Background
Globally, the prevalence of tuberculosis (TB) among prisoners is greater than the general population [
1]. Although the burden is well noticed, prisons are often overlooked by national health sectors and are not included in the national health statistics [
2]. Living in congregate settings where both TB patients and uninfected inmates frequently crowded increases the risks of contracting infection and developing multi-drug resistant TB [
3,
4].
Although prisons have healthcare centers that provide diagnosis and treatment of TB for both prisoners and prison staffs in Ethiopia, the services are provided through referral systems to outside healthcare centers [
5‐
8]. The infection control practices in the receiving healthcare centers is usually poor. A study conducted on the practice of TB infection control among healthcare workers in Ethiopia showed that only 38% of them practiced the control plans properly [
9]. Laboratories are often inadequate and delays both screening and referral systems [
10]. These all worsens the rate of transmission and become important reservoir of TB infection for inmates and general population.
Different factors might be attributable for the higher prevalence of TB in prisons. A study conducted in eastern Ethiopian prisoners showed that the knowledge of prisoners on the cause of TB was poor. Only 1.6% of the prisoners knew the causes of TB [
11]. Other studies conducted in prisoners in Ethiopia reported that the risks of developing tuberculosis were associated with undernutrition [
12,
13], illiteracy [
14], smoking [
12,
13,
15], increased duration of imprisonment [
6,
13], overcrowding and poor ventilation [
13,
16], reproductive age (15–44 years) and urban residence before imprisonment [
17], contact history with TB patients [
8,
12,
14,
17,
18] and previous TB infection [
15].
Systematic screening of contacts and high-risk groups is one of the pillars of the global End TB strategy [
19]; but the health services of prisons are often overlooked and underfunded creating opportunities for prisons to receive, concentrate and disseminate TB within and to outside population [
2,
20]. Despite the integration of TB care in prisons in the national TB prevention and control programs of Ethiopia, there is no systematic screening of detainees on admission for potential infectiousness to prevent transmission, disability and death [
4]. Therefore, prisoners infected and contagious for TB might be added to crowded cells and became sources of infection [
20].
Results from Ethiopian prisons showed a varying prevalence rates; ranging from 1.8 to 19.4% [
7,
15] and diagnoses were highly dependent on microscopy. Nevertheless, through strong screening and use of sensitive laboratory tests, the prevalence could be potentially higher than the reported one. Lack of national summarized data could be one of the reasons for the poor implementation of TB prevention and control programs in prisons. Except individual studies with varying prevalence rates, there are no national prevalence studies conducted in prison settings in Ethiopia. Appropriate estimates of the disease is essential to formulate health service plans most fitted for prisons settings. Therefore, this systematic review and meta-analysis was done to estimate the pooled prevalence of TB for better understanding of the burden of TB in Ethiopian prisons.
Discussion
The majority of studies included in this analysis reported a high TB prevalence rates in prisoners in Ethiopia. Although the ten studies contained information on the prevalence of TB, none of the studies provided information on the prevalence of latent TB in prisoners. The included studies used various inclusion criteria, diagnostic assays and sampled only suspected cases or did not include on treatment patients. Most of the studies were conducted in the Amhara and SNNP regions with no information about the burden in the capital (Addis Ababa), Afar, Gambella and Benishangul Gumuz regions.
TB in prisons is often overlooked and continued to be a major public health problem in many settings, particularly in countries with a high incidence of TB, including Ethiopia [
2]. To the best of our information, this is the first meta-analysis conducted in Ethiopia to determine the pooled prevalence of TB among prisoners. Accordingly, the pooled prevalence estimate of TB among prisoners in Ethiopia was 8.33% (95% CI; 6.28, 10.63,
p < 0.001). This pooled prevalence is comparable with reports from South Africa prison with a prevalence rate of 8.8% [
24].
The pooled prevalence result, on the other hand, was less than prevalence reports from Zambian prisons (22.7%) [
25] and Democratic Republic of Congo (17.7%) [
26]. This might be due to the laboratory tests used to diagnose TB. In Ethiopia, TB diagnosis relies on microscopy which has low detection rate and 50% of the included studies used microscopy as a confirmation of TB. In addition, screening was solely dependent on symptoms of cough and none of the included studies used chest X-ray as a screening and diagnosis of TB. Moreover, screening during admissions to prisons was not done for which it could increase the prevalence by participating larger numbers of prisoners. A study from Tigray [
8] also reported that half of pulmonary TB cases were left undiagnosed in prisons which supports the evidence of poor case detection rate in prisons.
Our findings are relatively higher than reports from similar settings in Brazil (4.5%) [
27], Malawi (1.4%, 5.1%) [
28,
29] and Cameroon (3.3%) [
30]. A previous series report on HIV and related infections in prisoners also reported a pooled prevalence of 5.3% in Eastern and Southern Africa (where Ethiopia is located) [
31], relatively lower than our finding. The relative higher pooled prevalence in our study could be attributed for the small size of prison population participated in the study which is pooled using only 4086 prisoners from the ten studies.
The pooled point prevalence was also estimated from nine studies. From the analysis, the point prevalence was found to be 888 per 100,000 (95% CI; 531–1333). Although in agreement with reports from Brazil prisons with a point prevalence of 917/100,000 [
27], it was three times higher than the point prevalence of the general population in Ethiopia (277/100,000) [
32]. Recent reports showed a national prevalence of 200/100,000 [
33], four times lower than our finding. On the other hand, lower prevalences were reported from prison settings; 696/100,000 in Malawi [
28], 341/100,000 in Turkey [
34] and 215/100,000 in France [
35]. The lower prevalences in these countries could be attributed for strong TB control strategies, low incidence rates and established good health systems both in prisons and general population. Other studies from prison settings in South Africa [
24], Zambia [
25], and Cameroon [
36] reported much higher point prevalences; 8772, 4005 and 3197; respectively).
In addition, subgroup analysis was done from quality effects model to see the differences in prevalence rates by type of diagnostic tests used. Accordingly, significant difference was observed across diagnostic tests. The pooled prevalence of TB using microscopy alone was 6.59% (95% CI: 3.96–9.50; I2 = 77%; p < 0.001) while the pooled prevalence of TB using microscopy combined with culture or GeneXpert was 8.57% (95% CI: 4.94–12.60; I2 = 87%; p < 0.001). This is clearly known that microscopy combined with sensitive tests including culture and molecular tests improves the detection rate of TB.
In meta-analyzing prevalence studies, studies with smaller or larger prevalence reports could affect the pooled result by giving wider confidence intervals and variance instability. The sensitivity analysis of our study showed that two smaller studies (Fuge et al. 2016 & Zerihun et al. 2015) having smallest and largest prevalence rates, respectively, were the prime determinants of the pooled result. We, therefore, combined them with other studies in to a weighted average to minimize the variance; so that larger studies and studies with lesser variation could have greater weight and vice versa in the final combined estimate by using the random effects model (see % weight column in Fig.
3).
This study has certain limitations. Included studies were cross-sectional and limited in number (only ten studies) providing only snapshots of the situation at a particular moment in time and fail to capture the dynamic nature of the prison population. Diagnostic methods of TB were also varied between studies which could affect the pooled results. Moreover, lack of information and data from some regions including the capital Addis Ababa, made it difficult to generalize the findings. In addition, this study was based only on published peer-reviewed studies and important data might be missed from unpublished studies and grey publications.
Conclusions
The pooled prevalence of tuberculosis among prisoners in Ethiopia is expectedly high. This high prevalence could explain the spread of TB within prisons and between prisoners and varies communities. Thus; attention should be given to prevent the transmission and emergence of drug resistance TB both in inmates and general population. Moreover, no evidence was found on the situation of latent TB and evidences on the effect of HIV on the occurrence of TB were conflicting. Further studies covering large scale prison population are needed to design effective diagnostic, treatment and preventive methods. Strengthening prison healthcare centers with manpower and infrastructure, screening during admission, periodic screening for TB symptoms and active case finding, trainings for prisoners and prison staffs on TB infection prevention, supporting diagnosis with CXR and culture and/or molecular tests, screening for latent infections and prompt treatments can be used as an immediate response to curve the burden and to make prisons not further be reservoirs.