Food insecurity measured with the HFSSM6 is strongly associated with a positive food insecurity screening using the HVSS2, introduced for household food insecurity screening during WIC certification visits in California in 2019. The specificity (probability of screening negative [i.e., food secure] given that one is food secure on the HFSSM6) of the HVSS2 as administered in WIC services was high. The negative predictive value (the proportion of negative screening results [i.e., food secure] that are also negative on the HFSSM6) was moderate-to-high at the different study time points. Sensitivity (probability of screening positive [i.e., food insecure] given that one is food insecure on the HFSSM6) of the HVSS2 as administered in WIC services was low, and positive predictive value (the proportion of positive screening results [i.e., food insecure] that are food insecure on the HFSSM6) was moderate-to-high at the different study time points. The present study evaluated the performance of the HVSS2 against both the 30-day and 12-month time frame versions of the HFSSM6, and found similar screening performance, suggesting that the instrument time frame does not have a substantial impact on the screening performance of the HVSS2. While the high specificity of the HVSS2 may help WIC to avoid directing referrals and other additional supports to households who are not food insecure, higher sensitivity of food security screening during the administration of WIC services is critical to providing appropriate referrals to other social services and to ensure those in greatest need of additional support are aware of those resources.
The concordance between the results of the HVSS2 and the HFSSM6, except for lower sensitivity, identified in this study extends prior research showing adequate performance of HVSS2 as a food insecurity screening tool in health care contexts, particularly pediatric care [18, 27], to WIC services. A prior study conducted in a pediatric dental setting identified high sensitivity (0.95) and specificity (0.84) of the HVSS2 against the HFSSM6 [28]. The reported sensitivity is much higher than in the present study (0.29–0.34 for food insecurity), and the specificity similar to the present study (0.90–0.93 for food insecurity). Differences in sensitivity may be due to differences in how questions were administered (in person during a dental visit compared to over the phone with WIC staff), or differences in the time between the two assessments (during the same dental visit compared to between 0 and 30 days between assessments). These differences in the timing and method of data collection for the two studies may differentially prime study participants to respond to the two instruments more or less similarly, and could explain the lower HVSS2 sensitivity found in the present study. Another study reported that a single item screener for food insecurity used in a pediatrics practice was found to have moderate sensitivity (0.59) and high specificity (0.87) [29], suggesting that short screening instruments for food insecurity may have consistently high specificity among families with young children.
The positive predictive value of the HVSS2 was moderate-to-high for food insecurity (0.57–0.83). Negative predictive value of the HVSS2 was moderate-to-high for food insecurity (0.54–0.74). Because predictive values are influenced by the prevalence of the condition being screened for [24], these results are expected. The high specificity of the HVSS2 identified in this study, and prior studies [28, 29], contributes to a high negative predictive value of the screening (i.e. the proportion of negative screening tests that do not have food insecurity according the HFSSM6) when performed in a population with a low prevalence of food insecurity (31.1% among the LACWS sample) and a lower negative predictive value when performed in a population with a higher prevalence of food insecurity (50.5% among the CVB sample) [24]. Similarly, the low sensitivity of the HVSS2 identified in this study, in contrast to the high sensitivity reported in a prior study [28], leads to a moderate positive predictive value (i.e. the proportion of positive screening tests that have food insecurity according to the HFSSM6) when performed in a population with a moderate prevalence of food insecurity (LACWS sample) and higher positive predictive value when performed in a population with a higher prevalence of food insecurity (CVB sample) [24]. In the LACWS sample, a small proportion of families who screened as food secure with the HVSS2 were identified as having very low food security with the HFSSM6 (3.9%), while in the CVB sample a higher proportion of those who screened as food secure with the HVSS2 were identified as having very low food security with the HFSSM6 (13.3%). This could be attributable to participants being more hesitant to report food insecurity to WIC staff during WIC services than to research staff over the phone (LACWS sample) or via an online data collection instrument (CVB sample).
Patterns of household food insecurity identified across the three time-points of the WIC Cash Value Benefit Study (between April 2021 and May 2022) found that over half of households had stable or improving food security (78.8% of households using HVSS2-derived patterns, 56.9% of households using HFSSM6-derived patterns). The screening performance of the HVSS2 for identifying longitudinal patterns of food insecurity was comparable to the cross-sectional assessments at all study time points (low sensitivity, high specificity, high positive predictive value, moderate negative predictive value). This finding is important, because household food insecurity varies in both severity and persistence, with relatively few households in the general US population experiencing persistent food insecurity [30, 31]. Given the much higher prevalence of food insecurity among WIC-participating households than non-participating households [12] and prior research that has reported that periods of transition between household food insecurity severity categories [19] and persistent household food insecurity may be acutely detrimental to children’s nutritional and growth outcomes [20], the ability to identify patterns of persistence and changes in household food insecurity may be critical to the WIC program maximizing its positive nutritional impacts [17].
This study has numerous strengths, including data from a randomly sampled survey and a longitudinal study that recruited all study-eligible individuals. The LA County WIC Survey is a large, recurrent study using a randomly selected sampling frame of WIC-participating households in LA County. Both the cross-sectional and the longitudinal studies included high quality food insecurity assessments with the HFSSM6. Survey and WIC administrative data allowed detailed characterization of the included households. The study also had limitations. Both measures of food insecurity used in this study are at the household level and therefore cannot be used to definitively identify food insecurity for WIC-participating women or children, only the households of WIC-participating women or children. The 6-item USDA HFSSM, a validated shortened version of the full 18-item HFSSM, was used to reduce survey respondent burden; however, the full 18-item HFSSM would be needed to assess whether the HVSS2 can identify food insecurity for specific individuals within WIC-participating households. Missing data for time 2 and time 3 in longitudinal assessment of food insecurity patterns could have altered how household trajectories were classified, and this assessment would benefit from replication in a population in which both food insecurity instruments are used in tandem at multiple time points. The population served by WIC in Los Angeles County is majority Hispanic, and all are low-income, and therefore generalization of the performance of the HVSS2 identified in this analysis should be done cautiously for demographically distinct populations. All WIC services during the study period were administered remotely due to the COVID-19 pandemic, which could impact the types of households participating in WIC and therefore the generalizability of the results, and it is possible that the performance of household food insecurity screening may differ between in-person and telephone-based assessments.
A qualitative study conducted prior to the COVID-19 pandemic reported that food insecurity screening was generally acceptable to caregivers of young children, and that WIC was the most commonly used resource to address household food insecurity among this population [32]. Given the broad reach of the WIC program among children under age 5 in the US [33, 34], embedding food insecurity screening in WIC services may represent a critical means of increasing the proportion of households with young children receiving food insecurity screenings who can then be appropriately referred to further food and social assistance [35]. Further, given that WIC programs systematically document data at the state-level, which is of great public health utility [36], food insecurity screening as part of WIC services may represent both a way of strengthening referrals to other services to improve household food security and of tracking the prevalence of household food security over time.