Cumulative Household Food Insecurity Is Associated with Adverse Eating Behaviors Among Youth Exposed to Gestational Diabetes in Utero Authors: Keally Haushalter, Jaimie N. Davis, Myles S. Faith, Yeyi Zhu, Erin A. Hudson, Alexis S. King, Erica P. Gunderson Published: 2026 ABSTRACT Background/Objectives: Food insecurity has been associated with adverse eating patterns, which increase the risk for overweight and obesity. However, most studies measure food insecurity and eating behaviors cross-sectionally, not assessing whether experiencing food insecurity over time influences eating behaviors. Therefore, this study assessed the relationship between cumulative food insecurity and food approach and food avoidance eating behaviors and evaluated the effect of breastfeeding intensity and duration on this relationship. Methods: The Study of Women, Infant Feeding and Type 2 Diabetes after Gestational Diabetes Pregnancy (SWIFT) is a prospective longitudinal cohort of 1033 mothers diagnosed with gestational diabetes and their youth. Breastfeeding intensity and duration were assessed at research visits and through mailed monthly surveys from birth to one year postpartum. The SWIFT study in Youth (SWIFT-Y) was a cohort of 701 SWIFT offspring that consented to participate in a research visit 11–16 years postpartum. Youth’s BMI was measured at 11–16 years old through a research visit or their electronic health record. At each of the four research visits (6–9 weeks, one year, two years, and 11–16 years postpartum), participants were categorized as being food secure (0) or food insecure (1). Scores across the four visits were summed (range 0–4), reflecting the number of assessments one was classified as food insecure. The Child Eating Behavior Questionnaire was used to measure the prevalence of eight eating behaviors in the youth, with higher scores indicating that the behaviors were more apparent. According to the measure, the behaviors were categorized into two groups: food approach and food avoidance. Food approach consists of youth’s food responsiveness, emotional overeating, enjoyment of food, and desire to drink, and has been positively associated with a child’s BMI. Food avoidance includes the youth’s satiety responsiveness, slowness in eating, emotional undereating, and food fussiness, and is negatively associated with BMI. Linear regressions assessed whether cumulative food insecurity was associated with eating patterns, and exploratory analysis analyzed whether this relationship differed by breastfeeding duration (150,000) [24,25]. Mothers also reported how many individuals lived in their homes by writing a numeric answer to the question, “How many persons currently live in your home, including yourself?” [24,25]. The participant’s poverty-to-income ratio was calculated using the household size, income, and the respective poverty guideline from the year the survey was completed, by dividing their income by the poverty level for their respective household size. Following the USDA guidelines, at each visit, participants were then classified as food secure if they answered, “We have enough food to eat and the kinds of food we want,” and had an income-to-poverty ratio greater than two. If participants answered, “We have enough food to eat, but NOT always the kinds of food we want to eat”, “Sometimes we don’t have enough food to eat”, or “Often we don’t have enough food to eat”, they were classified as food insecure regardless of income-to-poverty ratio [27]. Finally, in line with the guidelines, if participants selected “We have enough food to eat and the kinds of food we want,” but had an income-to-poverty ratio equal to or less than two, they were classified as food insecure [27]. For each visit, participants were assigned a one if they were food insecure and a zero if they were food secure. To assess cumulative food insecurity, the scores were summed across all four visits for each participant, ranging from zero to four. A zero indicated that the participant was never food insecure, while a four indicated that the participant was food insecure at all four visits (i.e., a higher score indicates experiencing food insecurity more). 2.3. Child Eating Behaviors Child eating behaviors were reported by mothers using the CEBQ at SWIFT-Y research visit 1 [11]. The CEBQ consists of 35 questions in which parents indicate on a five-point Likert scale if their child “never”, “rarely”, “sometimes”, “often”, or “always” does the given behavior [11]. The answers are converted to numeric values, in which doing the behavior more often is scored higher (never = 1, rarely = 2, sometimes = 3, often = 4, always = 5) [11]. Eight domains (food responsiveness, emotional overeating, enjoyment of food, desire to drink, satiety responsiveness, slowness in eating, emotional undereating, and food fussiness) are derived from the questions, and a higher mean domain score indicates the child doing the domain behavior more frequently [11]. In line with previous research [13], mean scores of food responsiveness, emotional overeating, enjoyment of food, and desire to drink were added together to create the composite mean score for “food approach” (range 4–20), with higher scores reflecting doing the behaviors more frequently. Satiety responsiveness, slowness in eating, emotional undereating, and food fussiness were summed to create the composite mean score for “food avoidance” (range 4–20), with higher scores indicating that the child did the behaviors more frequently. Since its original development and validation among parents of youth aged 6–7 years old, the measure has been validated in parents of youth aged 6–16 years old, demonstrating that it has internal consistency and reliability in an older sample [30]. In addition, among the older sample, the measure displayed similar relationships with BMI (positive associations among all food approach domains and negative associations among all the food avoidance domains, except for emotional undereating) [30]. In the analytic sample, the food approach domain had a Cronbach alpha value of 0.73 and the food avoidance domain had a Cronbach alpha value of 0.53 [31]. 2.4. Youth’s BMI Youths’ height and weight measurements were obtained from two sources. If the youth completed their SWIFT-Y research visit (2022–2025, 11–16 years old), their height and weight were measured by research staff (n = 275). Youths’ height was measured to the nearest 0.5 cm using a stadiometer, and their weight was measured to the nearest 0.1 kg using a calibrated research digital scale [25]. If the child did not attend their research visit, their electronic health record was used to obtain their height and weight, which were assessed at a Kaiser Permanente Northern California routine well-child outpatient visit using standard clinical practices (n = 46). Measurements within a one-year range (six months before or after) of when the youth assented to participate in the study were used, and if the child had multiple visits that occurred within that range, the visit closest to the assent date was selected. The youth’s BMI percentile was determined using the Centers for Disease Control and Prevention age- and sex-specific growth charts [32]. 2.5. Breastfeeding Intensity and Duration Breastfeeding intensity and duration were assessed through detailed infant feeding surveys that mothers completed at the first three in-person SWIFT research visits (6–9 weeks, one year, and two years postpartum), which were administered by research staff, and through returned surveys that were mailed to them monthly for the first year [24,25]. In each survey, mothers reported the average number of breastmilk feeds, formula feeds, and total feeds in the past week [24,25]. Mothers also reported if they ever breastfed, if they were currently breastfeeding, and if they were not currently breastfeeding, when and why they discontinued breastfeeding [24,25]. To quantify lactation intensity and duration, as described in previous SWIFT publications, participants reported the average number of breastmilk feeds during the past seven days and were assigned a score between zero and one each month based on the percentage of feeds that month that were breastmilk [24,25,33]. A score of zero indicated that none of the feedings were breastmilk, and a score of one indicated that 100% of the feedings were breastmilk. The lactation duration and intensity ratio summary score was created by summing the monthly scores from birth to 12 months to provide a single continuous measure of breastfeeding intensity and duration across the first year [24,25,33]. The score ranged from zero to 12, with a higher score indicating more intense and longer breastfeeding [24,25,33]. Breastfeeding intensity and duration were also dichotomized according to the World Health Organization recommendations and in line with other SWIFT analyses [22,23,24,25,34,35,36,37]. Participants were classified into the “exclusive breastfeeding” group if 100% of their feeds were breastmilk from birth to six months and they were not fed any water, other liquids, or complementary foods. Participants were classified as “not exclusive breastfeeding” if <100% of their feeds were breastmilk from birth to six months, including if they were fed water, other liquids, or complementary foods. Breastfeeding duration was dichotomized as