Factors associated with tuberculosis in persons deprived of liberty in Espírito Santo

ABSTRACT OBJECTIVES To calculate the rate of tuberculosis cases per prison unit in Espírito Santo; present the individual, clinical, and institutional characteristics of the cases in persons deprived of liberty (PPL); and analyze the association between these characteristics and treatment outcome in this population. METHODS The study included cases of tuberculosis in the PPL of Espírito Santo from 2014 to 2016. Rate calculation, descriptive analysis and hierarchical logistic regression were performed considering the individual, clinical and institutional levels. RESULTS The rate of diagnosed cases per prison unit in the state ranged from 0 to 17.3 cases per 1,000 inmates. Of all reported cases, 218 (72.6%) healed, 21 (7.0%) dropped out, 1 (0.3%) died of tuberculosis, 2 (0.7%) died from other causes, 56 (18.7%) transferred the treatment site and 2 (0.7%) developed drug-resistant tuberculosis. The adjusted analysis showed that supervised treatment ensures success (CR = 0.29; 95%CI 0.01–0.76). CONCLUSIONS The study highlighted the importance of knowing the TB treatment outcome in the PPL to implement measures to reduce failure, and the contribution of supervised treatment in this process.


INTRODUCTION
The prison environment favors disease outbreak in persons deprived of liberty (PPL)a consequence of confinement and lack of health services access. Besides the diseases already observed in the general population, dermatoses, mental illness, trauma, infectious diarrhea, STDs, and respiratory diseases such as pneumonia and tuberculosis (TB) are often prevalent in PPL 1 .
Monitoring TB indicators in the PPL of Brazil showed an increase in the incidence rate from 627.6 cases to 904.9 cases per 100,000 prisoners from 2007 to 2013, while mortality reached 18.0 deaths per 100,000 prisoners in 2007 and 16.0 deaths per 100,000 prisoners in 2013 2 . Some factors effectively contribute to the PPL high endemicity of tuberculosis. They may be related to individuals and their living conditions before incarcerationbeing young, male and with low education; using drugs; having associated injuries; exhibiting previous treatment for TB and having a history of incarceration; or to incarceration -overpopulation, poorly ventilated and poorly lit cells, frequent exposure to the bacillus in a confined environment, lack of information and lack of health services access in prison 3 . Thus, besides quantifying TB cases in PPL, we must investigate which factors permeate this context and interfere in the dynamics of disease outbreak and treatment. This study aimed to calculate the rate of tuberculosis cases per prison unit in Espírito Santo (ES) from 2014 to 2016; to present the individual, clinical, and institutional characteristics of the cases; and to analyze the association between these characteristics and the interruption of tuberculosis treatment in this population.

METHODS
The sample of this study comprised the tuberculosis cases registered in persons deprived of liberty in the State of Espírito Santo from 2014 to 2016, including new cases, relapses, readmission after dropping out, transfer or unknown entry method ("does not know"). Espírito Santo has 34 prison units, which in 2014 housed 16,234 prisoners, accounting for an imprisonment rate of 417.9 PPL per 100,000 inhabitants, higher than the country's rate-299.7 per 100,000 5 . These units refer to provisional, closed, semi-open and mixed prison systems, excluding police stations or the like.
The study variables were categorized into hierarchical levels based on the models proposed by Maciel 6 and the literature on the subject. Thus, the theoretical model used consisted of the levels and respective variables ( Figure 1).
The variable studied was the tuberculosis treatment outcome, classified by the Programa Nacional de Controle da Tuberculose (National Tuberculosis Control Program) as: cure, dropping out, death from tuberculosis, death from other causes, transfer, multidrug-resistant tuberculosis, treatment change, failure, and primary default. In the hierarchical logistic model, this variable was reclassified as success (cure) and failure (other categories).
Data were collected from the Sistema de Informação de Agravos de Notificação (SINAN -Health Information System) and the Levantamento Nacional de Informações Penitenciárias 5 (Infopen -National Penitentiary Information Survey), as well as databases and reports from the State health and justice departments, stored in a single database for statistical analysis.
Statistical analysis estimated the TB diagnosis rate per prison unit in Espírito Santo, considering only the cases diagnosed during incarceration, and excluding transfers and unknown entry method ("does not know"). Then a descriptive analysis was performed according to the variables ranked by level and categorized per TB treatment outcome.
Logistic regression estimated the gross and adjusted odds ratio (OR) and respective 95% confidence intervals (95%CI). Multivariate analysis used the hierarchical model defined according to the established levels. Initially, the variables associated with the interest outcome (p ≤ 0.20) were included in the model, to account for possible confounding variables. In the modeling stage, independent variables were included from the distal to the proximal level. Variables with p < 0.20 were kept at each hierarchical level and in the final model for adjustment. In the final model, a significance level of 5% was adopted to determine the variables associated with treatment failure.
Hierarchical models are used when distal factors influence intermediate factors, which, in turn, influence more proximal factors, directly impacting the outcome. Thus, this model allows including variables in their sequential order, even with dependency on observations 6 . Statistical analyses were performed using the R programming, version 3.4.1.

RESULTS
The rate of tuberculosis cases diagnosed in Espírito Santo's prison units from 2014 to 2016 ranged from 0 to 17.3 cases per 1,000 prisoners. Units with no TB cases were mostly from the countryside and had an occupancy rate below 100%; the Centro de Triagem de  We evaluated these cases for their individual, clinical and institutional characteristics ( Table 1) and categorized them according to treatment outcome (success and failure).
The cases were mostly male, young (18 to 29 years), mixed race, with five to eight years of schooling. Women and individuals aged 30 to 39 years, with one to four years of schooling, showed more treatment failure.
Among comorbidities and associated conditions, the most prevalent was alcoholism, reported in 24% of cases, followed by Aids (6.0%), diabetes (2.0%) and mental illness (1.2%). Results show that most treatment failures occurred in cases with Aids and mental illness. The percentage of cure (success) was similar between alcohol users (72.2%) and non-users (74.2%).
The most observed types of admission were new case (71.7%) and pulmonary tuberculosis (88.3%). Cases of extrapulmonary TB showed less treatment success when compared to pulmonary TB. Most X-ray examinations were suspected, while sputum smear and sputum culture tests were positive. Those treated with medication showed a success rate above 80%, approaching 60% in cases without the supervised dose.
Regarding institutional aspects, most cases occurred in provisional units, managed by the state, with an occupancy rate between 100.00 and 199.99%. The infrastructure assessment showed the prevalence of cases in institutions built for incarceration, with an expanded health (74.7%), education (70.0%) and visit (46.0%) structure. We also highlight the presence of a differentiated expanded health team, comprising a nursing team, general practitioner, dentist, social worker and psychologist or occupational therapist, or psychiatrist, or physiotherapist for 63.0% of cases. The work rate was 0.10 to 14.99% in the prison units where 65.6% of TB cases were diagnosed, and the most prevalent education rate was 15.00 to 29.99%, representing 32.3% of them. Table 2 details these data by treatment outcome (success and failure). Table 3 shows the characteristics that remained in the final model. The adjusted analysis showed that supervised treatment for tuberculosis prevents treatment failure (CR = 0.29; 95%CI 0.01-0.76).  In the period studied, the CTV had the highest rate of TB diagnosed cases, with 17.3 per 1,000 persons deprived of liberty. The CTV is singular if compared with other institutions, working as a gateway to the ES prison system and therefore has a high turnover, with an average of 690 entries each month 7 . The established flow foresees that these newcomers remain for a maximum period of 48 hours in the unit, awaiting transfer according to profile and legal situation, identified by an initial registration. However, this deadline is not met, with prisoners remaining for up to 30 days 8 .
According to the Protocolo de Controle de Tuberculose Pulmonar da População Prisional (Protocol for Controlling Pulmonary Tuberculosis in the Prison Population) of Espírito Santo based on the Ministry of Health guidelines, an active search for respiratory symptoms, using a specific screening form, must be performed when entering the prison system 9 -which may be responsible for the greater number of TB cases identified in the CTV compared with the other units. Once transferred, these individuals already have an initial screening for TB. However, although the diagnosis occurs during incarceration, we must reflect on the transmission mechanism of the TB bacillus and its incubation time. From our data, it is unclear if the prison unit was the place of TB contagion, since the PPL could already be infected, progressing to the active disease during this period. Although not the source, the prison environment would have contributed to the disease evolution 3 . Genetic studies could provide a better understanding on the dynamics of M. tuberculosis transmission, enlightening this issue. Knowledge of how the microorganism is dispersed in the prison and general population aids in disease prevention and control strategies 10 . According to the Ministry of Health, comparative data on TB treatment outcome in 2015 showed cure, dropping out, and transfer rates of 73.2%, 7.7% and 5.5% for PPL, while in the general population they reached 71.0%, 9.8% and 5.0%, respectively 12 . Our data on cure (72.6%) and dropping out (7.0%) rates were similar to this study, but the transfer rate (18.7%) was superior, probably due to the logistics adopted by the ES Department of Justice, which elected one prison unit as the gateway to the prison system, from which the allocation is made.

Tuberculosis cases recorded in ES
Frequent transfers between units influence follow-up loss; thus, communication between safety and health professionals must be agile and constant, aiming to minimize treatment drop out and ensuring that the medical records accompany the patient when changing prisons. The Ministry of Health states that mobility in the prison system increases the risk of TB infection, given the circulation between different penal institutions, health services and the general population, during and after incarceration 13 . Although the rate of incarcerated women increased by 10 Performing directly observed treatment (DOT) was associated with treatment success, reaffirming the beneficial effect of DOT shown in other studies 19 . One research showed that supervised treatment reduces unfavorable tuberculosis treatment outcomes by 25%, and that patients who did not receive DOT were more likely to abandon treatment (OR = 0.62; 95%CI 0.57-0.66), die from TB (OR = 0.68; 95%CI 0.61-0.77) and have unknown treatment results (OR = 0.71; 95%CI 0.66-0.76) 20 . The Ministry of Health recommends directly observed treatment for PPL, as it strengthens the bond with the health professional, guarantees quick access to health service regarding adverse effects, prevents using medication as an element of exchange and pressure, offers to exchange information and establish care, among others 13 .
The ministry also lists as main obstacles to implementing control strategies in the prison environment: the lack of knowledge about TB by PPL and security professionals; undervaluation of TB symptoms by PPL, hindering early diagnosis; difficulty of PPL access to health services, and its scarcity of human and financial resources; low adherence of PPL to treatment and prevention measures; and the stigmatization and segregation created by TB, considering the importance of protection by belonging to a group 13 .
Despite the challenges, the study showed that we must examine the components influencing the outcome of tuberculosis treatment in PPL, to implement measures that provide cure and reduce failure rates. Knowing what leads to drop out contributes to interrupt the chain of transmission of the disease and its prevention, not only among the PPL, but between family members and workers who circulate in the prison environment.