The high burden of tuberculosis (TB) in African prisons is a challenging phenomenon [1–3].

It has been related to several factors, including the fact that prisoners are mostly young men originating from disadvantaged neighbourhoods with a high TB incidence, dilapidated build- ings with poor inside ventilation, overcrowding, malnutrition, and, in some areas, high preva- lence of human immunodeficiency virus infection (HIV) [4].

The difficulty in ensuring continued comprehensive TB care in prison leads to delay in TB diagnosis and to discontinuity in TB treatment. The latter favours the emergence of drug-resis- tant strains, which may further disseminate due to prison promiscuity and to the lack of facili- ties to perform cultures and drug sensitivity tests. After they are released, prisoners contribute to the spread of TB (including drug-resistant TB) in the communities into which they reinte- grate [5,6]. Therefore, TB in prison is a public health concern for both the prisoners and the community, which makes it essential to develop comprehensive TB prison programs that include prevention, diagnosis, treatment and sentinel surveillance [7–9].

In Coˆte d’Ivoire, a West African country with 34 prisons harbouring more than 12,300 inmates, a TB prison program has been developed since 1999 by the national health authori- ties, in collaboration with MSF (Me´decins Sans Frontière, 1999–2005) and ESTHER (Ensem- ble pour une Solidarite´ The´rapeutique Hospitalière En Re´seau, from 2008 until today) [10]. This program includes TB case detection using the microscopic examination of acid-fast bacilli (AFB) in prisoners with TB-symptoms, first at their admission in prison and whenever they show up at the prison infirmary thereafter. Chest X-ray, TB culture and GeneXpert tests are not included in the program.

The aim of the present study was to estimate the prevalence of pulmonary TB among inmates at the Correctional and Detention Facility of Abidjan (Maison d’Arrêt et de Correction d’Abidjan, MACA), the largest prison of Coˆte d’Ivoire, 16 years after the prison TB program was set up.

Methods 

Setting

MACA was built in 1980. It was completely evacuated during the 2010–2011 civil conflicts, and then gradually re-filled afterwards. In 2015, its inmate population was approximately 4,600, for an originally scheduled capacity of 1,500 prisoners. It is organized into five areas: the administration area, the women’s detention area, the minors’ detention area, the men’s deten- tion area, and the infirmary. The women area comprises only one building, as well as the minors area. The men’s area comprises four buildings: (i) a building for remand prisoners; (ii) a short-term detention building, (iii) a long-term detention building and; (iv) a detention building for personalities. The infirmary comprises three rooms: an outpatient room, an isola- tion room for inpatients with confirmed TB, and a room for other inpatients.

Design and population

We conducted a systematic cross-sectional study at three buildings of MACA: the men’s long- term detention building, the women’s detention building, and the infirmary inpatients ward. In the latter, inpatients at the isolation room already treated for confirmed TB when the survey begun were excluded. All other prisoners from the three buildings who signed the informed consent were included. The reasons for choosing the three buildings were the following. We chose in priority the men’s long-term detention building, and excluded the men’s remand building, because we aimed at focusing on prisoners with long sentences, ie those with the lower turnover rate and the higher prison-related TB exposure. For women, we did not distin- guish between long and short sentences, because the overall number of female prisoners at MACA is low and all are in the same building. Finally, we excluded the minors area because running a survey in this area raises specific reglementary issues.

Clinical procedures

During a pre-survey phase that started in November 2014 and ended in February 2015, infor- mation and awareness meetings about the study were organized in each cell of the targeted buildings by the study investigators, in close collaboration with previously trained peer educators.

The survey then took place From March 1^st to September 30^th 2015. During this period, inmates were sequentially extracted from their cells by groups of 25 and taken to the infirmary outpatient room, at a pace of one group every two days. Each prisoner went through the fol- lowing stages.

First, they had an individual confidential interview and medical examination by one of the study physicians. At this stage, they were asked for consent for their personal data to be anony- mously recorded in the survey database, as well as for undertaking TB smear testing and HIV testing. Participants were offered the possibility to accept participating in the TB survey and refuse HIV testing.

Second, for those who accepted participation, socio-demographic, anamnestic and clinical data were recorded on a standardized form. All participants then underwent a chest X-ray and a first sputum collection, and were taken back to their cell.

Third, participants were taken back 48 hours later to the infirmary outpatient room for a second sputum collection. Those who accepted HIV testing also had a blood collection at this time.

Biology

HIV screening was performed using the rapid test Determine HIV-1/2¹ (Alere, Matsuhidai Matsudo Shi Chiba, Japan). HIV positivity confirmation was performed using a second rapid test, HIV-1/2 STAT-PAK¹ (Chembio, Medford, NY). All first and second sputum samples were microscopically examined after auramine stain- ing, and all first sputum samples were systematically put in mycobacterial culture, irrespective of patient’s clinical and radiological data. A culture was also performed on the second sputum sample for participants who met at least one of the following criteria: (i) current cough lasting more than two weeks, (ii) abnormal chest X-ray images, (iii) HIV seropositivity, and/or (iv) positive microscopy of the first sputum sample.

Mycobacterial cultures were systematically performed on Lowenstein-Jensen (LJ) solid medium (Biorad, Marnes-la-Coquette, France) and Bactec MGIT 960 liquid medicum (Bec- ton Dickinson Microbiology System, Sparks, NV, USA) [11,12]. Mycobacterial isolates were tested for drug susceptibility using Bactec MGIT 960 SIRE Kit (Becton Dickinson, Franklin Lakes, New Jersey, USA) [12]. The drugs tested were isoniazid, rifampicin, ethambutol and streptomycin. 

The interpretation of chest X-rays was performed by two independent experiment readers.

TB case definition

The diagnosis of TB was documented as confirmed (positive culture for M. tuberculosis com- plex), probable (positive microscopy with at least one acid-fast bacilli [AFB] per field, and/or chest X-ray images suggestive of TB), or possible [(i) at least four clinical signs or symptoms among the following: cough lasting for more than two weeks, fever lasting for more than two weeks, recent loss of appetite, chest pain, objectified body mass index lower than 18.5 Kg/m² or objectified fever with body temperature 2:37.5˚C on physical examination; or (ii) at least three clinical signs or symptoms and HIV seropositivity or positive microscopy with less than one AFB per field].

A mycobacterial strain was defined as pan-susceptible if it was susceptible to isoniazid, rifampicin, ethambutol and streptomycin and as multidrug-resistant if it was resistant to at least isoniazid and rifampicin.

All patients documented with TB received a TB treatment as per the National Tuberculosis Control Program guidelines [13,14].

Statistical analysis

The association between confirmed TB and other variables was analysed using univariate and multivariate logistic regression. All variables associated with TB with a significance level S 0.25 in univariate analysis were included in multivariate analysis. Statistical analyses were carried out using the SAS 9.3 software. Study questionnaire and data used for the analysis are available on Supporting Information (S1 Questionnaire and S1 Dataset).

Ethics

This study was approved by the National Ethics Committee for Health and Life Sciences, the National Tuberculosis Control Program and the Ministry of Health of Coˆte d’Ivoire. Signed informed consent was obtained from all participants. The interviewers informed the inmates that refusal to participate would impact neither their relationship with the prison team nor their medical care.

Each participant was assigned an anonymous number. A participant register was used to ensure correspondence between the identification data (surname, forenames and date of birth) and the anonymous number of each participant. Only the study physician and coordina- tor had access to this register.

n = number; % = percentage; SD = Standard Deviation; IQR = Interquartile Range

https://doi.org/10.1371/journal.pone.0181995.t001

Table 2. Symptoms, signs and test results.

n = number; % = percentage; Kg/m² = Kilogram per square meter

*excavations (n = 2), alveolar opacities in the apex (n = 11), alveolar opacities in other settings (n = 25) and pleural effusions (n = 5). Missing values = 13 for Chest X-ray and 39 for Microscopy and culture.

https://doi.org/10.1371/journal.pone.0181995.t002

Tests (Table 2)

Chest X-rays were performed in 930 participants (99%). Forty-six (5%) were deemed not inter- pretable, all of them from inmates at the men’s building. Of the remaining 884, 43 (5%) were considered suggestive of TB, on the basis of excavations (n = 2), alveolar opacities in the apex (n = 11), alveolar opacities in other settings (n = 25) and pleural effusions (n = 5). Of note, 13 out of these 43 participants had declared a past history of pulmonary TB.

Table 3. Tuberculosis cases.

n = number; % = percentage

(1) Confirmed TB: positive culture for M. tuberculosis complex

(2) Probable TB: negative culture or culture unavailable, positive microscopy with at least one acid-fast bacillus [AFB] per field, and/or chest X-ray images highly suggestive of TB

(3) Possible: TB (i) at least four clinical signs or symptoms suggestive of TB, or (ii) at least three clinical signs or symptoms suggestive of TB and HIV

seropositive or positive microscopy with less than one AFB per field

https://doi.org/10.1371/journal.pone.0181995.t003

A total of 1821 sputum specimens were collected in 904 participants (96%). Of these 1821 specimens, 1272 had AFB testing and 1251 were cultured. Five hundred and thirty-six inmates (59%) had one AFB testing, 368 (41%) had two AFB testings, 551 (61%) had one sputum cul- ture, and 350 (39%) had two cultures. Overall, 19 inmates had at least one positive culture (including 8 with positive AFB smear), and 119 had at least one positive AFB smear with a nega- tive culture or a culture non available. Of the latter, 115 (97%) had less than one AFB per field.

TB cases (Tables 3 and 4)

Overall, 88 (9.3%) participants met the TB case definition, including 19 (2.0%) with confirmed TB, 40 (4.2%) with probable TB and 29 (3.1%) with possible TB. Of the 88 inmates found with TB, 84% had been in prison more than 12 months. Of the 19 cases of confirmed TB, 18 were found in the men’s building and one among the infirmary inpatients. Three out of these 19 patients had none of the sign or symptoms listed in Table 2. Seven out of these 19 patients had X-Ray images suggestive of TB.

Table 4. Tuberculosis documentation.

AFB: acid fast bacilli TB: tuberculosis

* among the following ones: cough lasting for more than two weeks, fever lasting for more than two weeks, recent loss of appetite, chest pain, objectified body mass index lower than 18.5 Kg/m²

https://doi.org/10.1371/journal.pone.0181995.t004

Table 5. Factors associated with confirmed tuberculosis.