Our results showed that a diet rich in fruit and vegetables, consistent with higher intakes of vitamin C, vitamin B9, vitamin K, and dietary fiber was associated with a lower probability of SARS-CoV-2 infection, as reflected by seroprevalence data. We also observed that higher intakes of dairy products and calcium were associated with higher odds of SARS-CoV-2 infection. More specifically, higher intakes of vitamins C, K (from plant-based foods), and B9, as well as fruits and vegetables were associated with a lower probability of a symptomatic SARS-CoV-2 infection.
To the best of our knowledge, our study was the first prospective study to use objective serology data, coupled with detailed dietary assessment, to examine the associations between usual diet and the odds of having been infected by SARS-CoV-2. A prospective Spanish study in the SUN cohort observed an association between a better adherence to the Mediterranean diet and a lower risk of SARS-CoV-2 infection . Nevertheless, the outcome assessment in this study was based on self-reported questionnaires collecting SARS-CoV-2 diagnostic tests, and not seroprevalence. On the other hand, a study in the UK-Biobank reported a protective association for higher intakes of coffee and vegetables and higher odds of infection associated with higher intakes of processed meat . However, this study was conducted only among participants who sought COVID-19 testing, and based on a non-detailed dietary assessment, using a simplified 17-item questionnaire.
Vitamin C is a well-known actor of the immune system and could help prevent SARS-CoV-2 infection notably through an enhanced innate immune response with the maintenance of the epithelial barrier defending against pathogens and the promotion of phagocytosis (regulation of neutrophil function, notably increased chemotactic response) [6, 11, 34]. Although rare in developed countries, severe vitamin C deficiency (scurvy) has been observed to predispose individuals to infections, especially of the respiratory tract, like pneumonia . Yet, randomized trials have mostly failed to demonstrate a beneficial effect of vitamin C regarding the common cold, upper respiratory tract infections (URTI), or pneumonia, showing only a weak inconsistent effect on the incidence of infection, but a clearer effect on the duration of symptoms [6,7,8,9,10, 34]. However, these trials usually did not consider dietary intakes of vitamin C, which may have limited the observation of effects if baseline nutritional status was already adequate . Moreover, a prospective cohort study among Swedish adults observed a decreased risk of URTI associated with higher dietary intakes of vitamin C in women . Finally, some studies were performed in the context of COVID-19 but focusing on hospitalized patients, observing low vitamin C status in critically ill patients , and suggesting improved outcomes with high-dose vitamin C treatment . Largely consistent with the current evidence , our results suggest that higher vitamin C intakes as part of the regular diet may decrease the risk of SARS-CoV-2 infection, probably through an enhanced early immune response.
Vitamin B9 (folate) has also been suggested as a potential helper in preventing SARS-CoV-2 infection, through a role in innate immunity and in particular for natural killer cell cytotoxicity, which may reduce the risk of infection, as evidenced in an intervention study in older adults . Evidence more specifically related to SARS-CoV-2 from computer simulation studies suggested that vitamin B9 could prevent the virus from entering the cells through an inhibition of furin, a protease involved in the activation of the spike protein, and from replicating through an inactivation of the protease 3CLpro . Thus, our results are consistent with those mostly mechanistic data; strong direct evidence of the potential of vitamin B9 to prevent infection remains limited at present.
Vitamin K is known to play a role in the regulation of blood coagulation and calcification of bone and vessel tissue. In addition, vitamin K could be involved in the regulation of innate immunity, with a stimulation of apoptotic cell clearance by phagocytes, and of inflammatory responses . Hence, it has been suggested that vitamin K could provide some benefits against COVID-19 complications [17, 39]. Our data suggest that it could also help prevent the infection by SARS-CoV-2.
The potential effect of dietary fibers on the risk of SARS-CoV-2 infection is likely resulting from their interaction with the gut microbiota, producing short-chain fatty acids (SCFA) and promoting a diverse and balanced community. The importance of the gut microbiota for efficient and balanced immune and inflammatory responses is well-established. In particular, SCFA contribute to the innate immune response through enhanced production of reactive oxygen species and phagocytosis, apoptosis, and modulation of neutrophil recruitment . Studies in animals have shown the importance of gut microbiota and dietary fiber intakes in the response against viral infection [41, 42]. Hence, although direct evidence in humans is scarce, our results are in line with the mechanistic evidence linking dietary fibers, the gut microbiota, and the immune function.
Consistent with our results observed for vitamins C, B9, K, and dietary fiber, we also observed a decreased susceptibility of SARS-CoV-2 infection with higher intakes of fruit and vegetables. Fruit and vegetables indeed provide a mix of these highly relevant compounds for the immune function (as discussed above), and also other bioactive compounds such as polyphenols . Our findings regarding the consumption of vegetables are consistent with those of the recently published prospective study in the UK-Biobank . A retrospective study in pregnant women also suggested a moderate effect of fruit and vegetable intake in the prevention of URTI . In contrast, a study observed that countries with higher consumptions of fruit had higher infection and mortality rates by COVID-19 but it was based on an ecological design, which did not allow the adjustment for individual factors that may cofound the studied associations . The fact that we observed associations with several compounds provided by fruit and vegetables makes it difficult to disentangle the specific effect of each compound, which will need to be explored in experimental studies and randomized trials.
Whereas most of the associations we observed were in line with mechanistic evidence and current hypotheses regarding the potential role of nutrition in the prevention of SARS-CoV-2 infection, we also observed an unexpected association between higher intakes of dairy products (especially milk) and calcium and a higher likelihood of SARS-CoV-2 infection. Indeed, dairy product intake is usually considered as beneficial for the immune system, anti-inflammatory processes, and the response to infections . These findings, along with those on fruits and vegetables, are somehow consistent with a Spanish prospective study , reporting that lower odds of COVID-19 infection are associated with a higher adherence to Mediterranean diet (i.e., a diet with high intakes of fruits and vegetables and limited intakes of dairy products). A possible mechanism that could partly explain this increased risk involves a calcium-sensitive interaction of SARS-CoV-2 with the ACE-2 receptor (its entry point in human cells). Hence, more available calcium could ease the entry of SARS-CoV-2 in the host cells . However, this result remains unclear and warrants further investigation.
Finally, other nutrients that were expected to have a potential beneficial effect against SARS-CoV-2 infection (e.g., selenium, zinc, copper, iron, vitamins A, D, and E) [6, 47] did not display associations with the risk of SARS-CoV-2 infection in our study. This could potentially relate to the nutritional status of our participants for these nutrients, with perhaps an appropriate status for immune function in most, and a relatively reduced range of intakes among participants. These nutrients could also be involved to mitigate the severity of symptoms or the prognosis once infected, which was beyond the scope of this study focusing on the susceptibility to infection. As regards vitamin D, meta-analyses have shown a potential benefit of a vitamin D supplementation for COVID-19 outcomes and mortality ; similarly, a meta-analysis showed a trend for a deleterious association between low serum 25(OH)D levels and COVID-19 related health outcomes . However, diet is not the major contributor to vitamin D intake, compared with sun exposure and the use of dietary supplements or medication. Specific investigation of vitamin D accounting for all its sources constitutes a perspective of this research work. Likewise, we did not observe associations with the indicators of overall diet quality. This tends to suggest a specific effect of some components playing a key role in immune function and the prevention of SARS-CoV-2 infection, which may be diluted when considering these overall indicators.
Main strengths of our study include the comprehensive assessment of SARS-CoV-2 seroprevalence with highly sensitive assays in a large sample, independent of whether or not the participant sought testing (contrary to studies using PCR results retrieved from medical records) or even had symptoms (that might not be specific to SARS-CoV-2 infections), and the detailed characterization of dietary intakes of participants prior to the pandemic (at least 6 and an average of 10.2 validated 24 h dietary records) allowing us to relate habitual diet to SARS-CoV-2 infection with a prospective design. The presence of antibodies indicates that a person has been infected with SARS-CoV-2, but does not indicate the exact date of the infection. Therefore, to guarantee as much as possible a prospective design, dietary data were considered until February 2020, the start of the epidemic in France. Yet, some limitations should be acknowledged. First, studies have suggested that the ELISA test has imperfect sensitivity (85–90%) [50, 51] and that anti-SARS-CoV-2 antibodies may decrease over time, which may have resulted in misclassification in a way that a person who has been infected with the virus might have not been positive according to the ELISA test; however, the date of collection of dried-blood spot kits was between May and October 2020 (i.e. not a very long time-frame since the beginning of the pandemic in France). Furthermore, the extent to which participants have actually been exposed to the virus is difficult to assess; hence, our results might be confused by behavioral factors, such as an increase awareness and application of protective measures in people who are more health-conscious and have a more balanced diet. To mitigate this potential bias, our models included numerous covariates related to the risk of being exposed to the virus (e.g., employment status during the lockdown, presence of children at home, index reflecting the application of protective measures). In addition, some data were not available in our study. In France, the collection of racial and ethnic data is generally not permitted in the framework of cohort studies, thus, these data were not available. Moreover, quantitative data about the use of dietary supplements (relevant when investigating specific nutrients such as vitamin D) was not yet available and thus, has not been included in this study. Details about subtypes of vitamin K (K1 and K2) were not available; even though these 2 subtypes might have similar biological functions, differences pertain to bioavailability and tissue distribution . However, because dietary sources of vitamin K1 and K2 are distinct (plant or animal sources, respectively), we used these as proxy. Next, the seronegative and seropositive groups differed at baseline according to several characteristics that constitute potential confounders susceptible to influence the studied associations in both directions. For instance, older age and higher physical activity levels were associated with healthier diets and lower risk of infection (which would tend to strengthen the observed associations), but the higher educational level was associated with healthier diets and higher risk of infection (which would tend to attenuate the observed associations). However, even though residual confounding (linked to unmeasured factors or inaccuracies in data assessment) cannot be totally ruled out, we largely accounted for potential confounding by adjusting all models for a wide panel of covariates (e.g., age, sex, educational, professional and socio-economical levels, smoking status, behaviors during the pandemic…) and by matching for age, sex and residential area in sensitivity analyses. Finally, as a long-term study on nutrition and health, NutriNet-Santé includes more women and individuals with overall higher socioeconomic status and healthier lifestyle and dietary habits compared to the general French population. This may limit the generalizability of our findings and might also have resulted in a smaller range in dietary intakes, thus a loss of statistical power.