Effectiveness of seasonal inactivated influenza vaccination in Japanese schoolchildren: an epidemiologic study at the community level

ABSTRACT

Influenza vaccination is the most effective method for preventing influenza virus infection. The incidence of influenza is higher in schoolchildren than other age groups. This study evaluated the effectiveness of seasonal inactivated influenza vaccination (IIV) in a community population of schoolchildren during two seasons. This study was a cross-sectional survey of public schoolchildren based on data collected in the 2012/2013 and 2014/2015 seasons. The questionnaire was distributed to all public schoolchildren of target grade in a survey area, and 7945 respondents were included in the analysis. The vaccination status and influenza onset were defined based on the self-reported questionnaire by parents or guardians. Generalized linear mixed models were used to adjust clustering within schools and individual covariates and calculate odds ratios (ORs) and 95% confidence intervals (CIs) for associations between vaccination status and influenza onset. The influenza incidence was higher in the 2015 than the 2013 survey (25% versus 17%), although the vaccination rates were comparable between the two seasons. Receiving one- or two-dose vaccination was more protective against influenza than non-vaccination in both the 2013 (OR, 0.77; 95%CI, 0.65–0.92) and 2015 (OR, 0.88; 95%CI, 0.75–1.02) surveys. Full vaccination was also more protective in both the 2013 (OR, 0.75; 95%CI, 0.62–0.89) and 2015 (OR, 0.86; 95%CI, 0.74–1.00) surveys. Seasonal IIV was protective against influenza for Japanese schoolchildren in a community-based real-world setting. The difference in clinical effectiveness of IIV between the two seasons was likely due to the antigenic mismatch between the circulating and vaccine strains.

KEYWORDS:

• Antigenic mismatch
• generalized linear mixed model
• schoolchildren
• trivalent influenza vaccine
• vaccine effectiveness

Introduction

Influenza is an acute respiratory illness that is associated with morbidity and mortality, although most cases are self-limiting. Influenza virus spreads through communities in local outbreaks, and influenza vaccination is the most effective method for preventing influenza virus infection and its potentially severe complications.¹ Ministry of Health, Labour and Welfare of Japan states that vaccination against influenza suppresses the onset to some extent and prevents from becoming severe symptoms, and it is highly effective for people who have high possibility of getting severe symptoms if they become ill, such as elderly or those with underlying diseases.² Although the coverage rate of influenza vaccination is increasing in Japan, the rate for children <13 years old has been reported to be approximately 60%.³ Importantly, the incidence of influenza is higher in schoolchildren than other age groups,^4,5 and schoolchildren can play an important role in the transmission of influenza in a community.^6–8 Therefore, vaccination intervention in schoolchildren can protect other unimmunized community residents^9-11 or their household members^12-14 against influenza. Furthermore, the most effective strategy for preventing an influenza epidemic in a community is vaccination of healthy children, especially schoolchildren. Accordingly, estimating clinical vaccine effectiveness (VE) of influenza in schoolchildren is essential from a public health perspective.

A recent systematic review¹⁵ estimated that the VE rates for laboratory-confirmed influenza and influenza-like illness in children aged 2–16 years were 64% and 28%, respectively. Estimated VE rates can vary, depending on not only time, place, and the match between the vaccine and circulating strains but also study design, including diagnostic tests and outcome measurements. The VE of influenza in schoolchildren can also be influenced by vaccination rates and the epidemic status in a school. Accordingly, VE is likely to vary among schools and communities. However, few epidemiological studies investigated the VE of influenza in all schoolchildren within an entire study area.^16,17

Epidemiologic studies are required to investigate the VE at the community level. Therefore, we aimed to assess the effectiveness of seasonal inactivated influenza vaccination (IIV) in a large-scale general population of schoolchildren during two seasons.

Materials and methods

Study design and participants

This study was based on the ToMMo Child Health Study, which was a cross-sectional survey of public schoolchildren in Miyagi Prefecture of Japan. The detailed study design was previously reported.^18,19 The present study included data from the 2013 (2012/2013 season) and 2015 (2014/2015 season) surveys administered in June of the survey year that retrospectively collected information on influenza vaccination status and influenza onset of the previous season. Ethical approval was obtained from the Institutional Review Board of Tohoku University Graduate School of Medicine (No. 2012–1-278). Informed consent to participate in the present study was assumed by the completion and return of the questionnaire.

Among a total of 425 public schools, including elementary, junior high, and special-needs schools (233 and 192 in the 2013 and 2015 surveys, respectively) and 27 municipalities (13 and 14 in the 2013 and 2015 surveys, respectively), we surveyed children in grades 2 (age, 7–8 years), 4 (age, 9–10 years), and 6 (age, 11–12 years) at elementary school and grade eight (age, 13–14 years) at junior high school. Children born between April 2000 and March 2008 were included. The questionnaire was distributed to a total of 31 231 children (12 742 and 18 489 in the 2013 and 2015 surveys, respectively), all of whom were in the target grades across the target municipalities. The questionnaire was completed by parents or guardians at home. A total of 8448 (27.1%) questionnaires (4074 and 4374 from the 2013 and 2015 surveys, respectively) were returned to our laboratory by mail. A total of 503 children were excluded due to being outside the target grades (74 and 121 from the 2013 and 2015 surveys, respectively), missing data on sex (38 and 51 from the 2013 and 2015 surveys, respectively), and missing information on vaccination status (30 and 41 from the 2013 and 2015 surveys, respectively) and influenza onset (37 and 154 from the 2013 and 2015 surveys, respectively). Finally, 7945 respondents (25.4%, 3912 and 4033 from the 2013 and 2015 surveys, respectively) were included in the final analysis.

Definition of vaccination status and influenza onset

Information on vaccination status and influenza onset was collected based on the following responses of the parent or guardian to the two following questions:

From May 1, 2012, to April 30, 2013 (in the 2013 survey) or from May 1, 2014, to April 30, 2015 (in the 2015 survey), was your child vaccinated for influenza?

1. None

2. Vaccinated once during this period

3. Vaccinated twice during this period

From May 1, 2012, to April 30, 2013 (in the 2013 survey), or from May 1, 2014, to April 30, 2015 (in the 2015 survey), did your child contract influenza?

1. Yes

2. No

In Japan, receiving the influenza vaccine twice, 2–4 weeks apart, from October to December is recommended for children aged 6 months to 12 years. For children aged over 13 years, one-time vaccination per year is recommended. During the influenza season, children in eighth grade were 12 or 13 years old and were defined as fully vaccinated in the present study if they received vaccination once or twice for that season. A single dose was 0.25 ml for children aged 6 months to 2 years and 0.5 ml for children aged over 3 years. Until the 2014/2015 season, trivalent IIV was used in Japan. Influenza vaccine strains were A/California/7/2009(X-179A) for A(H1N1)pdm09, A/Texas/50/2012(X-223) for H3N2, and B/Massachusetts/02/2012(BX-51B) for type B Yamagata lineage in the 2012/2013 season and A/California/7/2009(X-179A) for A(H1N1)pdm09, A/New York/39/2012(X-233A) for H3N2, and B/Massachusetts/2/2012(BX-51B) for type B Yamagata lineage in the 2014/2015 season.

According to the Infectious Agents Surveillance Report from the National Institute of Infectious Diseases, the percentages of circulating influenza strains in Japan were 76%, 2%, and 12% for A (H3N2), A (H1N1), and type B, respectively, in the 2012/2013 season²⁰ and 85%, 1%, and 14% for A (H3N2), A (H1N1), and type B, respectively, in the 2014/2015 season.²¹

Statistical analysis

We used chi-squared test to compare the proportions of categorical variables between groups. To calculate odds ratios (ORs) and 95% confidence intervals (CIs) for the association of vaccination status with the onset of influenza, we fitted generalized linear mixed models because the vaccination rate and epidemic status varied among the schools participating in the surveys. We computed the model with a logit link function that included a random effect for schools that the participants were enrolled in and fixed effects for individual covariates such as sex and school grade. In model 1, children with complete and partial vaccination (vaccination dose, once or twice) were compared with unvaccinated children (vaccination dose, none). In model 2, children with complete vaccination (vaccination dose, twice for school grades 2, 4, and 6 and once for school grade 8) were compared with children with partial and no vaccination (vaccination dose, once or none for school grades 2, 4, and 6 and none for school grade 8). VE was calculated as follows: (1 − adjusted OR) × 100%. For additional analyses of robustness, we adjusted for the history of asthma. History of asthma was defined based on the response of parents or guardians to the following question: “Has your child ever had asthma?”

Results

The characteristics of the participants in the 2013 and 2015 surveys are presented in Table 1. Briefly, 2320 of 3912 (59%) children in the 2013 survey and 2310 of 3902 (59%) in the 2015 survey received one or two vaccinations. The vaccination rates among the survey participants were comparable between the 2013 and 2015 surveys; however, the incidence of children who contracted influenza was higher among the participants of the 2015 survey than those of the 2013 survey (991/3902 [25%] versus 652/3912 [17%]).

Table 1. Characteristic of study participants.

	Vaccination							Influenza onset
	Twice		Once		None			Contracted		Not contracted
Survey 2013	(n = 1712)		(n = 608)		(n = 1592)		P	(n = 652)		(n = 3260)		P
Sex, n (%)
Boy	806	(42)	299	(16)	814	(42)	0.067	335	(17)	1584	(83)	0.19
Girl	906	(45)	309	(16)	778	(39)		317	(16)	1676	(84)
School grade, n (%)
2nd	631	(58)	80	(7)	373	(34)	<0.001	165	(15)	919	(85)	0.082
4th	583	(53)	110	(10)	416	(38)		207	(19)	902	(81)
6th	437	(44)	117	(12)	447	(45)		173	(17)	828	(83)
8th	61	(9)	301	(42)	356	(50)		107	(15)	611	(85)
History of asthma, n (%)
Presence	357	(47)	118	(16)	283	(37)	0.08	125	(16)	633	(84)	0.88
Absence	1355	(43)	490	(16)	1309	(42)		527	(17)	2627	(83)
	Vaccination							Influenza onset
	Twice		Once		None			Contracted		Not contracted
Survey 2015	(n = 1632)		(n = 678)		(n = 1723)		P	(n = 991)		(n = 3042)		P
Sex, n (%)
Boy	832	(41)	352	(17)	853	(42)	0.50	522	(26)	1,515	(74)	0.12
Girl	800	(40)	326	(16)	870	(44)		469	(24)	1,527	(77)
School grade, n (%)
2nd	578	(51)	123	(11)	423	(38)	<0.001	301	(27)	823	(73)	0.006
4th	484	(49)	107	(11)	405	(41)		227	(23)	769	(77)
6th	466	(42)	115	(10)	531	(48)		295	(27)	817	(73)
8th	104	(13)	333	(42)	364	(45)		168	(21)	633	(79)
History of asthma, n (%)
Yes	328	(40)	150	(18)	343	(42)	0.45	226	(28)	595	(72)	0.028
No	1304	(41)	528	(16)	1380	(43)		765	(24)	2447	(76)

History of asthma was defined based on the response of the parent or the guardian to the following question: “Has your child ever had asthma?”

P values were computed by the chi-squared test.

Vaccination once or twice was more protective against influenzas than non-vaccination in both the 2013 (OR, 0.77; 95%CI, 0.65–0.92) and the 2015 (OR, 0.88; 95%CI, 0.75–1.02) surveys in model 1 (Table 2). The estimated VE were 23% and 12% in the 2013 and 2015 surveys, respectively, according to model 1. Furthermore, full vaccination was more protective against influenza in both the 2013 (OR, 0.75; 95%CI, 0.62–0.89) and the 2015 (OR, 0.86; 95%CI, 0.74–1.00) surveys in model 2. The estimated VE were 25% in survey 2013 and 14% in survey 2015, respectively, according to model 2. After adjusting for the history of asthma, the same results were produced.

Table 2. Association between influenza vaccination and the onset of influenza.

	Model 1				Model 2
	Vaccination	n* (%)	OR	95%CI	Vaccination	n* (%)	OR	95%CI
The 2013 survey
	≥1 (n = 2320)	354 (15%)	0.77	0.65–0.92^§	Complete (n = 2013)	296 (15%)	0.75	0.62–0.89^§
	0 (n = 1592)	298 (19%)	1.00	Reference	Incomplete (n = 1899)	356 (19%)	1.00	Reference
The 2015 survey
	≥1 (n = 2310)	544 (24%)	0.88	0.75–1.02	Complete (n = 1965)	451 (23%)	0.86	0.74–1.00
	0 (n = 1723)	447 (26%)	1.00	Reference	Incomplete (n = 2068)	540 (26%)	1.00	Reference

* number (%) of children with influenza

We fitted generalized linear mixed models to adjust clustering within schools and individual covariates including sex and school grade.

Model 1: Children with complete and partial vaccination (vaccination dose, once or twice) were compared with unvaccinated children (vaccination dose, none).

Model 2: Children with complete vaccination (vaccination dose, twice for school grades 2, 4, and 6, once for school grade 8) were compared with children with partial and no vaccination (vaccination dose, once or none for school grades 2, 4, and 6, none for school grade 8).

OR, odds ratio; CI, confidence interval

^†P < 0.05, ^§P < 0.01.

Discussion

We estimated the clinical effectiveness of IIV in Japanese schoolchildren in a community-based real-world setting and found that seasonal IIV reduced the incidence of influenza in schoolchildren in both the 2012/2013 and the 2014/2015 seasons.

Our study showed that the VE of one- and/or two-dose vaccination against influenza were 12%–23% compared with non-vaccination and that the VE of complete vaccination ranged 14%–26% compared with partial and/or non-vaccination. However, the estimated VE in the current study were lower than those reported in previous studies,¹⁵ which may be due to differences in the study design, including the diagnostic tests and outcome measures. In general, the VE for influenza-like illness is lower than that for laboratory-confirmed influenza¹⁵ because of the overlap in the presentation of influenza and influenza-like respiratory infections. The probability of influenza diagnosis in the present study was likely to be lower than that of laboratory-confirmed influenza. However, we predict that the true rate of influenza in the present study was higher than that based on the self-reporting of influenza-like illness because almost all children with influenza-like illness in a season with high influenza rates visit a clinic and are tested with a rapid influenza diagnostic test kit in Japan.^17,22,23 Furthermore, VE could also be affected by study design and setting. According to the studies on VE of influenza in 2014/2015 season from Japan, test-negative-case control design estimated VE rate at 33% (None versus Twice)²⁴ and 26% (age 6–12 years).²⁵ In contrast, epidemiological study which was based on a self-reported questionnaire estimated VE rate in 2014/2015 at 13%.²⁶ The VE rates reported by epidemiological studies including entire study area were comparable to that of the current study.^16,17,26 Therefore, one reason for the lower VE rate in the present study might be differences in the study design.²⁷

The VE in the 2014/2015 season was lower than that in the 2012/2013 season, although the vaccination rates were almost equal between the two seasons. According to the Infectious Agents Surveillance Report from the National Institute of Infectious Diseases, the circulating vaccines were antigenically similar to the vaccine strains used in the 2012/2013 season.²⁰ In contrast, >70% of the AH3 circulating vaccines were antigenically different from the 2014/2015 vaccine strain A/New York/39/2012,²¹ and A (H3N2) accounted for 85% of all circulating influenza strains in the 2014/2015 season. It is conceivable that the VE against influenza was lower due to a mismatch between the circulating vaccines and the vaccine strains.²⁷ Therefore, the lower VE of the influenza vaccine in the 2014/2015 season may be attributed to the antigenic mismatch of the AH3 component of the vaccine with the circulating strains. Unfortunately, no studies from Japan have compared VE between the 2012/2013 and 2014/2015 season in the same study.

There is insufficient evidence on whether one- or two-dose vaccination is associated with favorable outcomes against influenza.^15,28 The present study revealed that two-dose vaccination tends to be more protective against influenza than one-dose vaccination. Therefore, it is our belief that two-dose vaccination is necessary to protect against influenza. Conversely, one study reported that the VE of one-dose vaccination differs between influenza A and influenza B;²⁹ therefore, future investigation is needed to determine the association between VE and vaccine dose.

The strength of the present study was inclusion of a general population of schoolchildren in a community. In addition, the current study design had the advantage of including all public schools in the study area; therefore, our study includes both vaccinated and unvaccinated children within the same population. With a generalized linear mixed model, schools of the participants were furthermore included for random effect. However, there are several limitations in our study, which should be considered during interpretation of the results. First, the diagnostic methods for influenza and date of vaccination and onset were not confirmed. The onset of influenza was determined based on a self-reported questionnaire administered to the parents or guardians of the schoolchildren. To prevent the spread of influenza, Ministry of Education, Culture, Sports, Science and Technology has instructed schools to instruct schoolchildren with influenza-like symptoms during the period of the epidemic to receive a medical diagnosis immediately.³⁰ Under the School Health and Safety Act, children who have been diagnosed with influenza by doctors need to be suspended from school and their parents have to report to school.³¹ Those children who were reported to have contracted influenza by their parents or guardians in the current study were most likely tested by a rapid influenza diagnostic test kit, given that nearly all children with influenza-like illness during a high season visit a clinic to be tested in Japan.^17,22,23 Rapid influenza diagnostic test kits widely used in Japan have high sensitivities of 88–100% and specificities of 94–100%.^24,29 Therefore, we convinced that whether they have influenza was obvious. Second, there is a possibility that children who contracted influenza within 13 days after vaccination were included. Previous studies commonly excluded subjects who were vaccinated between 0 and 13 days before the onset of influenza³² as it may take 10–14 days for the immunity to build up fully after the vaccination. However, we believe that the impact of not excluding these schoolchildren in the current study was minimal, as their predicted percentage was reported to be less than 0.5%.³³ Third, a response bias due to a low response rate should be acknowledged. Parents or guardians who responded to the survey might differ in varying ways from those who did not respond. Fourth, the history confirmation was based on the parents’ or guardians’ memory. Therefore, recall bias may have affected the results due to the way in which the vaccination information had been collected, although we had investigated it at the end of the influenza season.

In conclusions, seasonal IIV was protective against influenza in Japanese schoolchildren in a community-based real-world setting. The reduction in the clinical effectiveness of IIV in one of the study seasons was likely due to the antigenic mismatch between the circulating and vaccine strains.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Financial Support

The authors have indicated that they have no financial relationships relevant to this article to disclose.

Acknowledgments

We would like to thank the Miyagi Prefectural Board of Education and the municipal boards of education in Iwanuma, Watari, Yamamoto, Shiroishi, Natori, Kakuda, Zao, Shichikashuku, Ogawara, Murata, Shibata, Kawasaki, Marumori, Kesennuma, Tome, Kurihara, Higashimatsushima, Osaki, Shichigahama, Taiwa, Osato, Shikama, Kami, Wakuya, Misato, Onagawa, and Minamisanriku for participating in the study.

Additional information

Funding

This work was supported by the MEXT Tohoku Medical Megabank Project and Japan Agency for Medical Research and Development (AMED) [grant number JP15km0105001].