Seasonal Influenza Vaccination at School A Randomized Controlled Trial Sharon G. Humiston, MD, MPH, Stanley J. Schaffer, MD, MS, Peter G. Szilagyi, MD, MPH, Christine E. Long, MPH, Tahleah R. Chappel, MS, Aaron K. Blumkin, MS, Jill Szydlowski, BA, Maureen S. Kolasa, RN, MPH This activity is available for CME credit. See page A3 for information.

Background: Influenza vaccination coverage for U.S. school-aged children is below the 80% national goal. Primary care practices may not have the capacity to vaccinate all children during influenza vaccination season. No real-world models of school-located seasonal influenza (SLV-I) programs have been tested.

Purpose: Determine the feasibility, sustainability, and impact of an SLV-I program providing influenza vaccination to elementary school children during the school day.

Design: In this pragmatic randomized controlled trial of SLV-I during two vaccination seasons, schools were randomly assigned to SLV-I versus standard of care. Seasonal influenza vaccine receipt, as recorded in the state immunization information system (IIS), was measured. Setting/participants: Intervention and control schools were located in a single western New York county. Participation (intervention or control) included the sole urban school district and suburban districts (five in Year 1, four in Year 2). Intervention: After gathering parental consent and insurance information, live attenuated and inactivated seasonal influenza vaccines were offered in elementary schools during the school day. Main outcome measures: Data on receipt of Z1 seasonal influenza vaccination in Year 1 (2009– 2010) and Year 2 (2010–2011) were collected on all student grades K through 5 at intervention and control schools from the IIS in the Spring of 2010 and 2011, respectively. Additionally, coverage achieved through SLV-I was compared to coverage of children vaccinated elsewhere. Preliminary data analysis for Year 1 occurred in Spring 2010; final quantitative analysis for both years was completed in late Fall 2012.

Results: Results are shown for 2009–2010 and 2010–2011, respectively: Children enrolled in suburban SLV-I versus control schools had vaccination coverage of 47% vs 36%, and 52% vs 36% (po0.0001 both years). In urban areas, coverage was 36% vs 26%, and 31% vs 25% (po0.001 both years). On multilevel logistic analysis with three nested levels (student, school, school district) during both vaccination seasons, children were more likely to be vaccinated in SLV-I versus control schools; ORs were 1.6 (95% CI¼1.4, 1.9; po0.001) and 1.5 (95% CI¼1.3, 1.8; po0.001). Conclusions: Delivering influenza vaccine during school is a promising approach to improving pediatric influenza vaccination coverage.

Trial Registry: ClinicalTrials.govNCT01224301 (Am J Prev Med 2014;46(1):1–9) & 2014 Published by Elsevier Inc. on behalf of American Journal of Preventive Medicine

From the Department of Pediatrics (Humiston), Children’s Mercy Hospitals and Clinics, Kansas City, Missouri; the Department of Pediatrics (Schaffer, Szilagyi, Blumkin), the Center for Community Health (Long, Chappel), the Division of Health Policy and Outcomes Research (Szydlowski), University of Rochester Medical Center, Rochester, New York; and the Health Services Research and Evaluation Branch (Kolasa), National Center for Immunization

and Respiratory Disease, Centers for Disease Control and Prevention, Atlanta, Georgia Address correspondence to: Sharon G. Humiston, MD, MPH, Children’s Mercy Hospital and Clinics, Division of Emergency and Urgent Care, 2401 Gillham Road, Kansas City MO 64108. E-mail:[email protected]. 0749-3797/$36.00 http://dx.doi.org/10.1016/j.amepre.2013.08.021

& 2014 Published by Elsevier Inc. on behalf of American Journal of Preventive Medicine

Am J Prev Med 2014;46(1):1–9

1

2

Humiston et al / Am J Prev Med 2014;46(1):1–9

Introduction

Objective and Hypothesis

I

During the 2009–2010 and 2010–2011 influenza vaccination seasons, a pragmatic, community-based RCT was conducted in Monroe County NY to determine the feasibility and impact of a sustainable SLV-I program to provide vaccination to elementary school children during the school day. Key features include utilization of a private mass vaccinator and third-party billing to simulate a real-world SLV-I. The study hypothesis was that seasonal influenza vaccination coverage would be higher among children enrolled in schools that offered SLV-I than in control schools without SLV-I.

nfluenza disease among children leads to a substantial number of emergency department and primary care visits, as well as hospitalizations.1–4 Because of the burden of influenza disease among children5–7 and its spread from children to adults,8–10 the Advisory Committee on Immunization Practices (ACIP) gradually expanded its routine recommendation for pediatric influenza vaccination to cover children aged o6 months to 5 years.11 In 2008, the ACIP recommendation for seasonal influenza vaccination included all children aged 5 to 18 years, beginning with the 2008–2009 season, if feasible, but no later than the 2009–2010 season.12 This expanded recommendation resulted in the need for annual influenza vaccination of an additional 26 million children within a relatively short time frame (i.e., fall and winter) each year. Studies have documented important barriers to influenza vaccination in primary care.13–16 Even though childhood influenza vaccination coverage has increased since publication of the universal recommendation, coverage remains low relative to the Healthy People 2020 goal level of 80%.17 Among children aged 6 months to 17 years, seasonal influenza vaccination coverage was 44% in the 2009–2010 season and 51% in the 2010–2011 season.18 Surveys performed subsequent to the recommendation for universal influenza vaccination have shown that most pediatricians support school-located vaccination for influenza19 and believe the availability of alternative vaccination delivery settings can increase vaccination coverage.20 Schools are a key alternative delivery setting. School-located vaccination is defined as any immunization program that takes place on school grounds before, during, or after school hours.21 School-located vaccination for influenza (SLV-I) provides an alternative to vaccination in primary care offices that obviates the need for both children and parents to travel to the office for a vaccination-only visit. Because schools have up-to-date parent contact information, they can notify parents directly about the need and opportunity for vaccination. Schools have a potential stake in influenza vaccination because immunization may decrease influenza-associated illness and absenteeism.22–25 Despite these positive aspects, SLV-I poses logistical challenges, including vaccine planning and ordering, storage and handling, communication, obtaining informed consent, and establishing ways to cover costs for personnel and vaccines.26–32 Much of the U.S. medical literature on SLV-I derives from manufacturer-supported trials33–35 or the pandemic influenza vaccination campaign21 in which the federal government covered the cost of the pandemic vaccine.

Methods The Research Subjects Review Boards of the University of Rochester and the Monroe County Department of Public Health approved this study in 2009.

Study Design and Setting A randomized, controlled cluster design was utilized, with allocation of schools within a school district to either SLV-I or control. The schools were located in Monroe County NY (2010 population 744,34436) comprising Rochester—the third most populous city in New York State—and surrounding suburbs. The schools were stratified by location (suburban, urban) before randomization because previous studies showed suburban children to have higher influenza immunization coverage than urban children.37

Recruitment and Participants Monroe County NY school district superintendents were asked to have their districts participate in the study (18 suburban, 1 urban). Among suburban school districts, five participated in 2009 and four participated in 2010. The Rochester City School district, the only urban district in Monroe County, participated both years. Participating schools (stratified by location, i.e., suburban, urban) had similar race/ethnicity and similar proportions of students eligible for free meals as did nonparticipating schools (data not shown). During each of the 2 years, after assignment to the intervention or control group, each school participated to the end of that year and contributed data in accordance with their randomization (i.e., intention-to-treat) (Figure 1). Data were analyzed for students in kindergarten through fifth grade from all participating schools.

Randomization and Blinding SLV-I versus control randomization. Twice as many schools were assigned to SLV-I than to standard of care (control, i.e., no SLV-I). In Year 1, among suburban districts, nine schools were randomly allocated to SLV-I and four to no-SLV-I; suburban schools were not stratified by student enrollment size. The single urban district agreed to have 18 of its 40 elementary schools participate. The participating urban schools were stratified into two groups by student enrollment size because of the greater logistic burden associated with conducting vaccine clinics in larger www.ajpmonline.org

3

Humiston et al / Am J Prev Med 2014;46(1):1–9

10 schools in 4 districts allocated to control (4129 children) 4129 children did not receive intervention

0 lost to follow-up 0 discontinued intervention

0 lost to follow-up 0 discontinued intervention

Analyzed data from 21 schools in 5 districts (8361 children)

Analyzed data from 10 schools in 4 districts (4129 children)

Figure 1. CONSORT flow diagram: Year 1 schools; larger was defined as having 4466 students (the median number of students in participating schools in Year 1). The project biostatistician then randomized schools to SLV-I (12 schools) or control (6 schools) within each size strata. Randomization assignments from Year 1 were kept constant in Year 2 (Figure 2). A single suburban district dropped out after Year 1. To replace the three intervention schools from that district in Year 2, additional schools that had not participated in Year 1 were recruited from a district that had participated in Year 1. All assigned schools contributed to the analysis. High- versus low-intensity parent notification randomization. Within SLV-I schools, several modalities were used to notify parents of the need for influenza prevention and the SLV-I project (e.g., website, automated telephone messaging, written materials). This study included a nested controlled trial of high-intensity versus lowintensity parent notification, with schools randomized to each level of intervention. High-intensity notification included (1) more methods of communication (e.g., SLV-I staff explained the program to parents at school visitation nights in Year 1) and (2) more frequent communication (e.g., having consent forms go home in students’ backpacks twice rather than once). Because the primary outcome— influenza immunization coverage—did not show significant differences based on this randomly assigned variable, in this paper the high- and low-intensity notification groups were combined. Blinding. Although it was not possible to blind researchers or participant schools to assignment, the study team did not communicate with control schools during the intervention period.

Intervention Vaccines. Both live attenuated (LAIV) and inactivated (TIV) seasonal influenza vaccines were offered each year. In 2009, a separate H1N1 (pandemic flu) vaccine was also recommended,38 but not offered through this project. SLV-I Personnel. Three groups collaborated on this project. (A) A mass vaccinator, experienced in vaccinating in nonschool settings, administered vaccines in schools. The mass vaccinator: 1. managed influenza vaccine ordering, storage, vaccine delivery

per routine procedures; 2. contacted each intervention school to schedule and arrange space requirements for the Vaccine Day clinic; January 2014

Nursing staff working for the mass vaccinator performed in-school vaccinations. (B) A logistical coordination team, funded by a grant from the CDC (part-time clerical and project coordination staff and a medical director totaling approximately 1.5 FTEs), developed and distributed notification materials for parents, managed all aspects of the consent process, and responded to questions about vaccines and Vaccine Day Clinics from parents and school staff. (C) A variety of school personnel worked with the mass vaccinator to schedule and arrange for space for Vaccine Days, and arranged and coordinated the movement of children from classroom to vaccination area. While all participating schools had a registered school nurse on staff, because of the already high demands on school nurses this project was designed to function without a specified role for school nurses. Each school and each school nurse could determine the role autonomously. Communication to parents. Parents were notified about the SLV-I project through (1) pre-recorded telephone messages, (2) a project website, and (3) print materials sent home with students or mailed to students’ homes either with report cards or separately. Materials included an introductory letter, LAIV and TIV Vaccine Information Statements, and a blank consent form. Schools provided their most up-to-date directory information (name, date of birth, address, telephone number, and name of guardian) for all enrolled students in order to organize mailings. Consents. In Year 1, consent forms were available only in English. In Year 2, they were simplified and were also available in Spanish. Consent forms were sent to parents in student backpacks 6 to 8 weeks before the school’s first scheduled Vaccine Day. Parents were asked to return signed consents to their children’s teachers 43 weeks before Vaccine Day. The logistical coordination team collected consent forms, reviewed each for completeness, and contacted parents for missing information (e.g., insurance information, medical history) or—when unable to reach parents— returned consents to schools. Teachers sent these incomplete consents home with the appropriate children. Before Vaccine

Enrollment

21 schools in 5 districts allocated to intervention (8361 children) 8361 children received intervention

ment to administer Vaccine for Children (VFC) vaccine; and 4. arranged for payment from third-party payers.

Allocation

31 schools randomized (18 urban, 13 suburban; 12,490 children)

3. obtained permission from the New York State Health Depart-

Analysis Follow-up

Analysis Follow-up

Allocation

Enrollment

6 school districts agreed to participate (1 urban, 5 suburban)

5 school districts agreed to participate (1 urban, 4 suburban) Randomization from Year 1 maintained.* 32 schools participated (18 urban, 14 suburban; 12,876 children)

21 schools in 4 districts allocated to intervention (8481 children) 8481 children received intervention

11 schools in 5 districts allocated to control (4395 children) 4395 children did not receive intervention

0 lost to follow-up 0 discontinued intervention

0 lost to follow-up 0 discontinued intervention

Analyzed data from 21 schools in 4 districts (8481 children)

Analyzed data from 11 schools in 5 districts (4395 children)

Figure 2. CONSORT flow diagram: Year 2 *A single suburban district dropped out after Year 1. To replace the three intervention schools from that district in Year 2, additional schools that had not participated in Year 1 were recruited from a district that had participated in Year 1.

4

Humiston et al / Am J Prev Med 2014;46(1):1–9

Day at each school, the logistical coordination team members provided the school with a list of students with valid consent forms who were eligible for vaccination and sent the mass vaccinator a count of eligible children by school. Vaccine Days (or clinics) were conducted from November 3, 2009, to December 18, 2009, and from November 2, 2010, to November 18, 2010. These dates late in the influenza vaccination season were chosen so students would have ample time to have been vaccinated in primary care. On each school’s Vaccine Days, the mass vaccinator’s team included at least one registered nurse and a clerical staff. When children arrived in the clinical area, the team checked the consent forms in collaboration with school staff, confirmed each child’s identity, and administered LAIV or TIV based on parental request as indicated on the consent form (provided that the medical history did not contraindicate use of LAIV). For the 2009–2010 season, two Vaccine Days per school were held during the fall of 2009. Because of low attendance at the second Vaccine Day, in 2010 only one Vaccine Day was offered per school. Recording vaccinations. Vaccinations given in school were recorded on a three-ply form. The mass vaccinator retained the original, transcribed it into their database, and uploaded the data into the New York State Immunization Information System (NYSIIS). NYSIIS is accessible to all primary care providers. Two copies of the vaccination record were given to the vaccinated child to take home—one for parents’ records and the other for the child’s primary care provider.

Sample Size An a priori sample size calculation required Z288 students per stratum (study/control and suburban/urban), to detect a difference of 15% in the primary outcome measure (Z1 influenza vaccination in the relevant season) at a two-sided alpha of 0.05 and statistical power of 90%, with an intra-class correlation of 0.02 among students.

Data Collection and Analysis Independent variables included study group (SLV-I, control) as well as grade (kindergarten [K] through Grade 5); location (urban, suburban); and school percent eligibility for free lunch. Previously, socio-demographic characteristics have been found to be associated with influenza vaccination coverage.18,39,40 School and school district also were included as independent variables because of potential confounding by leadership or personnel. Additionally, pre-intervention comparability of the SLV-I and control schools were assessed by examining NYSIIS data on influenza vaccination coverage in 2008 (prior to the SLV-I program) for students in the future SLV-I versus control schools. The primary outcome measure was receipt of Z1 seasonal influenza vaccination in Year 1 (15 August 2009 to 15 January 2010) and, separately, in Year 2 (15 August 2010 to 15 January 2011). Each child in the school directory files at the beginning of Year 1 and Year 2, separately, was counted as a potential vaccinee, and for neither intervention nor control schools was there additional ascertainment of school census later in the school year. In Spring 2010 (for Year 1) and 2011 (for Year 2) NYSIIS was reviewed to assess influenza vaccination dates for each child in the school directory files.

Analysis Randomization was at the school level, but the unit of analysis was at the student level. The analysis accounted for the lack of independence between students within the same schools and schools within the same district by using multi-level logistic regression models. Each model has a level for the student, school, and district. This type of clustered RCT is now used widely. Models predicting receipt of Z1 influenza vaccination included the following covariates: study group (SLV-I or not); location (urban, suburban); school district, school, and child’s grade. Preliminary data analysis for Year 1 occurred in Spring 2010; final quantitative analysis for both years was completed in late Fall 2012.

Results Participating School Populations Table 1 summarizes the school populations by study group (SLV-I or control); location (urban, suburban); and project year (2009–2010 and 2010–2011 vaccination seasons). When comparing Year 1 SLV-I and control schools, suburban SLV-I schools had a lower proportion of students eligible for free lunch and a higher proportion who were white; urban SLV-I schools had a higher proportion of students eligible for free lunch and a lower proportion that were white. Influenza vaccine coverage in 2008 (prior to the SLV-I program) was similar for students in the future SLV-I versus control schools in both suburban (29% vs 30%, p¼0.5), and urban schools (22% vs 24%, p¼0.2). Pre-intervention immunization coverage was higher among suburban than urban schools. The trial concluded according to the study plan and all analyses were by original assigned groups.

Influenza Immunization Coverage in SLV-I Versus Control Schools SLV-I versus control schools. Table 2 shows the proportion of students in grades K–5 receiving influenza vaccine by study group, location, project year, and receipt of vaccination at school or elsewhere (primarily physician offices). Children attending SLV-I schools had higher vaccination coverage (by 6%–16%) than children attending control schools, depending on year and suburban/urban location. Vaccination coverage was higher (po0.001) among SLV-I schools for both suburban (47% vs 36% in Year 1, 52% vs 36% in Year 2) and urban (36% vs 26% in Year 1, 31% vs 25% in Year 2) populations. Sites for influenza immunizations. In suburban schools, in Year 1 fewer children from SLV-I schools than control schools were vaccinated at sites outside the SLV-I program; however, this was not the case in Year 2. www.ajpmonline.org

5

Humiston et al / Am J Prev Med 2014;46(1):1–9

Table 1. Descriptions of school populations in Grades K–5 by randomization, location, and year, % unless otherwise noted 2009–2010 vaccination season: Year 1 SLV-I TOTAL

Control

Subtotal

2010–2011 vaccination season: Year 2 SLV-I

Control

Subtotal

21

10

31

21

11

32

9

4

13

9

5

14

4005

1476

5481

4208

1760

5968

21

25

—

19

28

—

White

86

73

—

81

74

—

Black

6

15

—

5

11

—

Hispanic

4

8

5

9

SUBURBAN No. of schools No. of students Eligible for free lunch Race/ethnicity

CI¼1.4, 1.9, po0.001) and 1.5 (95% CI¼1.3, 1.8, po0.001) (Table 3). That is, after adjustment, children in SLV-I schools had 50%–60% higher odds of influenza immunization than in non–SLV-I schools each year. Of interest, suburban children and younger children remained more likely to be vaccinated than their counterparts, irrespective of SLV-I.

Discussion Study Summary

During both intervention years, influenza vaccination URBAN coverage was significantly No. of schools 12 6 18 12 6 18 higher in SLV-I than control schools based on analNo. of students 4356 2653 7009 4273 2635 6908 ysis of data recorded in Eligible for free lunch 87 78 — 94 88 — the statewide immunizaRace/ethnicity tion registry. This effect occurred in both suburban White 11 16 — 10 12 — and urban school districts. Black 65 58 — 65 60 — To our knowledge, this is Hispanic 22 23 21 21 the first reported randomized controlled trial of Other race/ethnicity 2 3 4 7 SLV-I in the U.S. in which SLV-I, school-located vaccination for seasonal influenza a mass vaccinator administered vaccinations during In urban schools, similar proportions were vaccinated in-school vaccine clinics and billed for services (billing elsewhere in both years. Among control schools, vacciis detailed in a separate publication).41 Despite nation coverage was similar in Year 1 to Year 2 obstacles presented by needing consent and insurance (suburban: 36% both years, urban: 26% and 25%, information prior to vaccination, SLV-I schools had 6% respectively), suggesting little if any secular trend in to 16% higher vaccination coverage than control seasonal influenza vaccination coverage over this short schools, and the odds of vaccination were 50%–60% time period. higher for students in SLV-I than control schools. Disparities between suburban and urban schools. Although this intervention did not lead to achievement The SLV-I program appeared to have a similar level of of goal levels of coverage or an increase in school age impact in both suburban and urban schools, so disparvaccination sufficient to have, by itself, a major impact ities in immunization coverage between suburban on the overall community level of influenza disease, and urban school persisted despite SLV-I implethis trial did produce an impact that exceeded most mentation. interventions recommended to increase childhood vaccination coverage, for example, sending immunization Multivariable Analysis reminder messages or decreasing missed opportunities to vaccinate during visits.42–46 Because of the imporOn multilevel logistic analysis with three nested levels (student, school, school district) during the two vaccitance of influenza vaccination to the health of the nation seasons, the ORs for receipt of vaccination for individual and the community, we believe the improvechildren in SLV-I versus control schools were 1.6 (95% ment demonstrated in this trial is considerable. The Other race/ethnicity

January 2014

4

4

9

6

6

Humiston et al / Am J Prev Med 2014;46(1):1–9

hepatitis B and varicella.47,48 More recently, SLV-I has been studied. In 2006, King 2009–2010 2010–2011 et al.22 reported a controlled vaccination season: Year 1 vaccination season: Year 2 trial of SLV-I with LAIV in elementary schools in four SLV-I Control p* SLV-I Control p* geographically diverse states. Suburban Unlike the model presented Vaccinated anywhere 47 36 o0.0001 52 36 o0.0001 here, the vaccine was provided repeatedly (not just at Vaccinated at school 18 0 15 0 1–2 Vaccine Days) and at no Vaccinated elsewhere 29 36 37 36 charge, obviating the need to Urban collect or use insurance information. Almost half of Vaccinated anywhere 36 26 o0.0001 31 25 o0.001 students in these SLV-I Vaccinated at school 10 0 9 0 schools were vaccinated. Vaccinated elsewhere 26 26 22 25 Carpenter and colleagues28 * p-values from a multilevel logistic model with three nested levels: student, school, and school district described an SLV-I program SLV-I, school-located vaccination for seasonal influenza in which influenza vaccination was offered at no charge to all students in the Knox County TN public school primary outcome was receipt of Z1 dose of seasonal system in autumn 2005. The manufacturer donated influenza vaccine by January 15 of the vaccination LAIV and the county health department bore costs season. It is not known for how many of the vaccinated associated with the purchase of TIV, parent education, children less than age 9 this was the only dose received. and vaccine administration. The program vaccinated Thus, for both intervention and control group children, 56% of all elementary school students. In a 2006 SLV-I the level of protection may have been less than the trial without controls, three Minnesota counties offered proportion vaccinated. In future studies, this and the LAIV at no charge to all healthy school children. Health effect of school-located influenza vaccination on comdepartment staff conducted two vaccination clinics per munity levels of influenza disease burden should be elementary school at specified times during school hours. evaluated. The vaccine was either donated by the manufacturer or paid for by the federally funded Vaccines for Children Previous Studies (VFC) program. Almost half (47%) of these elementary School-located vaccination has been used in the last students were vaccinated.29 Hawaii conducted a large 20 years to increase vaccination coverage against uncontrolled SLV-I trial during 2007–2008, offering Table 3. Influenza vaccination odds from multilevel logistic regression model with three LAIV and TIV at no charge nested levels: student, school, and district for public and private school students in Grades K 2009–2010 2010–2011 through 8. Most (90%) vaccination season vaccination season schools in the state particiCategory OR (95% CI) p OR (95% CI) p pated and 60,760 students (46%) were vaccinated. As Intervention vs control (ref) 1.6 (1.4, 1.9) o0.001 1.5 (1.3, 1.8) o0.001 with this trial, younger Suburban vs urban (ref) 1.5 (1.3, 1.8) o0.001 2.0 (1.0, 3.8) 0.05 students were more likely Grades vs kindergarten (ref) to be vaccinated than older cohorts.49 With reference Grade 1 1.0 (0.9, 1.1) 0.92 1.0 (0.9, 1.2) 0.81 only to the 2009 pandemic Grade 2 0.9 (0.8, 1.0) 0.22 1.0 (0.9, 1.1) 0.64 influenza vaccination proGrade 3 0.9 (0.8, 1.0) 0.05 0.8 (0.7, 1.0) 0.02 gram, the DHHS Office of the Inspector General Grade 4 0.8 (0.7, 0.9) 0.001 0.9 (0.7, 1.0) 0.02 reported that “selected SLV Grade 5 0.8 (0.7, 0.9) o0.001 0.8 (0.7, 0.9) o0.001 sites vaccinated an average Table 2. Students with Z1 influenza vaccine dose in state registry by year, location, and randomization, % unless otherwise noted

www.ajpmonline.org

Humiston et al / Am J Prev Med 2014;46(1):1–9

of 28% of enrolled students in 1 day, ranging from 14 percent to 45 percent.”21 Viewed as a whole, these studies demonstrate that SLVI has tremendous potential to increase vaccination coverage of school-aged children. However, SLV-I has usually been performed without billing because most programs were manufacturer-supported or, in the case of the H1N1 influenza pandemic of 2009–2010, school-located vaccination campaigns provided federally purchased H1N1 vaccine free of charge.50 The RCT described here provides important new information on SLV-I using an established business model of vaccine delivery and demonstrates that SLV-I can achieve substantially higher vaccination coverage than standard of care. It should be noted that about twice as many students in intervention schools were vaccinated outside the school, presumably in primary care (New York pharmacists do not vaccinate young children). Parents’ attachment to primary care, where they have a long-term relationship with the provider, remained intact despite the convenient on-site school option. Concerns that school-located vaccination would merely shift vaccination from primary care to school were not substantiated in this trial.

Ascertainment This study’s primary outcome measure, influenza vaccination, was assessed exclusively by using information from the NY State Immunization Information System (NYSIIS). Children’s names on the schools’ rosters were used for the NYSIIS search. In some cases, the mass vaccinator reported delivering in-school influenza vaccine to people whose records could not be found in NYSIIS, probably because of small variations in identifying data (e.g., spelling of the name). Thus, this study may underestimate immunization coverage by, at most, 2 percentage points in Year 1 and 1 percentage point in Year 2 (a bias to the null). Since January 2008, the NYS Immunization Registry Law required healthcare providers to use NYSIIS to report all immunizations administered to children/youth aged o19 years, along with immunization histories.51 While all providers may not have complied with this law, it is unlikely there would be a bias in accuracy of NYSIIS data in SLV-I versus control schools for vaccinations received in primary care. This is further supported by the 2008 baseline equivalence in influenza vaccination coverage of schools that went on to be randomized to SLV-I or control.

Other Limitations First, the denominators of children in SLV-I and control schools were based on school censuses; these may have January 2014

7

been incomplete. However, schools were assigned to SLV-I or control status randomly and there is no apparent reason why there would be more errors in school records of either SLV-I or control schools. Second, the project cannot be classified as a pure effectiveness study because the logistical coordination team personnel were hired to fulfill SLV-I programmatic needs. Third, the generalizability of this study is limited in that all participating schools were located in one New York county; also, differences in SLV-I versus control school vaccination coverage may not be generalizable to settings with very high baseline influenza vaccination coverage. And finally, it is not clear if the increase in vaccination coverage in intervention schools was due to on-site immunizations, increased notification of parents, or both. However, in other studies, parent education alone has had a weak influence and in this study higher levels of notification did not have a dose effect.

Implications There is a growing body of evidence that SLV-I could help to improve influenza vaccination coverage among schoolaged children, and this study demonstrates that during two influenza vaccination seasons, vaccination coverage was significantly higher in SLV-I than control schools, in both suburban and urban settings. Nevertheless, despite SLV-I, influenza vaccination coverage did not reach levels close to the Healthy People 2020 goal of 80%. To reach the 2020 goal of influenza vaccination coverage among school-aged children, public health officials will need further refinements of both primary care–based and school-located vaccination against influenza, particularly in inner-city settings. Needed refinements in SLV-I include better mechanisms for gathering consent and insurance information and improved mechanisms for educating parents about the importance of influenza vaccination for school-aged children. The CDC, the National Association of County and City Health Officials, the National Association of School Nurses, and the American School Health Association each offer strategies for groups planning to start SLV-I programs (www. immunize.org/school-vaccination/). Strategies to implement SLV-I need further field testing and evaluation.

Conclusion In this RCT of SLV-I, substantially higher influenza vaccination coverage was found among elementary school students in schools that offered in-school seasonal influenza vaccination than among students in standardof-care control schools. SLV-I is a promising approach to address the problem of inadequate pediatric influenza vaccination coverage.

8

Humiston et al / Am J Prev Med 2014;46(1):1–9

This study was supported by the CDC National Center for Immunization and Respiratory Diseases (CDC NCIRD) Grant 055215-002, “School-Based Influenza Immunization Program.” This support did not include the purchase of any vaccine. The CDC NCIRD Project Officer, Mrs. Kolasa, met with the team regularly and contributed to the study design, implementation, analysis, and manuscript preparation. The authors wish to thank Dr. Andrew Doniger, William Russell, Kim DiMattia, and the many school personnel who helped implement the school located vaccination program. The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the CDC, U.S.DHHS. SGH is a consultant to the Immunization Action Coalition. No other financial disclosures were reported by the authors of this paper.

References 1. O’Brien MA, Ukeyi TM, Shay DK, et al. Incidence of outpatient visits and hospitalizations related to influenza in infants and young children. Pediatrics 2004;113(3 Pt 1):585–93. 2. Miller EK, Griffin MR, Edwards KM, et al. New Vaccine Surveillance Network. Influenza burden for children with asthma. Pediatrics 2008;121(1):1–8. 3. Neuzil KM, Mellen BG, Wright PF, Mitchel EF Jr, Griffin MR. The effect of influenza on hospitalizations, outpatient visits, and courses of antibiotics in children. N Engl J Med 2000;342(4):225–31. 4. Bourgeois FT, Valim C, Wei JC, McAdam AJ, Mandl KD. Influenza and other respiratory virus-related emergency department visits among young children. Pediatrics 2006;118(1):e1–e8. 5. Molinari NA, Ortega-Sanchez IR, Messonnier ML, et al. The annual impact of seasonal influenza in the U.S.: measuring disease burden and costs. Vaccine 2007;25(27):5086–96. 6. Poehling KA, Edwards KM, Weinberg GA, et al. New Vaccine Surveillance Network. The underrecognized burden of influenza in young children. N Engl J Med 2006;355(1):31–40. 7. Neuzil KM, Zhu Y, Griffin MR, et al. Burden of interpandemic influenza in children younger than 5 years: a 25-year prospective study. J Infect Dis 2002;185(2):147–52. 8. Poehling KA, Szilagyi PG. Not just for kids: new paradigms for vaccine delivery in pediatrics. Acad Pediatr 2009;5:293–4. 9. Fiore AE, Epperson S, Perrotta D, Bernstein H, Neuzil K. Expanding the recommendations for annual influenza vaccination to school-age children in the U.S. Pediatrics 2012;129(S2):S54–S62. 10. Longini IM Jr. A theoretic framework to consider the effect of immunizing schoolchildren against influenza: implications for research. Pediatrics 2012;129(S2):S63–S67. 11. Neuzil KM, Fiore AE, Schieber RA. Evolution of the pediatric influenza vaccination program in the U.S. Pediatrics 2012;129(S2): S51–S53. 12. Fiore AE, Shay DK, Broder K, et al. Centers for Disease Control and Prevention (CDC); Advisory Committee on Immunization Practices (ACIP). Prevention and control of influenza: recommendations of the Advisory Committee on Immunization Practices (ACIP), 2008. MMWR Recomm Rep 2008;57(RR–7):1–60. 13. Saville AW, Albright K, Nowels C, et al. Getting under the hood: exploring issues that affect provider-based recall using an immunization information system. Acad Pediatr 2011;11(1):44–9.

14. Gidengil CA, Rusinak D, Allred NJ, Luff D, Lee GM, Lieu TA. Financial barriers to implementing combination vaccines: perspectives from pediatricians and policy makers. Clin Pediatr (Phila) 2009;48(5): 539–47. 15. Paul IM, Eleoff SB, Shaffer ML, Bucher RM, Moyer KM, Gusic ME. Improving influenza vaccination rates for children through year-round scheduling. Ambul Pediatr 2006;6(4):230–4. 16. Bardenheier BH, Yusuf HR, Rosenthal J, et al. Factors associated with underimmunization at 3 months of age in four medically underserved areas. Public Health Rep 2004;119(5):479–85. 17. HealthyPeople.gov: 2020 topics and objectives. Immunization and infectious diseases. www.healthypeople.gov/2020/topicsobjectives2020/ objectiveslist.aspx?topicid=23. 18. CDC. Recent influenza vaccination trends across influenza seasons; Influenza vaccination coverage estimates by age group—U.S., 2007– 2008 through 2010–2011. www.cdc.gov/flu/professionals/vaccination/ trends-summary.htm. 19. Kempe A, Wortley P, O’Leary S, et al. Pediatricians’ attitudes about collaborations with other community vaccinators in the delivery of seasonal influenza vaccine. Acad Pediatr 2012;12(1):26–35. 20. Keane VA, Hudson AR, King JC Jr. Pediatrician attitudes concerning school-located vaccination clinics for seasonal influenza. Pediatrics 2012;129(S2):S96–S100. 21. Wright S, Deputy Inspector General for Evaluation and Inspections. Memorandum report: 2009 H1N1 school-located vaccination program implementation, OEI-04-10-00020 [document]. USDHHS. oig.hhs. gov/oei/reports/oei-04-10-00020.pdf. 22. King JC Jr, Stoddard JJ, Gaglani MJ, et al. Effectiveness of school-based influenza vaccination. N Engl J Med 2006;355(24):2523–32. 23. Hull HF, Ambrose CS. The impact of school-located influenza vaccination programs on student absenteeism: a review of the U.S. Literature. J Sch Nurs 2011;27(1):34–42. 24. King JC Jr, Beckett D, Snyder J, Cummings GE, King BS, Magder LS. Direct and indirect impact of influenza vaccination of young children on school absenteeism. Vaccine 2012;30(2):289–93. 25. Davis MM, King JC Jr, Moag L, Cummings G, Magder LS. Countywide school-based influenza immunization: direct and indirect impact on student absenteeism. Pediatrics 2008;122(1):e260–e265. 26. Bobo N, Etkind P, Talkington K. School-located vaccination clinics: NASN, NACCHO, ASTHO summit—November 16–17, 2010, meeting summary [document]. www.nasn.org/portals/0/2011_11_21_nasn_ naccho_astho.pdf. 27. Middleman AB, Tung JS. School-located immunization programs: do parental preferences predict behavior? Vaccine 2011;29(19):3513–6. 28. Carpenter R, Lott J, Lawson BM, et al. Mass distribution of free, intranasally administered influenza vaccine in a public school system. Pediatrics 2007;120(1):e172–e178. 29. Hull HF, Frauendienst RS, Gundersen ML, Monsen SM, Fishbein DB. School-based influenza immunization. Vaccine 2008;26(34):4312–3. 30. Schieber RA, Kennedy A, Kahn EB. Early experience conducting school-located vaccination programs for seasonal influenza. Pediatrics 2012;129(S2):S68–S74. 31. Lott J, Johnson J. Promising practices for school-located vaccination clinics—part I: preparation. Pediatrics 2012;129(S2):S75–S80. 32. Daley MF, Kempe A, Pyrzanowski J, et al. Vaccination and billing outcomes from a multi-site school-located vaccination program for adolescents [presentation]. National Immunization Conference. Washington DC, March 30, 2011. 33. Lott J, Johnson J. Promising practices for school-located vaccination clinics—part II: clinic operations and program sustainability. Pediatrics 2012;129(S2):S81–S87. 34. National Association of County and City Health Officials. The FluMists for Schools Program in Carroll County, MD. NACCHO CASE study, Feb 2007. www.naccho.org/topics/HPDP/infectious/ immunization/upload/Carroll-County.pdf.

www.ajpmonline.org

Humiston et al / Am J Prev Med 2014;46(1):1–9 35. National Association of County and City Health Officials. The FluMists for Schools Program in San Bernardino County, CA. NACCHO CASE study, Feb 2007. www.naccho.org/topics/HPDP/ infectious/immunization/upload/San-Bernadino.pdf. 36. National Association of County and City Health Officials. The FluMists for Schools Program in Knox County, TN – Evidence of positive public-private partnerships. www.naccho.org/topics/HPDP/ infectious/immunization/upload/Carroll-County.pdf NACCHO CASE study, Feb 2007. www.naccho.org/topics/HPDP/infectious/immuniza tion/upload/Knox-County.pdf. 37. U.S. Census 2010. 2010.census.gov/2010census/data/index.php. 38. Poehling KA, Fairbrother G, Zhu Y, et al. New Vaccine Surveillance Network. Practice and child characteristics associated with influenza vaccine uptake in young children. Pediatrics 2010;126(4):665–73. 39. Zimmerman LA, Bartlett DL, Enger KS, Gosney K, Williams WG. Influenza vaccination coverage: findings from immunization information systems. BMC Pediatr 2007;7:28. 40. Szilagyi P, Vann J, Bordley C, et al. Interventions aimed at improving immunization rates. Cochrane Database Syst Rev 2002(4):(CD003941). 41. Yoo BK, Humiston SG, Szilagyi PG, Schaffer SJ, Long C, Kolasa M. Cost effectiveness analysis of elementary school-located vaccination against influenza—results from a randomized controlled trial. Vaccine 2013;31(17):2156–64. 42. Bordley WC, Chelminski A, Margolis PA, Kraus R, Szilagyi PG, Vann JJ. The effect of audit and feedback on immunization delivery: a systematic review. Am J Prev Med 2000;18(4):343–50.

43. Szilagyi PG, Bordley C, Vann JC, et al. The effect of patient reminder/ recall interventions on immunization rates: A review. JAMA 2000;284 (14):1820–7. 44. Szilagyi PG, Schaffer S, Shone L, et al. Reducing geographic, racial, and ethnic disparities in childhood immunization rates by using reminder/recall interventions in urban primary care practices. Pediatrics 2002;110(5):e58. 45. Szilagyi PG, Rodewald LE, Savageau J, Yoos J, Doane C. Improving influenza vaccination rates in children with asthma: a test of a computerized reminder system and an analysis of factors predicting vaccination compliance. Pediatrics 1992;90(6):871–5. 46. Szilagyi PG, Rodewald LE, Humiston SG, et al. Reducing missed opportunities for immunizations: easier said than done. Arch Pediatr Adolesc Med 1996;150(11):1193–200. 47. Koff RS. Hepatitis B school-based vaccination programmes in the USA: a model for hepatitis A and B. Vaccine 2000;18(S1):S77–S79. 48. Hall S, Galil K, Watson B, Seward J. The use of school-based vaccination clinics to control varicella outbreaks in two schools. Pediatrics 2000;105(1):e17. 49. Effler PV, Chu C, He H, et al. Statewide school-located influenza vaccination program for children 5–13 years of age, Hawaii, USA. Emerg Infect Dis 2010;16(2):244–50. 50. Vogt TM, Wortley PM. Epilogue: school-located influenza vaccination during the 2009–2010 pandemic and beyond. Pediatrics 2012;129(S2): S107–S109. 51. New York State Immunization Information System (NYSIIS). www. health.ny.gov/prevention/immunization/information_system/.

Did you know? According to the 2013 Journal Citation Report, published by Thomson Reuters, the 2012 Impact Factor for AJPM is 3.945, which ranks it in the top 10% of PH and GEH Journals, and in the top 12% of GM and IM Journals.

January 2014

9