ABSTRACT
Background
The role of diet on Coronavirus Disease 2019 (COVID-19) is emerging. We investigated the association between usual diet before the onset of the pandemic and risk and severity of subsequent SARS-CoV-2 infection.
Methods
We included 42,935 participants aged 55–99 y in 2 ongoing cohort studies, the Nurses’ Health Study II and Health Professionals Follow-up Study, who completed a series of COVID-19 surveys in 2020 and 2021. Using data from FFQs before COVID-19, we assessed diet quality using the Alternative Healthy Eating Index (AHEI)-2010, the alternative Mediterranean Diet (AMED) score, an Empirical Dietary Index for Hyperinsulinemia (EDIH), and an Empirical Dietary Inflammatory Pattern (EDIP). We calculated multivariable-adjusted ORs and 95% CIs for Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) infection and severity of COVID-19 after controlling for demographic, medical, and lifestyle factors.
Results
Among 19,754 participants tested for SARS-CoV-2, 1941 participants reported a positive result. Of these, 1327 reported symptoms needing assistance and another 109 were hospitalized. Healthier diets, represented by higher AHEI-2010 and AMED scores and lower EDIH and EDIP scores, were associated with lower likelihood of SARS-CoV-2 infection (quartile 4 compared with quartile 1: OR: 0.80; 95% CI: 0.69, 0.92 for AHEI-2010; OR: 0.78; 95% CI: 0.67, 0.92 for AMED; OR: 1.36; 95% CI: 1.16, 1.57 for EDIH; and OR: 1.13; 95% CI: 0.99, 1.30 for EDIP; all P-trend ≤ 0.01). In the analysis of COVID-19 severity, participants with healthier diet had lower likelihood of severe infection and were less likely to be hospitalized owing to COVID-19. However, associations were no longer significant after controlling for BMI and pre-existing medical conditions.
Conclusions
Diet may be an important modifiable risk factor for SARS-CoV-2 infection, as well as for severity of COVID-19. This association is partially mediated by BMI and pre-existing medical conditions.
Keywords: dietary quality, SARS-CoV-2 infection, COVID-19 severity, prospective cohort study, long-term diet, Mediterranean diet, alternate healthy eating index, inflammatory diet, hyperinsulinemia
See corresponding editorial on page 1464.
Introduction
Coronavirus Disease 2019 (COVID-19), the disease caused by the novel Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), has become a major global public health crisis causing nearly 5 million deaths globally (1). Research has moved rapidly in defining the biological features of the virus, developing vaccines, and finding therapeutic agents to treat SARS-CoV-2 infection (2). In the meantime, several factors have been identified that increase risk of SARS-CoV-2 infection and its severity, including older age and presence of pre-existing medical conditions such as diabetes, high blood pressure, obesity, and cardiovascular disease (3). Despite this, the roles of modifiable factors such as diet in susceptibility to SARS-CoV-2 infection and its severity are not well understood. Further, whether diet can modify the higher risk of COVID-19 that is associated with the presence of comorbidities remains unknown.
An important defense against SARS-CoV-2 infection is a well-coordinated immune response from the host (4, 5). Diet may play a crucial role in supporting the immune system's response to SARS-CoV-2 infections, and could consequently modulate the severity of infection (6). For example, higher adherence to a Mediterranean-style dietary pattern was associated with a lower risk of respiratory infections (7, 8), and was recently found to be inversely associated with COVID-19-related deaths in an ecological study of 23 countries (9). Also, a healthy diet is associated with a lower risk of several chronic diseases that are strong risk factors for SARS-CoV-2 infection (10, 11).
We therefore investigated the association between usual diet before the onset of the pandemic and risk and severity of subsequent SARS-CoV-2 infection using data from 2 large cohorts, the Nurses’ Health Study (NHS) II and Health Professionals Follow-up Study (HPFS). To better capture the complexity and interaction of multiple dietary factors, we evaluated 2 widely used a priori diet quality indices (12)—the Alternative Healthy Eating Index (AHEI)-2010, emphasizing overall diet quality (13), and alternative Mediterranean Diet (AMED) score, assessing adherence to a Mediterranean Diet–style dietary pattern (14)—and 2 empirical hypothesis-oriented indices assessing the insulinemic [Empirical Dietary Index for Hyperinsulinemia (EDIH)] and inflammatory [Empirical Dietary Inflammatory Pattern (EDIP)] potential of diet (15, 16). We are especially interested in the 2 empirical hypothesis-oriented indices—EDIH and EDIP—because hyperinsulinemia and a proinflammatory status were shown to contribute to disease severity and mortality in COVID-19 (17, 18).
Methods
Study population
The NHS II is a prospective cohort study in 116,429 female nurses, aged 25–42 y, that began in 1989 (19). The HPFS began in 1986 and includes 51,529 men in health professions who were 40–75 y old at cohort baseline (20). In both cohorts, participants completed questionnaires at baseline and every 2 y thereafter on demographic, lifestyle factor, and health-related information.
In 2020, we invited participants who had returned the most recent primary cohort questionnaires—2019 in the NHS II (n = 55,925) and 2020 in the HPFS (n = 8900)—to complete a supplementary COVID-19 online series of surveys regarding their experiences during the COVID-19 pandemic (19) (Supplemental Figure 1). The COVID-19 survey collected information on lifestyle factors, current occupational status, demographic factors, personal protective equipment (PPE) use, whether participants were tested for SARS-CoV-2 infection and the corresponding results if applicable, presence of COVID-19 symptoms, hospitalization for COVID-19, treatment during hospitalization if applicable, interaction with presumed or documented COVID-19 cases, and concern about COVID-19.
In the NHS II, a total of 55,925 women were invited to participate in the COVID-19 study, and 39,564 (70.7%) women completed the baseline online COVID-19 survey sent in May 2020. Repeated follow-up COVID-19 surveys were sent out until April 2021 through 2 phases: in phase 1, monthly questionnaires were sent out to all participants until 84 d after the baseline survey was returned. Additional weekly questionnaires were sent to frontline health care providers. In phase 2, all participants received the questionnaire quarterly, whereas additional monthly questionnaires were sent to frontline health care providers. In total, 7 repeated COVID-19 surveys were sent to all participants and an additional 14 surveys were sent to frontline health care providers.
In the HPFS, 8900 men were invited to participate in the COVID-19 study. A total of 4415 men (49.6%) completed the baseline COVID-19 questionnaire sent in September 2020. These participants were then asked to fill out 2 follow-up surveys in January and April 2021.
By the end of the COVID-19 study (April 2021), 38,061 participants had returned the final follow-up questionnaire (34,375 in the NHS II and 3686 in the HPFS). The follow-up rate for the COVID-19 survey was 86.5%.
The study protocol was approved by the Institutional Review Boards of the Brigham and Women's Hospital and the Harvard TH Chan School of Public Health.
Assessment of dietary quality scores
Since 1991 in the NHS II and 1986 in the HPFS, dietary information has been updated every 4 y using validated, self-administered, semiquantitative FFQs (21–24). Each FFQ listed standard portion sizes for >150 foods and asked participants to record intake frequency over the past year, with 9 possible responses ranging from “never or less than once per month” to “six or more times per day.” Average daily nutrient intake was calculated by multiplying the frequency of intake by the nutrient content of each food and summing nutrient values across all foods.
We computed 4 dietary quality scores for each participant using FFQ data collected during the 2 most recent available 4-y data cycles completed before the 2020 COVID-19 survey—2011 and 2015 for the NHS II, and 2010 and 2014 for the HPFS. These scores include the AHEI-2010, AMED, EDIH, and EDIP (Supplemental Table 1).
The AHEI-2010 was based on 11 foods and nutrients that are predictive of chronic disease risk (13). Each component was scored between 0 and 10 with higher scores being assigned to higher intakes of vegetables, fruit, whole grains, nuts/legumes, long-chain fatty acids, and PUFAs, moderate alcohol consumption, and lower intake of sugar-sweetened beverages, red/processed meat, trans fatty acids, and sodium. The total AHEI-2010 score ranged from 0 to 110, with higher scores indicating a higher-quality diet.
The AMED score, an indicator of adherence to a Mediterranean-style diet, was calculated as a sum of 9 components (14). For vegetables, fruits, nuts, whole grains, legumes, fish, and the ratio of monounsaturated to saturated fat, a score of 1 was given to intakes above the median. For red and processed meat consumption, a score of 1 was given to intake below the median. For alcohol intake, a score of 1 was given for moderate consumption (between 5 and 15 g/d for women, 10–25 g/d for men). If participants did not meet the criteria to receive a score of 1 for a given component, they received a score of 0. The component scores were summed to obtain an overall AMED score ranging from 0 to 9 with higher scores corresponding to higher adherence to a Mediterranean diet.
The EDIH was developed to assess the insulinemic potential of the whole diet. The score was derived as a weighted sum of 18 food groups (see Supplemental Table 1) that have been described in detail elsewhere (16). In brief, 13 food groups were positively associated with C-peptide concentrations, a stable marker of insulin resistance and secretion, whereas 5 food groups were inversely associated. The weight assigned for each index component was the regression coefficient derived from the stepwise linear regression model to predict circulating C-peptide (16). Higher scores indicate a hyper-insulinemic diet, whereas lower scores indicate a hypo-insulinemic diet.
The EDIP was developed to assess the overall inflammatory potential of the diet and was constructed in a similar way as the EDIH. The EDIP included 18 food groups (see Supplemental Table 1) that were most predictive of 3 plasma inflammation markers: IL-6, C-reactive protein, and TNF-α-receptor 2. Of the 18 food groups, 9 food groups were anti-inflammatory, whereas 9 food groups were proinflammatory. Higher scores indicate proinflammatory diets, whereas lower scores indicate anti-inflammatory diets (15).
Assessment of nondietary factors
Height and weight were reported at baseline, and weight was updated biennially. BMI (in kg/m2) was calculated as weight in kilograms divided by height in meters squared. Participants self-reported average time spent weekly on 7 physical activities using validated questionnaires every 2–4 y (25). Total physical activity in metabolic equivalent of tasks (MET)-hours per week was calculated by multiplying the MET score for, and hours per week spent in, each activity and summing over all activity items (26). These analyses used the most recently updated BMI and physical activity data collected in each cohort. Geocoded mailing addresses were linked to 2010 census tracts to obtain information on the following measures of socioeconomic status (SES): census tract median family home value, median family income, and population density.
SARS-CoV-2 infection and severity
Our primary outcome was self-reported SARS-CoV-2 infection, including positive results from an antigen or antibody test. We also classified SARS-CoV-2-positive participants into 4 categories using a modified WHO clinical progression scale (27): 1) asymptomatic; 2) symptomatic; 3) independent (persistent cough, sore throat, loss of taste, or loss of smell); symptomatic, assistance needed (shortness of breath or difficulty breathing, fever, muscle aches, or digestive symptoms); and 4) hospitalization. The secondary outcome was symptomatic SARS-CoV-2 infection derived using a method similar to that of Menni et al. (28). The final prediction algorithm included age and reported COVID-19 symptoms including fever, sore throat, muscle aches, loss of taste, loss of smell, and other symptoms consistent with COVID-19 infection (29).
Statistical analysis
Of 43,979 (39,564 in the NHS II and 4415 in the HPFS) participants who completed the baseline COVID-19 survey, we excluded those who did not complete an FFQ in 2011 or 2015 for the NHS II, and 2010 or 2014 for the HPFS (before completing the baseline COVID-19 survey). We further excluded participants with unreliable dietary data [e.g., extreme energy intake in FFQ (men: <800 or >4200 kcal/d; women: <600 or >3500 kcal/d), left ≥70 FFQ items blank] and those with inconsistent self-reported infection status (e.g., reported being hospitalized owing to COVID-19 but tested negative for SARS-CoV-2 infection), leaving 42,935 (38,615 in the NHS II and 4320 in the HPFS) participants in the final analysis (Supplemental Figure 2). To better represent long-term exposures and to reduce within-person variation, we computed the mean of each dietary index from the 2 FFQs (30). In women and men separately, participants were grouped into quartiles by absolute scores for each dietary pattern index, and data were then pooled together.
Multivariable logistic regression models were used to examine the association of each of the 4 dietary quality scores with the likelihood of SARS-CoV-2 infection. We adjusted for potential confounders in sequential models: model 1 was adjusted for age (continuous), sex (women or men), and race (white or nonwhite); model 2 was further adjusted for smoking (never, past, or current), physical activity (continuous), total energy intake (continuous), census tract median family income (continuous), census tract median family home value (continuous), census tract population density (continuous), concern about COVID-19 (yes or no), interaction with people other than patients with presumed or documented COVID-19 (yes or no), and frontline health care providers and PPE use (not frontline health care providers, frontline health care providers without adequate PPE, and frontline health care providers with adequate PPE); model 3 was further adjusted for BMI (continuous), history of high cholesterol (yes or no), history of high blood pressure (yes or no), and presence of other pre-existing medical conditions (diabetes, heart attack, or cancer; yes or no). Missing covariates were replaced with medians. Tests for trends were conducted using the median of each index quartile as a continuous variable.
For our primary analysis of SARS-CoV-2 infection status, we first restricted the analysis to those who had reported test results for SARS-CoV-2 infection. We then used inverse probability weighting (IPW) to account for the probability of receiving a test given the potential of differences between people who reported test results for SARS-CoV-2 infection and those who had not. The IPW was a function of the dietary quality score being evaluated, age, sex, race, being a frontline health care provider, interaction with people other than patients with presumed or documented COVID-19, census tract median family income, census tract median family home value, census tract population density, and reported COVID-19 symptoms. We also evaluated the associations between dietary quality scores and symptom-based SARS-CoV-2 infection status using the multivariable logistic regression models adjusting for the same set of covariates.
In the analyses of the severity of SARS-CoV-2 infection, we ran logistic regression analysis using a data augmentation method with a single reference group (tested negative and without symptoms) shared by multiple case subtypes (tested positive without any symptoms; tested positive with only independent symptoms; tested positive with assistance-needed symptoms; tested positive and was hospitalized) (31). Statistical significance of heterogeneity across different degrees of severity was evaluated by the likelihood ratio test. In sensitivity analyses, we limited this analysis to only those who tested positive. According to the modified WHO clinical progression scale, participants who tested positive without symptoms received a score of 1 and were considered as the reference group; 3 case groups received scores from 2 to 4 in order of increasing severity, including tested positive with only independent symptoms, tested positive with assistance-needed symptoms, and tested positive and was hospitalized. We also ran ordinal logistic regression in analysis of the progression scale among tested-positive participants. Partial proportion odds models were adopted to separate covariate parameters across the logits for model effects exhibiting a lack of proportionality.
We conducted subgroup analyses for associations between dietary quality scores and SARS-CoV-2 infection by age, smoking history, being a frontline health care provider, being concerned about COVID-19, interaction with people with presumed or documented COVID-19, BMI, pre-existing medical conditions, history of high blood pressure, history of high cholesterol, and presence of diabetes. Tests for interaction were obtained using the Wald test of cross-product interaction terms between each index, modeled as a continuous variable, and the potential effect modifier, modeled as a categoric variable. Given the issue of multiple testing across the various categories, we adjusted the P-interaction for multiple testing by Bonferroni correction.
Analyses were conducted using SAS version 9.4 (SAS Institute); P values < 0.05 were considered statistically significant.
Results
Of the 42,935 participants who responded to the COVID-19 survey and had available dietary data, 90% were women, 97% were Caucasian, and 44% had reported ≥1 test result for SARS-CoV-2 infection or antibodies. At baseline of the COVID-19 study (March 2020), the mean ± SD age was 66.5 ± 6.1 y. In women, 10% completed no more than 3 COVID-19 questionnaires, 53% completed 6 or 7 COVID-19 questionnaires, and 22% completed ≥10 COVID-19 questionnaires during the follow-up. In men, 76% completed all 3 questionnaires.
Compared with participants with lower dietary quality (higher EDIH and EDIP scores and lower AHEI-2010 and AMED scores), participants with higher dietary quality had a lower BMI and were more likely to be physically active, as well as have a higher SES. At the same time, these participants were less likely to smoke or have pre-existing medical conditions (Table 1). Compared with participants who did not report test results, participants who reported test results were more likely to be frontline health care providers, have interacted with others with presumed or documented COVID-19, and have a higher SES (Supplemental Table 2). We observed the strongest correlation between the a priori diet quality indices: AHEI-2010 and AMED (rs = 0.74). The 2 empirical hypothesis-oriented indices—EDIH and EDIP—showed weaker correlations with each other (rs = 0.58) and the 2 a priori diet quality indices (|rs |≤ 0.45) (Supplemental Table 3).
TABLE 1.
Characteristics of 42,935 COVID-19 survey participants by quartiles of dietary quality scores in the Nurses’ Health Study II and Health Professionals Follow-up Study1
| AHEI-2010 | AMED | EDIH | EDIP | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Overall | Q1 | Q4 | Q1 | Q4 | Q1 | Q4 | Q1 | Q4 | |
| Quartile median, women/men | 52.1/55.1 | 80.6/81.1 | 2.0/2.0 | 6.5/6.5 | 0.0/0.0 | 0.5/0.7 | −1.3/−0.5 | 0.8/0.2 | |
| Age, y | 66.5 ± 6.1 | 66.2 ± 6.2 | 66.8 ± 5.8 | 66.2 ± 6.1 | 66.8 ± 6.0 | 66.8 ± 6.0 | 66.2 ± 6.1 | 66.2 ± 6.1 | 66.8 ± 5.9 |
| Received COVID-19 tests, % | 46.0 | 44.5 | 46.5 | 45.0 | 46.9 | 47.9 | 45.2 | 47.4 | 45.5 |
| BMI, kg/m2 | 27.4 ± 6.0 | 29.2 ± 6.7 | 25.4 ± 4.8 | 29.0 ± 6.6 | 25.8 ± 5.2 | 25.2 ± 4.7 | 29.9 ± 6.8 | 25.8 ± 5.0 | 29.3 ± 6.7 |
| Energy intake, kcal/d | 1789 ± 531 | 1788 ± 540 | 1831 ± 497 | 1497 ± 450 | 2091 ± 500 | 1680 ± 511 | 2040 ± 536 | 1831 ± 526 | 1842 ± 551 |
| Total activity, MET-h/wk | 30.6 ± 32.1 | 22.0 ± 26.6 | 40.6 ± 36.2 | 21.0 ± 25.4 | 41.1 ± 37.1 | 37.0 ± 35.3 | 25.3 ± 29.4 | 36.0 ± 34.6 | 26.0 ± 30.1 |
| Smoking status, % | |||||||||
| Past smoker | 32.3 | 29.5 | 34.4 | 30.0 | 33.5 | 37.6 | 28.5 | 37.3 | 27.4 |
| Current smoker | 2.0 | 3.2 | 1.0 | 3.2 | 1.1 | 2.1 | 2.1 | 1.7 | 2.3 |
| History of cancer, % | 4.2 | 4.5 | 3.9 | 4.4 | 3.8 | 4.0 | 4.4 | 4.0 | 4.7 |
| History of heart attack, % | 0.5 | 0.7 | 0.4 | 0.7 | 0.4 | 0.4 | 0.6 | 0.3 | 0.7 |
| History of high blood pressure, % | 25.2 | 31.2 | 18.8 | 29.5 | 20.9 | 19.1 | 32.2 | 21.1 | 30.7 |
| History of hypercholesterolemia, % | 27.8 | 31.5 | 24.5 | 30.3 | 25.1 | 23.6 | 31.5 | 24.7 | 31.2 |
| History of diabetes, % | 5.4 | 7.0 | 3.2 | 7.4 | 3.5 | 2.2 | 9.7 | 2.5 | 9.0 |
| Frontline health care providers, % | 28.1 | 28.3 | 27.5 | 29.0 | 27.3 | 27.8 | 28.6 | 28.3 | 28.0 |
| Community interaction with people with presumed or documented COVID-19, % | 8.6 | 7.8 | 9.4 | 7.8 | 9.2 | 8.8 | 8.6 | 8.9 | 8.2 |
| Expressed concern about COVID-19, % | 72.8 | 71.3 | 74.4 | 70.7 | 74.5 | 74.6 | 71.0 | 74.1 | 71.3 |
| Median family home value, $ (2010 census) | 291,766 ± 211,999 | 233,968 ± 170,253 | 355,774 ± 238,005 | 245,813 ± 178,032 | 333,498 ± 230,757 | 341,607 ± 233,548 | 242,375 ± 178,745 | 320,537 ± 222,689 | 264,253 ± 197,244 |
| Median family income, $ (2010 census) | 84,366 ± 32,931 | 77,161 ± 28,484 | 91,456 ± 35,776 | 78,847 ± 29,378 | 89,110 ± 35,009 | 90,338 ± 35,634 | 77,918 ± 29,617 | 88,837 ± 34,411 | 79,792 ± 30,944 |
| Population density: people per sq km, n (2010 census) | 1196 ± 3380 | 966 ± 2654 | 1471 ± 4019 | 1029 ± 2885 | 1359 ± 3753 | 1410 ± 4072 | 1000 ± 2810 | 1304 ± 3796 | 1116 ± 3118 |
Values are means ± SDs or percentages standardized to the distribution of age, with the exception of age itself. Quartiles were determined in women and men separately. AHEI, Alternative Healthy Eating Index; AMED, alternative Mediterranean Diet; COVID-19, Coronavirus Disease 2019; EDIH, Empirical Dietary Index for Hyperinsulinemia; EDIP, Empirical Dietary Inflammatory Pattern; MET, metabolic equivalent of tasks; Q, quartile.
A total of 1941 participants (4.5%) reported a positive test for SARS-CoV-2 infection or antibodies. Higher dietary quality, represented by higher AHEI-2010 and AMED as well as lower EDIH and EDIP scores, was associated with a lower risk of SARS-CoV-2 infection in the fully adjusted model (model 3). The multivariable-adjusted ORs comparing the top with the bottom quartile were 0.80 (95% CI: 0.69, 0.92; P-trend = 0.001) for AHEI-2010; 0.78 (95% CI: 0.67, 0.92; P-trend = 0.003) for AMED; 1.36 (95% CI: 1.16, 1.58; P-trend = 0.0001) for EDIH; and 1.13 (95% CI: 0.99, 1.30; P-trend = 0.01) for EDIP. When we accounted for potential selection bias using IPW, results remained unchanged (Table 2). Furthermore, higher dietary quality was also associated with lower risk of symptomatic SARS-CoV-2 infection, and the strongest association was observed for EDIH (quartile 4 compared with quartile 1: OR: 1.18; 95% CI: 1.06, 1.31) (Supplemental Table 4).
TABLE 2.
Associations (ORs and 95% CIs) between dietary quality scores and risk of SARS-CoV-2 infection in the Nurses’ Health Study II and Health Professionals Follow-up Study1
| Q1 | Q2 | Q3 | Q4 | P-trend2 | Continuous3 | |
|---|---|---|---|---|---|---|
| AHEI-2010 | ||||||
| Cases/noncases,4n | 576/4213 | 492/4358 | 454/4476 | 419/4766 | 1941/17,813 | |
| Model 15 | 1 (ref.) | 0.83 (0.73, 0.95) | 0.75 (0.66, 0.86) | 0.65 (0.57, 0.75) | <0.0001 | 0.83 (0.80, 0.88) |
| Model 26 | 1 (ref.) | 0.87 (0.76, 0.99) | 0.82 (0.71, 0.93) | 0.75 (0.65, 0.86) | <0.0001 | 0.88 (0.84, 0.93) |
| Model 37 | 1 (ref.) | 0.90 (0.79, 1.02) | 0.86 (0.75, 0.98) | 0.80 (0.69, 0.92) | 0.0014 | 0.90 (0.86, 0.95) |
| Model 3 + IPW8 | 1 (ref.) | 0.90 (0.79, 1.04) | 0.87 (0.76, 1.01) | 0.81 (0.69, 0.94) | 0.0051 | 0.91 (0.86, 0.96) |
| AMED | ||||||
| Cases/noncases,4n | 495/3678 | 540/4756 | 489/4689 | 417/4690 | 1941/17,813 | |
| Model 15 | 1 (ref.) | 0.85 (0.74, 0.97) | 0.78 (0.69, 0.90) | 0.67 (0.58, 0.77) | <0.0001 | 0.85 (0.82, 0.90) |
| Model 26 | 1 (ref.) | 0.90 (0.78, 1.02) | 0.86 (0.74, 0.99) | 0.76 (0.65, 0.89) | 0.0008 | 0.89 (0.84, 0.94) |
| Model 37 | 1 (ref.) | 0.90 (0.79, 1.03) | 0.87 (0.76, 1.01) | 0.78 (0.67, 0.92) | 0.0032 | 0.90 (0.85, 0.95) |
| Model 3 + IPW8 | 1 (ref.) | 0.92 (0.81, 1.06) | 0.90 (0.78, 1.05) | 0.79 (0.67, 0.94) | 0.0091 | 0.90 (0.85, 0.96) |
| EDIH | ||||||
| Cases/noncases,4n | 382/4739 | 466/4439 | 524/4335 | 569/4300 | 1941/17,813 | |
| Model 15 | 1 (ref.) | 1.30 (1.13, 1.50) | 1.48 (1.29, 1.70) | 1.61 (1.41, 1.85) | <0.0001 | 1.18 (1.12, 1.23) |
| Model 26 | 1 (ref.) | 1.23 (1.07, 1.42) | 1.38 (1.20, 1.60) | 1.45 (1.25, 1.68) | <0.0001 | 1.14 (1.08, 1.20) |
| Model 37 | 1 (ref.) | 1.21 (1.04, 1.40) | 1.35 (1.16, 1.56) | 1.36 (1.16, 1.58) | 0.0001 | 1.12 (1.06, 1.18) |
| Model 3 + IPW8 | 1 (ref.) | 1.24 (1.07, 1.44) | 1.38 (1.19, 1.60) | 1.40 (1.20, 1.64) | <0.0001 | 1.12 (1.06, 1.18) |
| EDIP | ||||||
| Cases/noncases,4n | 451/4648 | 449/4553 | 501/4287 | 540/4325 | 1941/17,813 | |
| Model 15 | 1 (ref.) | 1.02 (0.89, 1.17) | 1.22 (1.06, 1.39) | 1.31 (1.14, 1.49) | <0.0001 | 1.10 (1.06, 1.16) |
| Model 26 | 1 (ref.) | 0.99 (0.86, 1.13) | 1.13 (0.99, 1.30) | 1.17 (1.02, 1.34) | 0.0023 | 1.06 (1.01, 1.11) |
| Model 37 | 1 (ref.) | 0.98 (0.85, 1.12) | 1.11 (0.97, 1.28) | 1.13 (0.99, 1.30) | 0.0109 | 1.04 (0.99, 1.10) |
| Model 3 + IPW8 | 1 (ref.) | 0.99 (0.86, 1.15) | 1.09 (0.95, 1.26) | 1.13 (0.98, 1.31) | 0.0104 | 1.04 (0.99, 1.09) |
AHEI, Alternate Healthy Eating Index; AMED, alternative Mediterranean Diet; COVID-19, Coronavirus Disease 2019; EDIH, Empirical Dietary Index for Hyperinsulinemia; EDIP, Empirical Dietary Inflammatory Pattern; IPW, inverse probability weighting; PPE, personal protective equipment; Q, quartile.
P-linear trend across quartiles was calculated using the median of each quartile as a continuous variable.
Continuous analyses for a 1-SD increment.
Noncases number for models 1–3.
Model 1 was adjusted for age (continuous), sex (women or men), and race (white or nonwhite).
Model 2 was further adjusted for smoking (never, past, or current), physical activity (continuous), total energy intake (continuous), census tract median family income (continuous), census tract median family home value (continuous), census tract population density (continuous), concern about COVID-19 (yes or no), interaction with people other than patients with presumed or documented COVID-19 (yes or no), and frontline health care providers and PPE use (not frontline health care providers, frontline health care providers without adequate PPE, and frontline health care providers with adequate PPE).
Model 3 was further adjusted for BMI (continuous), history of high cholesterol (yes or no), history of high blood pressure (yes or no), and presence of other pre-existing medical conditions (diabetes, heart attack, cancer; yes or no).
IPW: probability of receiving a COVID-19 test was modeled using the dietary quality score being evaluated, age, sex, race, being a frontline health care worker, interaction with people other than patients with presumed or documented COVID-19, census tract median family income, census tract median family home value, census tract population density, and occurrences of COVID-19-related symptoms. We then weighted the study population by the inverse of the probability of receiving a COVID-19 test and then adjusted for covariates in model 3.
Higher adherence to the AHEI and AMED and lower adherence to the EDIP and EDIH were associated with lower likelihood of severe infection. In multivariable model 2, participants with 1-SD higher scores of the AHEI-2010 and AMED were 20%–22% less likely to be hospitalized owing to SARS-CoV-2 infection. On the other hand, participants with 1-SD higher scores of the EDIH and EDIP had a 23%–37% higher likelihood of hospitalization. Further adjustment for BMI and pre-existing medical conditions largely attenuated these associations which were no longer significant (Figure 1). In sensitivity analyses with tested-positive participants only, we also observed an inverse association between higher dietary quality and severity of COVID-19 before adjusting for the BMI and pre-existing medical conditions (Supplemental Figure 3).
FIGURE 1.
Associations (ORs and 95% CIs) between 1-SD increments of dietary quality scores and COVID-19 severity in the Nurses’ Health Study II and Health Professionals Follow-up Study. *Model 2 was adjusted for age (continuous), sex (women or men), race (white or nonwhite), smoking (never, past, or current), physical activity (continuous), total energy intake (continuous), census tract median family income (continuous), census tract median family home value (continuous), census tract population density (continuous), concern about COVID-19 (yes or no), interaction with people other than patients with presumed or documented COVID-19 (yes or no), and frontline health care providers and PPE use (not frontline health care providers, frontline health care providers without adequate PPE, and frontline health care providers with adequate PPE). †Model 3 was further adjusted for BMI (continuous), history of high cholesterol (yes or no), history of high blood pressure (yes or no), and presence of other pre-existing medical conditions (diabetes, heart attack, cancer; yes or no). ‡Continuous analyses for a 1-SD increment. §Tested negative and without symptoms (n = 10,231). ¶Tested positive without any symptoms (n = 270). #Tested positive with only independent symptoms (persistent cough, sore throat, loss of taste, or loss of smell) (n = 235). **Tested positive with assistance-needed symptoms (shortness of breath or difficulty breathing, fever, muscle aches, or digestive symptoms) (n = 1327). ††Tested positive and was hospitalized (n = 109). ‡‡P-heterogeneity across different degrees of severity was evaluated by the likelihood ratio test. AHEI, Alternate Healthy Eating Index; AMED, alternative Mediterranean Diet; EDIH, Empirical Dietary Index for Hyperinsulinemia; EDIP, Empirical Dietary Inflammatory Pattern; PPE, personal protective equipment.
The associations between dietary quality scores and likelihood of a positive test were not significantly modified by lifestyle factors, knowledge of exposure to SARS-CoV-2 virus, or any pre-existing medical conditions (Table 3).
TABLE 3.
Associations (ORs and 95% CIs) between dietary quality scores and risk of a positive COVID-19 test by demographic, lifestyle, medical, and personal characteristics in the Nurses’ Health Study II and Health Professionals Follow-up Study1
| Characteristics | Cases/noncases, n | AHEI-2010 | AMED | EDIH | EDIP |
|---|---|---|---|---|---|
| Age, y | |||||
| <65 | 884/6995 | 0.91 (0.84, 0.98) | 0.90 (0.83, 0.98) | 1.12 (1.04, 1.22) | 1.04 (0.97, 1.12) |
| ≥65 | 1057/10818 | 0.90 (0.84, 0.96) | 0.90 (0.84, 0.98) | 1.13 (1.05, 1.21) | 1.05 (0.98, 1.12) |
| Pinter2 | 0.23 | 0.38 | 0.56 | 0.37 | |
| Smoking | |||||
| Never | 1231/10777 | 0.92 (0.86, 0.98) | 0.91 (0.85, 0.98) | 1.13 (1.06, 1.21) | 1.05 (0.99, 1.12) |
| Ever | 710/7036 | 0.87 (0.80, 0.95) | 0.89 (0.81, 0.97) | 1.11 (1.02, 1.21) | 1.03 (0.95, 1.11) |
| Pinter2 | 0.14 | 0.24 | 0.36 | 0.72 | |
| Frontline health care worker | |||||
| No | 1213/12475 | 0.91 (0.85, 0.97) | 0.91 (0.84, 0.97) | 1.13 (1.06, 1.21) | 1.04 (0.98, 1.11) |
| Yes | 728/5338 | 0.88 (0.81, 0.96) | 0.90 (0.82, 0.98) | 1.12 (1.03, 1.22) | 1.06 (0.98, 1.14) |
| Pinter2 | 0.71 | 0.63 | 0.34 | 0.84 | |
| Concern about COVID-19 | |||||
| No | 630/4467 | 0.89 (0.82, 0.98) | 0.88 (0.80, 0.97) | 1.09 (1.00, 1.20) | 1.05 (0.96, 1.14) |
| Yes | 1311/13346 | 0.91 (0.85, 0.97) | 0.92 (0.86, 0.98) | 1.15 (1.08, 1.22) | 1.05 (0.99, 1.11) |
| Pinter2 | 0.57 | 0.61 | 0.48 | 0.85 | |
| Community interaction with people with presumed or documented COVID-19 | |||||
| No | 1586/15956 | 0.90 (0.85, 0.95) | 0.91 (0.85, 0.96) | 1.13 (1.07, 1.20) | 1.05 (1.00, 1.11) |
| Yes | 355/1857 | 0.89 (0.79, 1.01) | 0.86 (0.75, 0.98) | 1.09 (0.96, 1.24) | 1.03 (0.92, 1.15) |
| Pinter2 | 0.15 | 0.89 | 0.08 | 0.19 | |
| BMI, kg/m2 | |||||
| <25 | 673/7292 | 0.89 (0.82, 0.97) | 0.91 (0.83, 1.00) | 1.14 (1.04, 1.25) | 1.05 (0.97, 1.14) |
| 25 to <30 | 682/5997 | 0.87 (0.60, 1.28) | 0.90 (0.82, 0.99) | 1.12 (1.02, 1.22) | 1.04 (0.95, 1.13) |
| ≥30 | 586/4524 | 0.90 (0.81, 0.99) | 0.88 (0.79, 0.98) | 1.12 (1.02, 1.23) | 1.06 (0.96, 1.15) |
| Pinter2 | 0.86 | 0.97 | 0.87 | 0.92 | |
| Pre-existing medical conditions | |||||
| No | 1063/23131 | 0.90 (0.84, 0.96) | 0.89 (0.83, 0.96) | 1.10 (1.03, 1.18) | 1.07 (1.00, 1.14) |
| Yes | 878/17863 | 0.92 (0.86, 0.99) | 0.92 (0.85, 1.00) | 1.14 (1.06, 1.23) | 1.01 (0.94, 1.08) |
| Pinter2 | 0.38 | 0.36 | 0.64 | 0.19 | |
| History of high blood pressure | |||||
| No | 1428/13409 | 0.90 (0.85, 0.96) | 0.90 (0.84, 0.96) | 1.11 (1.05, 1.19) | 1.07 (1.01, 1.14) |
| Yes | 513/4404 | 0.89 (0.80, 0.99) | 0.91 (0.80, 1.02) | 1.16 (1.05, 1.28) | 0.98 (0.90, 1.08) |
| Pinter2 | 0.53 | 0.41 | 0.58 | 0.12 | |
| History of high cholesterol | |||||
| No | 1390/12771 | 0.90 (0.84, 0.95) | 0.89 (0.83, 0.95) | 1.11 (1.04, 1.18) | 1.05 (0.99, 1.11) |
| Yes | 551/5042 | 0.90 (0.82, 1.00) | 0.92 (0.83, 1.02) | 1.18 (1.06, 1.30) | 1.04 (0.95, 1.14) |
| Pinter2 | 0.79 | 0.54 | 0.44 | 0.83 | |
| Type 2 diabetes | |||||
| No | 1829/16863 | 0.89 (0.85, 0.94) | 0.90 (0.85, 0.95) | 1.13 (1.07, 1.20) | 1.05 (1.00, 1.10) |
| Yes | 112/950 | 1.01 (0.79, 1.28) | 0.96 (0.75, 1.23) | 1.05 (0.85, 1.31) | 1.01 (0.82, 1.25) |
| Pinter2 | 0.82 | 0.60 | 0.43 | 0.93 | |
Continuous analyses for a 1-SD increment. The multivariable model was adjusted for age (continuous), sex (women or men), race (white or nonwhite), smoking (never, past, or current), physical activity (continuous), total energy intake (continuous), census tract median family income (continuous), census tract median family home value (continuous), census tract population density (continuous), concern about COVID-19 (yes or no), interaction with people other than patients with presumed or documented COVID-19 (yes or no), frontline health care providers and PPE use (not frontline health care providers, frontline health care providers without adequate PPE, and frontline health care providers with adequate PPE), BMI (continuous), history of high cholesterol (yes or no), history of high blood pressure (yes or no), and presence of other pre-existing medical conditions (diabetes, heart attack, cancer; yes or no). AHEI, Alternate Healthy Eating Index; AMED, alternative Mediterranean Diet; COVID-19, Coronavirus Disease 2019; EDIH, Empirical Dietary Index for Hyperinsulinemia; EDIP, Empirical Dietary Inflammatory Pattern; PPE, personal protective equipment.
P-linear trend across quartiles was calculated using the median of each quartile as a continuous variable.
Discussion
In a substudy in 2 large cohorts of 42,935 health professionals who responded to a COVID-19 survey and had available data on usual long-term diet before the survey, we found that long-term adherence to healthy dietary patterns, represented by higher AHEI-2010 and AMED as well as lower EDIH and EDIP, was associated with a lower risk of both SARS-CoV-2 infection and disease severity. BMI and pre-existing medical conditions could be mediators for COVID-19 severity but not for SARS-CoV-2 infection itself.
Emerging evidence supports a role of diet in the development of COVID-19. Using data from a smartphone-based COVID-19 symptom study, we observed that a recent diet characterized by healthy plant-based foods was associated with lower risk and severity of COVID-19 (10). Similar to our findings, a recent population-based case–control study in health care workers found adherence to a plant-based diet, which was assessed retrospectively through self-identified dietary patterns, was associated with lower odds of moderate-to-severe COVID-19 (32). Another retrospective cross-sectional study in COVID patients who visited the respiratory emergency department in Iran also found that a healthier dietary pattern assessed by a 16-item FFQ was associated with less severe COVID-19 symptoms (33). An earlier ecological study of 23 countries in the Organisation for Economic Co-operation and Development supported a role for the Mediterranean diet in the prevention of COVID-19 (9). After adjusting for country-level indicators of income, education, housing, environment, life satisfaction, and physical inactivity, the authors found a negative association between Mediterranean diet adherence and COVID-19-related deaths (9).
Adherence to the AHEI-2010 and AMED indicates higher fruit and vegetable consumption that may lead to an enhanced immune cell profile (34). Moreover, a diet rich in these foods has been inversely associated with the severity of several respiratory diseases (35). In addition, adherence to the AHEI-2010 and AMED increased the consumption of polyphenols which could mitigate the exaggerated inflammatory and prothrombotic milieu associated with severe COVID-19 illness owing to their anti-inflammatory, antioxidant, and antithrombotic properties (36). An overall healthy diet, therefore, could plausibly play a role in preventing SARS-CoV-2 infection and reducing the severity, if infected (37).
We also examined the EDIH and EDIP in relation to COVID-19, because diets with high insulinemic and inflammatory potential could increase the risk of SARS-CoV-2 infection and worsen the severity (38, 39). Studies with mice models demonstrated that insulin can influence lung mechanical parameters, including tissue resistance and elastance, and hyperinsulinemia can induce bronchoconstriction. Hence, hyperinsulinemia could be a driver of lung dysfunction arising from virus infections (40). A high insulinemic diet directly contributes to sustained hyperglycemia and hyperinsulinemia and the SARS-CoV-2 virus can replicate more rapidly in settings of elevated glucose concentrations in ex vivo studies (18). Response to the SARS-CoV-2 infection can also cause excessive production of proinflammatory molecules, with an abnormal inflammatory response, termed the cytokine storm (41). Thus, a cytokine storm might be exacerbated by diets with high insulinemic or inflammatory potential. Recently, a meta-analysis of 56 studies including 8719 COVID patients found that patients with severe COVID-19 had higher concentrations of inflammatory markers than those with mild disease (42).
A notable strength of our study was that we had access to validated and repeated measures of long-term diet. In addition, information on SARS-CoV-2 infection and symptoms, as well as comprehensive information on covariates, were captured in a timely manner. At the same time, all our study participants were health professionals which allowed us to capture high-quality health information (43). However, the results of the present study need to be interpreted in the context of several limitations. First, we only invited participants who completed the most recent survey to the COVID-19 study, which could lead to selection bias. However, we have high active follow-up rates in our main cohorts (∼90%). Second, we had no information on fatal COVID-19 cases and the outcomes were based on self-report data. However, our study population were dedicated health professionals followed for decades already. Third, there were only a small number of hospitalized COVID-19 cases, which limited our power to observe associations. Last, although we carefully adjusted for several confounders, the possibility of residual confounding due to socioeconomic, lifestyle, and health conditions cannot be ruled out given the observational nature of our study. Specifically, we cannot rule out the possibility that participants who chose a healthier diet might also have been more careful in lowering their exposure to the SARS-CoV-2 virus.
In conclusion, we found that a higher-quality diet was associated with a lower risk of SARS-CoV-2 infection and its severity. At the same time, a diet with a higher inflammatory and insulinemic potential was associated with a higher risk of SARS-CoV-2 infection and severity. Although we carefully adjusted for several confounders, the possibility of residual confounding cannot be ruled out. Still, our results suggest that dietary quality may be important for lowering the burden and severity of COVID-19. Because our data were mainly collected before the time when COVID-19 vaccines became available, further research is warranted to investigate whether the observed associations change or are modified with time, treatments, and vaccination status during the pandemic.
Supplementary Material
Acknowledgments
We would like to thank the participants and staff of the Nurses’ Health Study II and Health Professionals Follow-up Study for their valuable contributions.
The authors’ responsibilities were as follows—WCW, SAS-W, and SNB: designed the research; YY, WM, EKA, and MD: analyzed the data or performed statistical analysis; WM: performed the technical review; FH, ATC, QS, and JR-E: provided conceptual advice; and all authors: acknowledge full responsibility for the analyses and interpretation of the report and read and approved the final manuscript.
Notes
Author disclosures: The authors report no conflicts of interest
Supported in part by NIH grants U01 CA176726 and U01 CA167552 and National Heart, Lung, and Blood Institute grant U01HL145386 . ATC is the Stuart and Suzanne Steele MGH Research Scholar. ATC and WM were supported by the Massachusetts Consortium on Pathogen Readiness (MassCPR).
Supplemental Figures 1–3 and Supplemental Tables 1–4 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.
Abbreviations used: AHEI, Alternate Healthy Eating Index; AMED, alternative Mediterranean Diet; COVID-19, Coronavirus Disease 2019; EDIH, Empirical Dietary Index for Hyperinsulinemia; EDIP, Empirical Dietary Inflammatory Pattern; HPFS, Health Professionals Follow-up Study; IPW, inverse probability weighting; MET, metabolic equivalent of tasks; NHS, Nurses’ Health Study; PPE, personal protective equipment; SARS-CoV-2, Severe Acute Respiratory Syndrome Coronavirus 2; SES, socioeconomic status.
Contributor Information
Yiyang Yue, Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA, USA.
Wenjie Ma, Clinical and Translational Epidemiology Unit, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA; Division of Gastroenterology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA.
Emma K Accorsi, Center for Communicable Disease Dynamics, Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA.
Ming Ding, Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA, USA.
Frank Hu, Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA, USA; Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA; Channing Division of Network Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA.
Walter C Willett, Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA, USA; Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA; Channing Division of Network Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA.
Andrew T Chan, Clinical and Translational Epidemiology Unit, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA; Division of Gastroenterology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA; Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA; Channing Division of Network Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA.
Qi Sun, Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA, USA; Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA; Division of Women's Health, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA.
Janet Rich-Edwards, Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA; Division of Women's Health, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA.
Stephanie A Smith-Warner, Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA, USA; Department of Epidemiology, Harvard TH Chan School of Public Health, Boston, MA, USA.
Shilpa N Bhupathiraju, Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA, USA; Channing Division of Network Medicine, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA.
Data availability
Data described in the article, code book, and analytic code will be made available upon request pending application and approval. Further information, including the procedures to obtain and access data, is described at https://nurseshealthstudy.org/researchers (contact E-mail: nhsaccess@channing.harvard.edu) and https://sites.sph.harvard.edu/hpfs/for-collaborators/.
References
- 1. Wu D, Wu T, Liu Q, Yang Z. The SARS-CoV-2 outbreak: what we know. Int J Infect Dis. 2020;94:44–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Liu C, Zhou Q, Li Y, Garner LV, Watkins SP, Carter LJet al. Research and development on therapeutic agents and vaccines for COVID-19 and related human coronavirus diseases. ACS Cent Sci. 2020;6(3):315–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Wolff D, Nee S, Hickey NS, Marschollek M. Risk factors for Covid-19 severity and fatality: a structured literature review. Infection. 2021;49(1):15–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Hosseini A, Hashemi V, Shomali N, Asghari F, Gharibi T, Akbari Met al. Innate and adaptive immune responses against coronavirus. Biomed Pharmacother. 2020;132:110859. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Catanzaro M, Fagiani F, Racchi M, Corsini E, Govoni S, Lanni C. Immune response in COVID-19: addressing a pharmacological challenge by targeting pathways triggered by SARS-CoV-2. Signal Transduct Target Ther. 2020;5(1):84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Suardi C, Cazzaniga E, Graci S, Dongo D, Palestini P. Link between viral infections, immune system, inflammation and diet. Int J Environ Res Public Health. 2021;18(5):2455. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Maiorino MI, Bellastella G, Longo M, Caruso P, Esposito K. Mediterranean diet and COVID-19: hypothesizing potential benefits in people with diabetes. Front Endocrinol. 2020;11:574315. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Moscatelli F, Sessa F, Valenzano A, Polito R, Monda V, Cibelli Get al. COVID-19: role of nutrition and supplementation. Nutrients. 2021;13(3):976. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Greene MW, Roberts AP, Frugé AD. Negative association between Mediterranean diet adherence and COVID-19 cases and related deaths in Spain and 23 OECD countries: an ecological study. Front Nutr. 2021;8:591964. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Merino J, Joshi AD, Nguyen LH, Leeming ER, Mazidi M, Drew DAet al. Diet quality and risk and severity of COVID-19: a prospective cohort study. Gut. 2021;70(11):2096–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Schwingshackl L, Hoffmann G. Diet quality as assessed by the Healthy Eating Index, the Alternate Healthy Eating Index, the Dietary Approaches to Stop Hypertension score, and health outcomes: a systematic review and meta-analysis of cohort studies. J Acad Nutr Diet. 2015;115(5):780–800.e5. [DOI] [PubMed] [Google Scholar]
- 12. Tabung FK, Brown LS, Fung TT. Dietary patterns and colorectal cancer risk: a review of 17 years of evidence (2000–2016). Curr Colorectal Cancer Rep. 2017;13(6):440–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Chiuve SE, Fung TT, Rimm EB, Hu FB, McCullough ML, Wang Met al. Alternative dietary indices both strongly predict risk of chronic disease. J Nutr. 2012;142(6):1009–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Fung TT, Hu FB, McCullough ML, Newby PK, Willett WC, Holmes MD. Diet quality is associated with the risk of estrogen receptor–negative breast cancer in postmenopausal women. J Nutr. 2006;136(2):466–72. [DOI] [PubMed] [Google Scholar]
- 15. Tabung FK, Smith-Warner SA, Chavarro JE, Wu K, Fuchs SF, Hu FBet al. Development and validation of an empirical dietary inflammatory index. J Nutr. 2016;146(8):1560–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Tabung FK, Wang W, Fung TT, Hu FB, Smith-Warner SA, Chavarro JEet al. Development and validation of empirical indices to assess the insulinaemic potential of diet and lifestyle. Br J Nutr. 2016;116(10):1787–98. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Gustine JN, Jones D. Immunopathology of hyperinflammation in COVID-19. Am J Pathol. 2021;191(1):4–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Codo AC, Davanzo GG, Monteiro LB, de Souza GF, Muraro SP, Virgilio-da-Silva JVet al. Elevated glucose levels favor SARS-CoV-2 infection and monocyte response through a HIF-1α/glycolysis-dependent axis. Cell Metab. 2020;32(3):437–46.e5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Bao Y, Bertoia ML, Lenart EB, Stampfer MJ, Willett WC, Speizer FEet al. Origin, methods, and evolution of the three Nurses’ Health Studies. Am J Public Health. 2016;106(9):1573–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Rimm EB, Giovannucci EL, Willett WC, Colditz GA, Ascherio A, Rosner Bet al. Prospective study of alcohol consumption and risk of coronary disease in men. Lancet. 1991;338(8765):464–8. [DOI] [PubMed] [Google Scholar]
- 21. Yue Y, Petimar J, Willett WC, Smith-Warner SA, Yuan C, Rosato Set al. Dietary flavonoids and flavonoid-rich foods: validity and reproducibility of FFQ-derived intake estimates. Public Health Nutr. 2020;23(18):3295–303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Yuan C, Spiegelman D, Rimm EB, Rosner BA, Stampfer MJ, Barnett JBet al. Validity of a dietary questionnaire assessed by comparison with multiple weighed dietary records or 24-hour recalls. Am J Epidemiol. 2017;185(7):570–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Yue Y, Yuan C, Wang DD, Wang M, Song M, Shan Zet al. Reproducibility and validity of diet quality scores derived from food-frequency questionnaires. Am J Clin Nutr. 2022;115(3):843–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Al-Shaar L, Yuan C, Rosner B, Dean SB, Ivey KL, Clowry CMet al. Reproducibility and validity of a semiquantitative food frequency questionnaire in men assessed by multiple methods. Am J Epidemiol. 2021;190(6):1122–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Wolf AM, Hunter DJ, Colditz GA, Manson JE, Stampfer MJ, Corsano KAet al. Reproducibility and validity of a self-administered physical activity questionnaire. Int J Epidemiol. 1994;23(5):991–9. [DOI] [PubMed] [Google Scholar]
- 26. Ainsworth BE, Haskell WL, Leon AS, Jacobs JD, Montoye HJ, Sallis JFet al. Compendium of physical activities: classification of energy costs of human physical activities. Med Sci Sports Exerc. 1993;25(1):71–80. [DOI] [PubMed] [Google Scholar]
- 27. WHO Working Group on the Clinical Characterisation and Management of COVID-19 infection . A minimal common outcome measure set for COVID-19 clinical research. Lancet Infect Dis. 2020;20(8):e192–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Menni C, Valdes AM, Freidin MB, Sudre CH, Nguyen LH, Drew DAet al. Real-time tracking of self-reported symptoms to predict potential COVID-19. Nat Med. 2020;26(7):1037–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Rich-Edwards JW, Ding M, Rocheleau CM, Boiano JM, Kang JH, Becene Iet al. American frontline healthcare personnel's access to and use of personal protective equipment early in the COVID-19 pandemic. J Occup Environ Med. 2021;63(11):913–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Tabung FK, Wang W, Fung TT, Smith-Warner SA, Keum N, Wu Ket al. Association of dietary insulinemic potential and colorectal cancer risk in men and women. Am J Clin Nutr. 2018;108(2):363–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Wang M, Spiegelman D, Kuchiba A, Lochhead P, Kim S, Chan ATet al. Statistical methods for studying disease subtype heterogeneity. Stat Med. 2016;35(5):782–800. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Kim H, Rebholz CM, Hegde S, LaFiura C, Raghavan M, Lloyd JFet al. Plant-based diets, pescatarian diets and COVID-19 severity: a population-based case–control study in six countries. BMJ Nutr Prev Health. 2021;4(1):257–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Tavakol Z, Ghannadi S, Tabesh MR, Halabchi F, Noormohammadpour P, Akbarpour Set al. Relationship between physical activity, healthy lifestyle and COVID-19 disease severity; a cross-sectional study. Z Gesundh Wiss. 2021; Feb 4 (Epub ahead of print; doi: 10.1007/s10389-020-01468-9). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Hosseini B, Berthon BS, Saedisomeolia A, Starkey MR, Collison A, Wark PABet al. Effects of fruit and vegetable consumption on inflammatory biomarkers and immune cell populations: a systematic literature review and meta-analysis. Am J Clin Nutr. 2018;108(1):136–55. [DOI] [PubMed] [Google Scholar]
- 35. Li W, Liu Q, Chen Y, Yang B, Huang X, Li Yet al. Effects of indoor environment and lifestyle on respiratory health of children in Chongqing, China. J Thorac Dis. 2020;12(10):6327–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Angelidi AM, Kokkinos A, Katechaki E, Ros E, Mantzoros CS. Mediterranean diet as a nutritional approach for COVID-19. Metabolism. 2021;114:154407. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Calder PC. Nutrition, immunity and COVID-19. BMJ Nutr Prev Health. 2020;3(1):74–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Zabetakis I, Lordan R, Norton C, Tsoupras A. COVID-19: the inflammation link and the role of nutrition in potential mitigation. Nutrients. 2020;12(5):1466. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Lim S, Bae JH, Kwon H-S, Nauck MA. COVID-19 and diabetes mellitus: from pathophysiology to clinical management. Nat Rev Endocrinol. 2021;17(1):11–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Santos A, Magro DO, Evangelista-Poderoso R, Saad MJA. Diabetes, obesity, and insulin resistance in COVID-19: molecular interrelationship and therapeutic implications. Diabetol Metab Syndr. 2021;13(1):23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Leung C. Clinical features of deaths in the novel coronavirus epidemic in China. Rev Med Virol. 2020;30(3):e2103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42. Ji P, Zhu J, Zhong Z, Li H, Pang J, Li Bet al. Association of elevated inflammatory markers and severe COVID-19: a meta-analysis. Medicine (Baltimore). 2020;99(47):e23315. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Belanger MJ, Hill MA, Angelidi AM, Dalamaga M, Sowers JR, Mantzoros CS. Covid-19 and disparities in nutrition and obesity. N Engl J Med. 2020;383(11):e69. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data described in the article, code book, and analytic code will be made available upon request pending application and approval. Further information, including the procedures to obtain and access data, is described at https://nurseshealthstudy.org/researchers (contact E-mail: nhsaccess@channing.harvard.edu) and https://sites.sph.harvard.edu/hpfs/for-collaborators/.