Study Design, area and period
A cross-sectional study was conducted on four prisons in North Gondar Administrative Zone to estimate the prevalence of smear positive pulmonary tuberculosis rates among prison inmates. The study was conducted from February to April 2015, at 4 prisons, namely Debark, Dabat, Chilga and Gondar towns of North Gondar zone, northwest Ethiopia. North Gondar zone has 21 weredas with 539 kebeles and 2 town administrations. According to the 2007 Ethiopian census, North Gondar has a total population of 2,929,628 (1,486,040 men and 1,443,588 women). In this zone, there are one zonal and 21 wereda police stations. There are 4 large prisons that could house about 1500–3500 prisoners. They detain mainly sentenced and some pre-trial prisoners from several surrounding weredas.
The source population for this study was all of the four prison inmates present in North Gondar zone. However, only prison inmates who had cough for ≥ 2 weeks during the study period were recruited and included as the sample population.
Inclusion/exclusion criteria and study variables
Prison inmates who were willing to participate and had 2 weeks and above duration of cough were included in the study. On the other hand, prison inmates who had 2 weeks and above duration of cough but were unable to produce sputum, and prison inmates who were on anti-TB treatment and/or provided incomplete information were excluded from the study. The prevalence of smear positive pulmonary tuberculosis was used as the dependant variable whereas age, sex, smoking status, educational status, residence, marital status, incarceration time, history of previous contact with active TB cases, frequency of imprisonment, nutritional status, number of prisoners per cell, sharing food and other materials, pervious treatment for tuberculosis, window opening practice, HIV status, occupation, and duration of cough were used as independent variables.
Sample size and sampling technique
The sample size was determined using the following single population proportion formula: N = z2p (1 − p)/w2, where N = the number of TB suspected prison inmates, Z = standard normal distribution value at 95 % CI which is 1.96, P = the prevalence of pulmonary tuberculosis among prison inmates (10.4 %, previous prevalence report from Gondar prison), W = the margin of error, taken as 4 %. Accordingly, the sample size calculated was 223. When the record of prisons was reviewed, the average prison inmates in the 4 prisons of north Gondar were found to be 3900. As this source population is less than ten thousand, the sample size was corrected using the formula nf = in/1 + in/N, where nf is the final sample size (223) and N the source population. Considering a 10 % no response rates the final sample size of 235 was determined, but all inmates who fulfilled the inclusion criteria were enrolled and the final sample size was 282.
At the time of the study about 3900 inmates were detained in the four prisons of the north Gondar zone. A mass screening strategy was used to identify PTB suspects. This strategy provides an equal chance of selecting eligible individuals and reduces the chance of losing PTB suspects. First, all prisoners were collectively questioned about the presence of cough; then prisoners with cough were individually interviewed about the duration of the cough. Prisoners who had a cough history of ≥ 2 weeks were included in the study. In this case, a total of 282 prisoners were selected to participate in the study.
Socio-demographic data collection
Once eligible inmates were recruited for the study, socio-demographic characteristics were collected. Information on socio-demographic characteristics, imprisonment, number of prisoners per cell, history of previous treatment for TB, previous exposure to TB, window opening practice, cigarette smoking, and others were collected using a structured and pretested questionnaire.
Sputum sample collection, processing and staining
Spot-morning-spot sputum samples were collected following Standard Operation Procedures (SOPs) of the University of Gondar Hospital Laboratory. Briefly, the sputum samples were collected using dry, clean, leak proof, translucent, and screw-capped plastic containers with a capacity of 30 ml. The sputum samples were collected by the principal investigator from each participant, and the sample was transported to the College of Medicine and Health Sciences (GCMHS) Microbiology Laboratory for processing and examination. Smear was prepared by taking a portion of the purulent part of the sputum sample on a glass slide. The sample was spread on a glass slide and allowed to air dry, and then heat-fixed by passing the preparation through a flame 2 to 3 times. Each smear was stained with Auramine O staining procedure. Briefly, 0.1 % Auramine O solution was flooded on the smears and allowed to stain for 20 min. The stain was rinsed with water, drained and decolorized with 0.5 % acid-alcohol for 3 min. The preparation was rinsed with water until the macroscopically visible stain was washed away and drained. The smears were flooded with 0.5 % potassium permanganate solution for 1 min to minimize non-specific fluorescence and washed with water followed by air dry and microscopic examination. Stained slides were examined under 20x, and 40x magnifications of Primo Star iLED, light emitting diode (LED) fluorescence microscopy (FM) for AFB. The AFB appeared bright yellow against dark back ground materials. All AFB positive samples were also re-examined by staining another batch of smears with Ziehl-Neelsen staining method. Briefly, smears were flooded with carbonfuchsin solution and heated from the underneath until steaming and allowed to stand for 5 min. After washing with tap water, smears were decolorized with acid alcohol for 1 min and washed with tap water and counter-stained with methylene blue for 30 s, washed and air dried. Slides were examined under 100x oil immersion objective with bright field illumination.
GeneXpert MTB/RIF assay
All AFB positive sputum samples were further examined using the GeneXpert MTB/RIF (Cepheid Gene Xpert system) procedure to detect rifampicin (RIF) resistance. In brief, sample reagent was added to a sputum sample in a 2:1 (V/V) proportion and mixed by hand shaking. The preparation was incubated at room temperature for 15 min. Two ml of the treated sputum sample was transferred to the cartridge using a sterile pipette and the GeneXpert MTB/RIF machine run was initiated. After about 2 h, the result was interpreted and displayed by the machine as MTBC detected/not-detected and “RIF resistance detected or not detected”.
In this study, all recruited participants were screened for HIV infection after getting consent from each inmate. Concerning HIV testing, nurses working at the Providing Initiative Counseling and Testing (PICT) clinics of the prison were involved. Prisoners with positive sputum smears were placed on an anti-tuberculosis treatment as recommended by the National Tuberculosis and Leprosy Control Program (NTLCP) guidelines , while smear-negative patients were given a 10-day course of broad spectrum antibiotic treatment.
Body weight was determined to the nearest 0.1 kg using an electronic digital scale, and height was measured to the nearest 0.1 cm. Body mass index (BMI) was used to determine the nutritional status of the prisoners as under nutrition (BMI < 18.5 kg/m2) and normal (BMI = >18.5 kg/m2) .
Data quality was maintained using a questionnaire translated from English to Amharic. Pre-testing of the questionnaire was done for completeness and appropriateness before data collection. The reliability of the findings was guaranteed by implementing quality control measures throughout the whole processes of the laboratory work. To assure the quality of the sputum specimen, proper sputum collection, handling and processing procedures were used. The quality of newly prepared reagents was determined by both positive and negative slides. The performance of the microscopy was checked by both negative and positive stained slides. All slides were seen by the principal investigator and senior laboratory workers after staining all preparations with Auramine O stain. To check the accuracy, Auramine O positive slides were also re-examined by preparing corresponding smear and stained with Ziehl-Neelsen stain. Moreover, discordant results between Auramine O and Ziehl-Neelsen microscopy were subjected to GeneXpert procedures.
Data processing and analysis
Data was checked for completeness, cleaned manually, and entered and analyzed using the Statistical Package for the Social Sciences (SPSS) version 20. All variables of the study were initially tested for association with smear positivity by using the binary logistic regression model. Variables which showed statistically significant association with smear positivity by the binary logistic regression model were put into the multivariable analysis model to check if the association existed after controlling against all of the rest of the variables, and a P-value less than 0.05 was considered as statistically significant.