Food Insecurity is Associated with Additional Severe Maternal Hardships during Pregnancy: Results from the 2012-2014 California Maternal Infant Health Assessment

Background: Assess the associations between food insecurity and ten additional maternal hardships experienced during pregnancy. Methods: Data on 14,274 low-/middle-income women from the statewide-representative 2010-2012 California Maternal and Infant Health Assessment were used to estimate food security status and prevalence of additional maternal hardships. Multinomial logistic regression was used to assess the associations between food security status—secure, marginal, low and very low—and these hardships. Results: Food insecurity was common (23%) among all pregnant women in California. Among women with incomes at or below 400% of the income to federal poverty guideline ratio, nine of ten hardships were independently associated with food security status; only the respondent or someone close to the respondent having a problem with alcohol or drugs was not independently associated with food security status after adjusting for socioeconomic factors. Husband/partner losing a job, depressive symptoms, not having practical support and intimate partner violence were consistently associated with marginal, low and very low food security status. Each additional severe maternal hardship a woman experienced during pregnancy was associated with a 36% greater risk of reporting marginal food security (Relative Risk Ratio 1.36, 95% CI: 1.27, 1.47), 54% for low food security (Relative Risk Ratio 1.54, 95% CI: 1.44, 1.64), and 99% for very low food security (Relative Risk Ratio 1.99, 95% CI: 1.83, 2.15). Conclusions: Food security status was strongly linked with several serious maternal hardships that could jeopardize maternal and/or infant health. implications: that

state prevalences have not been reported. Food insecurity is a common economic hardship faced by pregnant women but is often not evaluated in clinical settings. Few real-time surveillance systems directly assess the prevalence of food insecurity, the degree of its severity, or attendant risks during pregnancy.
Taking a life-course approach, maternal hardships experienced during pregnancy may have short-and long-term health consequences 10 . Severe maternal hardships are associated with elevated levels of stress, poor eating behaviors, and weight gain 14,15 that may affect the mother's health later in her life. For the infant, these in uence birth outcomes 16 , and children exposed in utero to severe maternal stress have been shown to have poor stress management later in life 17,18 and to have developed insulin resistance 19 .
Food insecurity is a relatively common hardship that may co-exist with a number of additional severe maternal hardships such as other nancial hardships, adverse life events, and emotional stressors 4,20 . Despite this, the association between food security status and additional severe maternal hardships during pregnancy has not, to our knowledge, been explored using a population-based sample. It is important to document the prevalence and severity of food insecurity and accompanying maternal hardship to inform policies and identify the resources needed by programs serving food-insecure pregnant women. The objective of this paper was to assess the extent to which food insecurity was accompanied by additional severe maternal hardships during pregnancy among a representative sample of women who had live births in California during 2010-2012.

Methods
MIHA 11 is an annual, statewide cross-sectional survey of a representative sample of California women who recently had a live birth, excluding women under the age of 15 years, non-residents, and women with multiple births greater than three.
MIHA is conducted by the California Department of Public Health using federal Title V funds.
MIHA uses random sampling, strati ed on county/region of residence, African-American race, and, during the study period of interest, enrollment in the Women, Infant and Child Supplemental Nutrition program. Survey data was collected from 20,480 women who recently gave a live birth during 2010-2012. MIHA maintained an annual response rate of approximately 70% over the three years. The MIHA data were weighted to represent all California-resident women 15 years of age and older in California with a singleton, twin or triplet live birth during each survey year. We excluded 3,108 sampled women with household incomes above 400% of the Federal Poverty Guidelines (FPG) (e.g., $74,000 for a family of three in 2011) 21 slightly higher than the median household income for California of $70,400 22 because few women (< 0.6%) with household incomes > 400% FPG experienced food insecurity. In addition, women with missing household income information were initially excluded, although these values were later imputed for sensitivity analyses (n = 1,479), had missing food security information (n = 112), or were in a stratum with a single sampling unit (n = 2). These 15,779 observations were used to estimate the overall prevalence of food security status. Women were excluded from additional analyses if they reported a racial/ethnic group other than White, Black, Latina, or Asian/Paci c Islander (n = 899) or had incomplete maternal hardship information (n = 606). The nal analytic sample consisted of 14,274 women.

Dependent Variable-Food Insecurity
During the study period, the MIHA survey incorporated the validated 6-item food security scale developed by the United States Department of Agriculture 23 , slightly modi ed to ask about the woman's experience with food security during her most recent pregnancy rather than the past 12 months (Table 1). 4 Not responding a rmatively to any of the six questions indicated the woman was food secure during pregnancy. One a rmative response indicated marginal food security. More than one a rmative response indicated food insecurity, further de ned as low food security (2-4 a rmative responses) or very low food security (5-6 a rmative responses). ; Latinas were the only group with enough US-born and foreign-born women to create two groups based on nativity. There were too few women of American Indian or "other" race to include in these analyses.
Apart from food insecurity, MIHA assesses ten additional severe maternal hardships including nancial hardships, adverse life events, and emotional stressors. Participants responded to the statement: "Here are a few things that might happen to some women during their pregnancies. Please tell us if any of these things happened to you during your most recent pregnancy." All response options were yes/no. Financial hardships during pregnancy included: the respondent losing a job or the respondent's husband/partner losing a job, or the respondent reporting being homeless or not having a regular place to sleep. Adverse life events during pregnancy included: the respondent becoming separated or divorced, someone close to the respondent having an alcohol or drug problem, someone close to the respondent going to jail, or the respondent experiencing intimate partner violence. Emotional Stressors during pregnancy included: not having someone she could turn to for emotional support (e.g., someone to listen to or comfort her when needed), not having someone she could turn to if she needed practical help (e.g., like getting a ride somewhere or help with shopping or cooking a meal), and experiencing depressive symptoms. We also created a continuous variable of the total number of maternal hardships (0-10) each woman experienced during her pregnancy.

Analysis
Prevalence estimates and con dence intervals of maternal characteristics and food security status were estimated using survey-weighted tabulations. Chi squared test was used to assess the bivariate relationship between food security status and each maternal hardship. Multinomial logistic regression was implemented to estimate the relative risk ratios for the ten severe maternal hardships in relation to being marginally food secure, low food secure, or very low food secure compared to food secure. The model accounted for year of survey, maternal race/ethnicity, age, education, language spoken at home, number of people in the household, income as the percent of the poverty guidelines, insurance coverage, and marital status.
Missing income was imputed using ordered logistic regressions across ve imputations and sensitivity analyses were conducted with the imputed datasets to determine if our ndings were robust after accounting for the missing values. All socioeconomic/demographic, hardship, and food security data were included in the imputation model. The imputed datasets contained an additional 875 women. All analyses were conducted using Stata (version 12.0, StataCorp, College Station, TX).
The research for this study was conducted with approval from two Institutional Review Boards: the California Department of Health and Human Services, and the University of California, San Francisco.

Results
Among women with household incomes ≤400% FPG, 65.1% reported being fully food secure during pregnancy ( Figure 1).
The prevalence of food insecurity was 23.4%; with 16.1% experiencing low food security and 7.3% experiencing very low food security. An additional 11.5% reported marginal food security. Table 2 provides a description of the analytic sample of pregnant women with household incomes ≤400% FPG who participated in MIHA 2010 -2012 and who had complete information on the variables that were studied. A higher percent of women reported food insecurity who: were Latina or Black compared to white, younger compared to older age, had a lower educational level compared to college graduate; primarily spoke Spanish compare to English at home; had ve or more people compared to four or less dependent on the household income; had income below compared to above the federal poverty level; were either uninsured or lacked private insurance compared to those with private insurance; or were not married compared to married. For each of the ten maternal hardships that were assessed, a stepwise gradient in hardship prevalence was observed by food security status (Figure 2). The prevalence estimates for each maternal hardship were statistically signi cantly higher for food-insecure women compared to women who were food-secure (p<0.05). The most common maternal hardship for any category of food insecurity status was prenatal depressive symptoms, with over one-half (54.4%) of women with very low food security reporting depressive symptoms during pregnancy, and one-third of women with low or moderate food security reporting those symptoms (34.8% and 31.8%, respectively). Three out of four women who reported food insecurity experienced at least one additional maternal hardship (data not shown).
The results from the multinomial logistic regression for the independent associations between food security status and the ten additional severe maternal hardships are shown in Table 3. The relative risk ratio for each maternal hardship on food security status, after adjusting for demographic and socioeconomic factors, was statistically signi cant for all but one hardship; having someone close who had an alcohol or drug problem. Having a husband or partner lose a job, experiencing depressive symptoms, having no practical support, and experiencing intimate partner violence were independently associated with food insecurity consistently across all levels, with the point estimate much higher among the very low food secure group. The respondent losing her job was associated with both low and very low food security. Not having emotional support, the respondent or someone close to the respondent going to jail, and homelessness were associated only with very low food security. While elevated, the respondent or someone close to the respondent having a problem with alcohol or drugs was not signi cantly associated with any level of food insecurity. Our sensitivity analysis, which included women with imputed household incomes, produced consistent results (data not shown). Table 3 Adjusted relative risk ratios for the associations between several maternal hardships and food security status, adjusting for socioeconomic characteristics among postpartum women with incomes ≤400% FPG in California, 2010 -2012 (n=14,274) Marginally

Discussion
Food insecurity was very common in this large, representative sample of California women with household incomes below 400% of the federal poverty guidelines; greater than a third of women reported either marginal food security or being food insecure during pregnancy. Furthermore, this study found food insecurity was seldom the only severe maternal hardship faced by pregnant women. As expected, prevalence of food insecurity during pregnancy was higher among low-income women; however, food insecurity was reported among women with household incomes as high as 300-400% of the federal poverty guidelines. This may suggest that federal, state and local assistance programs should relax strict income criteria for program participation. During pregnancy, many families experience dynamic employment and nancial changes 2 . Women from low-income households as well as those who were not married, lived in larger families, and reported additional maternal hardships were associated with greater risk for food insecurity.
Similar to those who reported low or very low food security, women who reported marginal food security during pregnancy were less educated, had lower-income, were not married, and more likely to report maternal hardships compared to their food secure counterparts. These results support the notion that women facing marginal food security are not only distinct from women who are fully food secure but have vulnerabilities comparable to those experienced by women reporting food insecurity 3,5,24 , and therefore marginal food security should be assessed, monitored, and addressed in addition to food insecurity 25 , especially during pregnancy.
While it is unsurprising that food insecurity is associated with an array of maternal hardships, the prevalence and magnitude of the cooccurrences has not been previously measured. The vast majority of women who experienced marginal, low or very low food security reported one or more additional maternal hardships. Most notable were depressive symptoms, job loss, lack of practical support, and intimate partner violence as these were among the most prevalent and were associated with all levels of food insecurity. At greatest risk were the over seven percent of women who reported very low food security during pregnancy. This group had the highest prevalence of all maternal hardships reported. These ndings suggest that pregnant women experiencing food insecurity not only face a threat to their nutritional status but also to their overall well-being and the health of their fetus. The strong and consistent association between levels of food security and number of maternal hardships, independent of demographic and socioeconomic factors, may suggest that food security status is a re ection of major adverse life events and could be used for screening for additional social needs.
A causal relationship between food security status and severe maternal hardships was not the focus of this analysis and is di cult to determine; causal direction cannot be inferred. However, the strong association between food security status and severe maternal hardships suggest that both must be addressed during prenatal care, nutritional services, and other social services. An excellent example of organized screening efforts are guidelines from the American Academy of Pediatrics (AAP) 26 and the American College of Obstetrics and Gynecology (ACOG) 27 that promote screening and referrals for social determinants of health, food security, maternal depression, maternal hardships, and to bolster emotional, physical and social support, although consistency of screening implementation is not known. The California Department of Public Health also recommends that providers screen for these social issues and food security during pregnancy within the comprehensive Initial and Trimester Assessment and Care Plan Program 28 through the Comprehensive Perinatal Services Program. The current AAP policy statement encourages pediatricians to promote food security for all families with young children in pediatric settings by screening for food insecurity and other social needs at routine health maintenance visits as well as to advocate at the local, state and federal level for policies and programs that support acquisition of nutritious foods for all families pediatrics 29 . These guidelines for screening must be instituted as a standard of care within the prenatal care setting to ensure social needs of pregnant women are being met. Until we have an effective preventive approach, we must put in place and bolster routine screening and referral services-not just at public hospitals/clinics but at most prenatal care sites, and nutrition programs like Women, Infant and Children Supplemental Nutrition Program (WIC) should have the resources to screen and provide referrals on a wide range of maternal hardships. Social policies could address the attendant stress-related outcomes and mental and somatic health consequences that result from these severe maternal hardships by ensuring enrollment in safety net programs and designing new interventions, as done in other industrialized countries, to shield pregnant women from falling into or worsening their poverty condition. 30 Points of interventions and referral must be identi ed to assist with communication among programs.
Our cross-sectional analysis may have been subject to selection bias, as the severely food insecure women and the women who faced the most hardships may have consistently chosen not to participate in the survey. Our analysis would subsequently underestimate the prevalence of food insecurity in pregnancy and potentially its co-occurrence with additional hardships. Additionally, our data is roughly eight years old and examines only Californian women which reduces its generalizability. However, state representative data on pregnant women and hardships is di cult to procure, and has not been subsequently analyzed. Additionally, while have been changes in the California economy, it is unclear if the relationships between food insecurity and hardships would change drastically over time.
A key strength of this study, is it is the rst to our knowledge that measured maternal hardships in pregnancy in a representative sample and examined the cooccurrence of hardships with food security status. While this paper did not seek to nor identify causal relationships between food security status and these various hardships, by measuring and highlighting the co-occurrence of food insecurity and hardships, this paper contextualizes pregnant women in California for programs that aim to help them.

Public Health Implications
Targeted screening for severe maternal hardships is necessary to identify needed services and provide referrals, especially among socioeconomically disadvantaged maternity populations. An example of the importance of real time screening is the Prenatal Risk Overview, a screening mechanism created by the Minneapolis Health Department that was used to assess the prevalence of psychological risk during pregnancy among low-income women seeking prenatal service in four community clinics 31 . Use of this screener between 2005 and 2007 identi ed high frequency of psychological risk among pregnant women: 75% had a lack of social support, 48% housing instability, 32% food insecurity, 25% drug use, 23% smoking and alcohol use, 18% depression, 9% physical/sexual abuse and 7% partner violence. Sixty percent were classi ed as high risk in one or more domains. Research has found that among African-Americans, each additional negative life event has been associated with lower gestational age, and higher levels of event distress were associated with lower birth weight 16 . Therefore, the number of hardships faced is itself a predictor of adverse outcomes, over and above the nature of any particular hardship. More work is needed to help pregnant women address each of the several maternal hardships that they may face during pregnancy.
Food insecurity is a very common adverse pregnancy condition, and has been associated with nancial hardship, stress and mental health, as well as poor nutrition. Food insecurity has persisted at the same levels for the last decade 4,11 , suggesting that food insecurity must be addressed with multifactorial strategies 3 . Federal programs that address food insecurity and well-being are more critical than ever when one in every three pregnant women of moderate-or low-income face food insecurity. Because food insecurity is often not an isolated problem, programs that provide services to pregnant women need to not only screen for food insecurity but other severe hardships as well, to bolster referrals and help inform future interventions. Strategies must address social inequities and eliminating root causes of health disparities.
Interventions are needed that go beyond acute situations and change the cycle of poverty, food insecurity, hardship and health outcomes for future generations.

Declarations
Ethics approval and consent to participate