Adjusted HRs of COVID-19 risk and severity according to hPDI scores
| Low hPDI (Q1; n=148 143) | Intermediate hPDI (Q2–Q3; n=296 286) | High hPDI (Q4; n=148 142) | P for trend | |
| hPDI score, median (P25–P75) | 45 (43–47) | 51 (49–52) | 56 (55–58) | |
| COVID-19 risk | ||||
| No of events/person-months | 8739/839 747 | 15 733/2 026 824 | 7359/1 022 078 | — |
| Incidence rate (10 000 person-months; 95% CI) | 104.1 (101.9 to 106.2) | 77.6 (76.4 to 78.8) | 72.0 (70.4 to 73.7) | — |
| Age-adjusted model | 1.00 (Ref) | 0.85 (0.82–0.87) | 0.80 (0.78–0.83) | <0.001 |
| Multivariable model 2 | 1.00 (Ref) | 0.85 (0.83–0.87) | 0.81 (0.78–0.83) | <0.001 |
| Multivariable model 3 | 1.00 (Ref) | 0.91 (0.89–0.93) | 0.91 (0.88–0.94) | <0.001 |
| COVID-19 risk (positive test) | ||||
| No of events/person-months | 1423/869 664 | 2829/2 081 970 | 1350/1 046 887 | — |
| Incidence rate (10 000 person-months; 95% CI) | 16.4 (15.5 to 17.2) | 13.6 (13.1 to 14.1) | 12.9 (12.2 to 13.6) | — |
| Age-adjusted model* | 1.00 (Ref) | 0.86 (0.83–0.90) | 0.79 (0.75–0.83) | <0.001 |
| Multivariable model 2* | 1.00 (Ref) | 0.87 (0.84–0.91) | 0.80 (0.76–0.84) | <0.001 |
| Multivariable model 3* | 1.00 (Ref) | 0.88 (0.85–0.92) | 0.82 (0.78–0.86) | <0.001 |
| Severe COVID-19 | ||||
| No of events/person-months | 187/871 995 | 390/2 086 790 | 163/1 049 476 | — |
| Incidence rate (10 000 person-months; 95% CI) | 2.1 (1.9 to 2.5) | 1.9 (1.7 to 2.1) | 1.6 (1.3 to 1.8) | — |
| Age-adjusted model | 1.00 (Ref) | 0.66 (0.56–0.77) | 0.45 (0.36–0.57) | <0.001 |
| Multivariable model 2 | 1.00 (Ref) | 0.66 (0.57–0.78) | 0.45 (0.36–0.57) | <0.001 |
| Multivariable model 3 | 1.00 (Ref) | 0.77 (0.66–0.91) | 0.59 (0.47–0.74) | <0.001 |
HRs and 95% CI for COVID-19 risk and severity. COVID-19 risk defined using a validated symptom-based model. COVID-19 or an RT-PCR positive test report. COVID-19 severity was defined based on hospitalisation with requirement of oxygen support (methods, online supplemental file).
Cox proportional hazards models were stratified by calendar date at study entry, country of origin and 10-year age group (age-adjusted model).
Multivariable model 2 was further adjusted for sex (male, female), race/ethnicity (white, black, Asian, other), index of multiple deprivation (most deprived <3, intermediate deprived 3–7, less deprived >7), population density (<500 individuals/km2, 500–1999 individuals/km2, 2000–4999 individuals/km2 and ≥5000 individuals/km2) and healthcare worker status (yes with interaction with patients with COVID-19, yes without interaction with patients with COVID-19, no).
Model 3 was further adjusted for presence of comorbidities (diabetes (yes, no), cardiovascular disease (yes, no), lung disease (yes, no), cancer (yes, no), kidney disease (yes, no)), body mass index (<18.5 kg/m2, 18.5–24.9 kg/m2, 25.0–29.9 kg/m2 and ≥30 kg/m2), smoking status (yes, no) and physical activity (<1 day/week, 1–2 days/week, 3–4 days/week, ≥5 days/week).
*Inverse probability-weighted analyses were conducted to account for predictors of obtaining RT-PCR testing (presence of COVID-19-related symptoms, interaction with a COVID-19 case, healthcare worker, age group and race). Inverse probability-weighted Cox proportional hazards models were stratified by 10-year age group and date with additional adjustment for the covariates used in previous models.
hPDI, healthful Plant-Based Diet Index; RTPCR, reverse transcription PCR.