Effectiveness of two and three mRNA COVID‐19 vaccine doses against Omicron‐ and Delta‐Related outpatient illness among adults, October 2021–February 2022

Abstract Background We estimated SARS‐CoV‐2 Delta‐ and Omicron‐specific effectiveness of two and three mRNA COVID‐19 vaccine doses in adults against symptomatic illness in US outpatient settings. Methods Between October 1, 2021, and February 12, 2022, research staff consented and enrolled eligible participants who had fever, cough, or loss of taste or smell and sought outpatient medical care or clinical SARS‐CoV‐2 testing within 10 days of illness onset. Using the test‐negative design, we compared the odds of receiving two or three mRNA COVID‐19 vaccine doses among SARS‐CoV‐2 cases versus controls using logistic regression. Regression models were adjusted for study site, age, onset week, and prior SARS‐CoV‐2 infection. Vaccine effectiveness (VE) was calculated as (1 − adjusted odds ratio) × 100%. Results Among 3847 participants included for analysis, 574 (32%) of 1775 tested positive for SARS‐CoV‐2 during the Delta predominant period and 1006 (56%) of 1794 participants tested positive during the Omicron predominant period. When Delta predominated, VE against symptomatic illness in outpatient settings was 63% (95% CI: 51% to 72%) among mRNA two‐dose recipients and 96% (95% CI: 93% to 98%) for three‐dose recipients. When Omicron predominated, VE was 21% (95% CI: −6% to 41%) among two‐dose recipients and 62% (95% CI: 48% to 72%) among three‐dose recipients. Conclusions In this adult population, three mRNA COVID‐19 vaccine doses provided substantial protection against symptomatic illness in outpatient settings when the Omicron variant became the predominant cause of COVID‐19 in the United States. These findings support the recommendation for a third mRNA COVID‐19 vaccine dose.

75D30121C11519, 75D30121C11529, 75D30121C11909, 75D30121C12246, 75D30121C12279, and 75D30121C12339. At Vanderbilt, the project was also funded by the National Center for Advancing Translational Sciences (NCATS) Clinical Translational Science Award (CTSA) Program, award number 5UL1TR002243-03. At Pittsburgh, the project was also supported by the National Institutes of Health through grant UL1TR001857.
Conclusions: In this adult population, three mRNA COVID-19 vaccine doses pro- Studies with active enrollment such as the US Flu Vaccine Effectiveness Network (US Flu VE Network) provide access to specimens for research purposes including whole genome sequencing and access to data not available in medical records including risk factors for SARS-CoV-2 infection. 6 To assess the impact of a third dose in the context of emerging variants with immune evasion 7

| SARS-CoV-2 status
Participants were tested for SARS-CoV-2 by reverse-transcription polymerase chain reaction tests using respiratory specimens collected for clinical or research purposes. We classified participants with a pos-

| COVID-19 vaccination status
COVID-19 vaccination status was verified using electronic medical records, immunization information systems, and vaccination record cards. Participants considered vaccinated with two doses were those who received two mRNA vaccine doses ≥14 days before illness onset (two-dose). To be considered for the two-dose analyses, participants must have received doses ≥16 days apart for Pfizer-BioNTech vaccines and ≥23 days apart for Moderna vaccines. Participants considered vaccinated with three doses were those who received three mRNA vaccine doses, where the third dose was given ≥7 days before illness onset (three-dose). 4 Participants who received a third dose before the recommended ≥150 days after the second dose were also considered three-dose recipients but excluded from sensitivity analyses. Three-dose recipients included both immunocompromised participants who received a third dose as a primary series and otherwise healthy participants who received a third dose as a booster. Those who did not report vaccine receipt and had no documentation of an mRNA COVID-19 vaccination before illness onset were defined as unvaccinated. We excluded participants who self-reported COVID-19 vaccination but were missing verified documentation of doses received.

| Statistical analyses
We limited analyses to adults aged ≥18 years. Using the test-negative design, 10 we compared the odds of two-or three-dose mRNA COVID-19 vaccination among COVID-19 cases versus test-negative controls using logistic regression. VE was calculated as (1 À adjusted odds ratio) Â 100%. Regression models were adjusted for variables identified a priori including study site, age, and illness onset week.
Sex, race and ethnicity, illness onset to specimen collection interval, self-reported high-risk exposure, self-reported chronic medical condition, and self-reported prior SARS-CoV-2 infection were evaluated as model covariates using a change-in-estimate (≥5% change in odds ratio) forward stepwise approach. In addition to covariates included a priori, prior SARS-CoV-2 infection was included in the final regression model because its inclusion changed the OR by 8%. All other potential confounders examined changed the OR by <1%. Network. Due to co-circulation of the Delta and Omicron variants between December 10 and 19, 2021, we excluded participants without sequenced viruses with onset dates during this period for variantspecific estimates. We also assessed potential waning immunity among two-dose recipients by comparing VE of those who received their second dose 14-149 days versus ≥150 days prior to illness onset during each variant predominant period.
We conducted several subgroup analyses where three-dose VE was stratified by self-reported high-risk exposure status, self-reported chronic medical condition, self-reported prior SARS-CoV-2 infection, days between illness onset and specimen collection date, and selfreported presence of fever with cough or shortness of breath during the Delta and Omicron predominant periods. Analyses by illness onset to specimen collection interval were performed to identify bias resulting from potential false negative SARS-CoV-2 test results among participants who presented for care or testing later than those presenting 0-2 days after illness onset. 10 Analyses by symptoms were performed to evaluate VE among persons with potentially more severe illness compared to those without fever paired with cough or shortness of breath, indicating more mild illness. Over the entire study period, participants who self-reported a high-risk exposure or reported fever were more likely to test positive (Table 1). Additionally, participants who were aged ≥65 years, identified as White non-Hispanic or other race non-Hispanic compared with Black non-Hispanic or Hispanic, self-reported a chronic medical condition, and did not self-report a fever were more likely to receive a third vaccine dose ( Table 2). Among two-dose recipients, the median interval between receipt of a second dose and illness onset date was 225 days (range 14-386); 13% and 87% had received a second mRNA vaccine dose 14-149 days or ≥150 days prior to illness onset, respectively (data not shown). The median interval between third dose receipt and illness onset was 53 days (range: 7-230) (data not shown).

| Vaccine effectiveness by subgroup
Self-reported high-risk exposure status, self-reported presence of a chronic medical condition, self-reported prior laboratory-confirmed SARS-CoV-2 infection, longer interval from illness onset to respiratory specimen collection, and self-reported presence of fever with cough or shortness of breath did not change three-dose VE during the Delta variant predominant period (Table 4). However, during the period when the Omicron variant predominated, three-dose VE point estimates tended to be lower but with overlapping confidence intervals among those who had a high-risk exposure, a chronic medical condition, a prior SARS-CoV-2 infection, or CLI that included fever. During the Delta period, 4% of cases and 14% of controls had prior infection compared with the Omicron period when 15% of cases and 20% of controls had prior infection.

| DISCUSSION
This investigation adds to early evidence of effectiveness of a third mRNA vaccine dose against laboratory-confirmed SARS-CoV-2 infection among adults seeking outpatient care and clinical testing for CLI symptoms during the pandemic wave predominated by the Omicron variant. 3,[11][12][13][14] However, three-dose effectiveness among adults was lower during the Omicron predominant period than during the pandemic wave associated with the Delta variant. Similar to analyses of large electronic medical record databases or data from SARS-CoV-2 testing sites, three-dose VE in this analysis was higher against Delta than against Omicron-related illness. 3,11,12 Findings from the US Flu VE Network are also consistent with higher estimates of two-dose VE when the second dose was given less than 5 months before current illness onset compared with at least