Is the modified household food security survey (HFSS) questionnaire a practical tool for screening food insecurity? Evidence from northwest of Iran

Adequate supplies of food and food security (FS) are the fundamental aspects of human societies, and considered one of the pivotal factors of individual and social health [1, 2]. FS is the result of efficient food system operating. An efficient food system positively affects all aspects of FS [3]. The four dimensions of FS are food availability, access to food, utilization, and stability. Food availability indicates if sufficient food is readily at people’s disposal. Access to food makes sure whether all people have enough resources to supply the food they require. Utilization includes the food preparation, food distribution, water, sanitation, and health care practices. Stability is another dimension of FS, which guarantees continuation of the three other dimensions over time [4].

Food insecurity (FI) is defined as “limited or uncertain access to nutritionally enough and safe foods, or limited ability to acquire acceptable foods through socially admitted ways” [5, 6]. It is estimated that more than 852 million people worldwide do not have access to sufficient food for healthy and active lives; approximately nine million of these people are in developed nations and the remaining part live in the developing countries [7]. There is evidence indicating that about 20% of the Iranian population suffers from an inadequate dietary intake of energy and protein, while micronutrient deficiencies appears to be much higher [8, 9] . This condition can negatively affect quality of life, and adversely influence physical, social and mental well-being [10‐12]. Therefore, providing useful and practical tools for assessment of FS is required to help with screening FS and consequently taking appropriate measures accordingly [8, 13]. Various indirect indicators, such as income, food consumption and nutritional situation have been previously used to monitor FI status in different countries [14‐18]. Furthermore, well-designed questionnaires which are considered to be reliable, simple and low cost resources have been applied for direct assessment of FS [15, 19‐28]. Radimer/Cornell questionnaire (12-item) has been found to be a highly reliable tool for assessing FI in Households with children [25]. In addition, one of the widely used measures of FS status is the 18-item Household Food Security Survey (HFSS) [20, 26]. Blumberg et al. [20] described an abbreviated version of the HFSS consisting of only six items (short questionnaire). They found that the short questionnaire may be adequate to identify a household’s level of FS when survey resources do not permit use of the full scale. They suggested that this measure maintains many of the essential indicators of FS, despite its abbreviated form [20]. For example, this questionnaire does not rely exclusively on specific measures of intake insufficiency. Other indicators capture self-perceived nutritional inadequacy, household food depletion, disrupted eating habits, and the repetitive pattern of decreased food consumption [20]. Collectively, Blumberg et al. [20] indicated that this form was a short but potentially valid and sufficiently beneficial instrument for national surveys and some state/local programs. Due to the high prevalence of FI in Iran, the use of simple, low-cost, rapid and useful tool for assessing FS status is necessary. Most of the previous studies have indicated that it is essential to determine the applicability of questionnaires in every country they are to be applied. This is due to the social and cultural diversities in various countries [8, 22, 29‐31]. Accordingly, the aim of the present study was to assess the applicability of the short questionnaire (adapted from the HFSS questionnaire) for screening FI and to evaluate the prevalence of FI in northwest of Iran.

Table 1 shows the basic characteristics of the study participants. The study sample included 285 males (57%) and 215 females (43%), with mean age of 35.3 (SD 13.7) years. Total prevalence of overweight and obesity was 52.4% (95% CI: 46.1–58.3). In addition, 27.8% (95% CI: 21.9–33.4) of the participants were underweight. The average number of persons per household was 5.6 (SD 1.2) and the mean income per capita was $ 395 (SD 80.6). 19.6% (95% CI: 14.3–25.9) of the household heads had attended school for less than 5 years, and 13.0% (95% CI: 10.1–19.9) of them had the lowest level of occupations. Based on dietary records, the prevalence of hunger and hidden hunger in the area was 30.8% (95% CI: 25.8–39.2) and 46.0% (95% CI: 41.3–50.8), respectively. Overall, 23.2% (95% CI: 18.1–28.2) of the subjects were classified as food secures.

According to Table 2, in subjects who were insecure in terms of hunger, the average energy intake was 69.2% of DRIs (95% CI: 63.6–84.9). In subjects who were secure in terms of hidden hunger, the median intake of energy was 123.7% of DRIs (95% CI: 112.5–133.9) which was above the recommended amount (90–110% of DRIs) [9]. Furthermore, the average intake of dietary calcium in food insecure group was 68.7% of the DRIs (95% CI: 52.7–77.4), which was less than the recommended amount.

Affirmative responses to questions of the short questionnaire are shown in Supplemental Table 1. The findings demonstrated that more participants of the study (67.8%) gave affirmative answer to question six indicating that the majority of subjects could not afford to eat balanced meals. In addition, 39.4% of the participants gave the affirmative response to two or more questions and were considered as food insecure.

According to Table 3, Sensitivity, specificity and accuracy of short questionnaire for assessment of hunger were 92.8% (95% CI: 87.3–95.9), 84.2% (95% CI: 79.3–89.3), and 87% (95% CI: 84–90.2), respectively. These values were 21.6% (95% CI: 15.7–29.9), 92.3% (95% CI: 88.7–99.4), and 53.4% (95% CI: 47.9–59.8) for hidden hunger. Furthermore, the kappa agreement between two questionnaires in terms of hunger and hidden hunger was 76% (95% CI: 72–81, p <  0.0005) and 16% (95% CI: 11–29, p <  0.001), respectively. FI prevalence showed a positive association with household size (P for trend < 0.001), while there was a negative correlation between FI and income, educational level of the household head, and occupational position of the head (P for trend < 0.001; Table 4). FI significantly enhanced the risk of underweight (RR = 12.8, 95% CI: 7.8–23.9, p <  0.001), while it reduced the risk of overweight and obesity (RR = 0.39, 95% CI: 0.19–0.74, p <  0.001 and RR = 0.22, 95% CI: 0.11–0.68, p <  0.005, Table 5).

Association between the monthly income per capita and frequency of food groups / items is presented in Supplemental Table 2. The results indicated that consumption frequency of meat, dairy products, vegetable and fruit groups were directly correlated with monthly income per capita (r = 0.522, 0.309, 0.311, and 0.568 respectively, p-value < 0.01). Although the association between the consumption frequency of cereal and grain groups, and monthly income was insignificant (r = − 0.017 and − 0.073 respectively, p-value> 0.05), frequency of rice consumption was directly associated with income (r = 0.231, p-value < 0.05) and frequency of bread consumption was inversely correlated with income (r = − 0.421, p <  0.05).

The mean consumption frequency of meat, dairy products, fruits, vegetables, and rice was significantly lower in food insecure group than food secure group (p-value < 0.001), while the average frequency of bread consumption was higher in food insecure group compared to food secure group (p-value< 0.001) (Table 6). FI was significantly associated with a less frequent consumption of fruits, vegetables, dairy products, rice and meat (Supplemental Table 3).

This cross-sectional investigation was designed to evaluate the applicability of a short questionnaire for screening FI in northwest of Iran. In our previous study, we assessed some epidemiologic indicators of this short form questionnaire on the study population (n = 300) which was a different location in Tabriz (Asadabadi Medical Centre) [34]. While, in the present work, the target area was a large industrial company in northwest of Iran, with different people and larger sample size (i.e. n = 550). Moreover, in the current study, the correlation between the results of this questionnaire and the criteria theoretically related to FI was also assessed to determine the applicability of the short questionnaire in study population. In the study by Dastgiri et al. [34], dietary data was obtained using a 24-h food-recall questionnaire. Whereas we compared the data collected from the short questionnaire with information provided by dietary records in the present research.

The results of nutritional status based on dietary records in the study population revealed that the prevalence of hunger and hidden hunger was 30.8 and 46.0%, respectively. Overall 23.2% of the subjects were classified as food secure, indicating that FI was prevalent in the studied population. The meta-analysis of 31 studies in Iran demonstrated that the prevalence of FI was 49.2% in households, 67.0% in children, 61.0% in mothers, 49.0% in adolescents, and 65.0% in the elderly [14]. The previous studies on Iranian households have also shown that more than 20% of the population has insufficient energy and protein intakes, while micronutrient deficiency is estimated to be much higher [8, 47]. In addition, studies in Iran have demonstrated that the average intakes of vitamin B₂, vitamin A, and calcium are lower than recommended [47, 48]. In the present study, the average consumption of dietary calcium in the food insecure group was 68.7% (95% CI: 52.7–77.4), which was less than the recommended value. In addition, in subjects who were secure in terms of hunger and hidden hunger, the average rate of energy intake compared to DRIs was 123.7%, which was higher than the recommended intake (90–110% of DRI). This finding might demonstrate that a number of food secure subjects have been able to meet their micronutrients’ requirements through overeating; in other words, adequate intake of micronutrients was the result of overeating rather than following a balanced diet. This might indicate inappropriate diet and eating habits in the study population. The prevalence of FI was 39.4% based on the short questionnaire, which was higher than that reported in the study by Dastgiri and colleagues in Iran [34]. In addition, in the study by Gulliford and his colleagues [19] who also applied this questionnaire, FI incidence was 25.2% in Trinidad and Tobago. In the current study, the percentages of affirmative answers to the six questions were 25, 16, 24, 18.6, 16.6, and 68.8%, respectively, which were higher than the those reported in the studies performed in the United States and Caribbean studies [19, 29]. The majority of the participants (68.7%) gave affirmative responses to question six, which was well above the values reported by other similar studies [29, 34]. The results of our study demonstrated that, although sensitivity, specificity, and accuracy of the short questionnaire were significantly low for surveillance of FI in terms of hidden hunger, the questionnaire can adequately efficiently detect hunger or severe FI. Furthermore, the degree of agreement between the results of the short questionnaire and dietary records was good for screening FI in terms of hunger. Conversely, the agreement was poor for assessing FI in terms of hidden hunger. In line with the present study, Dastgiri et al. [34] reported that the epidemiologic indicators of the short questionnaire were acceptably high for assessment of FI in terms of hunger in the area of their study. Low sensitivity and accuracy of the short questionnaire for screening FI in terms of hidden hunger might be attributed to the limited questions of the questionnaire. Moreover, economic status is one of the most significant determinants of FI, and the questions of this questionnaire have been designed in a way to focus on the financial constraints; as a result, other important factors remain overlooked in this regard. Note that significant groups of people who are food insecure in terms of hidden hunger, do have economic and physical access to foods, but have poor food choices most likely due to low nutrition literacy and lack of awareness in selecting a healthy and balanced diet. Consequently, the absence of questions targeting the subjects’ knowledge on healthy eating habits and balanced eating patterns in this questionnaire probably reduced the value of epidemiological indicators of the short form for assessing FI in terms of hidden hunger.

To examine the applicability of the short questionnaire, the association between the results of this form, and the criteria theoretically related to FI as well as the consequences of household FI were assessed. Consistent with previous findings, our study revealed an association between FI and demographic and socio-economic status in the study population. Similar to some previous results [17, 43, 49], FI prevalence showed a significant positive correlation with household size, while there was a statistically significant negative association between FI and income, educational level, as well as the occupational position of the household head. In relation to the association between FI and BMI, several studies have found that FI is linked to BMI [19, 34, 50‐52]. Gulliford et al. [19] reported that FI was prevalent across all BMI levels. On the other hand, FI was significantly associated with underweight, but not with obesity. The findings of Dastgiri et al. [34] suggested that FI enhanced the risk of underweight and reduced the rate of overweight or obesity. In the study of Cheung et al. [53], FI was associated with increased BMI. A study demonstrated that although mild or moderate FI was associated with a higher rate of obesity, severe FI was related to the lower risk of obesity [54]. In the current study, FI significantly increased the risk of underweight while reducing the risk of overweight and obesity. Several studies have also shown that there is a correlation between FI and food consumption frequency [8, 46, 55]. In a study by Gulliford et al. [19], food insecure subjects were less likely to be frequent consumers of fruit or green vegetables or salads. In a recent study, FI negatively affected the fruit and vegetable consumption among families with individuals under 18 years [55]. Zerafati Shoae et al. [8] reported that as the severity of FI increased, the frequency of consumption of fruits, vegetables, dairy products, red meat and rice significantly diminished, while the frequency of consumption of potato and bread increased markedly. These observations are consistent with our results indicating that the mean consumption frequency of meat, dairy products, fruits, vegetables, and rice was significantly lower in the food insecure group, while the average consumption frequency of bread was significantly higher in the food insecure group compared to the food secure group. In addition, the subjects who were food insecure or unable to afford a balanced diet were less likely to eat fruits, vegetables, dairy products, rice, and meat.

Consistent with previous studies, the present research lends general supports to the applicability of the short-form questionnaire in screening FI in terms of hunger in Iranian social, economic, and cultural context. However, our findings indicated that the scale works poorly in assessing FI in terms of hidden hunger, and further research is required to improve the applicability of the scale in this regard. Additionally, our findings demonstrated that there was a strong correlation between the results of the short-form questionnaire and the criteria theoretically related to FI. Furthermore, in agreement with the results of the previous studies, the present research, using the short form, found that abnormal BMI could be one of the consequences of FI. On the other hand, inappropriate frequency of food consumption was also observed as a result of FI in the present study. On the whole, the findings of this study indicated that the short-form questionnaire was a simple, low-cost, rapid, and practical tool for screening severe FI. According to the high prevalence of FI, macroeconomic policies should be implemented and necessary measures must be taken to help people meet their nutritional requirements; these measures could include promoting the nutritional knowledge of household members as well as supporting the households [56].

It can be concluded that FI was frequent in north-west of Iran. The findings indicated that the short questionnaire was a practical tool for screening FI in terms of hunger. Although this questionnaire was a weak tool for assessing all dimensions of FI, it may be used as a simple, low-cost, rapid and practical tool for the screening and surveillance of severe FI in similar areas. Further investigations are needed to validate the instrument for national surveys and state/local programs.