Risk of symptomatic severe acute respiratory syndrome coronavirus 2 infection not associated with influenza vaccination in the 2019–2020 season

The association of influenza vaccine and severe acute respiratory syndrome coronavirus 2 (SARS‐CoV‐2) infection was assessed by test‐negative design using data collected for a study of outpatient COVID‐19‐like illness with onset dates from June to September 2020. Multivariable logistic regression models examined the association between receipt of 2019–2020 influenza vaccine and PCR‐confirmed SARS‐CoV‐2 with adjustment for potential confounders. Receipt of influenza vaccine during the 2019–2020 influenza season was not associated with increased odds of SARS‐CoV‐2 infection in adults (aOR 0.83, 95% CI 0.63 to 1.10) or children (aOR 0.92, 95% CI 0.47 to 1.80).

Methodological flaws were identified that biased the results toward a positive association. 3 The author subsequently published a letter confirming that this study provided no evidence for a link between influenza vaccination and COVID-19 risk. 4 However, anti-vaccine advocates continue to raise concerns that influenza vaccination may increase the risk of COVID-19. Subsequent studies have failed to confirm a positive association between influenza vaccination and SARS-CoV-2 infection. A recent systematic review assessed the relationship between influenza vaccination and COVID-19 in seven studies. 5 One study was a COVID-19 prediction model that was not intended to estimate the independent risk from influenza vaccination. 6 The other studies found either no association or a modest reduction in risk of SARS-CoV-2 infection among influenza vaccine recipients after adjusting for potential confounders. A subsequent cross-sectional analysis of patients tested for SARS-CoV-2 in Israel found that influenza vaccination in 2019-2020, 2018-2019, or both seasons was associated with reduced risk of SARS-CoV-2 infection. 7 The test-negative design (TND) is a robust method for estimating influenza vaccine effectiveness (VE), and yields a valid estimate of VE under most scenarios. 8 Prospective studies of influenza vaccination and coronaviruses using the TND have been reported from Europe and Canada, but not the United States.

| METHODS
This was a secondary analysis of data from a COVID-19 epidemiology study. Patients with an acute illness were prospectively enrolled with verbal consent. All participants had results from clinically ordered reverse transcription polymerase chain reaction (RT-PCR) testing for SARS-CoV-2. To ensure complete and accurate immunization records, we restricted analyses to participants with at least two clinical encounters with a health system provider during the 36 months prior to study enrollment and any prior vaccine record in the local immunization registry (RECIN). RECIN exchanges data with the Wisconsin Immunization Registry and has been previously validated for nearly complete capture of influenza vaccinations. 10 Patients with an acute illness ≤10 days duration (onset to swab) were eligible if they met symptom criteria which included (1) loss of taste or smell or (2) cough or fever and at least one additional respiratory (shortness of breath, runny nose/nasal congestion, and sore throat) or gastrointestinal (abdominal pain, diarrhea, and vomiting) symptom. Participants with illness onset between June 1 and September 30, 2020 were included. We excluded participants who received 2019-2020 influenza vaccination <14 days prior to illness onset and participants who received 2020-2021 influenza vaccine any time prior to illness onset.
We assessed the association between influenza vaccination and COVID-19 using the TND. A case was defined as a study participant with a positive RT-PCR result for SARS-CoV-2 infection.
Controls had a negative result for SARS-CoV-2. Study samples were not tested for influenza, but there was no evidence of influenza circulation during the enrollment period. Multivariable logistic regression models examined the association between receipt of 2019-2020 influenza vaccine and PCR-confirmed COVID-19 with adjustment for potential confounders. Age as a spline and sample collection interval (days) were included a priori in all models. We assessed sex, high-risk condition (present or not present), and calendar time (month of onset) as potential confounders. Variables were included in the final model if they changed the influenza vaccine coefficient by ≥5%. Presence of a high-risk condition was defined by specific ICD-10 codes in the electronic health record from October 1, 2019, through the date of enrollment and was consistent with those used by the US Flu VE Network. 11 The primary analysis included all ages. Secondary analyses generated separate models for children (age 0 to <18 years) and adults (age ≥18 years).
Covariates for the age-stratified models were assessed separately, and age was included as a categorical variable for children. The Marshfield Clinic Research Institute Institutional Review Board approved this study.

| RESULTS
The analysis included 1736 symptomatic patients. The median age was 38 years, and 60% were female. The interval from symptom onset to sample collection was less than 5 days in 78%. Twenty-eight percent were positive for SARS-CoV-2 by RT-PCR, and 65% had received a 2019-2020 influenza vaccine ( Table 1). The peak month for illness onset was July with 594 enrolled. Almost half of all COVID-19 cases in the study occurred in September (n = 240, 49%).

| DISCUSSION
In this TND analysis, we found no positive association between  asymptomatic patients prior to surgical procedures). We assessed 2019-2020 influenza vaccination status using a previously validated vaccination registry to ensure accurate capture of vaccination status. 10 We also conducted secondary, age-stratified analyses, confirming children had similar findings to adults, with no observed impact of influenza vaccination on risk of SARS-CoV-2 infection. Influenza infections are a potential source of confounding in a test-negative analysis of SARS-CoV-2 infection, but this study was conducted during the summer when influenza was not circulating.
The timing of the study is a potential limitation. It was conducted 6-9 months after most influenza vaccines were administered. As a result, we were unable to assess short-term effects of influenza vaccination on risk of SARS-CoV-2 infection.
In conclusion, this study did not identify any association between influenza vaccination and symptomatic SARS-CoV-2 infection in adults or children. The results from this study and others support current public health recommendations for annual influenza vaccination.