This cross-sectional study was a part of the first author’s PhD dissertation that was done from April 2016 to December 2016. The study population consisted of all pregnant women who met the following criteria. The inclusion criteria of this study were; being literate and being able to read and write in Farsi, Nulliparous and multiparous, in their 24-28 weeks of gestation (As we know the second trimester is the most convenient trimester of pregnancy and women have many concern in the other trimesters such as fear of abortion and preterm birth in the first and third trimester respectively. Such condition can obviously affect their response to study questions; on the other hand they attend health centers for performing some laboratory tests like GTT and Hb/Hct between 24-28 weeks of gestation. For this reason we considered this period as the best time for our study), Lack of known medical conditions, and Consent to participate in the study. Women with these characteristics were excluded: Lack of cooperation to participate in the study or not completing the study questionnaires.

Sampling was done using cluster random sampling. Ilam province in southwest of Iran has 10 cities with 197,896 women in reproductive age. All city in Ilam province were divided in five geographic regions (Central, North, South, East, and West) and each region was considered as a cluster. Health centers in each region were selected randomly for sampling. The appropriate sample size for each center was calculated and identified considering main sample size based on the number of prenatal documents. Since the number of prenatal documents covered by each urban health center was different, proportional random sampling was used to determine the sample size of the all centers. After explaining the purpose of research, informed consent was taken from the sabjects and they were assured that their information will remain confidential and they could leaves at any time they desired. If the subject had the inclusion criteria, they would be interviewed and questionnaires were completed by the participants.

Sample size was calculated using the following formula: (n=669, P: 6%; d: 0.018) [18]. Considering subject attrition of 20%, the final sample size was estimated 837 participants. Also sample size was determined for each of the goals in thesis. The sample size determined for the relationship between the social support and food insecurity estimated 329 pregnant women with regard to the prevalence of 31% (P:31%, d:0.05) [6]. Also regarding to cluster sampling method and calculating the design effect equal 2, the sample size reached to 658 pregnant women. Given that the sample size determined for the main purpose of the thesis was higher than the target sample size, the larger sample size (837) was considered.

Data collection tools were four questionnaires. Demographic and obstetric characteristics; Socio-economic Status, Perceived social support and Household Food Insecurity Access Scale (HFIAS) questionnaires.

Demographic and obstetric characteristics questionnaire was designed by the research team regarding women’s pregnancy age, husband’s age, ethnicity, gestational age, number of pregnancies, interval between pregnancies, the risk of chronic diseases (diabetes, hypertension), history of premature birth, low birth weight history, history of abortion, history of high birth weight over 4500 g and using supplements, etc.

Socio-economic status questionnaire included 22 questions and was designed by the research team. It contained questions on pregnant mother’s education, spouse’s education, job, husband’s occupation, the number of people living at home, the number of working people at home, monthly household income, individual’s independent income, household expenses per month, type of housing and ownership type, etc.

The validity of the questionnaire was evaluated through face and content validity, and its reliability was assessed using internal consistency, Cronbach’s alpha (0. 794). According to the Lavsheh table, the questionnaire had a high content validity, as mean of the relevance; questions’ simplicity and clarity were 0.94, 0.98 and 0.98, respectively.

Multidimensional Perceived Social Support Scale (MSPSS) is a 12-item scale, which measures perceived support in 3 domains: family, friends and a significant other. The scale consists of 12-items in which samples are classified into 3 groups of low, moderate and high social support. The lowest and highest scores for each question are 1 and 7, respectively. Total scores of answering the questions fall in three categories: Low (scores 12-48), Moderate (scores 49-68) and high (scores 69-84) social supports. Validity and reliability of this questionnaire have been approved by Zimet GD et al., for the first time, coefficient alpha for subscales and scale ranged from 0.85 to 0.91 that indicates good stability [23]. This instrument also has been validated in Iran by Bagherian-Sararoudi R et al., as Cronbach’s coefficient has been found to be 0.92 for the scale and 0.89, 0.92 and 0.87, for the friends, significant others and family subscales, respectively [24].

The food security status was assessed using HFIAS, developed by USAID’s Food and Nutrition Technical Assistance (FANTA) project used in measuring food security in each household. The scale consisted of 9 items in which samples were classified into 4 groups of food security, mild insecurity, moderate insecurity and severe insecurity. The lowest and highest scores for each question are zero and 3, respectively. Total scores of answering the questions fall in four categories: Food security (scores 0-1), Mild insecurity (scores 2-7), Moderate insecurity (scores 8-14) and Severe insecurity (scores 15-27) [25,26]. This instrument also has been validated in Iran by Mohammadi F et al., as the internal consistency of this questionnaire was 0.86 using Cronbach’s alpha coefficients [27].

Data were analysed using SPSS₁₉ software {IBM^© SPSS^© Statistics version 19 (IBM^© Corp., Armonk, NY, USA)}. The Kolmogorov Smirnov test was used to determine normal distribution of data. Continuous variables were expressed as mean ± SD according to distribution state. Categorical variables were expressed as numbers and percentage. Relationships between social support and food security were assessed by inferential statistics including Chi-square, and logistic regression. A p-value of less than 0.05 was considered significant throughout the study.

The Ethics Committee of SBMU approved the study design (SBMU. REC. 1394.112 dated September 7, 2015). Written informed consent was obtained from the participants after offering a comprehensive explanation of the study procedure.

Of 837 pregnant women, 103 were excluded from the study. So information was collected for 734 participants. The mean age of them was 28.73± 4.41 years (minimum and maximum ages of 18 and 35 years). The average number of pregnancies was 1.82±0.96 and 340 of the participants were primigravida. Considering the household members, 566 (77.2%) had 1 to 3 members. A total of 637 participants (86.8%) had one working person at home.

University graduated level had the highest frequency in participants and their husband’s with 50.8% and 48.2% respectively. The lowest and highest frequency of women’s career were self-employed 19 (2.6%) and Housewives 633(86.2%), respectively [Table/Fig-1].

Reviewing the status of participants’ Social support classes showed that 15.4%, 50.4% and 34.2% had low, moderate and high social supports, respectively. In addition results indicated that the mean±SD scores of the three subscale of MSPSS including the significant other, family and friends were (23.03±4.56), (22.39±5.00) and (16.04±6.55), respectively [Table/Fig-2].

In terms of household food security, the majority of participants (64.9%) had food security, 24.9%, 7.3% and 2.9% had mild, moderate and severe insecurities, respectively [Table/Fig-2].

Chi-square test results showed significant association between food security status and Social support, women’s education, husband’s education, women’s jobs, husband’s jobs, family size, household income, the average household expenditure and residence area (p <0.001) [Table/Fig-3].

Binary logistic regression analyses were used in investigating the association of social support with food insecurity. [Table/Fig-4] presents the estimated Odds Ratios (OR) and 95% CI of binary logistic regression analysis. The results of the logistic regression analysis (unadjusted) revealed that social support was a protective factor against food insecurity. The probability of household food insecurity was lower for women with moderate and high social support compared to those with lower social support. Subjects with moderate and high social support had 66% and 82% lower odds of food insecurity than those with lower social support (CI=0.229 – 0.546, OR=0.345) and (CI=0.116 – 0.303, OR=0.187). After adjusted for women’s age, husband’s age, women’s job, husband job, women’s education, husband’s education (Participants’ Demographic Data and their husband), results revealed that social support was a protective factor against food insecurity and participants with high social support had 73% lower odds of food insecurity than those with lower social support (CI=0.162 – 0.455, OR=0.272).

Model 3 has been adjusted for women’s age, husband’s age, women’s job, husband’s job, women’s education, husband’s education, household income, family size, the average household expenditure and residence area (Participants’ Demographic Data and their husband and household socioeconomic factors), the results revealed that social support was a protective factor against food insecurity and participants with moderate and high social support had 51% and 65% lower odds of food insecurity than those with lower social support [Table/Fig-4].

The result of this study showed that the majority of participants (64.9%) had food security while 35.1% had food insecurities among them 24.9%, 7.3% and 2.9%, corresponded to mild, moderate and severe insecurity, respectively. Various studies have shown the prevalence of food insecurity in developed countries to be between 10 to 11.2% while the corresponding percentage is between 7.5 to 73% in developing countries [28,29]. In Iran, according to different scales of household food security measurements, the overall prevalence of food insecurity was reported in the range of 20-60%. This percentage rises to 86% in female headed and low household income. It can be claimed that these results are acceptable by considering the overall situation of food security in Iran [2]. The studies regarding food insecurity on pregnant women are limited. Laraia BA et al., reported that the prevalence of food insecurity was 15% [9]. Recently, Yadegari L et al., have reported the prevalence of 30.9% in Rasht [8]. Factors related to food insecurity, including age, education of household head, loss or lack of permanent jobs, increasing the number of family members, loss of supporting food package, ethnicity and region’s eating habits [14]. In the present study, significant associations were observed between food security with parental education and occupation, family income and family size, while no significant relationship has been found between parental age and food security as people with higher education and income, lower household size, better socioeconomic status had food security which confirms the results of other studies [30-33]. It can be concluded, that by increasing parental education level, the knowledge, skills and attitudes regarding nutritional status will be improved. Furthermore, higher education provides better jobs and socioeconomic conditions [34,35]. Therefore, high quality nutrition will be more accessible. On the other side the number of children in the family is associated with increased training, school transportation, clothing and health care costs. The volume and number of food serving decreases by increasing household size, leading to food insecurity [7].

Researcher has also paid attention to the protective role, which a social support can have on household well-being, particularly the protective nature of social support on food security [36]. In this study, pregnant women had 15.4, 50.4 and 34.2 percent of low, moderate and high social supports scores, respectively. The mean score of perceived social support from husband, family and friends were 23.03±4.56, 22.39±5 and 16.04±6.55, respectively [Table/Fig-2]. The results revealed that the women in this study had high levels of social support. Spouse support had the highest rating on social support which was consistent with the results of Bahiraii A et al., [37]. The average score of perceived social support was calculated to be 61.46±13.26. In the study by Jamshidimanesh M et al., in Iran the average score of perceived social support is estimated to be 60.55 among pregnant women. In addition, they showed that participants can get higher perceived social support from partners rather than family and friends [38], which confirms the results of the present study.

In this study, the logistic regression analysis showed that social support played a protective role in food insecurity and people with high social support had 82% less food insecurity than those with low social support. The proportion of food insecurity reached 73% by adjusting variables such as gender, age, education and occupation of parents. This reduced amount is due to the influence of the mentioned variables. It should be noted that these results are consistent with those of Miller ME, who demonstrated that people with high social support had less food insecurity (about 51%) [20]. Martin KS et al., who investigated the relationship between social capital and household food security in the US which showed that social capital was significantly associated with food security and households with higher levels of social capital are less likely to experience hunger [39]. In contrast to the work of Martin KS et al., in the US, Hadley C et al., conducted their study on the relationship between food security and perceived social support in rural Tanzania. They reported significant associations between food insecurity and both ethnic group and material wealth, in addition researchers revealed, that higher social support is associated with food security, suggesting that it may protect against the occurrence of seasonal food insecurity [36]. Findings of their study are in agreement with those of the present study.

A study in Nigeria in 2011 showed that the level of social capital through membership and access to formal and informal organizations is effective in improvement of social welfare, particularly poverty reduction and food insecurity [40]. In addition, Walker JL et al., showed that dimensions of social trust had deep connection to food security and health so that high social capital had inverse association with household food insecurity and was positively correlated with health status [22].

One of the strengths of this study is that we clearly observed the importance of the psychosocial aspects of food security, especially social support. Studying food security in pregnancy is necessary because of it’s positive effects on pregnancy outcome. On the other hand the number of studies in this field is limited, and this study can be a beginning point for such studies in different populations. The limitation of this study was its cross-sectional nature. Therefore, it may not be realized that food insecurity is a temporary or chronic condition in the household.

The results of this study show that several factors such as socioeconomic and psychosocial factors, especially social support have significant relationships with household food insecurity. Therefore the design, implementation and evaluation of appropriate interventions about the food security in the individual and household levels are necessary by governments, policy makers and specialists. In addition social and cultural factors affecting food security must be studied on mother and foetus.