Design and Setting
The current study used cross-sectional baseline data from a community-based intervention program to improve paternal parenting practices and youth energy balance-related behaviors (Padres Preparados, Jóvenes Saludables) . Latino fathers or male caregivers (hereafter referred to as fathers) and early adolescents (10-14 years) were the primary research participants, however mothers or female caregivers (hereafter referred to as mothers) were also invited to attend educational sessions and complete evaluation data collection procedures as part of the intervention. Baseline data collection occurred from 2017 to 2020 at three churches and two Latino-serving non-profit community centers in the Minneapolis/St. Paul metropolitan area.
Participants were recruited using flyers, word of mouth, and announcements at community events. Eligibility criteria for fathers included self-identifying as Latino, speaking Spanish, and having meals with their early adolescent child at least three times in a week. Mothers self-identified as speaking Spanish and as spouses/partners of fathers who were parents of the youth participants. Three hundred sixty-nine families were interested in the Padres Preparados, Jóvenes Saludables study, and 277 families remained after screening for eligibility. Baseline data were analyzed in the current study from 106 father and mother couples who completed food security questions. All questionnaires were completed in Spanish. Fathers received cash compensation ($35) for participating in baseline data collection procedures. Fathers and mothers provided consent prior to participation. The study was approved by the University of Minnesota Institutional Review Board.
Trained researchers measured body weight and height of fathers and mothers using a digital scale (BWB-800 Scale, Tanita) and a stadiometer based on standard procedures . Measurements were completed twice to the nearest 0.1 kg weight and 0.1 cm for height. Body mass index (BMI) was calculated using the average weight and height (kg/m2).
Fathers and mothers individually reported their age, educational attainment, employment status, household income, number of years in the U.S., language spoken at home, and marital status. Age, educational attainment, employment status, and household income were dichotomized as < median age vs. ≥ median age, < high school vs. ≥ high school, full-time/self-employed vs. part-time/not-employed, household income < $25,000 vs. ≥ $25,000. Acculturation level was based on the number of years in the U.S. (<10 years, ≥10 and <20, ≥20 and <30, ≥30) and language spoken at home (only native, more native language than English, English equal to native language, only English, or more English than native language), which were combined and categorized as low, middle, or high acculturation levels .
Food security was assessed based on two questions in Spanish from the US Department of Agriculture 18-item Household Food Security Survey. The two questions were “Within the past 12 months, we worried about whether our food would run out before we got money to buy more” and “Within the past 12 months, the food we bought just didn’t last and we didn’t have money to get more”, with response options of often true or sometimes true vs. never true. Hager et al. reported that an affirmative response (often true or sometimes true) to at least one of these two questions showed 97% sensitivity and 83% specificity for detection of household food insecurity .
Participation in food assistance programs
Participation in any of the following food or financial assistance programs was assessed (yes/no): Special Supplemental Nutrition Program for Women, Infants & Children (WIC), Supplemental Nutrition Assistance Program-Education (SNAP-Ed), free or reduced-price meals at school, and the Minnesota Family Investment Program. Participation in any of the following nutrition education program was assessed (yes/no): SNAP-Ed, Expanded Food and Nutrition Education Program (EFNEP), WIC, and Cooking Matters.
Physical home food environment
Questions about home food availability and accessibility were adapted from a previous cross-sectional study of adolescents (Project EAT - Eating Among Teens) . Home food availability was assessed by summing coded responses to six questions regarding availability of fruits, vegetables, junk foods, soda pop, sweets, and potato chips (never = 1, sometimes = 2, usually = 3, always = 4) [32,33]. Scores for fruit and vegetable (FV) availability were the sum of the coded responses to the questions about availability of fruits and vegetables. Home FV accessibility was assessed by summing coded responses for three questions regarding accessibility of FV (never = 0, sometimes = 1, usually = 2, always = 3) [32,34]. Summed coded responses were dichotomized as < median value vs. ≥ median value.
Food-related activities and responsibilities
A variable regarding frequency of food-related activities with their child was based on four questions with coded response options; one assessed family meal frequency adapted from Project EAT  and the other three assessed frequency of food-related activities with their child adapted from Musher-Eizenman and Holub (2007) . These included family meal frequency (never = 0, 1-2 times = 1, 3-4 times = 2, 5-6 times = 3, 7 times = 4, >7 times = 5)  and frequency of three food-related activities with their child (planning meals, buying foods, preparing foods - almost never or never = 1, rarely = 2, sometimes = 3, most of time = 4, almost always or always = 5) . Questions assessing parents’ responsibilities for feeding their child were adapted from Mallan et al. (2013) . Five questions addressed feeding responsibilities, including responsibilities for planning meals, buying foods, preparing meals, making decisions about, and controlling what the child eats. Coded response options were almost never or never = 1, rarely = 2, half of the time = 3, most of the time = 4 and almost always or always = 5. The sum of the coded response options for food-related activities with the child and feeding responsibilities were dichotomized as < median value vs. ≥ median value.
Neighborhood safety was assessed based on agreement with two questions adapted from Bennett et al. (2007)  regarding perceptions of whether the crime rate makes walking in their neighborhood unsafe during the day and at night (strongly disagree = 1 to strongly agree = 4). Family stress was assessed using three questions about the importance of family relations, conflict between personal and family goals, and individualism among family members (not at all worried = 1 to extremely worried = 5) . These variables were represented by the sums of the coded responses, dichotomized as < median value vs. ≥ median value.
Demographic data were described using means and standard deviations (mean ± SD) or frequencies for categories for fathers and mothers. To determine the differences in demographic characteristics between fathers and mothers, t-test or Fisher’s exact tests were conducted.
Associations between food security and predictor variables were examined among fathers and mothers separately using three logistic regression models. Preliminary analyses were conducted to assess appropriateness of variables for inclusion in regression models, including Fisher’s exact tests to determine differences in sociodemographic characteristics, participation in food assistance programs, physical home food environment, food-related activities and responsibilities, and environmental factors by food security status. In the logistic regression models, food security status served as the dependent variable and appropriate covariates were included as independent variables based on the preliminary analyses and tests for multicollinearity. Neighborhood safety was included in model 1 for both fathers and mothers. Household income and participation in food assistance programs were additionally entered into model 2 for fathers, while home FV availability and family stress were entered into model 2 for mothers. All these five variables were entered in model 3 for both fathers and mothers.
A cross-tabulation of household food security, as reported by fathers and mothers, was examined to evaluate concordance within couples using Fisher’s exact test. Father and mother couples were dichotomized into perceived concordance or discordance in food security status. Concordance was defined as couples where both the father and mother reported food security status or both the father and mother reported food insecurity status. The predictors for couple concordance in reports of food security status were determined using concordance or discordance with distinct characteristics of the variables within couples. Logistic regression analysis was used to evaluate the predictors for discordance in reports of food security within couples. Fisher’s exact tests were used to determine appropriateness of covariates for inclusion in the logistic regression model.
The level of significance was set at p<0.05. Statistical analysis was performed using R version 3.5.1 and SPSS for Windows, version 26.