Racial and ethnic minorities historically have worse outcomes during and after disasters as they already face a range of economic, political, and social barriers [23, 24]. These include but are not limited to racial discrimination, exploitation, limited English-speaking proficiency, and lack of legal status .
The overall prevalence of food insecurity in California between 2017 and 2019 was 9.9% having low food security and 3.6% with very low food security . Overall national levels of food insecurity were higher during this time period for Hispanic households (15.6%).
Food insecurity has increased for American children and adults since the start of COVID-19 pandemic with elevations in every state . Data from the Census Pulse Household Survey suggest that overall food insecurity tripled from 9.4 to 29.5% for households with children during COVID-19. Similarly, we report dramatic increases in food insecurity in our low income Latinx cohorts in the San Francisco Bay Area. These population groups already had low rates of food security. Prior to COVID-19, 76.9% of LEAD families reported having food security but during COVID-19 only 38.1% report food security. Similarly, HEN food security declined from 79.6% of adults having food security to 34.2%. Decreases in child food security were also significant with 87.9% secure prior to COVID declining 56.9% in LEAD and 89.2 to 54.1% in HEN. Our results in HEN and LEAD correspond with the high food insecurity rates (only 36% with food security) reported by US adults living < 250% below the federal poverty line during the COVID-19 pandemic by Wolfson and Leung (2020) .
Our findings of high food insecurity among three Latinx cohorts in the San Francisco Bay Area suggest rapid growth of food insecurity in urban areas impacted by COVID-19 shutdowns. Latinx individuals are among the most likely to be pushed into poverty as a result of pandemic-related unemployment [6, 27]. A significant percentage of our participants did not have any adult household members with employment at the time of the interview (more than one third of HEN and LEAD households and a quarter of TAB), and unemployment is strongly associated with food insecurity [6, 28]. Additionally, Latinx individuals, such as the individuals in cohorts, who faced food insecurity prior to COVID-19 likely have more severe food insecurity during the pandemic and are particularly at risk for adverse outcomes .
Common coping strategies for food insecurity prior to COVID-19 include use of food pantries, borrowing money, going to homes of friends or family for meals, or sending children to relatives . Some, if not all of these tactics to buffer families against food insecurity, are more limited given the recommended physical distancing measures being taken to prevent COVID-19 transmission . Even when food pantry resources are available, Latinx families compared with other groups may face language barriers or concerns about legal status which limit use . Shame can also result in people hiding hunger rather than going to food pantries and further increasing food vulnerability .
Differences among Latinx cohorts
There were large differences between our Latinx cohorts in levels of food insecurity. Part of this can be explained by socio-demographic differences by cohort. HEN and LEAD were recruited primarily at Zuckerberg San Francisco General Hospital, the local safety net county hospital with TAB recruited primarily at UCSF Benioff, a tertiary care academic hospital. The women recruited for HEN and LEAD cohorts had lower levels of eduation at recruitment (a low percentage reported any education beyond high school at recruitment (20.6% for LEAD and 26.1% for HEN) versus a much higher 65.7% for TAB). Low education status is the greatest predictor of unemployment  and since the emergence of COVID-19, employment losses are associated with low education levels . Higher rates of household and child food insecurity in HEN and LEAD versus a much lower level in TAB (60% with high food security) parallel the lower education levels and potential earning power in these cohorts [6, 28].
Additionally, over 90% of mothers from the HEN and LEAD were preferred/dominant Spanish-speakers while 65% of the TAB mothers were primary English-speaking. Being fluent in English leads to greater job opportunities  also potentially contributing to food insecurity [6, 28]. Our previous studies with HEN and LEAD have also found that the majority of adults in these households are recent immigrants and foreign-born [17,18,19], potentially placing these families at a greater disadvantage in terms of accessing food and other government resources .
There was little difference in reported crowding indices (the number sharing bedrooms and bathrooms) between these three cohorts and although there were only associations between these metrics and food insecurity in TAB and HEN possibly based on differences in sample size (smaller sample size and less power in LEAD). Furthermore, we did not see any differences in food insecurity and education level or employment status for LEAD and HEN possibly because there was little variation in education level with the majority not having any advanced education beyond high school. We did not assess for underemployment common among Latinx adults but HEN and LEAD may have higher levels of underemployment . Meanwhile for TAB, the association between education level and employment status and food insecurity was highly significant. Those with less education had greater food insecurity, possibly due to the heterogeneity in education levels for this cohort.
Necessary interventions for Latinx families
Our data suggest the urgent need for interventions for Latinx families with children, particularly those who are Spanish-speaking and with low education and employment levels in the context of the COVID-19 crisis. Data from our TAB cohort points to the heterogeneity among Latinx families in the Bay Area and the potential need for targeted interventions in specific communities. Due to urgent need, some safety net clinics in California have had creative responses in terms of bringing food to families. More funding and development of these efforts should be prioritized by state and county legislators [38, 39]. Furthermore, individuals from underrepresented groups tend to not have enough accessible information in their preferred language and mode of communication . For instance, 29% of Latinx in the United States have no English fluency . There can also be distrust of the government and health agencies due to poor history and absence cultural competency [36, 40]. The best implementations for COVID-19 relief will be made through the inclusivity of cultural richness, such as having information in multiple languages, modes of communication, and a diverse workforce to implement these interventions and work with diverse communities.
Risk for COVID-19 infection with household food insecurity
Previous researchers have noted an increased risk for COVID-19 morbidity and mortality due to low quality nutrition, obesity and associated metabolic disease . While crowding factors and poverty are associated with risk for COVID-19 infection and food insecurity, food insecurity may be independently associated with increased risk for COVID-19 infection through poorer immune health and increased susceptibility to disease. Previous studies, particularly related to risk for HIV infection in the context of HIV exposure suggest that food insecurity may be an independent risk factor for infectious disease . For example, other studies suggest that household food insecurity puts children in a more susceptible position for tuberculosis infection . In multivariable analyses adjusting for crowding, unemployment and Spanish language use in the TAB cohort, food insecurity score trended towards association with household COVID infection (OR 1.25, 95%CI 0.97–1.61). Further studies are needed with larger cohorts to assess the role between food insecurity, poor nutritional status and risk for infectious disease, inlcuding COVID-19.