Study sites and period
The current study was carried out in four districts situated at the southern coastal regions of Bangladesh from January to March 2021. The coastal area of Bangladesh consisted of 19 districts accounting for 32% of the land area and 25.7% of the total population of Bangladesh . Geographic locations made this area vulnerable to different devastating natural disasters (e.g., floods, cyclones, tidal surges, river erosion) which have had an adverse impact on the socio-economic, water, sanitation, and hygiene problems (such as foodborne diseases like diarrhea) of populations living in this part of the country [32,33,34]. Thus, four coastal districts namely, Barishal, Patuakhali, Pirojpur, and Chattogram were randomly selected for the study.
Study design, participants, and sampling
A cross-sectional study was conducted among 300 meat handlers working at butcher shops from the selected areas (n = 4) of Bangladesh. A systematic random sampling technique was employed to select each butcher shop followed by an equal allocation of participants where there was more than one individual working in the butcher shop. Initially, research staff visited the respective authorities of the selected areas (city-corporation or municipality) and inquired about a documented list of locally operated butcher shops. Unfortunately, there was no official list but the local authorities provided an idea of the location of local butcher markets, and the research staff noted the location names and visited these areas accordingly. Seventy-five butcher shops from each of the four districts were selected randomly from the main butcher market areas. The butcher shops were small and had an average of three to four workers per shop. Participants were considered eligible for the study based on the following criteria: (i) participants who have direct contact with meat or meat handling surfaces, (ii) having at least 6 months working experience in butcher shops, and (iii) participants free from any disability and illness. All participants were enrolled by lottery techniques from each butcher shop carried out by research staff. The name of each worker in a particular butcher shop was written on a piece of paper, folded, and entered into a container with other pieces of folded paper with butcher shop worker names. A research staff member picked one name from the container and assessed the individual for study eligibility.
Sample size determination
A sample size of 384 was calculated using Cochran’s formula  by considering 50% prevalence of expected food safety knowledge and practice among meat handler as similar studies were lacking in Bangladesh. A 95% confidence interval (CI) and 5% margin of error between the sample and the underlying population was employed. Subsequently, we used a modified Cochran’s formula  for calculating adjusted sample size in a small population (assuming 1000 meat handlers work at butcher shops in the selected areas), which gave us a minimum sample size of 277. In anticipation of any missing data or incomplete surveys, the research staff enrolled more participants, and the final sample included in this study was 300.
Data collection procedures
Prior to data collection, the research protocols were reviewed and approved by the Research Ethical Committee (REC) of Department of Food Microbiology, Patuakhali Science and Technology University, Bangladesh (approval number: FMB:15/12/2020:04). Data collectors explained the purposes of the study to all participants and asked those willing to participate to sign a consent form or obtained verbal consent from those who were illiterate. Data were gathered by in-persons interviews of a structured questionnaire, which was based on previous literature [36, 37]. The English-version of the questionnaire was translated into the local language (Bengali) at the time of data collection. First translation of the questionnaire from English to Bengali was conducted by a bilingual translator, which was cross-checked by an independent bilingual research staff member. Further, a back-translation of the questionnaire was conducted by another, independent bilingual research staff member to check for consistencies and to avoid any bias in the questionnaire. The pre-test of the questionnaire was done among a randomly-selected small group of meat handlers (n = 15) to verify the applicability and clarity of the questionnaires, and the time needed for each interview. Anonymity of participants’ responses was ensured through the coding of questionnaires. Each interview took 15–20 min to complete at the butcher shop of participant/workers.
Study variables and measures
A questionnaire was adapted from previous studies [36, 37] with certain questions modified based on the specific socio-demographic status of the target sample (i.e., meat handlers who work at butcher shops) and the differing perspectives of persons living in Bangladesh versus the other countries where previous studies had taken place. Based on the pre-testing of the questionnaire, modifications were made to several items pertaining to food safety knowledge as test respondents noted that aspects of these questions or statements were unclear. For example, we added an item, “Anthrax can be transmitted by food” instead of “AIDS can be transmitted by food”. The questionnaire included a total of 43 questions with three sections as follows: (i) socio-demographic information, (ii) assessments of food safety knowledge and (iii) assessments of food safety practice.
In the first section, socio-demographic characteristics such as, gender, age, education level, years of experience, working hour per day, field of duty, income, employment status, having health certificate and attending food safety training of the respondents were included.
For assessing food safety knowledge, a set of 18 close-ended questions with three possible answers (such as “true,” “false,” and “do not know”) were used. The food safety knowledge questions included information on personal and food hygiene, cross-contamination of food, high-risk groups for food poisoning, and specific foodborne diseases and pathogens. To reduce the possibility of respondents selecting the correct answer by chance, the multiple-choice answers included the option “do not know.” One point was assigned for each “true” answer, with the other two answers (“false” and “do not know”) assigned zero score. The food safety knowledge score ranged between 0 and 18, and scores 9 or above was considered a good level of knowledge, and a score of below 9 was indicated as a poor level of knowledge .
The last section of the questionnaire dealt with food safety practices of the respondents emphasizing issues of personal hygiene, hand washing practices, and practices against food borne diseases and cross-contamination. A set of 15 questions was provided and the level of practices were assessed using a five-point ordinal scale (never = 0, rarely = 1, sometimes = 2, often = 3 to always = 4). For four questions, reversed scoring was employed (items 2, 3, 12, and 15). To avoid the possibility of respondents picking the correct answer by chance, the option “sometimes” was included. The score ranged between 0 and 60, and scores below 30 were recognized as poor practices, and scores 30 or above were considered as a good level of practice 
Data were analyzed by Statistical Package for the Social Sciences (SPSS) software (version 23.0). Descriptive statistics (e.g., response frequencies/percentage, means and standard deviations) were used to summarize variables of interest. Analytical statistics including bivariate analyses and multiple logistic regression models was used to identify factors associated with food safety knowledge and practices of meat handlers. All socio-demographic variables except gender and health certificates (excluded due to lack of variation of categories) were included in both univariable (unadjusted) and multivariable (adjusted) logistic regression models. Multicollinearity among the independent variables was checked using variance inflation factor (VIF) and tolerance. The mean VIF for adjusted model of Table 4 was 1.463 (Min VIF = 1.056, Max VIF = 2.432), while the mean VIF for adjusted model of Table 5 was 1.374 (Min VIF = 1.116, Max VIF = 1.751). Previous studies reported that a mean VIF of less than 10 is acceptable [38, 39]. The strength of association between independent variables (such as age, education level, and income) and dependent variables (level of food safety knowledge and practice) was assessed by odds ratios with 95% confidence interval (CI). And a p value of less than 0.05 was considered statistically significant.