Evaluation of Japanese people's perception of risk information for making decisions to receive influenza and rubella vaccinations

Abstract Background Generally, vaccination uptake in Japan lags behind World Health Organization targets. Objective This study aimed to understand how risk information and advice affect intention to receive vaccinations. Methods This study had a within‐subjects design. An online survey based on the Health Belief Model was sent to 2501 Japanese individuals (≧20 years) to assess the intention to be vaccinated for influenza and rubella after receiving minor and severe risk information and hypothetical advice about each vaccine. Regression analysis was used to measure changes in intentions to receive each vaccination after being provided with (1) risk information about each vaccine and (2) hypothetical encouragement and discouragement to be vaccinated. Main Outcomes The main outcomes included changes in vaccination intentions from baseline. Results Forty‐one percent (N = 1030) of those sent the survey completed it. At baseline, 43% and 65% of the respondents intended to have influenza and rubella vaccinations, respectively. Being provided with information about severe risks and susceptibility increased the intention to have the influenza vaccination among females in their 40s. Receiving inaccurate and discouraging information from one's mother significantly decreased the intention to have the rubella vaccination. Women 50 and older were more likely to intend not to have vaccination for rubella. Severe risk information decreased rubella vaccination intention in all age groups, except women in their 30s and 40s (p < .05). Conclusion For both vaccinations, older individuals demonstrated vaccine hesitancy. This group requires tailored messaging to help them understand their vulnerability (to influenza) and their role in transmission (for rubella) to encourage uptake of essential vaccinations. Patient or Public Contribution Members of the Japanese public responded to our online questionnaire on vaccination risk.


| INTRODUCTION
In Japan, influenza vaccination rates among different age groups of adults older than 20 years ranged between 40% and 50% during the 2017/2018 influenza season; rubella vaccination coverage among different age groups ranged between 29% and 90% in 2018. 1,2 Both rubella and influenza vaccinations are recommended by the Japanese government, but vaccination decisions are voluntary. Although several previous reports describe factors that might increase Japanese people's vaccination intentions, they focused primarily on sociodemographic factors, understanding of vaccination benefits, physician recommendations and concerns about side effects. [3][4][5] Forces that shape risk perceptions and methods that effectively promote vaccination uptake are not well understood in Japan.
Multiple studies have identified that concerns about the risks of receiving vaccinations, including side effects, pose a barrier to vaccination. [6][7][8][9][10] Even if due to incorrect information, mistrust or misunderstanding, vaccine hesitancy dampens immunisation uptake. [9][10][11] Documented primarily in the United States and Europe, 10,12 vaccine hesitancy appears to be multifactorial, attributable to underestimation of susceptibility, excessive anxiety about vaccine side effects and misinformation. [10][11][12] Recent studies indicate that vaccine hesitancy is also operative in Japan, 11 shaped by an anti-vaccination movement that undermines confidence in vaccines and weakens government recommendations, a preference for treating diseases rather than preventing them, 13 the costs of vaccination 3 and insufficient information from healthcare services. 4,5 Risk perception is considered to be a combination of probability and subjective judgement shaped by an individual's unique psychological, social, cultural and even political context. 14, 15 Matsui et al. 4 found that accurate information and understanding of individual susceptibility, vaccination and severity of seasonal influenza infection were associated with increased vaccination uptake. However, even when respondents feel sufficiently informed to decide whether to be vaccinated, concerns about safety may shape risk perceptions and uptake, as demonstrated by a study by Walter et al. 16  those who are unsure whether they been previously infected with or vaccinated against rubella and those who plan to travel overseas. 18 The Japanese government recommends that those who live with or have frequent contact with women of childbearing age be vaccinated for rubella because epidemics of rubella have occurred since 2002. 18,19 The aims of this study were to discover how exposure to different risk messages and social cues affect intentions to be vaccinated for influenza and rubella in Japan. Identification of differences in vaccination intentions between influenza and rubella could reveal how Japanese people appraise the risks of side effects alongside the benefits of being vaccinated. We designed risk messages around likely concerns about vaccination, including adverse reactions, consequences of remaining unvaccinated and social cues to action. 6 voluntary in Japan; however, Japanese immunisation law states that influenza vaccines be generally obtained at individual expense, preferably annually and timed appropriately to provide immunity during the peak influenza season, while the rubella vaccination is strongly recommended and provided free, with resultant immunity persisting for many years.
To minimize bias in respondent risk perception about influenza and rubella, we avoided the peak influenza season. A within-subjects design was used for the analyses; individuals under 20 years of age and medical professionals were excluded from the study. Variables included (1) demographics, (2) information-seeking behaviours regarding healthcare-and vaccination-related risks and (3) responses after receiving different risk information. Demographic questions included gender, age and education level to assess how these strata interpret the risks of vaccination. We also included having/not having a child or children under 20 years of age, based on evidence that having children affects influenza vaccination uptake. 3 We developed risk statements around influenza and rubella vaccinations for the survey guided by the Health Belief Model (HBM). 20 In the HBM, core constructs of perceived susceptibility to and severity of an adverse outcome, perceived benefits of and barriers to adopting a preventive action, cues to action and self-efficacy all shape intentions. If perceived benefits of a preventive action exceed perceived barriers, and if individuals perceive that they are susceptible to an adverse outcome, they are more likely to adopt a recommended preventive health action. In this study, we defined the adverse outcome as side effects of having vaccination, because several articles reported that concerns about side effects of having vaccination would lead to a lack of confidence or negative attitude towards vaccination. 6,[8][9][10] We defined 'severity' as the severity of the side effect, 'susceptibility' as the risk of experiencing infections and 'barrier' to adopt a preventive action as a barrier to having vaccination. If respondents saw vaccination as beneficial after being made aware of minor and severe risks and susceptibility, their intention to receive a vaccination would subsequently increase. Cues to action, such as messages in mass media campaigns, doctors' recommendations and advice from others, can also promote or discourage vaccination intentions. Self-efficacy is the level of a person's confidence in his or her ability to take action 21 and is associated with increased information-seeking behaviours as well as integration and more effective use of information. 22  and an adverse health risk faced by unvaccinated persons (susceptibility). Self-efficacy was assessed through questions inquiring about other health information-seeking behaviours. For cues to action, we asked respondents to consider two hypothetical nonmedical advice statements: (1) being discouraged by one's mother from getting the vaccine and (2) being encouraged to get the vaccine by a friend.

| Survey design
These scenarios were selected based on studies showing that parental advice (especially from one's mother) and peer influences have a strong influence on preventive health behaviours and vaccination uptake among young adults of childbearing age. [23][24][25][26] Peers have been shown to influence decision-making by parents about immunisations for their children. 24,26 It is unclear whether and how maternal influence over health decision-making wanes over the life course in Japan; our study attempted to address this literature gap by exploring whether and how maternal and peer advice influenced adult intentions.
Minor and severe risk explanations were designed by the authors based on information published on the home page of the MHLW, including the National Institute of Infectious Diseases ( Table 1). The information on the MHLW website is based on medical evidence, but worded for laypersons, with some probabilities presented as percentages, while others are qualitative, using phrases such as 'in quite rare cases'. Conveying vaccination risk information in this way seems to be common practice internationally, exemplified by information issued by the Centres for Disease Control and Prevention (CDC) in the United States.
We piloted the survey to measure the reliability of the questionnaire using a test-retest approach, adapting the questions to reach a sufficiently high Cronbach's α (α = .90) with a subgroup of respondents: .89 for the influenza vaccination and .82 for the rubella vaccination.
At baseline and after being presented with each explanation of risk or cue to action, respondents rated their intentions to receive each vaccine along a 6-point Likert scale, where 1 indicated no intention of having a vaccination and 6 indicated strong intention to have a vaccination. The results were dichotomized so that scores of 1, 2 and 3 were converted into 0 (no intention) and scores of 4, 5 and 6 were converted into 1 (intention).

| Data analysis
Respondents' vaccination intentions were examined using multiple logistic regression analyses using stated intentions before being

| Ethical considerations
Participation in this study was voluntary and anonymous; individuals could not be identified by researchers. Respondents were asked if they would like to participate in the online survey, and the questionnaire was distributed only to those who agreed for it to be sent.
Agreeing to receive a questionnaire did not constitute consent to participate. Respondents who entered the survey site were considered to have given agreement and informed consent after they YASUHARA ET AL.
| 2015 clicked to agree to participate and submitted their answers; they were free to opt out at any time. This study was approved by our institution's Ethics Committee (Approval #821).

| Participant profile
The questionnaire was electronically distributed to 2501 individuals. Of these, 1030 respondents (41%) completed the survey, of whom 515 were male and 515 were female. Macromill implemented an automatic cut-off for enrolment in each age group to ensure a 1:1 ratio of males to females and an even distribution among the five age groups: 20s, 30s, 40s, 50s and 60s and older (n = 103 per group). The respondents' mean age was 45

| Influenza vaccination
Intentions to have influenza vaccination (multiple logistic regression analyses) No gender association was found with intention to have the influenza vaccination. Overall, respondents in their 40s and 50s, regardless of gender, did not intend to have the influenza vaccination at baseline; intentions did not change after risk information or advice was presented (p < .05 for all associations; Table 2; Table S1).
As for statistical interactions, females in their 40s had higher influenza vaccination intentions after receiving severe risk information (p = .02, 34 of 103, 33%) and susceptibility information (p = .01, 37 of 103, 36%). Conversely, females 60 and older showed no T A B L E 1 Risk information in the online survey

Influenza vaccination Rubella vaccination
Minor risk 10%-20% of people receiving the influenza vaccination will experience eczema at the injection site. 5%-10% will have eczema all over the body, but it will resolve in 2-3 days One in several thousand people (0.05%) receiving the rubella vaccination will experience severe headache with cramping at the back of the neck, fever, nausea and/or vomiting (aseptic meningitis). For comparison, the allcause rate of aseptic meningitis among those not vaccinated for rubella is 2 in 100 (2%) Severe risk In rare cases, fever, headache, spasms, disturbance of motility and/or consciousness, shock, hives and difficulty breathing may occur several days to 2 weeks after having the influenza vaccination. These side effects may be serious and possibly lifethreatening In rare cases, idiopathic thrombocytopenic purpura may develop, which poses a risk of mild to excessive bruising and bleeding. This condition is associated with an unusually low level of platelets, which aid blood clot. Extremely rarely, platelet levels may fall so low that dangerous internal bleeding occurs, though effective treatments are available   Table S2).

Comparison of intentions between baseline and after provision of different risk information and advice (McNemar tests)
Intentions to have the influenza vaccination significantly increased from baseline after being provided minor risk information (9% increase, n = 41 more than baseline), susceptibility information (10% increase, n = 42 more than baseline) and a friend's comment (8% increase, n = 34 more than baseline; p < .01 for all associations). Even severe risk information and a mother's discouragement did not influence the respondents' vaccination intentions (Table S3).

| Rubella vaccination
Intentions to have rubella vaccination (multiple logistic regression analyses) Overall, gender was not associated with intention to receive the rubella vaccination. However, females (n = 515) were significantly more likely to refuse the rubella vaccination after being hypothetically discouraged by their mother (p < .01, 76%, n = 390; Table S1). Regarding statistical interactions, females in their 30s (n = 103) and 40s (n = 103) were significantly more likely than other age groups to intend to have the vaccination at baseline (74%, n = 77 and 59%, n = 61, respectively) and after any risk information was provided (p < .05 for all associations). However, females 50 and older (n = 206) were statistically significantly more likely than other groups to refuse the rubella vaccination after a mother's discouragement (83%, n = 172; both, p < .05). with children, 9% increase among those in their 40s with children, 4% increase among highly educated respondents with children and 6% increase among highly educated respondents in their 40s; p < .05 for all associations; Table S4).

Comparison of intentions between baseline and after provision of different risk information and advice (McNemar tests)
A number of inputs significantly decreased intentions to have the rubella vaccination from baseline, including severe risk information (7% decrease, n = 45 fewer than baseline), hypothetical maternal discouragement (56% decrease, n = 372 fewer) and even a friend's hypothetical encouragement to have the rubella vaccination (36% decrease, n = 242 fewer; p < .01 for all associations). Hearing a mother's discouragement first seemed to influence subsequent responses to a friend's encouragement. Therefore, a McNemar test was conducted on the results to assess the relationship between the mother's discouragement and the friend's encouragement. There was a significant increase in vaccination intention when the friend's encouragement was provided (44% increase, n = 130 more than after mother's discouragement; p < .01; Table S3).

| Comparison of information-seeking behaviours with vaccination intentions
For our measure of self-efficacy, approximately half of the respondents indicated that they engage in each of the health information-seeking behaviours surveyed. Over half of the respondents (n = 577, 56%) stated that they would compare the advantages and disadvantages of a medical procedure if treatment were required; 49% (n = 509) reported reading the warning labels on overthe-counter flu medication, 49% (n = 508) stated that they researched the risks of having an influenza vaccination and 62% (n = 643) stated that they researched the risks of having a rubella vaccination. Elderly individuals, as well as females more generally, were statistically significantly more likely to seek vaccination information and assess the risks and benefits (Table S5).

Influenza vaccination
Researching the risks of medical procedures was significantly related  Table S6).

Rubella vaccination
Comparing the advantages and disadvantages of having medical procedures was significantly related to intentions of having the rubella vaccination at baseline (38%, n = 394), after being provided minor (39%, n = 399) and severe risk information (36%, n = 370) and a mother's discouragement (18%, n = 189; p < .05 for all associations).
Those who had researched the risks of the rubella vaccination had significantly higher vaccination intentions even after being provided a mother's discouragement (p = .02, 21%, n = 213; Table S6).

| Demographic aspects of vaccine hesitancy
The influenza vaccination is required annually, at personal expense and is self-protective, particularly for the very young and elderly, while the rubella vaccination is needed much less frequently and offers direct, long-lasting benefits primarily to women of childbearing age and children. Despite these differences, older respondents ex- considerable barrier to efforts by the Japanese government to increase uptake and is of public health concern, as the elderly are at a much higher risk of serious illness and death from influenza if unvaccinated, and can spread rubella infection. 13,18 Surprisingly, this demographic group also reported being proactive in obtaining vaccination information and considering its benefits, but despite this apparent self-efficacy, they were still impervious to vaccination uptake.
Those in their 30s and 40s-particularly females, those with children and highly educated respondents-were more likely to express influenza vaccination intentions after receiving information on risk (even severe risk) and susceptibility. It is possible that this de-

| Peer and family influences
Remarkably, even for older generations whose mothers are likely infirm or deceased, maternal influence was stronger than any of the medical information provided about the rubella vaccination. Japanese YASUHARA ET AL.
| 2019 culture and family patterns may influence these responses. In East Asian cultures including Japan, people tend to respect the opinions of their elders because of their life experience 27 and also view themselves as interdependent with others in specific contexts. 28 Respecting one's elders still influences older generations: knowing that one's mother had a negative opinion of vaccination could decrease vaccination intention. However, considering that we did not observe any maternal influence on influenza intention among older age groups, maternal discouragement might instead provide a convenient excuse for refusal of the rubella vaccination, which would lead to vaccine hesitancy due to complacency. [10][11][12] Hypothetical advice from one's mother was intentionally based on a potentially unreliable source (her memory) as well as medically incorrect information. Regardless, a sizeable proportion of respondents of all ages reported that this advice would decrease their intention to receive the rubella vaccination. Respondents who were influenced by maternal advice appeared to accept vaccination information unquestioningly. A previous study has reported that parental influence, especially maternal influence, affects young adults' decision-making about preventive vaccinations for women's diseases. 25 Focusing on increasing support for rubella vaccination among mothers, especially mothers of grown children, may boost vaccination intention in those of reproductive age, but further research is required.
Peer influence on younger generations was not particularly evident in this study, as peer encouragement did not seem to increase vaccination intention in younger cohorts as it did among the middleaged with children at home, who are more likely to be exposed to information and opinions through school activities through their children and parental meetings at school. Peer influence seemed to have no effect on rubella vaccination intentions. Those with high information-seeking behaviours generally had higher vaccination intentions, especially for influenza. An exception was the elderly, who had high health information-seeking behaviour, but lower vaccination uptake. High self-efficacy among the elderly appeared to be influenced or reinforced by peers; therefore, providing information and vaccinations in gathering places like existing community resources for seniors might enhance vaccine uptake. 29

| Implications for vaccination campaigns in Japan
In Japan, a high-income nation with a well-developed vaccination programme, vaccine hesitancy may be driven by mistaken impressions about vaccinations, some of which may be driven by misinformation provided by peers or on the Internet; however, poor messaging, changing recommendations and outreach by vaccination campaigns may also play a role. [3][4][5]13 The Japanese Health Depart-

| Limitations
Use of an Internet survey excluded individuals who do not have online access or do not use computers or smartphones, potentially under-representing the very poor and very old. The survey relied on self-report, which may be unreliable and could include some social desirability bias (e.g., respondents may have over-reported intention to be vaccinated before receiving any vaccine information, which would diminish the true effect of providing risk information). Using a private marketing firm's database for the study population relied on their roster of registered participants, which may introduce some selection bias. However, the vast majority of Japanese people are highly computer literate, and the large number of responses across Japan indicates that geographic reach was good.
Other limitations pertain to study design and analysis. Due to changing policies and recommendations around vaccination, the proportions of unvaccinated individuals are not evenly distributed by age group in Japan, particularly for rubella, and some respondents probably had had adult rubella vaccination, which may have decreased the observed effect of the risk information provided. The way in which different types of information were provided (minor and severe risks, susceptibility, mother's and friend's advice) may have created an order or cumulative effect. Additionally, our questionnaire design made it impossible to ascertain whether the impact of advice from friends and family was attributable to the content of the advice or the relationship with the person giving the advice.
However, within-subjects designs have two advantages: Higher statistical power and a lower probability of failing to detect a true difference.
Subjective expressions such as 'the severity is high' and 'the probability of the outcome is so low' may have been understood differently by different respondents according to their risk tolerances and subjective interpretations of these phrases. Using congenital rubella syndrome as the example of susceptibility for the rubella vaccination may have skewed favourable intentions towards women of reproductive age. However, these are common ways to communicate vaccination risk information to the public used by major national healthcare organisations, such as the CDC and MHLW.
Additional research is needed to evaluate how the citizens of Japan and other countries perceive these messages.
Our analytical decision to collapse our 6-point Likert scales to a dichotomous scale introduced potential bias, but is a common approach in social research to allow for analysis of a discrete outcome of interest (in this case, vaccination intention or not). As Japanese people, like people from other East Asian nations, tend to choose the middle choices in questionnaires, 30 dichotomization was an attempt to allow respondents to feel comfortable with their response while capturing meaningful responses.

| CONCLUSION
Messages and information to increase vaccine uptake in Japan and elsewhere must target specific demographics with tailored approaches that target the main factors underlying vaccine hesitancy for specific vaccines and for each sociodemographic group, particularly the elderly, who show higher vaccine hesitancy than other groups despite higher susceptibility to many vaccine-preventable diseases. Risk information about severe side effects tends to decrease vaccine intention and should be delivered carefully and in the proper context (e.g., alongside benefits and susceptibility) to groups most concerned about these risks. Messages to motivate individuals to receive vaccines that may not benefit them directly, such as for rubella vaccine promotion among men and older individuals, will require inventive approaches, possibly by appealing to the sense of shared social obligation that animates Japanese culture.