MEZACOPH investigators, CSF and MCHC directors and promotoras, and the MCHC-certified diabetes educator met in a day-long work session to discuss local strengths, challenges, and educational needs. The group reached consensus on module subject areas, order, format, appropriate instructional styles, and community differences that might influence recruitment, retention, and program delivery. After this initial meeting, MEZACOPH investigators met separately with promotoras from each site in monthly meetings over an eight-month period to review draft instructional materials and to gain feedback on the approach, format, and translation of the developing curriculum. In these meetings, promotoras from both sites also provided examples of health education materials that had been well received in their communities.

Once the curriculum was developed, a bilingual MEZACOPH investigator conducted a day-long training session in Spanish with CSF and MCHC promotoras. Since the promotoras had contributed to the development of many of the materials incorporated into the curriculum, the training offered an introduction to the overall flow and format of the curriculum and instruction in the use of educational materials. Promotoras gained familiarity and comfort with the curriculum during and after the training through playing roles, practicing delivery, and having coworkers critique their style.

La Diabetes y La Unión Familiar is a 12-week program with 10 points of contact: three home visits, five educational sessions, and two celebratory events. Drawing on key concepts of Social Learning Theory, a theoretical model was developed to guide the program design process (Figure). Key concepts include the influence of the social environment in behavior change, the need for knowledge and skills to change behavior (behavioral capability), and the importance of building confidence in the ability to take action (self-efficacy). Intervention activities include teaching team-building and communication skills to build and reinforce intrafamily communication, collective esteem (so all family members accept and value the family as a group), and collective efficacy (which promotes confidence in the family's ability to make changes). Intervention activities also include providing information on food choices and physical activity so families can make informed choices. Key concepts related to diet, exercise, and family support are introduced and discussed through the use of pictorial flipcharts, educational games, food sampling and preparation, and low-level physical activities.

Family social behaviors such as cohesion, adaptation (exhibited by resiliency and problem solving), and support within the family influence family food choice and physical activity behaviors. The proposed outcome of targeting family social behaviors to change health behaviors is improved nutrient intake, activity level, and diabetes management or prevention for all family members. The program encourages family members to collectively set health-behavior goals, to overcome obstacles hindering healthy behaviors, and to develop a plan to sustain behavior changes.

Figure. A family-based diabetes control and prevention program at the U.S.-Mexico border.

The promotora instructional manual provides an overview of the goals and format of the program, a description of the objectives, and an outline of the activities and supplies needed for each point of contact. Each family is given a notebook that includes copies of the flipchart materials and pockets for handouts, recipes, and other memorabilia such as photographs and a graduation certificate.

To accommodate differences in promotora skills and community characteristics, curriculum delivery is flexible. Educational flipcharts and games are prescribed for specific modules, but foods/snacks, exercises, and supplemental activities (e.g., additional games and stories) are selected by the promotoras from appendices in the instructional manual.

The program sequence and content are described below:

All training, intervention, and evaluation materials and activities were produced and delivered in Spanish.

Impact of the intervention was assessed using the following: 1) a written preintervention and postintervention Knowledge, Attitudes, Beliefs, and Behaviors (KABB) questionnaire with 15 close-ended items that documented adult participants' self-reported knowledge of risk factors, dietary and exercise habits, perception of need to eat healthy foods and be active, collective (family) exercise habits, and collective efficacy to make behavior changes; and 2) postgraduation interviews. Questions were designed to track changes in the stated learning objectives of each of the educational sessions. Promotoras reviewed the questionnaires for readability and comprehension. A team of university Border Health ¡SI! investigators reviewed the individual questions and entire questionnaire for content and face validity. The design of the questionnaire was driven by community and university collaborators' intent to develop a user- and administrator-friendly instrument that the promotoras would continue to use to document program impact.

The McNemar test for paired categorical data and the Wilcoxon signed rank test for paired continuous data were used to compare preintervention and postintervention responses of 116 adult participants (48 Santa Cruz County and 68 Yuma County).

In the initial visit, promotoras administered the preintervention questionnaire. In the final home visit, an evaluator not identified with the intervention administered the postintervention questionnaire. To accommodate variations in literacy levels and to avoid bias, all administrators were instructed to use a straightforward objective style when reading questionnaires aloud in Spanish. Participants were instructed to make their selection independently.

In the second home visit (Cómo Están?) ) two weeks after graduation, promotoras were impressed by participants' comments emphasizing the psychosocial importance of the program. To further explore an outcome not assessed by the preintervention and postintervention questionnaire, MEZACOPH investigators decided to conduct follow-up interviews with a sample of participants. One year after graduation, a sample of 18 participants consented to an open-ended interview with a MEZACOPH investigator not identified with the intervention. Given the small sample and an interest in documenting all experiences within this case study, participants' statements, as recorded in writing by the interviewer, were grouped by common themes.

Four rounds of La Diabetes y La Unión Familiar were implemented in each intervention county, yielding 72 patients with diabetes and 177 support people, including children and grandchildren. Table 1 illustrates the distribution of participants by sex and age group; figures include all patients with diabetes, family members, and supporters. Twenty-five percent of participants were children (younger than 18 years). Depending on their age and interest, children participated in the games, listened to stories, and participated in discussions generated by the flipcharts. Younger children and infants played or were cared for in a separate child-care area in the multifamily group sessions.

Adult daughters and wives were the predominant participating supporters. Percentage of supporters by relationship to the family member with diabetes was: 22% daughters, 20% spouses (with 54% wives and 46% husbands), 15% sons, and 9% friends. Of the total adult participants, 87% attended three or more of the five educational modules, and 43% of the youth and adults attended eight or more of the 10 points of contact.

Sixty-one (53% of total participants) of the preintervention and postintervention response pairs were from the family member with diabetes.

Table 2 provides the percent of yes responses to a list of possible diabetes risk factors listed in the questionnaire. A family history of diabetes (heredity) and being Hispanic, overweight, inactive, and older than 45 years were introduced in the first educational session (Familias y Diabetes) as known risk factors for diabetes. Based on their outreach experience, promotoras reported that some clients believed that stress, fear, and contact with an individual with diabetes were also risk factors. This first session provided an opportunity to discuss these perceptions. Preintervention and postintervention test comparisons indicate a significant increase in participants who identified the five known risk factors and a significant decrease in those who indicated yes to stress, fear, and contact after the intervention.

Table 3 provides a comparison of participants' preintervention and postintervention responses to family efficacy questions. Participants responded to the following questions on a five-point scale with 1 = not very confident and 5 = very confident: How confident are you that your family can become more physically active? How confident are you that your family can eat healthier? After the intervention, participants report a significant increase in their perception of their family's efficacy to make specific behavior changes.

Questions of food intake and activity were changed during the course of the program in response to questionnaire administrators' report that participants may be misinterpreting questions. The initial questions asked about general intake of specific foods and participation in specific activities. Questionnaire administrators indicated that respondents did not believe that they had regular food and activity behaviors and answered based on their behavior over the previous week. To create a questionnaire responsive to the administrators' observations and to provide an evaluation instrument that allowed participants to report behavior change in their own terms, these questions were changed after two rounds of administration from "in general" to "in the last week." Given the change, responses to these two different sets of questions were analyzed separately (separate data table not provided). Despite the difference in question wording, these separate data sets do reflect similar patterns of change, which are described below:

Changes in knowledge, attitudes, behaviors, and beliefs were not different in a comparison of family members with diabetes and family members without diabetes.

Eighteen individuals (both patients and family members) from both sites were interviewed individually for approximately one hour. All statements could be grouped into one of three themes:

No other follow-up data were collected to assess long-term behavior change or retention of knowledge.