key: cord-0336293-pt1r3qph authors: Nguyen, A. T.; Arnold, B. F.; Kennedy, C. J.; Mishra, K.; Pokpongkiat, N.; Seth, A.; Djajadi, S.; Holbrook, K.; Pan, E.; Kirley, P. D.; Libby, T.; Hubbard, A. E.; Reingold, A.; Colford, J. M.; Benjamin-Chung, J. title: Evaluation of a city-wide school-located influenza vaccination program in Oakland, California with respect to race and ethnicity: a matched cohort study date: 2021-07-23 journal: nan DOI: 10.1101/2021.07.21.21260311 sha: 4265495d8614ff89150930258592c63971cadbdb doc_id: 336293 cord_uid: pt1r3qph Objectives: To evaluate the effectiveness of city-wide school-located influenza vaccination by race/ethnicity from 2014-2018. Methods: We used multivariate matching to pair schools in the intervention district in Oakland, CA with schools in West Contra Costa County, CA, a comparison district. We estimated difference-in-differences (DIDs) in caregiver-reported influenza vaccination coverage and laboratory-confirmed influenza hospitalization incidence. Results: Differences in influenza vaccination coverage in the intervention vs. comparison site were larger among White and Latino students than Asian/Pacific Islander (API), Black, and multiracial students. Concerns about vaccine effectiveness or safety were more common among Black and multiracial caregivers; logistical barriers to vaccination were more common among White, API, and Latinos. In both sites, hospitalization in 2017-18 was higher in Blacks vs. other races/ethnicities. All-age influenza hospitalization incidence was lower in the intervention site vs. comparison site among White/API individuals in 2016-17 and 2017-18 and Black older adults in 2017-18, but not in other groups. Conclusions: SLIV was associated with higher vaccination coverage and lower influenza hospitalization, but associations varied by race/ethnicity. SLIV alone may be insufficient to ensure equitable health outcomes for influenza. To reduce influenza transmission, the Advisory Committee on Immunization Practices 18 (ACIP) recommends annual influenza vaccinations for all Americans over 6 months of age, with 19 a target coverage level of 80% in non-institutionalized, non-elderly persons. 1 During recent 20 influenza seasons, all racial/ethnic groups experienced low vaccination levels that fell short of 21 this goal, with communities of color having the lowest coverage. While 49% of white adults 22 received an influenza vaccine during the 2018-19 season, only 39% of Black adults, 37% 23 Hispanic adults, and 44% of Asian adults were vaccinated. 2 24 Racial/ethnic disparities in vaccination coverage can be attributed to non-belief in the 25 utility of vaccinations, institutionalized racism, and distrust of medical institutions that contribute 26 to vaccine hesitancy among communities of color. 3-7 These inequities contribute to 27 disproportionately high influenza morbidity rates among disadvantaged racial/ethnic groups, 28 resulting in elevated rates of hospitalization and death. Prior research on the social determinants 29 of influenza hospitalization showed that Black/African Americans and Hispanics had higher 30 risks of hospitalization compared to Whites. 8-10 Differences in hospitalizations by race/ethnicity 31 are linked to other socioeconomic risk factors which disproportionately impact communities of 32 color, such as low household income or high residential density. 8-10 Increasing vaccination 33 coverage among marginalized groups may reduce race/ethnicity inequities in influenza morbidity 34 and mortality. 35 School-located influenza vaccination (SLIV) programs aim to increase vaccination 36 coverage levels among young children by providing free vaccination in schools. SLIV has the 37 potential to reduce barriers to vaccination that disproportionately impact communities of color. 38 Prior community-based interventions increased influenza vaccination coverage among 39 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint communities of color by reducing logistical barriers to vaccination through door-to-door and 40 street-based immunizations. 11 By increasing vaccination coverage, SLIV may contribute to herd 41 immunity and reduce influenza transmission community-wide, which may reduce racial/ethnic 42 disparities in influenza 12, 13 . Prior studies reported that SLIV programs were associated with 43 increased influenza vaccination coverage 14-20 and decreased school absences 14-17,20-22 and 44 student illness [14] [15] [16] [17] , but no studies have measured the differential impacts of large-scale SLIV 45 interventions by race/ethnicity. 46 We previously reported results from an evaluation of a city-wide SLIV program delivered 47 in elementary schools in Oakland, California from 2014-2018. 20 We found that the intervention 48 was associated with 7-11 percentage points higher vaccination coverage among school-aged 49 children and 17 to 37 lower incidence of influenza hospitalizations per 100,000 during influenza 50 seasons in which a moderately effective vaccine was being used. Here, we investigated whether 51 SLIV effectiveness varied by race/ethnicity in a pre-specified subgroup analysis. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 23, 2021. 70 We employed a matched cohort design to evaluate the effect of the intervention on 71 vaccination coverage and influenza hospitalization. We focused our study on public elementary 72 schools in Oakland Unified School District (OUSD, the intervention district). We excluded 73 private and non-district charter schools because pre-intervention data on school characteristics 74 was not available for them. 75 We selected West Contra Costa Unified School District (WCCUD) as the comparison 76 district, as it resulted in the closest school pair matches based on pre-intervention student 77 characteristics. Additional details on the matching procedure are described in Supplement 1. 78 Outcomes and Data Sources 79 We evaluated the association between SLIV and (1) influenza vaccination coverage 80 among school-aged children and (2) community-wide lab-confirmed influenza hospitalization. 81 Pre-Intervention District Comparison 82 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint To compare population characteristics between the comparison and intervention districts, 83 we obtained pre-intervention data on socioeconomic status, school enrollment, and race/ethnicity CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint Multiracial, and American Indian/Alaska Native populations. We calculated the cumulative 106 incidence of hospitalization each season using age-, race-, and ethnicity-specific population 107 estimates from the 2010 Census. We restricted analyses to influenza seasons using a pre-108 specified, data-driven definition based on local transmission patterns, as described in Supplement CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint We pre-specified subgroup analyses by race and ethnicity. Post-hoc we excluded Native 128 American students from the stratified analyses because the group was very small (N<20 per site 129 per year). We excluded survey responses that did not specify a student's race/ethnicity (2017: 130 Intervention N = 60 (2.67%), Comparison N = 90 (2.35%); 2018: Intervention N = 63 (2.60%), 131 Comparison N = 90 (2.20%)), 132 Laboratory-confirmed influenza hospitalization 133 To estimate incidence ratios, we fit log-linear modified Poisson models with an offset for 134 population size. 24 To account for pre-intervention differences in influenza hospitalization 135 incidence between districts, we estimated the difference-in-differences (DID) in cumulative We pre-specified stratification by age groups (non-elementary school aged individuals 140 (<= 4 years, >13 years) and older adults (>=65 years)). Because the number of elementary school 141 aged children who were hospitalized for influenza was small, there was inadequate statistical 142 power to estimate associations separately for this age group. 143 After examining the survey data, we combined White and Asian/Pacific Islander racial 144 categories due to low incidence among Asian/Pacific Islanders. We excluded Multiracial and 145 American Indian/Alaska Native groups from the analysis due to rare outcomes We excluded 146 hospitalization records that did not report race/ethnicity. 147 Prior to the intervention, the intervention and comparison districts had generally similar 148 demographic characteristics (Table 1) . However, relative to the comparison district, the 149 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 23, 2021. Intervention 37%) (Table S7) . 164 Vaccination coverage levels varied by race/ethnicity, when controlled for highest 165 caregiver education ( Figure S8 ). In all seasons, we observed lower vaccine coverage among 166 Black/African American and multiple race students across both districts. During the 2017-18 167 season, 40% (95% CI 24%, 59%) of Black/African American students were vaccinated, 168 compared to 65% (95% CI 57%, 72%) of Asian/Pacific Islander and 46% (95% CI 28%, 66%) of 169 White students in the comparison district. In the intervention district, 46% (95% CI 30%, 64%) 170 of Black/African American students were vaccinated, compared to 74% (95% CI 65%, 80%) of 171 Asian/Pacific Islander and 67% (95% CI 48%, 82%) of White students. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint Associations between SLIV and vaccine coverage varied between racial/ethnic groups 173 ( Figure 1 , Table S9 ). In the first two seasons of SLIV there were no differences in vaccine 174 coverage for any group, apart from Asian/Pacific Islanders. In the 2016-17 season, there were 175 increases in vaccine coverage in all groups other than multiple race students. We observed higher 176 coverage levels in the intervention district relative to the comparison district among 177 Black/African American (9% higher in intervention versus comparison; 95% CI 2%, 17%), 178 Latino students (11%; 95% CI 5%, 17%),White (7%; 95% CI -1%, 14%), and Asian/Pacific 179 Islander (5%; 95% CI 0%, 9%) students. 180 In the 2017-18 season, we observed significantly higher influenza vaccine coverage 181 levels among Asian/Pacific Islander (9%; 95% CI 1%, 16%), Latino (13%; 95% CI 9%, 18%), 182 and White (21%; 95% CI 10%, 32%) students in the intervention district relative to the 183 comparison district. Vaccine coverage was higher among Black/African American (6%; 95% CI 184 -2%, 14%) and multiple race (6%; 95% CI -2%, 14%) students, with more vaccinated students in 185 the intervention district. Figure S12 ). We observed fewer all-age influenza hospitalizations in the intervention 218 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 23, 2021. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint because the presence of SLIV could have improved caregivers' ability to correctly remember 285 their child's vaccination history. 47,48 Additionally, we saw higher levels of non-response among 286 Black/African American students. If those who did not respond to the survey were also less 287 likely to have been vaccinated, our results may have under-estimated the true disparities in 288 vaccination between Black/African American students and other racial/ethnic groups. However, 289 we previously standardized vaccine coverage models by school-district race distributions and did . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint Two or more races 6 (6, 7) 6 (5, 7) Hispanic or Latino ethnicity 26 (25, 27) 33 (32, 35) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint Emerging Infections Program. 470 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 23, 2021. ; https://doi.org/10.1101/2021.07.21.21260311 doi: medRxiv preprint Immunization and Infectious Diseases | Healthy People Flu Vaccination Coverage, United States Health disparities and 325 prevention: racial/ethnic barriers to flu vaccinations Racial disparities in the utilization of 327 preventive health services among older women with early stage endometrial cancer Commentary-Anatomy of Racial Disparity in Influenza Vaccination Segregation in Primary Care as Usual Source of Care: Findings from the Exploring Health Disparities 333 in Integrated Communities Study Racial and ethnic disparities in medical and dental health, access 335 to care, and use of services in US children Social determinants of influenza hospitalization in the United States Project VIVA: A Multilevel Community-Based Intervention to Increase Influenza 344 Vaccination Rates Among Hard-to-Reach Populations in New York City The Best Way to Fight Flu: Inoculate Children -Scientific American Community Studies for Vaccinating Schoolchildren Against 349 Influenza Elementary School-Based Influenza 351 Vaccination: Evaluating Impact on Respiratory Illness Absenteeism and Laboratory School-Located Influenza Vaccination Decreases Laboratory Influenza and Improves School Attendance Uptake and impact of vaccinating school age children against influenza during a 356 season with circulation of drifted influenza A and B strains Seasonal Influenza Vaccination at School: A Randomized Controlled Trial School-Located Influenza Vaccinations: A Randomized Trial Impact of elementary school-located influenza vaccinations: A stepped wedge 363 trial across a community Evaluation of a city-wide school-located influenza vaccination program in 365 The legacy of Tuskegee and trust in medical care Tuskegee responsible for race differences in mistrust of medical care? Neighbourhood influence on the fourth dose of diphtheria-tetanus-pertussis 398 vaccination Understanding Vaccine Refusal Increasing Inner-City Adult Influenza Vaccination Rates: A Randomized 402 Increasing Childhood Influenza Vaccination Predicting Human Papillomavirus Vaccine Uptake in Young Adult 406 Women: Comparing the Health Belief Model and Theory of Planned Behavior Social-Cognitive Factors and Intentions to Receive HPV Vaccination Among Young Women? Clin Increasing influenza 412 vaccine uptake in children: A randomised controlled trial Influenza Vaccination: The Persuasiveness of Messages Among 414 People Aged 65 Years and Older How Structural Racism Works -Racist Policies as a 419 Root Cause of U.S. Racial Health Inequities Intersection of Bias, Structural Racism, and Social Determinants With Health Care 421 Racial/Ethnic Minority Populations During an Influenza Pandemic Racial/ethnic differences in multimorbidity development and chronic disease 426 accumulation for middle-aged adults Full and Partial Flu Vaccination Coverage in Young Children, Six Immunization Information 428 | FluVaxView | Seasonal Influenza (Flu) | CDC Estimates of Flu Vaccination Coverage among Children -United States Challenges to estimating vaccine impact using hospitalization data & 2020. Racial Disparities in COVID-19: Key Findings from Available Racism, and Structural Injustice: Equitable Allocation and 439 Distribution of Vaccines for the COVID-19 COVID-19 and Health Equity-A New Kind of Genetic Matching for Estimating Causal Effects: A General 443 Multivariate Matching Method for Achieving Balance in Observational Studies