According to results from 16 epidemiologic surveys conducted in East Africa, the prevalence of MDR among new cases ranges from 0.4% in Tanzania to 4.4% in Uganda, and among recurrent cases ranges from 3.9% in Tanzania to 17.7% in Uganda (Kidenya et al., 2014). As a proportion of total reported cases of TB, the number of confirmed MDR-TB cases ranged from 6/6828 (0.09%) in Burundi to 76/6784 (1.12%) in Rwanda (WHO, 2012a, b, c, d, e, f, g). In Tanzania, a total of 65,902 TB cases of all forms were notified in 2016, which is a 5.6% increase compared to cases reported in 2015, In the same year of 2016, a total of 196 MDR-TB cases were notified country wide among which 158 (82%) were started on MDR-TB treatment from 21 regions of Tanzania (Ministry of Health, Community development, Gender, 2018).

The World Health Organization (WHO) enlists number social determinants that are risks for the development resistance to anti-TB drugs, identifying inadequate therapy as the main cause of progression to MDR-TB (Desissa et al., 2018; Seyoum et al., 2014; World Health Organization, 2014). The major patient-related factor that determines the occurrence of MDR-TB is non-adherence to treatment (WHO, 2003). Reports show that only 48% of MDR-TB cases are successfully treated (Kibret et al., 2017; WHO, 2014b). This is attributed to a diverse set of factors, including co-infection with HIV, tobacco use, alcoholism, the later accelerating MDR-TB development through reduced adherence to treatment, prior TB treatment, and inadequate adherence mainly due to secondary factors such as length of treatment and adverse drug reactions (Desissa et al., 2018; Seyoum et al., 2014; World Health Organization, 2014). Several studies have shown a wide variation in predictors of treatment outcome among countries worldwide (Atif et al., 2017; Monadil et al., 2019; Tamary et al,. 2019). In Sub Saharan Africa, management of MDR-TB cases has become a challenge to the mostly insufficient healthcare systems of respective countries, especially when it is coupled with HIV co infection (Breen and Smith, 2004; Chattu and Kumaryy, 2015). In Tanzania the MDR-TB treatment success rate is estimated at 68%, lower than the WHO recommended rate of over 75% (WHO, 2018).

In order to achieve the 2020 milestones of the End TB Strategy, efforts to control TB must be intensified (Stop TB Partnership,  2018;  WHO,  2015). Understanding and mitigating the factors that lead to treatment failure among MDR-TB patients is a prerequisite to such efforts. Since the effect of some of the studied determinants for MDR-TB development are likely to have changed in an era of improved TB diagnostics and management, epidemiologic information on patient determinants that accelerate the development of MDR-TB is urgent. In the present study, we aimed to define demographic and clinical determinants for development of MDR-TB among TB patients in Tanzania.

Study area

The study was conducted at the Kibong’oto Infectious Diseases Hospital (KIDH), located in Siha District of Kilimanjaro region. The hospital provides care services to MDR-TB patients from all parts of Tanzania as referred cases. The diagnostic laboratory at KIDH which has completed external proficiency testing performs smear microscopy, Mycobacterial culture, Gene Xpert® MTB/Rif assay, and DNA probe for Mycobacterium tuberculosis complex - line probe assay (LiPA) employing Genotype MTBDR plus Ver 2.0 assay.

Study design

This was a prospective cross-sectional study conducted from April 2017 to December, 2019.

Inclusion and exclusion criteria

In this study, inclusion criteria were the presumptive MDR-TB diagnosis, being above 20 years of age and being a Tanzania resident.

Study variables

Demographic variables collected were gender, age, marital status, consumption of alcohol, cigarette smoking, occupation, education level and history of contact with MDR-TB patients. Additional clinical information regarding Treatment history, Smear, HIV status, Irregular treatment was extracted from patient files, for which written informed consent was obtained. The information collected was entered into structured data abstraction form (DAF).

Sample size determination

The minimum sample size for seroprevalence determination was estimated using the Epi Tools online sample size calculator using the formula [Z2·(p)·(1 − p)]/c2, where Z = 1.96 for 95% confidence level (CI), p = expected true proportion of 20.0%, and c = minimal tolerable error at 95% CI (0.05). Computing with the above formula gives a minimum sample size of 246. This study enrolled 428 participants to increase the power of statistical analyses.

Recruitment of participants

We applied consecutive  sampling  inviting  participants  coming  as referred presumptive MDR-TB patients to KIDH to participate in the study. The study included 428 patients from district and regional hospitals located all over the country who were previously diagnosed with TB, but failed treatment and referred to KIDH as presumptive MDR-TB cases. Among the 428 cases initially enrolled in the study, 100 suspected MDR-TB cases were excluded based on negative results with the Gene Xpert assay. An addition, 20 cases were excluded because of failure to grow in Lowenstein Jensen (LJ) media with glycerol (Hardy Diagnostics, CA, USA). Of the remaining samples, 8 gave invalid results in Line probe assay and thus excluded. After exclusion of the 128 cases, only 300 qualified for the study as confirmed TB cases.

Data collection tools and procedures

The data abstraction form (DAF) used in this study was developed by the investigators. Validity and reliability of the DAF were determined using IBM SPSS ver22.0 software (IBM® Corp., Armonk, NY, USA). It was first piloted on ‘dummy’ patients before commencement of the actual study. After the pre-test, necessary adjustments in phrasing were made accordingly. Demographic information on gender, age, marital status, consumption of alcohol, cigarette smoking, occupation, education level and TB contact history with MDR-TB patients, were collected. Information on relapse or return for treatment, Sputum Smear Score, HIV status, history of irregular treatment was extracted from patient files for which written informed consent was obtained and recorded in structured DAFs.

Collection and processing of sputum samples

Two sputum samples were collected from each patient in a sterile falcon tube (BD Falcon ^TM BD Biosciences, Indiamart, India). For all of the samples, GeneXpert® MTB/Rif assay (Cepheid, Sunnyvale, CA) was used to test the first sputum sample for molecular confirmation of MDR-TB (Eklund, 2013; Ioannidis et al., 2011). Direct sputum  smear microscopy was used to test the second sample for acid fast bacilli (AFB) using fluorescence microscopy (Ryu, 2015; Steingart et al., 2006). All positive sputum samples for AFB were cultured in solid Lowenstein-Jensen (LJ) medium with glycerol (Hi Media Laboratories, GmbH) enriched with 5-10% CO₂ for 8 weeks and examined every week to observe the presence of Mycobacterium growth. Positive growths were sub cultured on fresh L.J media with glycerol for further identification. M. tuberculosis was confirmed in cultures using measurements of growth rate, colony morphology, pigmentation and by using commercial biochemical tests (niacin assay, nitrate reduction, and catalase tests) based on their growth characteristics (Kent and Kubica, 1985). Recovered M. tuberculosis from L.J. medium with glycerol culture were used to perform Line Probe Assay (LiPA), employing Genotype MTBDR plus Ver 2.0 assay (Hain Life sciences, GmbH) according to the manufacturer’s instructions (Hainlifescience, 2015).

Gene Xpert MTB/RIF assay

Gene Xpert MTB/RIF assay (Cepheid, Sunnyvale, CA) was performed for all sputum samples as described by the manufacturer. Briefly, the Xpert sample reagent (SR) was added to the samples to be tested as per manufacturer’s instructions. The mixture was shaken vigorously for 15 min during the incubation time at room temperature. Two milliliter of the mixture was transferred to Xpert test cartridge. The cartridge was then loaded into Xpert device. Finally, the results were interpreted by the Gene Xpert DX system from measured fluorescent signals and displayed automatically after 90 min.

Line probe assay

Briefly, LiPA was performed as follows: M. tuberculosis DNA was extracted from recovered M. tuberculosis growth using GenoLyse DNA extraction kit followed by polymerase chain reaction (PCR) to amplify the rpoB gene encoding the β-subunit of RNA polymerase, katG gene encoding for the catalase peroxidase and the promoter region of the inhA gene encoding for the NADH enoyl ACP reductase for the detection of mutations and determination of MDR-TB. Primers and polymerase included in the MTBDRplus Ver. 2.0 assay kit were used for amplification of the genes. The H37Rv quality control M. tuberculosis strain was used as the positive control. The mutations in the rpoB gene associated with rifampin resistance and the mutations in katG and the inhA genes associated with isoniazid resistance among M. tuberculosis were interpreted and determined by the band patterns on the LiPA strips after reverse hybridization of the gene amplicons (Meaza et al., 2017).

Data analysis

Analyses were carried out using STATA version 13 (Stata Corp, College Station, TX, USA).  For patients’ baseline data, mean ± standard deviation (SD) and frequencies and proportions were used to report continuous and categorical variables respectively. Chi- square statistical test was used to assess the distributions of patients’ characteristics by MDR-TB status and whether such distributions were different between MDR-TB and non-MDR-TB cases. Accordingly, both bivariate and multivariate logistic regression analysis models were used to examine potential predictive factors associated with MDR-TB. Odds Ratio (OR) was used to present the effects of covariates to MDR-TB development. A p-value of 0.05 or less was considered as the level of statistical significance.

Ethical considerations

Ethical clearance was obtained from Kilimanjaro Christian Medical University College Research and Ethical Review Committee (CRERC). Permission from KIDH authority to conduct the study was also obtained. Before the investigation was performed, written informed consent was obtained from all study participants to allow for their clinical and demographic information to be used for research purpose.

Baseline characteristics of study participants

This study included a total of 300 patients who were confirmed to be TB cases by gene expert, sputum culture and LiPA. Out of these, 190 (63.3%) were males. The mean age of the patients was 18 years for males and 37.5 years for females. Majority of patients were in the age of between 30 and 49 years. Nearly half of the patients in this study were from the Northern Zone. In terms of occupation, it was found that majority, 253 (84.3%) of the patients were agro-pastoralists, traders or unspecified occupation (Table 1).

Distribution of patients’ characteristics by MDR-TB status

The HIV infection status of  all  patients  was  determined, and TB cases were also presented across their HIV infection status. Results presented in Figure 1 show that the proportion of patients with MDR-TB/HIV Co infection was 29.3%, while patients without HIV co-infection was 0.7%. Non MDR-TB with and without HIV co-infection were recorded among 1.3 and 68.7% of patients, respectively.

Association between demographic and clinical determinants of development of MDR-TB among HIV infected patients

Chi  square  analyses  of  associations   between   patient factors (demographic and clinical factors) and MDR-TB/HIV infection status. Results are presented in Table 2. Relapse after treatment, history of irregular treatment, cigarette smoking, drinking alcohol and being a truck driver were found to be statistically associated with MDR-TB/HIV positivity. Other factors showed no statistical association with MDR-TB/HIV infection status.

Independent demographic and clinical determinants of MDR-TB

Demographic factors associated with MDR-TB

In attempt  to  determine  risk  factors  for  MDR-TB,  data was fitted in logistic regression model. In crude analysis, alcohol drinking and cigarette smoking were found to be significant factors associated with MDR-TB (Table 3). In the multivariate logistic regression analysis, it was revealed that, the odds of having MDR-TB among patients who were cigarette smokers at least once a day were nearly 4 times the odds among patients who never smoke (aOR= 3.94; 95%CI:1.845-8.428, P<0.001), inferring that, cigarette smoking increases the chances of having MDR-TB by about 4 times. Likewise, MDR-TB patients who were alcohol drinkers to any extent were about ten times more likely to have MDR-TB than those who don’t drink alcohol at all (aOR= 9.98: 95%CI; 2.414-41.267, P=0.001). When clinical/treatment history variables were analyzed for their association with MDR-TB, it was observed that being HIV positive strongly increased the likelihood to have MDR-TB compared to their HIV negative counterparts (aOR= 2266.0: 95% CI; 407.5-2599.8, P<0.001). Compared to treatment failure, Relapse and non-adherence were found to be strong predictors for MDR-TB (aOR = 64.22: (95% CI; 12.786-322.507) P=0.001), and (aOR = 37.44: (95% CI; 3.895-359.836, P=0.002), respectively. History of irregular treatment in the intensive phase and smear culture grade were both not found to be predictors for MDR-TB. Results for this analysis are presented in Tables 3 and 4.

Clinical factors associated with MDR-TB

Table 4 shows results of clinical determinants of development of MDR-TB. Relapse (aOR= 64.22: 95%CI; 12.786-322.507, P=0.001),  Return  for  treatment  (aOR= 37.44: 95%CI; 3.895-359.836, P=0.002), being HIV positive (aOR=22.66: 95%CI; 407.5-2599.8, P=0.001), and irregular treatment (aOR= 2.0: 95%CI; 0.1-3.4, P=0.038) were factors that determined development of MDR-TB.

While it is understood that high rates of MDR-TB among newly diagnosed patients suggest  problems with TB control in the  past, high rates of MDR-TB among relapse cases is suggestive of  existing  problems in  the  TB control program including suboptimal treatment  regimens (Meaza et al., 2017). Inadequate management of TB cases, poor mechanisms to ensure adherence to treatment and shortage or absence of anti-TB drugs are factors previously reported to contribute to the high rates of drug-resistant TB (Nino et al., 2009; Weinstock et al., 2001; Zalesky et al., 1999). It is therefore urgent that responsible   ministries   of  health  in   High   TB  burden countries undertake rigorous restructuring of the systems and elements for improving diagnosis, management and monitoring of TB treatment to reduce the chances of progression to MDR-TB. Tobacco smoke has been reported to have various effects on the lung in both animal and human studies. More specifically, smoking has been associated with changes in pulmonary macrophages and lymphocytes that play a major role in cellular immunity. Tobacco smoking is known to compromise treatment efficacy by lowering immune mechanisms, hence compromising drug effectiveness and recovery (Carnevali et al., 2003; Leung et al, 2003). In line with our findings, other reports, including a report from the WHO, indicate that the use of alcohol increases the risk of developing MDR-TB due to poor adherence to treatment, increased risk of adverse drug effects and impaired immune responses due to oxidative stress (Liang et al., 2015; WHO, 2014a, c, 2018).

Patients involved in this study were those who failed treatment in primary and secondary health care facilities and provisionally diagnosed and referred as MDR-TB cases. Despite efforts being made by the Tanzanian Ministry of health to improve management of TB cases in every district and regional hospital, the cure rate remains a challenge partly because medications used in the treatment of MDR-TB take a long time of at least 20 months and are associated with a wide range of side effects, which is likely to compromise adherence to MDR-TB drugs. Congruent to findings in other studies, relapse could be due to many factors including delayed diagnosis,  poor  adherence   with   treatment,   treatment default and factors related to development of drug resistance (Deiss et al., 2009; Duarte et al., 2011; Dye and Williams, 2000; World Health Organization, 2011). These factors underscore the importance of consolidated and sustained TB control programs.

This study report demographic and clinical determinants for development of MDR-TB in a Tanzanian setting. The factors associated with development of MDR-TB are HIV infection, TB relapse, history of irregular treatment, cigarette smoking, alcohol abuse, treatment failure (return) as the most important independent factors that contributes to development of MDR-TB. These determinants affect the development from non MDR-TB to MDR-TB through different mechanisms that either compromise adherence or reduce host immune mechanisms. From our findings, consolidated and integrated TB/HIV control programs, including establishing strong diagnostic and drug delivery systems, patient management systems that improve adherence (DOTS), and TB resistance surveillance systems are recommended.

The authors have not declared any conflict of interests.

The authors appreciate the Kilimanjaro Christian Medical University College (KCMUCo’s) and Kibong’oto Infectious Diseases Hospital leadership for granting permissions to conduct this study and also the statistical advice provided by Gibson Kapanda.