Schools provide opportunities for close contact between students which aids the spread of influenza virus infections.1 School-age children play a significant role in the transmission of influenza within schools, families and communities.2,3 Influenza vaccination among school-age children has been recommended in many countries as a way to reduce the incidence of influenza infections4,5; indeed, some studies suggest that high vaccination coverage among school-age children can reduce the transmission of influenza among both vaccinated and unvaccinated school children through herd immunity.6,7
Since 2007, a school-based vaccination campaign has been administered in Beijing, providing seasonal trivalent influenza vaccines (IIV3) free-of-charge for elementary and high school students before the wintertime influenza seasons. However, no evaluation of vaccine effectiveness (VE) in preventing influenza illness in these schools has been conducted to date.
IIV3 recommended by World Health Organization for the 2014–2015 influenza season in the Northern Hemisphere (NH) was used in Beijing. This IIV3 contained hemagglutinin derived from an A/California/7/2009 (H1N1)-like virus, an A/Texas/50/2012 (H3N2)-like virus and a B/Massachusetts/2/2012-like (Yamagata lineage) virus.8,9 The virologic surveillance by Chinese National Influenza Center showed that influenza A(H3N2) was the predominant circulating influenza virus in the northern region of China during the 2014–2015 season,10 similar to most areas in the NH.11 Across the NH, circulating viruses were noted to be antigenically drifted from the vaccine virus and antigenically similar to a new A/Switzerland/9715293/2013 (H3N2)-like virus.12 Studies from the United States and other NH countries reported relatively low VE against A(H3N2) influenza viruses, presumably because of the poor match between the vaccine strain and circulating strains during 2014–2015.13–15
We report here on a case-control study to estimate the effectiveness of influenza vaccine in preventing influenza illness among school-age children in Beijing, China, during the 2014–2015 influenza season.
Study Design and Participants
This case-control study examines influenza outbreaks in elementary, junior high and high schools reported to Beijing Centers for Disease Control and Prevention (CDC) between November 1, 2014, and December 31, 2014.
School outbreaks of influenza illness were found through 2 existing syndromic surveillance systems in Beijing that monitor influenza-like illness (ILI) (measured or self-reported temperature ≥38°C with either cough or sore throat) and febrile illnesses of any etiology (measured or self-reported temperature ≥37.5°C). We used the existing outbreak definitions for each surveillance system:
- ILI outbreak: Ten or more epidemiological-linked ILIs identified in a school within 1 week; local CDC staff used a broad definition of potential links, including any opportunities for face-to-face contact in a classroom or any school setting, though (as noted below), most ILI outbreaks were identified within a single classroom or within several adjacent classes.
- Febrile outbreak: Ten or more febrile illnesses within a single school classroom within 2 days.
The local CDC staff were notified of suspected outbreaks fitting either criteria by school doctors; local CDC staff began field epidemiological investigations within the same day (typically within 2 hours). Epidemiological information for sick children absent from school was collected through telephone interviews with parents. We attempted to collect oral pharyngeal swabs that were collected from up to 10 symptomatic cases from each school where an ILI or febrile outbreak was reported. Priority was given to students who were currently sick and attending school. If this number of cases was less than 10, CDC attempted to collect respiratory specimens through home visits from sick children dismissed from school within the 7 days before the outbreak because of illness.
During the 2014–2015 influenza season, the school vaccination campaign was conducted from 15 October, 2014, to 30 November, 2014. Because vaccinees are typically considered immunized 14 days after vaccination,13 we examined school influenza outbreaks that occurred at least 14 days after the start of each school’s vaccination campaign.
The time period of a school outbreak began with the earliest illness onset date (index case) among cases within an outbreak and ended when no new cases with ILI or fever were found for 7 consecutive days (as determined by the investigating team based on daily contact with school teachers and staffs).
An influenza case was a student who had ILI or febrile illness connected with a reported school outbreak and was positive for an influenza virus by real-time reverse-transcription polymerase chain reaction (rRT-PCR) assay.
A control was a classmate of an influenza case who was fully asymptomatic (ie, had no symptoms of fever, cough or sore throat) during the period from the illness onset of the index case of the outbreak to the end of the outbreak.
Information of enrolled schools and students was collected using a standardized questionnaire. School information included the type of school (elementary, junior and high schools), locale (urban or rural areas), the total number of students and number of students who received IIV3 vaccination. Individual-level information on participants was provided by students and their parents, including age, sex, height, weight, presence of chronic medical conditions, symptoms and illness onset date. Because most students were identified at the start of their illness and no follow-up was conducted, we could not document whether illnesses were medically attended. Documented influenza vaccination records for 2013–2014 and 2014–2015 IIVs were collected by using Beijing Management System of Information on Immunization Program. Staff administering vaccinations entered influenza vaccination records directly into the electronic registry during each school’s campaign. Local CDC staff then abstracted vaccination records of cases and controls from the registry following each outbreak.
Oral pharyngeal swabs were tested by rRT-PCR in the district CDC collaborating laboratories managed by Beijing CDC using PCR procedures recommended by the World Health Organization Collaborating Center for Reference and Research on Influenza at the Chinese National Influenza Center.
Questionnaire and laboratory data were entered in duplicate using EpiData Software (EpiData Association, Odense. Denmark) and analyzed using SAS 9.3 software (SAS Institute, Inc., Cary, NC). Participant characteristics and vaccination status of cases and asymptomatic controls were compared using χ2 tests. We present VE estimates in 4 steps. First, we present unadjusted VE (1 − odds ratios; the odds of IIV3 vaccination among cases divided by the odds of IIV3 vaccination among controls) estimates using unconditional logistic regression models. Second, we present VE adjusted for participant characteristics (age, sex, locale, body mass index category and presence of 1 or more chronic conditions) and calendar time (onset week of index case). Third, because schools differed in vaccination coverage and may have also differed in factors associated with influenza exposure (eg, classroom crowding, student mixing and hand hygiene), we present VE adjusted for potential confounding and clustering effects by school by using a mixed effects logistic regression model.16,17 Fourth, we presented a fully adjusted VE by using a mixed effects logistic regression model to adjust the clustering effect of school, participant characteristics and calendar time. To aid in interpretation, we also report VE stratified by age group and by rural versus urban locale, given economic and demographic differences between schools in these neighborhoods. All statistical tests were 2-sided, and statistical significance was defined as P value <0.05 or the lower bound of the 95% confidence interval (CI) for VE >0.
Given recent observations from studies of children in the United States18,19 and China20 that the VE of a current season’s IIV3 is influenced by vaccination history, we also tested for potential effect modification by receipt of the prior season’s 2013–2014 IIV3. Similar to previous studies,21 we did so by including main effects for current and prior season vaccination status plus an interaction term into each model; because standard errors in such models are inflated,22 we used a P value of <0.2 for statistical significance.
This study was approved by the institutional review board and human research ethics committee of Beijing CDC. Informed consent was completed by parents by telephone before children’s participation. A signed informed consent was required from their parents or legal guardian for students <18 years of age.
Characteristics and Vaccination Status of Participants
From November 1, 2014, to December 31 of 2014, a total of 70 influenza outbreaks in elementary and high schools were identified in Beijing. According to the inclusion criteria, 43 school outbreaks were eligible for this study (Fig. 1); these outbreaks were reported from 38 schools in urban districts and 5 schools in rural districts, including 22 (51%) elementary schools, 15 (35%) junior high schools and 6 (14%) high schools. Most outbreaks affected a single school classroom; 12 (28%) of 43 outbreaks affected more than 1 class. The number of cases per outbreak ranged from 10 to 76 (median: 16). The average coverage rate of 2014–2015 IIV3 across students in the 43 schools was 48% (range across schools: 13%–99%). The timing of outbreak by school type is illustrated in Figure 2.
In total, 3323 students 6–18 years old were recruited across the 43 outbreaks; 3101 (93%) of these students were included in the study, excluding 222 with incomplete or missing 2014–2015 vaccination records. Among the 3101 study sample, 313 students with ILI or febrile illness for whom respiratory specimens were collected, 854 students with ILI or febrile illness but without respiratory specimens collected, 236 students with at least 1 symptom (although not meeting ILI or febrile illness case definitions) and finally 1698 with no illness symptoms from the date of illness onset of the index case through the end of the school outbreak (asymptomatic controls) (Fig. 1).
Among the 313 students with ILI or febrile illness for whom respiratory specimens were collected, we identified 210 rRT-PCR–confirmed influenza cases and 103 students who tested negative for influenza. Of the 210 confirmed cases, 195 had A(H3N2) influenza viruses; the remaining 15 also tested positive for influenza A, and although they were unsubtyped, are presumed to be A(H3N2) influenza viruses. The average time from individual illness onset to specimen collection was 2 days (standard deviation: ± 2 days) for the confirmed influenza cases as well as for the influenza-negative illnesses.
Although the median age was 12 years old for both cases and asymptomatic controls, there was a wider distribution of ages among controls; 54% of cases were 11–15 years old compared with 38% of controls. Cases were also more likely to be from rural areas (18% vs. 5%) and to have a chronic health condition (8% vs. 5%) (Table 1). Among both cases and asymptomatic controls, vaccination coverage was highest among students 11–15 years of age (cases: 56%, 63/113; controls: 48%, 303/634) and among those in rural areas (cases: 71%, 27/38; controls: 58%, 53/91) (Table 1). Vaccination coverage did not differ by sex, body mass index category or chronic conditions.
VE of 2014–2015 IIV3
The unadjusted VE of 2014–2015 IIV3 against influenza illness was −11% (95% CI: −49% to 17%) for all children (Table 2). After adjusting for participant characteristics only, the VE point estimate increased to 18%; after adjusting for school cluster only, the VE point estimate increased to 36%. The VE fully adjusted (using a mixed effects model) for participant characteristics, calendar time and school/cluster was 38% (95% CI: 12%–57%) (Table 2). This increase in VE after introducing adjustments was reflected across age groups. However, we observed notably lower VE point estimates among children of 11–15 years of age (Table 2).
VE for Combinations of 2013–2014 and 2014–2015 IIV3
After including the main effects for 2013–2014 IIV3 and 2014–2015 IIV3 in the fully adjusted model, a significant interaction effect between these vaccine exposures was noted (P = 0.12), suggesting effect modification by prior vaccination. Therefore, we estimated VE for 3 IIV3 combinations with no IIV3 in either year as the referent. In the fully adjusted model, the VE for receipt of 2014–2015 vaccination only, 2013–2014 vaccination only and vaccinations in both seasons was 54% (95% CI: 8%–77%), −18% (95% CI: −107% to 32%), and 29% (95% CI: −8% to 53%), respectively (Table 3).
As crowded places, schools provide an ideal environment for spreading influenza, which is why the Beijing city government began rolling out a school-located influenza vaccination program starting in 2007. However, with the exception of a cohort study of the effectiveness of the monovalent influenza A(H1N1)pdm09 vaccine among school-age children during the 2009 H1N1 pandemic,23 there have been no evaluations of influenza VE in Beijing schools. Therefore, we took advantage of ongoing investigations of ILI and febrile illness outbreaks in local schools to assess the preventive benefit of IIV3. A strength of our study is its focus on the preventive benefit of IIV3 against outbreaks of mild influenza illness in schools, which is a key objective of school-located vaccination programs and with few exceptions7,23 rarely evaluated for the mild influenza illnesses. However, as our study design focused on nonmedically attended influenza illness, it is unclear how our VE estimates compare to those reported on medically attended cases.
We identified 43 school-based influenza outbreaks from November 1, 2014, to December 31, 2014. The average vaccination rate among investigated schools was 48%, which was much higher than the IIV3 coverage of 9% observed among ambulatory patients 3–17 years of age in Beijing 2 years earlier (in 2012–2013).24 This is an encouraging finding and reinforces the potential value of school-located influenza vaccination programs as a way to achieve high vaccination coverage in a relatively short period of time.3
Our fully adjusted estimate of VE against influenza A(H3N2) illness using asymptomatic controls, which took into account variations in VE by age and other participant characteristics, by calendar time and by any cluster effect associated with schools, was 38%. The stepwise increase in VE we observed in our adjusted models indicate that there were likely confounding and substantial variations in VE by participant characteristics and schools, and the school clustering had the largest and statistically significant effect. We could not disentangle these confounding and variations in our study given the limited number of influenza cases and schools included. The null VE observed for the strata of children 11–15 years old, for example, was unexpected. Although others have also noted variations in VE when smaller age strata are examined (including null VE among children 13–15 years old in Japan in 2013–201425), examinations of VE among older children are rare.
Our estimated VE of 38% against influenza predominated by A(H3N2) influenza viruses among school children in 2014–2015 was higher than we expected given reports that the circulating A(H3N2) influenza viruses in China10 (including unpublished surveillance reports within Beijing) and other NH countries8,14 were drifted from the vaccine strain and given the low or nonsignificant VE estimates reported for 2014–2015 by Canada,15 the United Kingdom16 and the United States.13 However, direct comparison between our findings and these studies is difficult because of differences in study design and population. These other studies applied the test-negative design (with symptomatic influenza-negative controls vs. our asymptomatic controls) and focused on patients in contrast to our focus on illnesses identified in schools (of which many or most were likely not medically attended). We also do not know the genetic characterization of the A(H3N2) viruses infecting the students. When a study of ambulatory patients in the US narrowed their focus to 1 A(H3N2) genetic group (3C.3b), they observed a VE point estimate of 35% among older children and adults, similar to our finding; however, they noted null VE for another circulating group (3C.2a).26
Similar to recent studies, we found that current season VE was not independent of prior season vaccination. Our study adds to the small number of studies who have extended this finding to include children,20,27 since most prior research excluded children. When we examined VE for different combinations of current and prior vaccine exposure, we only observed significant VE for 2014–2015 IIV3 against influenza illness among children who were not vaccinated the previous season. Similarly, Ohmit et al18 conducted assessments of VE for ages 9 years old and older in the US for medically attended illnesses in 2012–2013 and in households with children in 2010–201119 and observed lower VE among those vaccinated in 2 consecutive seasons, compared with those vaccinated in the current season only.
Most notably, Skowronski et al14 observed a similar pattern of findings among patients 2 years old and older in Canada for the same vaccine years as our study; specifically, VE against influenza A(H3N2) among those who received the 2014–2015 influenza vaccine without prior vaccination in 2013–2014 season (43%) was higher than that among participants vaccinated with the same influenza A(H3N2) vaccine component in both 2013–2014 and 2014–2015 seasons (−15%). The antigenic distance hypothesis by Smith et al28 was presented as a possible explanation since the 2013–2014 and 2014–2015 influenza A(H3N2) vaccine components were homologous and may have interfered with immunogenicity.
There are several study limitations. First, because of possible asymptomatic or atypical influenza virus infections,29 some students who had been infected with influenza viruses might have been misclassified as controls, which could result in underestimation of VE. Second, our findings may not generalize beyond mild influenza illness, since we collected respiratory specimens from symptomatic children who were attending school or being cared for at home and likely missed more severe manifestations of disease. Third, although our study design was appropriate for given our focus on nonmedically attended illness, it is unclear how our VE estimates compare to those using the more commonly used test-negative design in medical settings. Nonetheless, other studies of children and adults that employed both healthy controls and influenza-negative controls reported similar VE estimates using both methods.30,31 Fourth, our evaluation focused on the initial month of local influenza circulation; therefore, it is unclear whether similar performance of IIV3 continued throughout the season. Fifth, we lack data on the antibody or cell-mediated immune response of children to IIV3; therefore, the mechanism of prior vaccination on the study season’s VE is unclear. Sixth, certainly, unvaccinated and vaccinated children may differ in multiple unmeasured ways (such as hygiene habits, household income, type of dwelling, nutrition, preventive care); so our results may be subject to residual and unmeasured confounding.
In conclusion, this mid-season assessment of influenza VE based on school influenza outbreaks indicated that the 2014–2015 IIV3 was modestly effective in protecting school-age children from influenza A(H3N2) virus illness in Beijing. Further research is needed to evaluate the protective benefit of annual school-located influenza vaccination campaigns to children and their families, teachers and staff and the broader community.
1. Finnie TJ, Copley VR, Hall IM, et alAn analysis of influenza
outbreaks in institutions and enclosed societies.Epidemiol Infect2014142107–113
2. Piedra PA, Gaglani MJ, Kozinetz CA, et alHerd immunity in adults against influenza
-related illnesses with use of the trivalent-live attenuated influenza
vaccine (CAIV-T) in children.Vaccine2005231540–1548
3. Gaglani MJEditorial commentary: school
vaccination: why worth the effort?Clin Infect Dis201459333–335
4. Rogier Bodewes D, Joost H, Guus FYearly influenza
vaccinations: a double-edged sword?Lancet Infect Dis20099784–788
5. Fiore AE, Epperson S, Perrotta D, et alExpanding the recommendations for annual influenza
vaccination to school
-age children in the United States.Pediatrics2012129suppl 2S54–S62
6. Barclay VC, Smieszek T, He J, et alPositive network assortativity of influenza
vaccination at a high school
: implications for outbreak
risk and herd immunity.PLoS One20149e87042
7. Pannaraj PS, Wang HL, Rivas H, et alSchool-located influenza
vaccination decreases laboratory-confirmed influenza
and improves school
attendance.Clin Infect Dis201459325–332
9. Feng L, Yang P, Zhang T, et alfor Chinese Center for Disease Control and Prevention; for Chinese Center for Disease Control and Prevention[Technical guidelines for the application of seasonal influenza
vaccine in China
(2014-2015)].Zhonghua Liu Xing Bing Xue Za Zhi2014351295–1319
12. Centers for Disease Control and Prevention (CDC)CDC Health Advisory Regarding the Potential for Circulation of Drifted Influenza
A (H3N2) Viruses.December 3, 2014Available at: http://emergency.cdc.gov/HAN/han00374.asp
. Accessed January 22, 2015
13. Flannery B, Clippard J, Zimmerman RK, et alCenters for Disease Control and PreventionEarly estimates of seasonal influenza vaccine effectiveness
- United States, January 2015.MMWR Morb Mortal Wkly Rep20156410–15
14. Skowronski D, Chambers C, Sabaiduc S, et alInterim estimates of 2014/15 vaccine effectiveness
A (H3N2) from Canada’s Sentinel Physician Surveillance Network, January 2015.Euro Surveill20152021022
15. Pebody RG, Warburton F, Ellis J, et alLow effectiveness of seasonal influenza
vaccine in preventing laboratory-confirmed influenza
in primary care in the United Kingdom: 2014/15 mid-season results.Euro Surveill20152021025
16. MacIntyre CR, Wang Q, Cauchemez S, et alA cluster randomized clinical trial comparing fit-tested and non-fit-tested N95 respirators to medical masks to prevent respiratory virus infection in health care workers.Influenza
Other Respir Viruses20115170–179
17. Janjua NZ, Skowronski DM, Hottes TS, et alSeasonal influenza
vaccine and increased risk of pandemic A/H1N1–related illness: first detection of the association in British Columbia, Canada.Clin Infect Dis2010511017–1027
18. Ohmit SE, Thompson MG, Petrie JG, et alInfluenza vaccine effectiveness
in the 2011–2012 season: protection against each circulating virus and the effect of prior vaccination on estimates.Clin Infect Dis201458319–327
19. Ohmit SE, Petrie JG, Malosh RE, et alInfluenza vaccine effectiveness
in households with children during the 2012–2013 season: assessments of prior vaccination and serologic susceptibility.J Infect Dis20152111519–1528
20. Fu C, Xu J, Lin J, et alConcurrent and cross-season protection of inactivated influenza
vaccine against A(H1N1)pdm09 illness among young children: 2012–2013 case–control evaluation of influenza vaccine effectiveness
21. Belongia EA, Simpson MD, King JP, et alVariable influenza vaccine effectiveness
by subtype: a systematic review and meta-analysis of test-negative design studies.Lancet Infect Dis201616942–951
22. Thompson MG, Clippard J, Petrie JG, et alInfluenza vaccine effectiveness
for fully and partially vaccinated children 6 months to 8 years old during 2011-2012 and 2012-2013: the importance of two priming doses.Pediatr Infect Dis J201635299–308
23. Wu J, Xu F, Lu L, et alSafety and effectiveness of a 2009 H1N1 vaccine in Beijing.N Engl J Med20103632416–2423
24. Yang P, Thompson MG, Ma C, et alInfluenza vaccine effectiveness
against medically-attended influenza
illness during the 2012-2013 season in Beijing, China
25. Shinjoh M, Sugaya N, Yamaguchi Y, et alKeio Pediatric Influenza
Research GroupEffectiveness of trivalent inactivated influenza
vaccine in children estimated by a test-negative case-control design study based on influenza
rapid diagnostic test results.PLoS One201510e0136539
26. Flannery B, Zimmerman RK, Gubareva LV, et alEnhanced genetic characterization of influenza
A(H3N2) viruses and vaccine effectiveness
by genetic group, 2014–2015.J Infect Dis20162141010–1019
27. Thompson MG, Clippard J, Petrie JG, et alInfluenza vaccine effectiveness
for fully and partially vaccinated children 6 months to 8 years old during 2011–2012 and 2012–2013: the importance of two priming doses.Pediatr Infect Dis J201635299–308
28. Smith DJ, Forrest S, Ackley DH, et alVariable efficacy of repeated annual influenza
vaccination.Proc Natl Acad Sci U S A19999614001–14006
29. Hsieh YH, Tsai CA, Lin CY, et alCIDER Research TeamAsymptomatic ratio for seasonal H1N1 influenza
infection among school
children in Taiwan.BMC Infect Dis20141480
30. Thompson MG, Li DK, Shifflett P, et alPregnancy and Influenza
Project WorkgroupEffectiveness of seasonal trivalent influenza
vaccine for preventing influenza
virus illness among pregnant women: a population-based case-control study during the 2010-2011 and 2011-2012 influenza
seasons.Clin Infect Dis201458449–457
31. Ferdinands JM, Olsho LE, Agan AA, et alPediatric Acute Lung Injury and Sepsis Investigators (PALISI) NetworkEffectiveness of influenza
vaccine against life-threatening RT-PCR-confirmed influenza
illness in US children, 2010-2012.J Infect Dis2014210674–683
Keywords:Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.
influenza; vaccine effectiveness; school; outbreak; China