Time to sputum culture conversion and its determinants among Multi-drug resistant Tuberculosis patients at public hospitals of the Amhara Regional State: A multicenter retrospective follow up study

Background In Ethiopia, Multi-drug resistant Tuberculosis (MDR-TB) is one of the major public health problems that need great attention. Time to sputum culture conversion is often used as an early predictive value for the final treatment outcome. Although guidelines for MDR-TB are frequently designed, medication freely provided, and centers for treatment duly expanded, studies on time to sputum culture conversion have been very limited in Ethiopia. This study was aimed at determining the time to sputum culture conversion and the determinants among MDR-TB patients at public Hospitals of the Amhara Regional State. Methods A retrospective follow up study was conducted between September 2010 and December 2016. Three hundred ninety two MDR-TB patients were included in the study. Parametric frailty models were fitted and Cox Snell residual was used for goodness of fit, which the Akaike’s information criteria was used for model selection. Adjusted hazard ratio (AHR) with a 95% confidence interval (CI) was reported to show the strength of association. Result Out of the 392 participants, sputum culture changed for 340(86.7%) during the follow up period. The median culture conversion time in this study was 65 (60–70 days). Alcohol drinking (AHR = 3.79, 95%CI = 1.65–8.68), sputum smear grading +2 (AHR = 0.39, 95%CI 0.19–0.79), smear grading +3 (AHR = 0.30, CI = 0.14–064), cavitations (AHR = 0.36, 95%CI = 0.19–0.68), and consolidation (AHR = 0.29, CI = 0.13–0.69) were the determinants of time to sputum culture conversion. Conclusion In this study, time to sputum culture was rapid as compared to 4 months WHO recommendation. Alcohol drinking, sputum smear grading, cavitations and consolidations were found to be the determinants of time to sputum culture conversion. Therefore, providing a special attention to patients who had baseline radiological finding is recommended, high bacillary load and patients with a history of alcohol intake at baseline should be given priority.

Introduction Tuberculosis (TB) which kills approximately 2 million globally every year has been the second leading cause of morbidity next to the human immune deficiency virus (HIV) [1,2]. The multidrug resistant (MDR) TB is an increasing global problem [3]. Proper treatment and enhanced case detection is needed to ensure good success rate and halt the emergency of MDR or extensively drug resistant (XDR) TB, as the risk is high among previously treated patients [4]. The treatment of MDR-TB takes 2 years (18-24 months), and monthly sputum culture is considered as a remarkable indicator of treatment effectiveness, especially during the intensive phase of treatment [5]. Knowing the time to sputum culture conversion is often used as an early predictive value for the final treatment outcome, especially in MDR-TB patients [2], and a delayed after four-month conversion was considered as a precondition for suspecting MDR-TB treatment failure [2].
About 88% of the anticipated MDR-TB cases were found in Brazil, China, India, the Russian Federation, and South Africa [7], while its impact worsened in Asia and Africa [8]. According to the 2015 global report, 28% of MDR-TB patients were found in the African region [2], Ethiopia ranked 15 th among the 27 high MDR-TB burden countries with more than 5000 estimated MDR-TB patients each year. The prevalence is 2.8% in newly diagnosed patients and 21% in patients under re-treatment [9].
The most common factors that affect time to sputum culture conversion are: MDR-TB category, HIV co-infection, presence of radiological findings, presence of chronic diseases, number of resistant drugs at initiation, number of active drugs taken, and therapeutic delay of greater than one month [10][11][12][13].
Delaying sputum culture conversion resulted in economic wastage by prolonging the duration of treatment, poor treatment adherence and consequently failing treatment. It is also associated with higher case fatality rates (50-80%) as a result of drug toxicity, leading to the emergence of XDR-TB [13]. Despite the provision of free treatment, monthly culture monitoring, and expansion of services, studies on time to sputum culture conversion in our country including the study area is very limited. Therefore, this study is aimed to determine the time to sputum culture conversion and its determinants among MDR-TB patients.

Study design and setting
An institution based retrospective follow up study was conducted from September 2010 to December 2016 among MDR-TB patients in three at the nine public hospitals of Amhara

Population and sample
Gondar, Debre-Markos and Borumeda referral hospitals were selected out of the nine hospitals in the region. All drug resistant TB patients who were culture positive at the start of treatment at least for two months and had follow up were included in the study. Patients who had incomplete data on the outcome variables were excluded. Culture positive MDR-TB patients followed for less than two months and were culture negative at baseline were excluded. A total of 392 MDR-TB patients which fulfilled the inclusion criteria took part in study.

Data collection procedures and variables
Time to sputum culture conversion was the response variable, where as socio-demographic characteristics (sex, age, residence, educational status, marital status), behavioral factors (smoking, alcohol drinking) and clinical characteristics(HIV co-infection, MDR category, MDR treatment regimen, type of resistance, presence of chronic diseases, clinical complications, radiological findings, therapeutic delays> 1 month, base line Body Mass Index (BMI), smear grading and functional status) were the explanatory variables.
Time to sputum culture conversion-was defined as the time from initiation of treatment of MDR-TB for a patient who had two negative consecutive cultures taken at least 30 days apart after initiation of treatment, censored was defined as when the outcome of interest has not been observed for an individual. This includes treatment stopped while culture was positive, died before conversion, transferred out before conversion, and study time completion before culture conversion. Therapeutic delay was defined as confirmed MDR-TB patient who started treatment after one month of MDR-TB diagnosis, previously not treated for TB was defined as a patient who had no prior anti-TB treatment or taken anti TB for less than one month. Previously treated case was defined as a patient who was treated for TB for one month or more, a patient who had under <18.5 kg/m2 body mass index was classified as low BMI, where as a patient who had !18.5kg/m2 body mass index was classified as normal BMI. Acid Fast Bacilli (AFB)Sputum smear microscopy was performed at baseline and on monthly follow-up using Ziehl-Nielsen staining. Results are reported based on the number of AFB: negative (no AFB/ 100 high-power fields [HPF]), scanty (1-9 AFB/100 HPF), 1+ (10-99 AFB/100 HPF), 2+ (1-9 AFB/HPF), and 3+ (>9 AFB/HPF).

Data management and analysis
Prior to data collection, records (both baseline and follow up) were reviewed and identified by their medical registration/card numbers. Trained data collectors (nurse and health officer) reviewed and extracted data from patient charts and registries, using a semi structured check list. The filled formats were checked for completeness by the principal investigator and three supervisors. The data were entered in to EPI info version 7 and exported to Stata version 14 statistical software for further analysis.
Descriptive summary statistics, like the median time of culture conversion and Kaplan Meier curve estimations were computed. Proportional hazard assumption (PHA) was checked using graphs and Schoenfeld residuals tests. The Cox regression model for bivariable and multi variable analyses was done. Model comparison was carried out based on Akaike Information Criterion (AIC). Then, parametric frailty models were fitted and compared to identify the potential determinants. Goodness of fit was checked by Cox Snell residual test. Final results were taken as significant at 5% level of significance. The Adjusted Hazard Ratio (AHR) with its respective 95% Confidence Interval (CI) was reported to show the strength of association.

Ethical issues
The ethics committee of Institute of Public Health, College of Medicine and Health Sciences, University of Gondar approved this study. Similarly, informed consent was not taken from the study participants directly as it is entirely secondary and no access to meet the study subjects. However, the committee gave accreditation for the approval and permission letter was obtained from Amhara regional state. Then permission was also requested from each hospitals and informed consent also waived by each respective hospitals for their patients. All data were de-identified prior to access by authors. The check list was kept securely in locked cabinets and the data base was password protected.

Baseline socio-demographic characteristics
About 428 pulmonary MDR-TB patients were reviewed. Among these, 36 (8.4%) were excluded from the study because of baseline negative culture and contaminated results. Thus, a total of 392 (91.6%) study units were included in the analysis. The majority of the patients were attended at the University of Gondar Referral Hospital; 242(61.7%) were followed at Borumeda hospital,113(28.8%), and the remaining at Debre-Markos Referral Hospital.

Time to sputum culture conversion
The participants were followed for a total of 1000.6 person months (83.4 person years) observation. The median follow up period was 65 days (IQR: 61-70). The cumulative probability of survival on initial conversion was 0.89 at the end of one month, 0.56 at the end of two months, and 0.19 at the end of four months (Fig 2).
The Schoenfeld global test of the functional status satisfies the PHA (P-value = 0.294). The mean survival time, on culture conversion for patients who were bedridden had a median conversion time of 87 days, whereas ambulatory and working patients had median conversion times of 63 and 56, respectively, and the difference was significant (p-value = 0.0318) (Fig 3).

Determinants of time to sputum culture conversion
Similarly, the proportional hazard assumption was checked by the Schoenfeld residual global test, and p-value showed 0.68. So, proportional hazard assumption was met. Based on Akaike Information Criterion (AIC), the univariate frailty with Weibull distribution and gamma frailty (AIC = 396.82) model was more efficient than Cox-proportional hazard (AIC = 1859.41), parametric exponential model (AIC = 545.05), and other frailty models (Table 3). Thus, the inclusion of shared frailty term showed no statistically significant variance and indicated that there was no any shared heterogeneity or shared unobserved variability among hospitals. But there was a statistically significant heterogeneity among individuals (variance = 1.56)( Table 4). Goodness of fit for the fitted model was also performed using the Cox Snell residual test and showed that the model was adequate (Fig 5).  The finding from the bivariate analysis showed that alcohol drinking, type of resistance, baseline sputum smear grading, and cavitations at baseline, baseline infiltration, baseline consolidation, chronic fibrotic change at the baseline, and baseline functional status were significantly associated with culture conversion time. However, in the multi-variable analysis, alcohol drinking status, baseline sputum smear grading, baseline cavitations, and consolidation remained statistically significant predictors of time to sputum culture conversion. The rate of sputum culture conversion among who didn't drink alcohol at baseline was 3.8 times as compared to alcohol drinkers(AHR = 3.8; 95%CI:1.7-8.7).

Discussion
In this study the median culture conversion time was 65 days. Previous baseline alcohol drinking status, baseline sputum smear grading, radiological findings, like cavity and consolidation were found to be the determinants of time to sputum culture conversion.
The median culture conversion time found in this study is in line with that of studies conducted in South Africa [14] and Georgia [15]. However, this finding showed a delayed conversion compared with a study conducted in Peru [16], which has a median conversion time of 59 days. In Peru [16]these DST results as well as other prior anti-TB drug exposures were taken into account when formulating the treatment regimen, but in our study DST was done when the start of treatment had some indication like previous history of MDR-TB, patients with HIV AIDS, failure with first line anti-TB, and contact with known MDR-TB treatment. In contrast, our study showed a rapid culture conversion time compared with multi centered studies conducted in five countries such as Peru, Latvia, Estonia, Russia, and the Philippines in which the median culture conversion time was 3 months (90days). Similarly, a study in Pakistan had a median conversion time of 191 days [13,17]. This might be due to the way outcome variables were defined. In our study culture conversion time was explained by two consecutive negative culture results, whereas this was done by five consecutive negative culture results in the multicenter study. The second possible reason might be that a very small sample was used in the Pakistan study (N = 85). The third reason could be the fact that in present study MTB/Rif Xpert was used upon a positive sputum smear microscopy and the detection of Rifampicin resistance to place on standardized treatment regimen right from the beginning rather than waiting for DST results [17].
In MDR-TB patient's, alcohol drinking resulted in delayed culture conversion time. This finding is in line with that of a multi centered study conducted in five countries like Peru, Estonia, the Philippines, Latvia, and Russia [13]. This might because a large number of MDR-TB patients who drink alcohol had cavitations in the multicenter study [13]. The second reason could be related to poor nutritional status, and the direct toxic effects of ethanol on the immune system or poorer adherence to anti-tuberculosis treatment [18,19]. The third reason might be that alcoholism causes drug resistance by decreasing the immunity status [20].
In MDR-TB patients smear +2 and +3 resulted in delayed culture conversion time. This finding is in line with that of a study conducted in Latvia [21], Korea [22], and Indonesia [23]. Patients with high smear grading had high bacillary load. Having a high bacillary burden suggests a stronger infectivity and requires a longer isolation period and a more intensive treatment. Thus, it needs time to clear the bacilli if the bacillary load is high [22]. The second possible explanation could be that the presence of a high bacillary load is better associated with a reduction in bacterial killing and sterilizing activity of anti-TB drugs. It may be natural that patients with higher colony counts take longer time to convert sputum cultures [24].
Multi drug resistance TB patients who had baseline cavitations had delayed conversion time. The negative association between time to sputum culture conversion and lung cavitations in our study is in line with those of studies conducted in China [25] and Pakistan [26]. The possible reason might be the presence of lung cavitations decrease penetration and antibacterial activity of drugs [27]. In contrast to our finding, studies conducted in South Africa found no significant association between lung cavitations at baseline chest X-ray and culture conversion time [28]. The difference might be explained by a larger sample size and the four fold lower proportion of HIV co-infected patients used in our study [28].
The presence of baseline consolidation on chest radiography was associated with a longer time to initial sputum culture conversion as compared with MDR-TB patients who had no baseline consolidation. Consolidation occurs through the accumulation of inflammatory cellular exudates in the alveoli and the adjoining ducts. The liquid can be pulmonary edema, inflammatory exudates, pus, inhaled water, or blood (from bronchial tree or hemorrhage from a pulmonary artery). This minimizes the response to MDR-TB treatment and may delay the culture conversion time [29]. In contrast to this finding, a study conducted in South Africa revealed that there was no significant relationship between baseline consolidation and time to sputum culture conversion. This difference might be explained by the use of a small sample (n = 56) and a large number (88%) of HIV positive participants used in the study conducted in South Africa, showing these groups were immune-compromised, since, a radiological finding in immune compromised persons like HIV may not have sufficient lung finding [28].
Sputum culture conversion in MDR-TB patients is an important routine indicator in monitoring treatment outcome. Achieving more rapid sputum culture conversion can increase patients comfort by reducing the duration of injectable drug use and simplifying a patient's therapy. In addition, from the public health perspective, reducing the time to sputum culture conversion is an important infection control measure because patients with MDR-TB and positive sputum cultures are infectious and may transmit the disease to other persons, family members, and health care providers [21]. The strength of this study is its incorporation of a multi-center site, and implementation of the frailty model estimation instead of a binary end point (logistic regression) for addressing the effect of unobserved heterogeneity, and handling interval censoring by using parametric survival analysis. Since the data were collected from secondary source, some important predictors like HGB, BCG scar, and income which had a significant association with time to sputum culture conversion in other studies were missed.

Conclusion
The time to culture conversion in our study is rapid as compared to WHO guidelines which recommend four months. Previous alcohol drinking, high baseline sputum smear grading, presence of cavitations and consolidation were found to delay time to sputum culture conversion.
Close follow-up and attention is important especially for those who were alcohol drinkers, have baseline cavitations, and consolidation. It is also useful to assess baseline sputum smear grading and provide great attention to those who have had a high bacillary load. Finally, we recommend that researchers conduct prospective follow up studies which include other independent variables (Anaemia, BCG scar, income).