Vaccine Mandates and Influenza Vaccination During the Pandemic

We conducted a longitudinal cohort study using enrollment and claims data from a regional nonprofit insurer in the Northeast with 13% to 30% of market share in the region.²⁴  State variation in influenza vaccine mandates provided a natural experiment to evaluate the effect of the mandate.

We included children aged 6 months to 18 years living in Massachusetts, New Hampshire, and Maine, with continuous commercial insurance coverage from March 2019 to May 2021. Individuals aged <6 months by September 1, 2019, or 19 years or older by April 1, 2021, were excluded. Children with a claim documenting a history of an allergic reaction to any vaccine (International Classification of Diseases, 10th Revision, code Z88.7) were also excluded.

The primary outcome of interest was receipt of any influenza vaccine in the 2020–2021 season, based on having 1 or more influenza vaccine Current Procedural Terminology/Healthcare Common Procedure Coding System codes (see Appendix) in claims from September 1, 2020, to March 1, 2021.

The main predictors in this study were living in a state that adopted an influenza vaccine mandate in 2020, county-level COVID-19 severity, and receipt of any influenza vaccine in the preceding influenza season (2019–2020). Children living in Massachusetts were exposed to a vaccine mandate in the 2020–2021 season, whereas those living in New Hampshire and Maine were never exposed to a mandate. Community COVID-19 severity was measured using county-level total cumulative confirmed case counts between the start of the pandemic (March 2020) and August 2020 (start of the 2020–2021 flu season) compiled by Johns Hopkins University²⁵  and connected to member data through zip codes. Case counts per 100 000 in the child’s zip code were converted to categorical variables (<500, 500–999, 1000–1499, 1500–1999, >2000) on the basis of previous studies.²⁶  Children with 1 or more influenza vaccine Current Procedural Terminology/ Healthcare Common Procedure Coding System codes in claims from September 1, 2019, to March 1, 2020, were categorized as receiving an influenza vaccine in the 2019–2020 season (ie, previous vaccination). We truncated the 2019–2020 and 2020–2021 flu seasons at the beginning of March because of overlap of the 2019–2020 season with the start of the COVID-19 pandemic in March 2020.

Covariates included member sex, age, race and ethnicity, overall Social Vulnerability Index (SVI)²⁷  score for child’s census tract of residence, presence of a chronic condition in 2019 based on the Pediatric Medical Complexity Algorithm,²⁸  child’s diagnosis of influenza in the 2019–2020 season, family member diagnosis of influenza in the 2019–2020 season, and family member diagnosis of COVID-19 between January 1, 2020, and September 1, 2020. The social constructs of race and ethnicity were included as potential confounders in our analyses because they have been shown to be associated with both influenza vaccination rates and community COVID-19 severity.^29–31  To measure race and ethnicity, the health plan uses a hierarchical combination of (1) self-reported data, (2) data from hospitals and health systems, and (3) predicted race and ethnicity based on the Rand Bayesian Improved Surname Geocoding.³²  Information on member census tract was used to link to the Center for Disease Control and Prevention SVI scores, representing a ranking of each census tract on 15 social factors meant to identify communities with decreased capacity to anticipate and recover from natural disasters and disease outbreaks.²⁷  SVI scores have been used as a measure of social risk and to characterize factors that may contribute to health disparities.^33,34  As in these studies, scores were categorized by quartile, labeled as low, low–moderate, moderate–high, and high. Race/ethnicity, SVI, and community COVID-19 severity did not exhibit multicollinearity on the basis of variance inflation factors <10. Previous diagnosis of influenza or COVID-19 infection in children and family members on the same insurance plan were identified using International Classification of Diseases, 10th Revision, codes (see Appendix).

We used means and frequencies to describe the characteristics of children who did and did not receive influenza vaccination in the 2020–2021 season. We assessed whether living in a state with an influenza vaccine mandate or community COVID-19 severity predicted influenza vaccine uptake in the 2020–2021 influenza season using an individual-level logistic regression model controlling for previous vaccination in the 2019–2020 season, child age, sex, race/ethnicity, presence of chronic conditions based on Pediatric Medical Complexity Algorithm, overall SVI category, child influenza diagnosis in the 2019–2020 season, family member with an influenza diagnosis in the 2019–2020 season, and family member with a COVID-19 diagnosis in 2020.

To assess whether living in a state with an influenza vaccine mandate or a county with elevated COVID-19 severity had a differential impact for previous vaccinators versus nonvaccinators (ie, effect modification by previous vaccination status), we included 2 interaction terms in the model. The first was between living in a state with an influenza vaccine mandate and receipt of influenza vaccine in the 2019–2020 influenza season. The second was between county COVID-19 severity category and receipt of influenza vaccine in the 2019–2020 influenza season. Using marginal effects methods,³⁵  we compared the difference in predicted probability of influenza vaccination in 2020 to 2021 between those living in mandate and nonmandate states for previous vaccinators and nonvaccinators, respectively, and then calculated the difference between previous vaccinators and nonvaccinators. We did the same to compare predicted probabilities of vaccination between those living in the counties with the highest COVID-19 severity (>2000 cases per 100 000) and the lowest COVID severity (<500 cases per 100 000) by previous vaccination status, and then the differences between previous vaccinators and nonvaccinators. Because previous studies have demonstrated a differential impact by race/ethnicity, we separately added another interaction term for mandate and white versus nonwhite race.³⁶  All analyses accounted for clustering between children within the same family sharing the same insurance plan.

Because influenza vaccines given at sites other than physicians’ offices may not be fully captured in claims data, a sensitivity analysis was conducted among children aged <3 years as of September 1, 2019. This age group is more likely to be vaccinated in a physician’s office than a school setting and most pharmacies will not vaccinate children aged <3 years.^37,38 

Of the 71 333 children in this study, the mean age was 9.7 years. The majority of the population was white and two-thirds of participants lived in Massachusetts (Table 1). More than half lived in a county where between 1500 and 2000 COVID-19 cases per 100 000 had been identified through August 2020.

Within the study population, 45% of individuals received influenza vaccines in both the 2019–2020 and 2020–2021 influenza seasons, 26% did not receive an influenza vaccine in either season, and 29% were vaccinated in 1 season but not the other (14.4% in the 2019 season only, 14.8% in the 2020 season only).

Overall, receipt of an influenza vaccine in the 2019–2020 season was a statistically significant predictor of vaccination in the 2020–2021 season (Table 2). Living in a state with a vaccine mandate was a significant predictor of vaccination in the 2020–2021 season, whereas the county COVID-19 severity was not. Young age (1–4 years), having a chronic medical condition, and having a family member who was diagnosed with influenza in the 2019–2020 season were also significant predictors of influenza vaccination in the 2020–2021 season. Black race was significantly associated with not receiving influenza vaccination in the 2020–2021 season.

The association between influenza vaccination in 2020 to 2021 and the main predictors of interest (vaccine mandate and COVID-19 severity) varied on the basis of whether the child had been vaccinated in the previous season (Table 3). Although both previous vaccinators and nonvaccinators had significantly higher probability of vaccination in mandate versus nonmandate states in the 2020–2021 season (47.7% vs 21.2%, respectively, for previous nonvaccinators, P < .001, and 78.2% vs 58.2% for previous vaccinators, P < .001), the mandate was associated with a 6.5 percentage point higher increase in vaccination rate among previous nonvaccinators than previous vaccinators (P = .014). There was no differential impact of mandate by race (interaction term P = .682). Among previous vaccinators, high (versus low) community COVID-19 severity was associated with significantly lower rates of influenza vaccination in 2020 to 2021 (72.1% vs 77.3%, P = .005). The impact of county-level COVID-19 severity on vaccination rates for previous nonvaccinators was not statistically significant (39.4% vs 43.2%, P = .125 in high versus low COVID-19 severity communities, respectively). The differential impact of COVID-19 severity on vaccination rates for previous nonvaccinators relative to previous vaccinators was not significant (interaction term P = .6).

The effect of living in a mandate state was similar in sensitivity analyses among children aged <3 years, with significantly higher vaccination rates in a mandate state, more so among nonvaccinators (Supplemental Tables 4 and 5). Among younger children, high county COVID-19 severity was not associated with lower rates of influenza vaccination among previous vaccinators.

This longitudinal child-level study found that, controlling for previous influenza vaccination, living in a state that adopted an influenza vaccine mandate was associated with higher probability of receiving an influenza vaccine during the 2020–2021 season compared with those living in states without a mandate. This impact was significantly greater among children who were not vaccinated against influenza in the previous season. Among previously vaccinated children, living in a county with high COVID-19 severity was associated with lower influenza vaccination rates in 2020 to 2021 compared with living in low COVID-19 severity counties.

An influenza vaccine mandate appears to be an effective but not sufficient strategy to achieve optimal vaccine coverage rates. Our findings are consistent with studies on influenza vaccine mandates in other states.^22,23  In New York City, cross-sectional vaccination rates for prekindergartners subject to a mandate increased by 6.3 percentage points relative to children not subject to the mandate. For Connecticut children aged 6 to 59 months, vaccination rates increased from 67.8% to 84% after an influenza vaccine mandate was implemented, a rate similar to ours among this age group. Among parents who have been willing to vaccinate, a vaccination mandate may be an additional stimulus to overcome barriers of convenience³⁹  to complete vaccination. For those not vaccinating previously, school-related mandates may be a strong lever. However, even in a state with a vaccine mandate, the predicted probability of influenza vaccination among previous nonvaccinators remained <50%. Caregiver vaccine hesitancy, specifically as it relates to influenza vaccination,^40,41  may limit the effect of vaccine mandates, and other approaches may be needed. Because of an ultimately mild influenza season, the Massachusetts influenza vaccination mandate was dropped in January of 2021.⁴²  However, because the original mandate deadline for vaccination was December 31, 2020,¹⁸  the bulk of the response should have occurred before it being rescinded, and the low severity of the 2020–2021 influenza season is unlikely to have affected vaccination rates.⁴³ 

Overall, we found that previous vaccinators had lower rates of subsequent vaccination in communities with high versus low COVID-19 severity. Parents in communities with high levels of COVID-19 may have been wary of making an additional trip to a clinic or pharmacy for vaccination, or may have faced limited availability of in-person visits. In the setting of more-severe disease outbreaks, parents may need targeted messaging and strategies to encourage safe and convenient venues to access influenza vaccines.

We found no difference by community COVID-19 severity in the probability of vaccination in the 2020–2021 season for children aged <3 years in sensitivity analysis, regardless of previous vaccination status. In contrast to the dampening effect of high community COVID-19 severity on vaccination rates for previous vaccinators that we saw among the full population, a lack of differential effect of COVID-19 severity for younger children may be attributable to greater influenza vaccination access for younger children compared with older children.^44,45  During the first year of the pandemic, when access to in-person care was limited, infants and young children were prioritized over older children for in-person visits at many clinics given the need for routine vaccination and developmental surveillance.⁴⁶ 

It is well established that COVID-19 disproportionately affected minority and lower socioeconomic status communities.^30,31  Our findings highlight another way in which COVID-19 may have worsened preexisting health disparities through lower rates of influenza vaccination access. We found that Black children and those living in communities with higher overall SVI scores were less likely to be vaccinated against influenza. These groups were also more likely to live in counties with elevated COVID-19 severity,⁴⁷  where we found influenza vaccination rates among previous vaccinators to be significantly lower than in low COVID-19 severity areas. Unlike previous studies,³⁶  this study found no differential impact of a mandate by race. Additional efforts should be made to improve messaging and increase accessibility of influenza vaccines within these communities to address the disproportionate morbidity and mortality caused by seasonal influenza.

Future studies are needed to assess the downstream effects of the COVID-19 pandemic on family decisions about influenza vaccination. Preliminary studies suggest that concerns about the COVID vaccine may have already negatively influenced influenza vaccination rates nationally.⁴⁸  Although mandates may encourage caregivers receptive to vaccines to vaccinate against influenza, they likely will not work as well among families with staunch antivaccine views.⁴⁹  Further work is needed to enhance communication and policy strategies incentivizing potentially hesitant caregivers to vaccinate children against influenza.

There were several limitations to this study. Results from a population with continuous private insurance in the Northeast may not generalize to publicly or intermittently insured children in the rest of the country. To account for population differences between study states, we controlled for numerous sociodemographic characteristics, and used a continuously enrolled population to enable a within-child pre–post design. Only about 30% of enrollees were excluded for not having continuous coverage. Response to the mandate may have depended on whether children attended public or private school or attended remotely, as done by 67% of students nationally in the fall of 2020,⁵⁰  but school data were not available for our study. Children who were vaccinated at sites that do not bill insurance (ie, schools or health departments) may have been miscategorized as not having received a vaccination. However, rates of vaccination in these alternate sites are low, particularly among younger children, and claims data have been used in other studies to measure influenza vaccination, including from pharmacies.^38,51 

In this longitudinal cohort study, children living in a state with a vaccine mandate during the 2020–2021 influenza season had a higher predicted probability of receiving an influenza vaccine than those living in states without a mandate. The impact of the mandate was greater among previous nonvaccinators. Previously vaccinated children had a lower predicted probability of receiving an influenza vaccine if they lived in a county with high COVID-19 severity. Strategies to improve uptake of influenza vaccination may have differential impacts based on previous vaccination acceptance and must account for community factors such as the COVID-19 pandemic.

Influenza vaccine Current Procedural Terminology codes: 90653, 90694, 90662, 90672, 90674, 90682, 90685, 90686, 90687, 90688, 90756

Influenza vaccine Healthcare Common Procedure Coding System codes: Q2034, Q2035, Q2036, Q2037, Q2038, Q2039, G0008

Influenza illness International Classification of Diseases, 10th Revision, codes: J09–J18

COVID-19 illness International Classification of Diseases, 10th Revision: U07.1, B34.2, B97.2, J12.81, Z86.16, M35.81, J12.82

COVID-19

coronavirus disease 2019

SVI

Social Vulnerability Index