Alcohol use disorder in individuals with tuberculosis is an important driver for poor tuberculosis treatment outcomes (16). In comparison to tuberculosis patients with no alcohol use disorder, those who have this problem, are faced higher rates of treatment failure, relapse, and death. Despite this and other impacts that AUD poses on individuals with tuberculosis, to the knowledge of researchers of the present study; there is no aggregate evidence on the average prevalence of AUD among this target population. The present meta-analysis study, therefore, intended to narrow the gap in evidence in this area by supplementing solid evidence on alcohol use disorder and its associated factors in TB patients. The evidence obtained will be of paramount importance for public health practitioners and policymakers.
We retrieved a total of 1965 titles that were screened carefully at multiple stages to provide twenty-seven studies (11, 14, 15, 17, 25, 33–54) that assessed AUD in 30654 tuberculosis patients. A difference in regional variation with difference in prevalence of AUD was reported to be from 11.2–62.5% in Russia (17, 25, 35, 37, 38, 48), from 23.3 to 63% in South Africa (41, 46, 47, 49, 50), from 4.4–18.8% in Ethiopia (33, 34, 51, 52, 54). The remaining studies were from United States (US) (36, 39), Estonia (42, 43), India (14, 44, 53), Thailand (45), Nigeria (40), Botswana (15) and Zambia (11).
Therefore it was necessary to have an average estimate for the prevalence of AUD in the global context and the current meta-analysis was therefore rooted in this justification. The average prevalence of alcohol use disorder among tuberculosis patients using the random effect model was found to be 30% (95% CI: 24.00, 35.00). This result was consistent with the global average prevalence of AUD among individuals living with HIV/AIDS (29.80%) (55).
However, the present finding was higher when compared with the average prevalence of AUD in individuals living with HIV/AIDS in Africa (22%) (56). It was also higher than the DSM-V 12 month prevalence of alcohol use disorder in the adult general population in the USA (13.9%)(57). Moreover, the finding was higher than the average prevalence of AUD in the European, Australian, and Ethiopian general population in which the AUD prevalence was 11.1% (58), 11.8%(59), and 23.86%(60). This could be the use of alcohol as a coping response for the psychological distress associated with the perceived severity of such life-threatening illness (58, 59).
On the contrary, the average prevalence of AUD in the present study was lower when compared with the prevalence of AUD in mental disorders (28–70%) (61). Individuals with mental illness are most of the time in poor judgment and insight towards their illness which could be responsible for the higher prevalence of AUD.
The average prevalence of AUD in male participants as reported by a few of the studies was 33.6% and higher than the average prevalence of AUD in females (11.67%). This was consistent with earlier studies in Canada(62), the East African countries (63), and the United Kingdom (64). The sociocultural expectations and influences between males and females could be responsible for this. Besides, differences in the neurochemistry of the brain between men and women like the higher release of dopamine in men than women with the same amount of alcohol intake could lead to the high level of AUD in men(65). However, the exact justification for such differences is the recommendation for future researchers.
The average prevalence of AUD was with a slight heterogeneity (I2 = 57%, p-value = 0.000) from the difference between the twenty-seven studies. For this reason, we did a sub-group analysis. Therefore we did a subgroup analysis and the average prevalence of AUD varied based on the continent of the study, the measurement tool for AUD, the type of study design, and the mean age of the participants.
The subgroup analysis based on the continent where the study was done showed a significant difference in the average prevalence of alcohol use disorder among tuberculosis patients. The average prevalence of AUD in tuberculosis patients was higher in Asia and Europe;37% (14, 17, 25, 35–39, 42–45, 48, 53) than the prevalence in US ; 24% (36, 39) and Africa ; 24% (11, 15, 33, 34, 40, 41, 46, 47, 49–52, 54). This was supported by earlier studies (66). Differences in the cultural context, variation in the availability of alcoholic drinks and socio-economic variants could bring the variation. Furthermore, the difference in the number of articles included in the subgroup could also be responsible.
The average prevalence of AUD was 36% in studies that do not report the assessment tool for AUD (35, 36, 38, 41–44, 51) higher than the prevalence in studies that utilized AUDIT (26%) (14, 15, 17, 25, 34, 39, 45–50, 52–54). This could happen due to the possibility of inclusion of mild levels of alcohol use and the overestimation of AUD in studies that did not report the measurement tool.
Besides, case control studies (15, 40, 42) provided higher prevalence of AUD (39%) than cross-sectional (11, 14, 17, 33, 34, 36, 37, 42, 44, 46, 50, 51)(30%), cohort (25, 35, 38, 41, 45, 47, 49, 52–54)(30%). The small number of studies included in the case control subgroup might affect the validity of the estimate and result in higher prevalence of AUD.
Finally, the mean age of the study participants included in the study was considered during the subgroup analysis and the average prevalence of AUD was 42% in studies with a mean age of the participants 40 years and above which is higher than the average prevalence of AUD in participants with a mean age of < 40 years (24%) and mean age not reported (27%). This was however in contradiction with the study finding by grant et al (67) in which the prevalence of AUD declines over the age of 40 years.
Regarding the factors associated with AUD, our qualitative synthesis showed that the socio-demographic factors such as male gender (14, 25, 36, 39, 49), age older than 35 years(14), being single, divorced or widowed (11, 14), being unemployed (11), being black American (36), colored ethnicity (49), low level of education (14, 49), no educational background (39), low level of income (< Rs 5000 per month) (14) and poverty (49) were related to AUD. Also, being on category-II tuberculosis treatment(relapse and treatment failure)(14), TB retreatment patient status and non-adherence to anti-TB medication(49), patients with chronic/relapsing form of tuberculosis (37), patients with perceived TB stigma(39), patients who feel ashamed of having TB(39), people close to you would avoid you because of TB(39), HIV-co-infection and low HIV CD4-count(39), having cavitary lesions on chest radiographic examination(36), and smear-positive and culture-positive types of TB(36) were also the reported clinical and tuberculosis related factors for AUD.
Difference between included studies in the meta-analysis
Due to the slight heterogeneity existed in the present meta-analysis brought by the variance between the twenty-seven studies; we did a subgroup analysis. The result from subgroup analysis showed that the measurement tool employed to screen AUD, the continent where the study was done, the mean age of the participants studied, and type of the study design were identified as sources of difference between the 27 included studies. Furthermore, a single study leaves out analysis was done to screen studies outweighing the overall result but the average prevalence of AUD was not outweighed by a single particular study. This study is the first of its type to assess the average prevalence of alcohol use disorder in tuberculosis patients. The use of a pre-determined search strategy to obtain eligible articles minimizes the reviewer's bias which increases the study quality. Besides, the implementation of subgroup analysis based on the measurement tool, the continent of the study, study setting, and mean age to identify the source of heterogeneity is also the strength of the present study. However, the use of a few studies in some groups of the subgroup analysis might affect the validity of estimate so that under or overestimation could occur. Moreover, the exclusion of articles published in non-English language might have also an effect on the magnitude of the average prevalence of AUD.