospital stay overtime is shown in (Fig. 1).

Predictors of longer in-hospital stay
In the bi-variable binary logistic regression, age, housing condition, occupation, pulmonary form of multidrug resistant tuberculosis, adverse drug effects, registration group, functional status at admission, and treatment initiating centers (hospitals) were significant at a P-value of 0.2.
In the multivariable regression analysis, functional status at admission, adverse drug reactions, pulmonary form of TB, and treatment initiating centers (hospitals) were significantly associated with longer hospital stay at a P value of 0.05. Patients who had

Discussion
In this study, the median length of hospital stay was 61 (IQR, 34 to 101) days. Low functional status, pulmonary form of MDRTB and adverse drug reaction were factors associated with longer hospital stays. Besides, significant LOS differences were also observed among treatment initiating centers.
The median length of hospital stay in this work was shorter than the WHO 2014 global TB report of 90 days [11], South Africa centralized hospital of 144 days [12], South Africa community based sites of 143 days [12], and Canada Ontario of 82 days [19]. This might be due to differences in health care system, the magnitude of drug resistant tuberculosis, including XDRTB cases and co-morbidities, like HIV. However, the median length of hospital stay of this attempt was longer than that of a study conducted in San Francisco (14 days) [8]. The possible reasons might be differences in treatment approaches. In the San Francisco study, MDRTB treatment was provided through outpatient follow ups, which decreased the length of hospital stays in the course of treatments.
Length of hospital stay was significantly different among treatment initiating center, ranging from 39.5 days at Debremarkos referral hospital to 72 days at Bourmeda. Thus, for patients who were treated at the University of Gondar comprehensive specialized hospital and Borumeda, the odds of longer hospital stay were 2.11 and 3.59 times higher compared to Debremarkos, respectively. This might be due to differences in professional expertise among hospitals. In addition, there is no clear criteria for MDRTB patient discharge from hospitals. This might leads to difference in the length of stays owing to delays of discharge. Similarly, LOS was documented different by overtime. As Figure 1 shows