Introduction: Finding and successfully treating all tuberculosis (TB) patients is the cornerstone of the Global Strategy to Stop TB. However, many patients in resource-limited countries remain undiagnosed. Prisons are a well-known source of undetected TB. Thus, there is a need to find feasible interventions to find and treat TB patients in these settings.

Objective: The objective of this study is to evaluate whether empowering and involving inmate peer educators in TB control has an impact on increasing TB case detection rate and improving treatment success in resource-limited prison settings.

Methodology: This is a matched cluster randomized control trial where randomization to the intervention and treatment groups will be carried out within pairs. Eight matched prison pairs will be randomly selected for this study in which eight prisons from each pair will be randomly assigned to the intervention and the remaining to the control group. Trained prison peer educators at the intervention sites will organize and provide education about TB every two weeks on a regular basis for one year. Peer educators will also perform routine TB screening, using a screening protocol to identify presumptive TB cases for a referral. Identified presumptive TB cases will then be linked to the prison health personnel for a referral to nearby hospitals. The TB diagnosis will be carried out at the referral sites using the routine direct smear microscopy and/or chest X-ray (Radiography). Tuberculosis case finding in the control sites will follow the existing referral system (self-referral to nearby hospitals) and the diagnosis will be undertaken using direct sputum microscopy and/or chest X-ray as in the intervention sites. The data will be entered using Epi Data entry version 3.1 software and analyzed using SPSS version 20.0. Considering prisons as a unit of analysis, the mean Case Detection Rate (CDR), Treatment Success Rate (TSR) and the percentage of patients symptomatic for > =3 months will be compared within pairs using the paired t-test or sign test as appropriate.

1. Background and justification

   Although the global incidence of tuberculosis (TB) has fallen by an average rate of 1.5% per year between the years 2000 and 2013, TB remains a major global health problem and is the second leading cause of death from infectious diseases [1,2]. An estimated 9 million people had developed TB in the year 2013 of which one-quarter was from Africa, the continent that also has the highest rates of cases and deaths relative to its population [1,3]. In the same year, only 5.7 million cases were both detected and notified to the National TB programs (NTPs), meaning that about 3 million people were "missed" by TB control programs [1]. The TB case detection rate (calculated as notifications of new and relapse cases divided by the estimated incidence per year) was only 64% at the global level and is below the World Health Organization (WHO) target (70%) [1].

   Tuberculosis disproportionately affects vulnerable populations such as prisoners. According to a review of studies from both high and low-income countries, the prevalence of TB in prisons was 3 to 200-fold higher than in the general population [4,5]. A number of factors such as overcrowding, inadequate ventilation, delayed case detection and the movement of prisoners from site to the site make prisons a high-risk environment for TB transmission [6]. The emergence of drug-resistant TB has also made TB control efforts very complicated in prisons. [7]. If untreated, prisoners with drug-resistant TB will be an additional health hazard for fellow prisoners and staff [8]. Furthermore, TB and resistant forms of TB can spill out into the community and create a public health hazard [9]. In sub-Saharan African prisons, the high burden of human immunodeficiency virus (HIV) infection and poverty make the problem much worse [10]. Health care services are also very limited and due to the lack of diagnostic facilities many TB patients remain undetected [11]. In a recent study from a South African prison, for example, the prevalence of undiagnosed TB was reported to be high at 3.5% [12].

   Ethiopia, which ranks 8th of the 22 countries with a high burden of TB [13], has implemented the directly observed treatment short-course (DOTS) program since 1992 [14]. Currently, the country has achieved the Millennium Development Goals (MDGs) of halving TB incidence, prevalence, and mortality by 2015 [1]. However, the case detection rate (CDR) remains low as in many other developing countries.

   The case detection rate of all forms of TB was reported to be 62% which is still below the WHO target [1]. Ethiopia is listed among the top ten countries that accounted for 74% (2.4% million) of missed TB cases [1]. Shortage of health workers, low service coverage, low health seeking behavior and lack of efficient TB diagnostics are some of the factors for the low case detection rate in the country [15].

   Tuberculosis detection is a huge challenge for Ethiopian prisons [7,16]. Because of lack of skilled health professionals and laboratory facilities, TB diagnosis relies on a referral of prisoners to health facilities outside prisons. As prisoners often do not recognize the symptoms of TB or use non-TB medicines to relieve symptoms, TB diagnosis is often severely late [17,18]. As a consequence, many prisoners with TB in the Ethiopian prisons remain undiagnosed [19]. Recently, we reported that at least half of symptomatic pulmonary TB cases were left undiagnosed and hence untreated in Northern Ethiopian prisons [20]. Some of these patients remained undiagnosed for more than two years [20]. This creates an opportunity for the transmission and dissemination of TB not only among prisoners but also to the general community. In addition, although accurate data is not available on TB treatment adherence, the conditions in Ethiopian prisons make the TB treatment supervision difficult and hence adherence to the full course of anti-TB drugs remains a challenge to TB patients [20]. This will lead to the emergence of drug-resistant strains, an important challenge for TB control. Thus, efforts are needed to improve TB case detection and treatment adherence in Ethiopian prisons. Epidemiological models show that employing active case finding through entry and exit screening, and frequent mass screening, using a combination of diagnostic tools (usually sputum smear microscopy and chest-X-ray), improves TB case detection in prisons [21]. However, resource constraints and lack of skilled health professionals limit its applicability in prisons of poor countries. Increasing the health service coverage and training health workers to fill the gap in short period seems a remote possibility. Therefore, there is a need to find alternative ways that could increase access to diagnostic and treatment services, ultimately increasing TB case detection and treatment success rate. In the current study, the investigators plan to undertake a cluster randomized controlled trial to assess whether empowering and involving inmate peer educators in the TB control improves TB case detection, treatment success rate, and pre-treatment symptom duration in a resource-limited prison setting.

2. Hypotheses

   Hypothesis 1: Case detection rate: Compared to the control sites, the TB case detection rate in the intervention sites will be increased by 50%.

   Hypothesis 2: Treatment success rate: Compared to the control sites, treatment success rate in the intervention sites will be increased by 10%.

   Hypothesis 3: Pre-treatment symptom duration: Compared to the control sites, the proportion of patients with pre-treatment symptom duration of >=3 months will be reduced by 30% in the intervention sites.

3. Methods 3.1. Study area and study population

   This study will be conducted in Northern Ethiopian prisons. Ethiopia had an officially registered prison population of 112,361 (136/100,000 persons) in 2010 [22], which is higher than the imprisonment rates observed in some sub-Saharan African countries such as in Kenya (121/100,000 persons), and Malawi (76/100,000 persons). The fact that prisons are overcrowded and inadequately ventilated means that Ethiopian prisons are ideal for the transmission of Mycobacterium tuberculosis [23] and urge appropriate interventions. This interventional study will be conducted in 16 main prisons (8 matched pairs) located in the Amhara and Tigray regions including Mekelle vs. Shire, Adawa vs. Tembien, Humera vs. Adigrat, Mychew vs. Alamata, Wukro vs Axum, Dessie vs. Woldya, Fenoteselam vs. Debremarkos, and Debretabor vs. Bahir Dar for a one year period from March 2016 to March 2017. This interventional study focuses only in prisons of the Northern Ethiopia considering feasibility. All the prisoners in the 16 selected prisons are study populations.

   3.2. Study design and inclusion criteria

   This is a matched cluster randomized controlled trial where randomization to the intervention and treatment groups will be carried out within pairs. Prisons will be matched into pairs on the basis of their location and/or the number of prisoners. Larger prison sites located in the main cities of Amhara and Tigray regions will be considered as eligible while small jails will be excluded. Accordingly, there are a total of 22 main prisons (11 matched pairs) eligible for the study in which the five matched pairs are in Tigray and the remaining in the Amhara region. All the five matched prison pairs from Tigray will be included in the randomization considering their proximity and hence feasibility. However, the investigators will undertake a random selection and will include three matched pairs from the Amhara region considering the feasibility and financial constraints.

   3.3. Randomization

   Once the required number of matched prisons pairs was fixed, the randomization was done within pairs in which eight prisons from each pair were randomly assigned to the intervention and the remaining to the control group. One of the investigators from Maastricht University who didn't have information about the study prisons performed the randomization by consulting the online randomization website; random.org. The investigators will then undertake follow-up and monitoring for a year (from March 2016 to April 2017) and outcomes will be measured 3.3.1. Procedure

   After the investigators have had discussions and agreements with the concerned bodies in the Amhara and Tigray regions, eligible prisons were selected according to our inclusion criteria. Then the investigators matched the selected prisons into pairs considering geographical location and/or the number of prisoners as described above. After the number of the required matched pairs was fixed, the randomization was done to the intervention and control group .

   3.4. Interventions

   3.4.1. Training inmate peer educators

   Before launching the intervention program, the investigators will discuss the objectives and procedures of the program with the prison administrators, the prison health personnel, and the prisoners. Inmate peer educators, about 3-6 from each prison, will be recruited in consultation with the prison health personnel. The investigators, in collaboration with the prison health personnel, will provide 3 days training for inmate peer educators using standard TB training materials. The training will focus on basic facts about TB: its cause, transmission, symptoms, diagnosis, prevention, treatment, and outcomes. Inmate peer educators will be trained on how to provide health education about TB, screen and identify presumptive TB cases, and support TB patients to adhere to the TB treatment. Moreover, inmate peer educators will be provided with leaflets and posters illustrating the cause, transmission, main symptoms of TB, and containing the information that TB is curable with proper treatment. Inmate peer educators will also be informed to teach and encourage their fellow inmates about the isolation of TB patients as much their setting allows.

   3.4.2. Educating prisoners and performing TB screening

   After the training, but before trained prison peer educators start their activities, the investigators will assess whether inmate peer educators are really capable of doing the expected activity. The investigators will make them teach a sample of trainees in front of trainers and will evaluate inmate peer educators using a checklist. Inmate peer educators will also be tested whether they are cable of identifying presumptive TB cases by providing them presumptive TB cases and non-presumptive TB cases and verifying their diagnosis by TB experts (trainers). Once they are tested for their capability, peer educators will organize inmates into groups and will provide education about TB, its prevention, and control every two weeks on a regular basis for one year. Peer educators will undertake routine TB screening using a standardized TB screening protocol to identify presumptive TB cases for a referral. Inmate peer educators will be provided with N95 masks to be used during the screening. Identified presumptive TB cases (those, at least, having had a cough for two weeks or more according to the recommendation of national TB guideline) will be referred by the prison health personnel to nearby hospitals.

   3.4.3. Tuberculosis treatment follow-up

   Trained inmate peer educators will follow and encourage their fellow patients to adhere to the prescribed TB treatment and to undertake follow-up sputum examination according to the national recommendation. Inmate peer educators will provide health education about the consequences of nonadherence to TB treatment using the prepared TB training module .

   3.5. Control Conditions

   At the control sites, there will not be involvement of inmate peer educators and no training will be provided. The TB screening and treatment activity will follow the existing referral system (self-referral to nearby hospitals). However, the prison personnel at these sites will be informed about the proposed intervention. The prison personnel will be informed that this is a trial and that the investigators don't know whether the intervention is effective or not. The investigators will also inform the prison personnel that other prison sites were selected by chance and the intervention will be implemented in the control prisons if the investigators found this trial effective and helpful. Moreover, the investigators will provide a standardized up-to-date referral protocol to the prison personnel at the control sites as in the intervention sites.

   3.6. Follow-up and monitoring

   Supervisors assigned to each intervention site (prison) will regularly follow the intervention progress, do a daily assessment of the activities and give a timely solution to the problems met. The principal investigator will also supervise all activities closely through a regular visit and phone call. Randomly selected prisoners from each intervention site will be asked short random questions to check whether they did get the intervention or not at random times during the intervention.

   An intervention check will also be done by collecting data on knowledge attitude and practices (KAP) of prisoners in the intervention sites using a standardized pre-tested KAP questionnaire through a face-to-face interview by trained prison nurses. Inmate peer educators will be involved in facilitating and interviewing of the baseline data, but the post-test will be performed by trained nurses only. Briefly, the KAP questionnaire will consist of questions on the etiology of TB, mode of transmission, prevention and on attitudes of prisoners such as whether they are afraid to get infected with TB bacilli or not. A sample of prisoners will be selected randomly with proportional allocation of the total number of prisoners in each intervention site and data on their KAP score will be assessed at baseline and the end of the intervention (after 12 months). Only prisoners that will be available in prisons during the intervention period will be eligible for the random selection. Finally, data will also be collected from the randomly selected prisoners from the control sites and the KAP scores of the two groups will be compared.

   Supervisors will also be assigned for each control site. They will neither encourage nor discourage the existing TB diagnosis and treatment program in these sites. But, supervisors will regularly follow the ongoing activities with equal frequency as in the intervention sites. The principal investigator will also supervise all activities closely through a regular visit and phone call.

   3.6.1. Contamination check

   Together with the supervisors, the principal investigator will undertake a random assessment at the control sites regularly every week throughout the intervention period to check whether some sort of new TB control methods or peer education has been started in the control sites as in the intervention sites. Then,the investigators will give timely solutions if possible or will consider the contamination during result interpretation.

   3.7. Blinding

   Neither the prison health personnel nor peer educators will be blinded to the allocation. However, the data analyzer and the laboratory personnel will be blinded.

   3.8. Demographic and clinical data collection

   Demographic and clinical information such as pre-treatment symptom duration will be collected from TB confirmed cases detected at both the intervention and control sites using a pre-tested structured questionnaire. Patients' sputum smear status, the radiological findings and treatment outcomes will be recorded from the referrals sites and Directly Observed Treatment Short Course (DOTS) centers in both groups. The investigators will also collect data on the size of the prison cells and on the number of pulmonary TB cases detected in the previous two years before the start of this intervention in both the intervention and control sites to consider for possible adjustments.

   3.9. Laboratory Methods and Diagnostic criteria

   For both the intervention and control conditions, TB diagnosis will be carried out at the referral sites using the direct smear microscopy and/or chest X-ray (the routine available diagnostic methods). Three sputum samples (spot-morning-spot) will be collected from each presumptive TB case and direct sputum microscopy will be performed using Ziehl-Neelsen staining technique as per the national guideline at the referral sites by experienced laboratory technologists. Chest X-ray will also be performed at referral sites. A presumptive TB case with one or more initial sputum smear examinations (direct smear microscopy) AFB-positive; or one sputum examination AFB-positive plus radiographic abnormalities consistent with active pulmonary TB as determined by a clinician will be considered as a smear-positive TB case according to the national guideline [24]. A patient having symptoms suggestive of TB with at least 3 initial smear examinations negative for AFB by direct microscopy, and with no response to a course of broad-spectrum antibiotics, and with radiological abnormalities consistent with pulmonary tuberculosis will be considered as a smear-negative pulmonary TB case [24]. All confirmed TB cases will be treated according to the national guideline. The treatment regimen for new TB cases consisted of two months intensive phase treatment with ethambutol, rifampicin, isoniazid and pyrazinamide followed by a continuation phase treatment for 4 months with isoniazid and rifampicin. Follow-up sputum examination will be conducted according to the national recommendation and outcomes (treatment success, failure, default, and death) will be recorded by the data collectors (trained prison nurses). A smear-positive TB patient with at least two negative smears including the last month of treatment will be reported as cured. A patient who finished the treatment but did not have a smear result in the last month of treatment and on at least one previous occasion will be reported as treatment completed. If a patient remained or became smear-positive at the end of 5 months or later, he/she will be reported as treatment failure. A patient who missed treatment for eight consecutive weeks after receiving treatment for at least 4 weeks will be reported as a defaulter. A patient who was transferred to another recording and reporting unit after receiving treatment for at least 4 weeks and whose treatment outcome was not reported to the referring district will be reported as transferred out. A patient who died while on treatment will be reported as dead irrespective of the cause of death [25].

   3.10. Sample size calculation

   The number of pair-matched clusters required for this study was determined using a formula suggested by RJ Hayes and S Bennett [26] for a pair-matched cluster randomized trials of unequal cluster size as shown below. The sample size was calculated based on the following assumptions: an estimated average annual pulmonary TB case detection rate (CDR) of 40% in the study prisons (estimated using unpublished review of the two years DOTs record from some study prisons), expecting to detect a 50% increment in case detection rate over the intervention period (from 40% in the control groups to 60% in the intervention groups), a power of 80%, a coefficient of variation of 0.25, and a significance level of 0.05. The number of clusters was fixed considering only the primary outcome (case detection rate). A coefficient of variation of 0.25 was considered since there was no full data for the matched prison pairs to calculate the exact value.

   C = 2+ (Zα/2 + ZB) 2 [X0 x Av (1/X0j) + X1 x Av (1/X1j) + K2 (X02 + X12)]/ (X0-X1)2

   Where:

   C = number of matched clusters Zα/2 = level of significance, 95% = 1.96 Zβ = power of the study, 80% = 0.84 X0 = the estimated average annual smear-positive case detection rate in the control sites = 40% X1 = the estimated average annual smear-positive case detection rate in the intervention sites = 60% K= coefficient of variation = 0.25 Av (1/X0j) = the mean of the reciprocals of the cluster sizes (person-years) in the control group = 0.001 Av (1/X1j) = the mean of the reciprocals of the cluster sizes (person-years) in the intervention group = 0.0014 This computation provides a cluster size of 8.6 pairs. Considering feasibility, the investigators chose 8 pairs.

   3.11. Study variables

   3.11.1. Outcome variables

   The primary outcome measure for this study is case detection rate while treatment outcome (success, failure, death) and pre-treatment symptom duration are the secondary outcomes of interest. Moreover, the investigators will also compare the KAP scores of the two groups at the end of the intervention.

   Case detection rate (CDR): Calculated as notifications of new and relapse cases divided by the estimated incidence per year. The mean CDR will be compared within each pair at the end of the intervention (after 12 months).

   Treatment success rate (TSR): Cure or treatment completion rate will be calculated as the number of patients cured or treatment completed divided by the total number of patients reported expressed as a percentage. Treatment success rate is the sum of cure and treatment completion rate.

   Pre-treatment symptom duration: Measured as the proportion of patients with pre-treatment symptom duration of >=3 months vs. < 3 months.

   3.11.2. Predictor variables

   • The presence/absence of the intervention
   • Socio-demographic characteristics (age, sex)
   • Seasonal variations (spring, winter, autumn, summer) 3.12. Operational definitions

   Jail: A facility that is used to temporarily detain those who are suspected or convicted of a crime.

   Inmate/prisoner: Somebody confined in a prison as a punishment for a crime or while waiting to stand trial.

   Peer education: The teaching or sharing of health information, values, and behavior in educating others who may share similar social backgrounds or life experiences.

   Prison: A facility that holds convicts who have committed crimes the legal system deems especially serious for more long term sentences.

   3.13. Plan for Analysis

   Collected data from questionnaire and laboratory analysis will be checked for completeness before entry to a database. The data will be entered using Epi Data entry version 3.1 software. Frequency analysis will be run to look into the range of values, identify missing data or possibly miscoded data. Then we will use SPSS 20.0 for Windows for analysis as recommended for a cluster level analysis (matched analysis). Considering prisons as a unit of analysis, the mean CDR, TSR and percentage of patients symptomatic for > 3 months will be compared within pairs using the paired t-test or sign test as appropriate. The investigators will also perform analysis by date to see the trends of the outcome measures over the intervention period.

   3.14. Quality assurance methods

   The questionnaire will be pre-tested to ensure its reliability and validity. Inmate peer educators will be thoroughly trained in TB screening procedures and treatment supervision. The supervisors assigned to each prison alone, or with the principal investigator, will supervise the activities, do a rapid assessment of the general activities, and give a timely solution to the problems met. Experienced laboratory professionals will be involved in the microscopic examination and quality control of the slides examined will be ensured at regional laboratories in each region.

   3.15. Ethical Consideration

   Ethical clearance will be taken from Ethical Review Committees of Mekelle University. A written co-operating letter will be taken from the prison- administrators of each region. The participants will be included in the study after they have provided written informed consent. Patients who are confirmed to have tuberculosis will be referred to get treatment. The confidentiality of both paper-based and electronic patient data will also be safeguarded throughout the research activities and even after the research is completed.

   3.16. Dissemination of the results

   Once analyzed and written up, the findings of the study will be submitted to Mekelle University, Maastricht University, Federal Ministry of Health and other potential partners. Moreover, key findings of the study will be presented at scientific meetings and conferences so as to lobby for a reformulation of national TB control policy with the necessary attention given in such settings. The written manuscript will also be submitted to peer-reviewed journals for publication.

4. Trail registration The protocol of this clinical trial will be registered at Clinicaltrials.gov website.

Inclusion Criteria:

• Age of 18 years or more.

Exclusion Criteria:

-