Feeding America's Bravest: Mediterranean Diet-Based Interventions to Change Firefighters' Eating Habits

Aims and hypothesis:

Recognizing the benefits and safety of the Mediterranean diet in CVD prevention, the investigators seek to develop behavioral change strategies in the fire service to modify the existing food culture based on key principles of the Mediterranean diet. Using education, participation and incentives, the investigators propose to motivate firefighters and their families to incorporate Mediterranean diet principles at work and home. The research design will allow us to measure the effectiveness of our Mediterranean Nutrition Interventions (MDNIs) in the "field settings" of the fire services for changing eating behavior and modifying CVD risk. Additionally, the cost-effectiveness of these strategies for future wider implementation will be examined. Specifics aims are:

1. Develop a multi-pronged, MDNI behavior change strategies including: diet/lifestyle education; discounted access to key Mediterranean diet foods; electronic education platforms and reminders; and targeted incentives. MDNI components will be refined via surveys, literature review and local/national firefighter input including labor/management and fire service focus groups. Hypothesis (1). Cost-effective, fire-service MDNIs will be developed for the career randomized controlled trial (RCT).
2. Cluster Randomized Diet Intervention-Phase I- the Indianapolis Fire Department (IFD) has over 1,000 members. All 45 IFD firehouses will be cluster randomized into two groups and then individual members will be consented regarding study participation. Group 1 will receive an active 12-month MDNI, while Group 2 will receive no intervention.
3. Cluster Randomized Diet Intervention -Phase II: Group 1 will cross-over to "self-sustained continuation" for 12 months to examine persistence of behavior change during this less intense, self-directed maintenance diet intervention. Group 2 will cross-over to receive the MDNI for 6 months (assessing a shorter MDNI), followed by a final 6 months of self-sustained continuation. Our existing modified Mediterranean diet score (mMDS) will be further integrated with other validated scores and nutrition questionnaires. Questionnaires, mMDS and clinical data will be collected longitudinally throughout both phases of the RCT.

Hypotheses (2-3). Both MDNI lengths will improve mMDS, reduce weight, and improve Cardiovascular Disease (CVD) risk profiles in career firefighters. The 12 month MDNI will produce greater persistence of adherence than a 6 month MDNI. Cost-effective methods that combine messaging and discounts for improving firefighters' diets will be developed and validated, and these low cost methods will be adaptable for widespread translation and implementation throughout the career fire service.

Implementation by Fire Service:

Our national surveys of career and volunteer firefighters demonstrate several common key facts. Firefighters would like more nutrition information from the fire service and want to learn more about healthy eating. Moreover, to succeed, the dietary approach must be acceptable to firefighters. The Mediterranean diet does not require completely giving up any food; and therefore, is easily adopted for long-term adherence. In our surveys, the Mediterranean diet had the greatest appeal among five proposed diet descriptions. In addition, the web-based tools, other educational materials and behavior change strategies proposed to develop will be low-cost and cost-effective. With the support of fire service partners, they should be practical for widespread implementation both on a local basis by individual fire departments and regionally/nationally by fire service organizations.

Project/Statistical analyses:

Data recording, storage, management, cleaning and basic analyses will be performed using SPSS. The data will be imported into Stata and SAS for more advanced statistical analyses such as multivariable regression and longitudinal data analysis, as needed. Differences in the mean values of a quantitative variable between two groups will be assessed using the independent t-test whereas differences in mean values among three or more groups will be examined using the analysis of variance (ANOVA) technique (or non-parametric Wilcoxon and Kruskal-Wallis tests, respectively, as appropriate). Differences in qualitative characteristics will be compared using the chi-square test, or Fisher's exact test, or McNemar's test (for paired comparisons), as appropriate. Statistical significance for all analyses will be p < 0.05, and all tests will be two-tailed.

RCT: During the first 12 months of the RCT, groups 1 and 2 will be compared directly on an intention to treat basis according to the randomization of each subject's fire house. This type of analysis- disregarding the personal level of compliance or engagement of any particular participant- is most rigorous and appropriate because our primary interest is to test the effect of the behavior change strategies. As several outcomes will be assessed on each participant throughout the study (3 annual medical exams and 5 semi-annual assessments (where weight, mMDS, etc. will be evaluated)), repeated measures techniques, such as mixed modeling and generalized estimating equation models, will be utilized to estimate changes over time on key outcomes. The investigators will take into consideration the specific variance-covariance structure of correlated measurements. After adjusting for baseline and time-dependent covariates, changes in the outcomes of interest will be assessed and compared among groups. The investigators expect our greatest power to be for within-group changes (paired comparisons).

For the RCT, sensitivity analyses have been performed and demonstrate that even in the worst case scenario where only 500 firefighters would be enrolled, the study is well-powered for intention to treat comparisons of MDNI vs usual care. Several conservative examples are given below for very small changes. However, consistent with previous studies, the investigators expect the MDNI to actually produce larger changes that would be more easily measured. The first example assumes the MDNI can achieve a small increase in the mMDS of 6 points (about 1.0 times the population's baseline SD based on our cross-sectional study of mMDS).

Barriers:

Several potential barriers to successful completion of the MDNIs have been identified and taken specific steps to prevent and surmount these barriers. Some question whether firefighters will follow the diet at home as well as work. First, our preliminary work demonstrates that firefighters eat better at home than work. Second, our proposal recognizes the family's importance and has incorporated measures to educate and involve family members who share meals and may do much of the shopping and cooking. Another challenge is MDNI cost for sustainability and wider implementation. Therefore, an economist has been added to our team with expertise in food and nutrition to ensure that the interventions developed are low cost, cost-effective and sustainable. MDNIs are very most cost-effective measures from a societal perspective. An additional challenge identified in the career randomized trial is the potential movement of firefighters from one station to another during the study period. This is one of the reasons the investigators chose the Indianapolis fire department: 90% of members are permanently assigned to a fire house. Thus, there will be minimal loss/change of firefighters' intervention assignments over the course of the project. Finally, our current proposal facilitates the challenge of national implementation by directly addressing both the career and volunteer fire services.

Human Subjects: Recruitment- Previously, the investigators have been very successful in participant recruitment for our FEMA-sponsored studies. For example, about 90% of IFD members approached have consented to our current FEMA cardiac imaging study. Additionally, the research proposal has been endorsed by significant fire service partners, which should increase trust and willingness to participate. The extent to which consented participants personally engage in the MDNIs is voluntary, and the additional time commitment required for the study is nominal- for the RCT: a weigh-in, blood pressure check, waist circumference measure and questionnaire completion every six months; and for the demonstration project only a web-based survey every six months. Moreover, our study designs are such that all firefighters who consent will receive the MDNI at some point and have an opportunity for health benefits. Therefore, no problems identifying and recruiting the necessary participants for the RCT and demonstration project are expected.

Confidentiality and IRB:

Any RCT data from the proposal transferred to Harvard is extracted in de-identified form without personal identifiers and is maintained in a confidential manner. PHI linkages to de-identified data are maintained in locked, safe locations at PSM's clinical facilities. Harvard researchers have access only to de-identified data ensuring that the research has a very low risk of breaching any confidentiality. For the demonstration project, consent forms and names will be linked to study codes to be used in all electronic data collection. The linkages to volunteer names will be maintained in a locked, safe location at Harvard. All electronic data will be restricted to authorized personnel and will be password protected. All protocols, questionnaires and procedures will be approved by Harvard's IRB and DHS before any participants are contacted/recruited.