A two-arm parallel cluster randomized controlled trial was used to examine the effectiveness of a community-based brucellosis prevention program over six months in a rural population in Ahar, East Azerbaijan, Iran. The initial baseline assessment was conducted in July 2016, and the post-intervention assessment was performed six months after the baseline measurements were taken.
The high incidence of brucellosis in the rural areas of Ahar (31-41 cases per 100,000)  and the crucial need for a community-based brucellosis prevention program  led to its selection as the study setting. Due to the strong relationships among neighboring households in rural communities, it was not possible to assign participants to the same group. Therefore, we performed a stratified randomized controlled trial using rural health houses as the units of randomization. Health houses are located in rural areas, far away from health centers .
Recruitment and sampling design
In total, 16 health houses (one per village) with a high prevalence of brucellosis during the prior two years were selected for the recruitment of participants. Figure 1 shows the study flow chart. Participants from rural communities at high risk for brucellosis were randomly recruited using household health records in the rural health houses. The inclusion criteria were as follows: (a)
participant had to be 15 years of age or older and (b) living in the villages for at least six months after enrollment. People who were unwilling to provide consent for data collection and employees in the health centers or the veterinary office of Ahar County were excluded from the study.
A multi-stage random sampling method was used to select villages in the county. Ahar was stratified into four regions: north, south, east, and west. Two health centers were randomly selected from each region, and eight health houses (two from each region) with a high prevalence of brucellosis during the prior two years  were selected from each center. In total, 400 participants were recruited, and the number of participants selected from each health house was based on the proportion of the population it covered.
Randomization was carried out after the baseline measurements were taken. The 16 selected health houses were randomly allocated into eight intervention and eight control groups (arms) using permuted blocks randomization (PBD) to guarantee balance in the number of the units allocated to each arm. The randomization sequence was created manually by a biostatistician using Microsoft Excel software [Excel command for random block sizes column: =rand()] to assign the health houses to the study arms using a 1:1 allocation ratio with block size of four. A colleague not connected to the study performed equal group random allocation.
The participants were recruited by an independent researcher using computer-generated random number schedules from recorded lists of the household health files at the health houses. Trained research assistants gathered the baseline measurements while group allocation concealment was implemented. The participants were also blinded to their group assignments.
Model for program planning
The PRECEDE-PROCEED model  was used for the design, implementation, and evaluation of the brucellosis prevention and control program. The model includes nine phases based on assessments (PRECEDE), which should be made before planning a health intervention, and an evaluation (PROCEED) to enable the measurement of the effectiveness of the intervention at each stage of implementation, as well as the immediate and long-term effects (see Figure 2).
The priority targets for the intervention were established in each phase of the assessment based on the importance and changeability of specific behavioral and ecological factors in determining brucellosis outcomes (phases 1-5). Specifically, the assessment process focused on the seven behavioral and six ecological highest-priority factors (See Table 1), and the evaluation (phases 7- 9) tracked the impact of the intervention on these factors.
Phase 1: Social assessment
Phase 1 of the PRECEDE-PROCEED model commenced by assessing the goals of the intervention. Diagnostic activities were performed to gather data to assess the health of the target population and identify the appropriate outcomes of the intervention.
Phases 2 and 3: Epidemiological, behavioral, and environmental assessments
We conducted an extensive literature review and informal discussions with selected key informants and health care staff from the Ahar Communicable Diseases Office, the Ahar Agriculture Jihad Office, and the Veterinary Office of Ahar County. These activities resulted in the identification of the issues affecting brucellosis prevalence, the steps that could be taken to reduce the impact of the disease, and the changes needed to prevent brucellosis. Using this information, the most significant behavioral and ecological risk factors for the intervention were determined based on their importance and changeability, resulting the selection of the seven behavioral and six ecological highest-priority factors, as noted above (See Table 1).
Phase 4: Educational and ecological assessments
This phase helped determine which intervention factors, if modified, would most likely result in behavior changes. These factors, classified as predisposing, enabling, and reinforcing factors, were identified through the literature review and key informant interviews with health workers, health care providers, experts on the surveillance and control of brucellosis in health centers, a veterinary specialist, and experts from agriculture organizations working on brucellosis in Ahar.
The results of this process were recently documented . Briefly, a standardized, structured questionnaire was used to gather information from the target population about potential routes of transmission to humans and practices for dealing with aborted animal fetuses and for processing and consuming milk and dairy products. This questionnaire included five parts, addressing the predisposing, reinforcing, enabling, environmental, and behavioral factors. The first part focused on predisposing factors that provide a reason or motivation to perform a behavior, including knowledge, attitudes, and self-efficacy about the prevention, transmission, and control of brucellosis. The knowledge section included 11 items that measured rural population awareness and understanding about the causes, modes of transmission, and protective behaviors of brucellosis. The attitudes were measured using a 15-item five-point Likert scale (strongly agree, agree, do not know, disagree, and strongly disagree). The self-efficacy section contained seven items with five-point Likert-type responses (very uncertain, uncertain, do not know, certain, and very certain). This section assessed the confidence of the participants in their ability to practice protective behaviors against brucellosis.
The second part of the questionnaire focused on reinforcing factors that encourage and support certain behaviors due to social support, rewards, and praise, such as encouragement from family members or friends to take preventive measures against brucellosis. The prime sources of reinforcement identified were positive feedback and advice from family, friends, and health workers. Six items were used to assess how encouraging behaviors could be repeated and sustained by family members or friends.
The third part assessed enabling factors that allow people to act based on financial support, resources, assistance, and services. This part measured access to materials and financial and educational resources, including access to personal protective equipment, such as gloves and masks; financial support for the renovation of barns and buildings; and veterinary services supporting regular animal vaccinations against brucellosis.
The fourth part, assessed environmental factors, including characteristics of the environment that facilitate the development of behaviors or the attainment of resources required for protection from brucellosis. This part examined environmental safety practices related to brucellosis transmission vehicles using various direct and indirect measures targeted at reducing the risk of brucellosis infection and creating protective measures for the environment (e.g., wearing protective equipment, like gloves and masks, before going to the barn).
Finally, the fifth part looked at behavioral factors, which evaluated using nine items, including working with pastoral livestock, regularly vaccinating animals against brucellosis, consuming pasteurized dairy products on a regular basis, and wearing protective clothes while working in barns.
Phases 5 and 6: Administrative and policy assessments and intervention
The predisposing, enabling, and reinforcing factors that influence behavior were analyzed using the results from the assessments in phase 3. Then, the appropriate strategies and interventions were matched with high-priority project changes according to the policies, resources, and organizational situations.
First, the enabling factors—which were considered the highest priority factors in this phase—were addressed, which required coordinated efforts and intersectional collaboration between local public health organizations, agricultural organizations, and veterinary organizations. The specific activities included continuously training health workers on brucellosis prevention, providing adequate resources for restoring and renovating barns, providing farmers with access to disinfectants and training in the proper use of those disinfectants, and compensating farmers whose animals died due to brucellosis.
Second, to address the predisposing factors, the knowledge, attitudes, and self-efficacy of the participants were promoted through the design and implementation of educational interventions, which were tailored using demographic variables. Third, the activities related to the reinforcing factors included encouraging more people to properly dispose of animal waste, giving information about how to bury aborted animal fetuses, and supporting regular animal vaccinations against brucellosis.
In addition, we selected an advocacy strategy in order to promote policy changes and prevention measures that would reduce the prevalence of brucellosis in Ahar. Specifically, the goal of the advocacy initiative was to decrease the prevalence of brucellosis by 10% and increase HRQOL, in cooperation with the health centers and veterinary and agricultural organizations, by the end of 2016.
The relevant stakeholders included anyone who would gain or lose from reaching the goal set for our advocacy efforts. We identified stakeholders through informal interviews and consultations with related organizations. The stakeholders were categorized into four groups: beneficiaries (farmers, household women, the Ahar primary health care network, and veterinary organizations), partners (the Ahar primary health care network, farmers, and veterinary and agricultural organizations), decision makers (the Ahar primary health care network, farmers, and veterinary and agricultural organizations), and adversaries (e.g. washing udders properly before milking). The advocacy approaches and persuasion techniques are described in Table 1.
For the advocacy initiative, activities included integrating veterinary and public health campaigns to increase public awareness of brucellosis and provide information about ways in which people could reduce the risk of infection. Close cooperation and coordination between all partners was crucial to success.
The following demographic data was collected from all participants: gender, age, marital status, educational qualifications, job, history of brucellosis, and family history of the disease. As described above, a standardized, structured questionnaire, which included five parts (covering predisposing, reinforcing, enabling, environmental, and behavioral factors related to brucellosis), was used . The questionnaires were filled out in writing and participants were also interviewed..
The sample size was calculated based on a standard deviation increase in knowledge (6.8) , as this was one of the most important variables. As such, a study with a power of 90% at 5% significance level would need 185 participants in each group. Considering the risk of attrition, 200 participants per group were sought. The number of participants selected from each health house was based on the proportion of the population it covered.
The characteristics of the participants were summarized as numbers, percentages, or means with standard deviations, where appropriate. If the continuous variables were not normally distributed, appropriate transformations were performed to achieve normal distribution.
A generalized mixed effects-model was used to analyze the data. This model used cluster randomization to incorporate random effects in order to reflect correlations among the observations of the members of the same health house. According to the distribution of our outcomes, the appropriate distribution and link functions were selected. For all parameters, 95% confidence intervals were defined, and two-sided p-values of less than 0.05 were considered statistically significant. All analyses were performed with the Statistical Package for the Social Sciences (SPSS) version 23 (SPSS Inc., Chicago, IL).