dcyphr | Evaluation of a city-wide school-located influenza vaccination program in Oakland, California, with respect to vaccination coverage, school absences, and laboratory-confirmed influenza: A matched cohort study


Vaccinating 50-70% of school-aged children may produce indirect protection of the general population. In this study, the researchers evaluated a city-wide school-located influenza (flu) vaccination intervention (SLIV). The campaign was implemented from 2014-2018 in ~95 preschools and elementary schools in northern Califronia. The researchers used a matched cohort design, matching the preschools with a pair that did not have the vaccine campaign. The researchers then estimated the impact of the campaign on three factors: student vaccination coverage, absences from school, and indirect effects on influenza hospitalizations in the general public. The researchers used a matching algorithm to match schools receiving the campaign and those that did not. They identified 22 school pairs and conducted surveys of students’ caregivers, estimated the incidence of influenza hospitalizations by school zip code, and analyzed absentee data from each school district. The researchers found that in districts that received the vaccine, student vaccine coverage and indirect protective effects on the general population were higher, while influenza hospitalizations were lower. In conclusion, the researchers found that the campaign was effective in increasing student vaccine coverage and also produced indirect protective effects in the general population.


The researchers wanted to evaluate a school-located influenza vaccine intervention (SLIV) based on the following criteria: student vaccination coverage, absences from school, and indirect protective effects on the general population.


Seasonal influenza significantly contributes to hospitalization and mortality, especially in infants and the elderly. Vaccination for all over 6 months of age has been recommended in the US since 2010. Vaccine efficacy varies based on seasonal differences in quality of the virus used in the vaccine, mutations in the virus, etc.


School-aged children make up the majority of influenza transmission, so they have been historically targeted in vaccination campaigns. It is estimated that by vaccinating 50-70% of school-age children, herd immunity/indirect protective effects can be produced in a population and effectively reduce influenza transmission.


School-located influenza vaccination (SLIV) programs have been proposed as a strategy to increase vaccine coverage in children and induce these protective effects. Previous studies have suggested that SLIVs can increase vaccination coverage, decrease school absences and student illnesses, and are a cost-effective strategy to do so. However, other studies have produced conflicting results, and there have been no prior studies rigorously evaluating the effectiveness of large-scale SLIVs, especially on community health outcomes.

In this study, the researchers report 4 years’ worth analysis of a SLIV program in ~95 schools in Oakland, California. The SLIV tried to increase student vaccine coverage and produce herd immunity effects in the general public. The researchers measured whether the intervention was associated with increased vaccine coverage and decreased influenza hospitalizations and school absences using a matched-cohort design and 3 sources of data.


Influenza vaccine coverage

Most pre-intervention demographics in the intervention and non-intervention school districts were similar. However, in the intervention site, median household income was lower, the proportion of black residents was higher, the percentage of Hispanic/Latino residents was lower, and the percentage of those with a bachelor degree was higher. Table 1 shows the demographic characteristics of the intervention and non-intervention school districts before the SLIV.


8,121 surveys in 22 schools were collected. Response rates were similar in matched pairs. In the first 2 years of the SLIV, influenza vaccine coverage did not differ significantly between intervention and non-intervention districts. However, after 2 years, coverage was higher in the intervention district, at 7% and 11% higher than the non-intervention districts in 2016-2017 and 2017-2018 respectively. Table 2 reports coverage data. Differences were statistically significant, even after adjusting for student race and parent education.


Influenza hospitalization

The incidence of influenza hospitalization was similar between intervention and non-intervention school districts during the 3 influenza seasons before the SLIV and during the first two years of the intervention. In the two latter years of the intervention, influenza hospitalization rates were lower in the intervention districts than in the non-intervention districts, especially in vulnerable populations (elementary school-age and individuals 65 years old and above). These results are shown in Figure 4.


In addition, the length of influenza hospitalization was 1-2.5 days lower in the intervention school districts for all ages. Influenza-related intensive care unit admissions and mortality were also lower in the intervention districts.


School absenteeism

Before the intervention, absences in the intervention vs. non-intervention districts per 100 days were 4.85 vs. 5.84 for all absences and 2.84 vs. 2.81 for influenza-related absences. Overall, illness-specific absence rates decreased in the intervention school districts for the first 2 years, and decreased to lower than that of the non-intervention school districts in the latter 2 years of the intervention. Intervention effects on school absence rate per 100 students can be observed in Figure 5.


In the first two years of the SLIV, the intervention offered the LAIV vaccine, which had low efficacy and did not impact any of the three factors analyzed by the researchers. However, in the latter 2 years when the intervention offered the IIV vaccine, it was moderately effective and reduced school absences and increased indirect protective effects in age groups not targeted by the intervention. In addition, the SLIV increased influenza vaccination coverage by 11% in students attending schools in the intervention school districts. Although this is lower than previous studies of SLIV, it may have been because pre-intervention coverage was higher than in other studies.


An important question is if SLIV interventions increase vaccination coverage in students who otherwise would not have been vaccinated. The findings of this study suggest that because coverage increased during the intervention, children who otherwise would not have been vaccinated were covered by the intervention. 

The increase of coverage in SLIV districts up to 64%, which is within the 50-70% range necessary for herd immunity, may have resulted in observed reductions in influenza hospitalization in populations that weren’t targeted by the SLIV. This is consistent with mathematical models, and suggests that even minimal increases in coverage by SLIV can produce protective effects in the general population.


SLIV intervention

The SLIV evaluated was the Shoo the Flu intervention conducted in public and charter schools in the Oakland Unified School District in Oakland, California. Information about the SLIV can be found here: http://shootheflu.org/.


Influenza vaccine effectiveness during the intervention

The vaccines used in the SLIVs were less effective in 2014-2015 and 2015-2016 and moderately effective in 2016-2017 and 2017-2018. Figure 1 shows vaccine effectiveness of the 2 types of vaccines used, IIV and LAIV.


Study design

The researchers used a matched cohort design to evaluate the SLIV intervention, analyzing data of vaccine coverage, influenza hospitalization, and school absence rates collected from 3 data sources. A matching algorithm was used to match schools in districts receiving the intervention and districts not receiving the intervention. The algorithm used mean enrollment, class size, parental education, academic performance, percentage of English language learners, and percentage of students receiving free lunches to pair schools. The algorithm was evaluated, and it was found that it produced good quality school matches.


Outcomes and data sources

2 surveys of student caregivers (parents, foster parents, guardians, etc.) were conducted to measure student influenza vaccination, and its type and provider. Influenza hospitalization data was collected within school districts receiving and not receiving the SLIV. School absence records were collected from each school.


Statistical analysis

All analyses were conducted in R.


Definition of influenza seasons

The researchers defined influenza season as beginning at 2 consecutive weeks in which the percentage hospital visits for influenza was above a certain cutoff, and ending at 2 consecutive weeks below the cutoff. The researchers used a 2.5% cutoff in this study.

The researchers estimated influenza vaccination coverage, influenza vaccine coverage, and laboratory-confirmed influenza hospitalization informed by various data from a variety of databases.


Offering SLIV to elementary schools in a large, urban school district was associated with a 7-11% increase in vaccination coverage, which resulted in reductions of illness-related absences and population-wide influenza-related hospitalizations. Even modest increases in coverage by SLIV may have protective effects on the entire community.