Tuberculosis treatment outcome and predictors in northern Ethiopian prisons: a five-year retrospective analysis

Globally, the burden of tuberculosis (TB) is higher in vulnerable populations such as prisoners and is reported to be up to 100 times higher than in the general population [1]. In Sub-Saharan African prisons, TB remains as one of the fastest growing epidemics [2, 3]. The high prevalence of human immunodeficiency virus (HIV) infection and the lack of well-organized TB diagnostic and treatment systems [2, 4] contribute to the disproportionate burden of TB in the Sub-Saharan African prisons.

Inadequate TB treatment will lead to the emergence of drug-resistant strains [5]. The prison situations are notorious for causing interruptions of TB treatment and occurrence of unfavorable outcomes [6]. In some prisons, up to 24% of the TB cases have been shown to harbor multidrug-resistant TB (MDR-TB) which makes TB control efforts very complicated [7]. In Russian and Brazilian prisons, 12% [8] and 8% [9] of the TB cases have been reported to default their treatment, respectively. In a Ugandan prison, 43% of the prisoners with TB had defaulted their treatment [10]. The TB treatment category, HIV co-infection, smoking, alcoholism, and a lack of family support have been indicated as factors affecting TB treatment success in prisons [8, 9, 11].

In Ethiopian prisons, TB treatment programs exist and are integrated within the national TB control program where the prison health staff provides treatment based on the national guidelines [12]. However, treatment outcome results and the potential factors for unsuccessful outcome are not well documented. According to a report in 2013 from the prison of North Gondar, the treatment success rate (TSR) of the prisoners ranged from 42 to 80% within the 10 years period [13]. However, this study was not comprehensive in that it did not assess the potential factors that might be affecting the treatment success. Evaluating the treatment outcome results and identifying the risk factors for the unsuccessful outcome will help to identify the gaps between the national TB treatment policy and practice in prisons and initiate evidence-based interventions. This study was designed to assess the treatment outcome of TB cases and identify risk factors for unsuccessful outcome in northern Ethiopian prisons.

In this study, the overall treatment success was found to be 94% for the prisoners that complete their treatment while in prison. However, all the prisoners transferred to other prisons or released during their treatment were not appropriately linked to health facilities and might be lost to treatment follow-up. TB treatment category, the year of treatment, and sputum non-conversion at the second-month check-up (for smear-positive cases) were statistically associated with unsuccessful treatment outcome.

The observed treatment success rate (94%) is slightly higher than the findings from similar studies in the general population of Tigray (89.2%) [18], northeast Ethiopia (90.1%) [19] and is in the range of the target level set by End TB Strategy (a TSR of ≥90%) [20]. It also remains remarkably higher than reports from Southern (85.2%) [21], Western (70.8%) [22] and Northwest (85.6%) [23] Ethiopia. More specifically, the loss to follow-up (2.5%) and death rates (1.4%) in our study are lower than the loss to follow-up and death rates recorded in the previous studies [18, 21, 22] which ranged from 3.2 to 35.5% and 3.3 to 58.8%, respectively. In all the above-mentioned studies, the TSR was calculated in a similar way with our study where the transferred out patients were not considered. The discrepancy with findings could partly be attributed to differences in settings that might account for the variation in the DOTs performance. In prisons, it easier to administer DOTs and monitor the progress as the patients are accessible to the prison health personnel [24]. In addition, we observed that the prison health personnel provided the TB drugs under strict supervision not only during the first 2 months but also during the 4 months of the continuation phase, which might result in an increased likelihood of better treatment outcome in prisons.

When compared with prison specific studies, the overall TSR in this study (94%) is higher than the findings from the prison in Gondar which ranged from 42 to 80% across the 10 years [13]. The time difference could not exclusively be the reason for this discrepancy because the TSRs in this study were still smaller than the TSR in our study during the years that overlapped with our study period. Other factors such as differences in the healthcare delivery service between the prisons of the two regions and the commitment of the prison health personnel in delivering the DOTS service might be possible reasons. Higher rates of loss to follow-up were reported elsewhere in prisons in Uganda (43.0%) [10], and Brazil (13.0%) [9]. Similarly, the 1.4% death rate observed in this study is lower than those recorded in Ugandan (5.0%) [10], South African (1.8%) [11] and Brazilian (2.0%) prisons [9]. One possible explanation for this variation could be the difference in the burden of HIV co-infection, which has been shown to be associated with unsuccessful treatment outcome [19, 25]. For example, 54% of the study participants in the South African prison [11] were co-infected with HIV [11] whereas only 11% of the TB cases had HIV co-infection in our study.

Our study demonstrates that the DOTS program is effectively functioning for prisoners that complete their TB treatment while in the study prisons. However, the absence of appropriate linkage for those prisoners transferred to other prisons or released during their treatment raises a public health concern. It is more likely for such patients to end up with a loss to follow-up or treatment failure if not death. Those patients who survived would be at high risk to develop of drug-resistant TB [10]. This also implies that the community would be at risk of being infected with a resistant strain from these patients. For the national TB program to be successful, governmental and non-governmental organizations (NGOs) should take the initiative in establishing improved linkage between the prison and public health facilities. For example, a mobile-based communication system involving the prison health personnel, the health professionals at a public health facility, and the released prisoners could be a good approach in tracking released prisoners and reducing loss to treatment follow-up [26].

In this study, as it could be expected, previously treated cases were more likely to have unsuccessful treatment outcome compared to new cases which is consistent with several previous reports [11, 18]. One of the reasons could be the high level of treatment failure and hence possible development of drug-resistant strains in retreatment cases [27]. In addition, patient-related behavior might also have contributed to the unsuccessful outcome; a study indicated patients that were already lost to follow-up previously could be reluctant and tend to interrupt their treatment again [28]. Due attention should be given to such patients to make sure that they are taking the full course of treatment and immediate referral should be done for MDR screening if they already had a treatment failure. Year of treatment was also associated with variations in treatment outcome; in general, those treated during the earlier year (September 2010–Augst 2011) were more likely to have unsuccessful outcome compared to the latest years, indicating improvements in the DOTS performance over years. This could be due to the increasing efforts of the national TB control program in introducing and implementing the TB/HIV collaborative activities in healthcare settings and prisons in recent years [29].

Unlike several previous reports [11, 19, 25], in this study, HIV co-infection was not associated with unsuccessful outcome. The difference in the sample size and burden of the co-infection rate might be possible reasons for this. In addition, the prison health personnel reported that they routinely supervise the TB treatment progress of prisoners including those co-infected with HIV which might have contributed to the improved outcome in both groups. Though it was not associated with unsuccessful outcome, body weight at initiation of anti-TB treatment (<50 Kgs) was found to be a significant predictor of death of the patients, which is in agreement with previous reports [30, 31]. The relationship between TB and malnutrition is bidirectional. Severe TB disease renders patients to be malnourished and malnourished individuals are at a high risk to develop severe TB diseases and end up with unfavorable treatment outcomes [32]. Hence, the death of patients might be attributed to the severity of the TB disease itself or due to the malnutrition and associated consequences. Closer nutritional monitoring and earlier initiation of nutrition support are important to rescue severely malnourished TB patients [32]. In smear-positive cases, sputum non-conversion at the second-month check-up was a predictor of unsuccessful outcome. This might be partly related to the drug resistance development. In 2015, the incidence of multidrug-resistant/rifampicin-resistant (MDR/RR-TB) in Ethiopia was estimated to be 2.7% among the new and 14% among retreatment TB cases [20]. In the Tigray region of Ethiopia, 55% of the presumptive MDR cases have been shown to harbor MDR strains [33]. We suggest that such prisoners should be immediately referred for DST.

Furthermore, an overview of the TB profile data showed that the majority of the patients in our study were EPTB and PTB− cases, which is consistent with the previous reports from the general population in Ethiopia [34, 35]. The exact causes for the high proportion of EPTB in Ethiopia remain unknown, however, the high potential for a wrong diagnosis (due to the poor diagnostic facility), and poor immunologic and nutritional status have been shown to be associated with high rates of EPTB and PTB− cases [36, 37]. This might suggest the need to incorporate a more accurate diagnostic test, such as the GeneXpert MTB/RIF assay, in Ethiopian prisons.

This study has some limitations mainly inherited from the retrospective design. As we relied on historical records, we were not able to add some important variables such as alcoholism, smoking, substance abuse, nutritional status, and lack of family support, which are known to be related to the variations in the TB treatment outcome [38]. Tracking the prisoners transferred between prisons was also not possible, as the information was not clearly indicated in the treatment recording protocol.

The mean treatment success rate of the prisoners in the study prisons was quite satisfactory when gauged against the target level set by the End TB Strategy. However, the lack of appropriate linkage and tracking systems for those prisoners transferred or released before their treatment completion would have a negative implication for the national TB control program as such patients might interrupt their treatment and develop drug-resistant TB. Being in a re-treatment regimen and sputum non-conversion at the second-month check-up were significantly associated with unsuccessful outcome among all forms of and smear-positive TB cases, respectively. The concerned authorities should take an urgent action to help establish improved linkages between the prison and public health facilities to prevent possible loss to follow-up of released prisoners and due attention should be given to the previously treated cases.