Article Text

Original research
Parental preference for influenza vaccine for children in China: a discrete choice experiment
  1. Shunping Li1,2,
  2. Tiantian Gong1,2,
  3. Gang Chen3,
  4. Ping Liu1,2,
  5. Xiaozhen Lai4,
  6. Hongguo Rong5,6,
  7. Xiaochen Ma5,
  8. Zhiyuan Hou7,
  9. Hai Fang5,8
  1. 1Centre for Health Management and Policy Research, School of Public Health, Shandong University, Jinan, China
  2. 2NHC Key Lab of Health Economics and Policy Research, Shandong University, Jinan, China
  3. 3Centre for Health Economics, Monash Business School, Clayton, Victoria, Australia
  4. 4School of Public Health, Peking University, Beijing, China
  5. 5China Center for Health Development Studies, Peking University, Beijing, China
  6. 6Institute for Excellence in Evidence-based Chinese Medicine, Beijing University of Chinese Medicine, Beijing, China
  7. 7School of Public Health, Fudan University, Shanghai, China
  8. 8Health Science Center-Chinese Center for Disease Control and Prevention Joint Center for Vaccine Economics, Peking University, Beijing, China
  1. Correspondence to Hai Fang; hfang{at}


Objectives To investigate what factors affect parents’ influenza vaccination preference for their children and whether there exists preference heterogeneity among respondents in China.

Design Cross-sectional study. A discrete choice experiment was conducted. Five attributes were identified based on literature review and qualitative interviews, including protection rate, duration of vaccine-induced protection, risk of serious side effects, location of manufacturer and out-of-pocket cost.

Setting Multistage sampling design was used. According to geographical location and the level of economic development, 10 provinces in China were selected, and the survey was conducted at community healthcare centres or stations.

Participants Parents with at least one child aged between 6 months and 5 years old were recruited and the survey was conducted via a face-to-face interview in 2019. In total, 600 parents completed the survey, and 449 who passed the internal consistency test were included in the main analysis.

Main outcomes and measures A mixed logit model was used to estimate factors affecting parents’ preference to vaccinate their children. In addition, sociodemographic characteristics were included to explore the preference heterogeneity.

Results In general, respondents preferred to vaccinate their children. All attributes were statistically significant and among them, the risk of severe side effects was the most important attribute, followed by the protection rate and duration of vaccine-induced protection. Contrary to our initial expectation, respondents have a stronger preference for the domestic than the imported vaccine. Some preference heterogeneity among parents was also found and in particular, parents who were older, or highly educated placed a higher weight on a higher protection rate.

Conclusion Vaccination safety and vaccine effectiveness are the two most important characteristics that influenced parents’ decision to vaccinate against influenza for their children in China. Results from this study will facilitate future policy implementations to improve vaccination uptake rates.

  • discrete choice experiment
  • influenza vaccine
  • children
  • parental preference
  • China

Data availability statement

Data are available upon reasonable request. Data are available upon request. To obtain data, please email Shunping Li (

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Strengths and limitations of this study

  • This is the first nationwide study to explore parental preference for influenza vaccine for their children using discrete choice experiments (DCE) in mainland China.

  • The experimental design and data analysis were conducted following the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) Conjoint Analysis Task Forces.

  • The external validity of DCE results cannot be testified, which is a common limitation of most DCE studies.

  • We did not differentiate barriers and facilitators among factors associated with the vaccination decision.


Influenza is an acute respiratory infection caused by influenza viruses and can result in substantial mortality.1 Among four types of influenza viruses, influenza A and influenza B can create epidemics.2 According to the WHO, annual epidemics of influenza can lead to 3–5 million cases of severe illness and about 290 000–650 000 respiratory deaths worldwide.2 In China, up to 88 100 seasonal influenza-associated respiratory excess deaths occurred each year from 2010 to 2014, accounting for 8.2% of deaths from respiratory diseases.3 All age groups can be affected by influenza, however, the prevalence of influenza among children under 48 months was highest (up to 33%).4 In central China, children under 5 years old accounted for 69% of inpatients owing to influenza-associated severe acute respiratory infections.5 The economic burden of influenza-associated outpatient and inpatient healthcare utilisation is substantial in China, particularly for young children.6 7

It is cost-effective or cost-saving to vaccinate against influenza.8 9 In China, two types of influenza vaccines have been licensed, including trivalent inactivated influenza vaccine (IIV) and tetravalent IIV; whereas the live attenuated influenza vaccine has not been approved.10 The vaccination rate in children 6 months to 18 years of age was 49% in the USA during the 2010–2011 influenza seasons,11 the vaccination rates in 2010 and 2011 in the Israeli paediatric population were 21.4% for children from 6 months to 2 years of age and 16.1% for children from 2 to 5 years of age.12 However, the vaccination coverage among children aged under 5 years was stable at a low level of 3%–4% from 2015 to 2019 in China.13 It is important to understand parental attitudes and preferences for vaccines and to explore key factors associated with parents’ decisions to vaccinate their children.

Identifying facilitators and barriers to influenza vaccination would be important to promote vaccination. A systematic review revealed that several facilitators for parents to accept influenza vaccination were belief in vaccine efficacy and influenza severity and susceptibility, perception of advantages of the school setting (eg, it is very convenient to vaccinate children in school) and trust in vaccines.14 In China, the barriers were complex. One study surveyed various populations and found that the most common reason for being unvaccinated in the influenza vaccine was worrying about the side effects.15 Another study that targeted at quadrivalent influenza vaccine for school-aged children showed that the pivotal barriers hindering parents from having their children vaccinated were fear of side effects and no perceived susceptibility.16 On the contrary, one study indicated that perceived severity and knowledge about influenza were not independently significantly associated with uptake.17

Children aged 6–59 months, recommended routine influenza vaccination strongly by WHO,2 are also among the priority vaccination groups stated by the Chinese Center for Disease Control and Prevention (CDC).10 However, the influenza vaccine for children has not been covered by China’s National Immunisation Programme. The decision to vaccinate against influenza for children mostly depends on parents’ views and preferences. Consequently, it is crucial to understand the factors affecting parents’ decisions to vaccinate their children which will help the government to implement more targeted vaccination promotion strategies, so as to improve the vaccination rate of influenza vaccine for the nation.

As a stated preference method, discrete choice experiments (DCEs) can simulate different hypothetical vaccination scenarios and elicit respondents’ preferences. DCEs have been widely used to estimate preference for vaccines,18 such as human papillomavirus, influenza and hypothetical vaccines.19–21 Although there exist some DCE studies on vaccines in China, respondents normally came from one particular province.22 23 This is the first nationwide DCE study on vaccination that aims to recruit respondents by involving parents from 10 provinces to understand the preference for influenza vaccination. This study aimed to address two research questions: (i) to elicit the preference of parents when choosing influenza vaccine for their children; (ii) to investigate whether there exists preference heterogeneity among respondents.


DCEs are increasingly used in health economics to identify and evaluate the participants’ preferences.24 DCEs can also be used to estimate participants’ willingness to pay as well as to predict programme uptake rates given a set of goods or services characteristics.25 26 In the DCE, a vaccine profile can be described by a series of attributes and their corresponding levels, and under the random utility theory, respondents choose the option with the highest utility from the alternatives presented.27 The DCE design and analysis were conducted following the checklist and reports of the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) Conjoint Analysis Task Forces.28–30

Survey design

Based on previously published literature,18 20 31 12 attributes were identified initially. To assess the appropriateness of these potential attributes and their levels and to further narrow down the number of attributes, four experts on vaccination were interviewed face-to-face in Jinan Maternity and Childcare Hospital. Two focus groups (n=12) were also conducted. One focus group included four parents only, and the other contained one vaccine expert, three parents and four health economics/DCE experts. They were asked to review and rank the list of attributes. Finally, five attributes were selected for this study (table 1). The attribute levels were also decided based on the influenza vaccine instructions and clinical randomised controlled trials evidence. They have been reviewed by experts and discussed in the focus group interviews.

Table 1

Attributes and attributes levels for DCE choice questions

A D-efficient design was developed using Ngene Software (, which yielded 60 choice sets that were further divided into six blocks to reduce respondents’ cognitive burden. To check for internal consistency, one choice set in each block was duplicated. Each respondent received one block randomly and was asked to answer 11 choice sets. For those who failed the consistency test, their data were excluded from the main analysis. Before completing DCE questions, respondents were also asked to rate the importance of five attributes.

Given vaccination is a voluntary decision, an opt-out option was included and implemented by using a two-stage response design to maximise the information gained from the respondents.32 In the first stage, the respondents were forced to choose between two hypothetical vaccinations. Then, they were asked to confirm whether they would vaccinate their preferred option from the first stage for their children.

In addition to DCE questions (which were presented in a hardcopy questionnaire), sociodemographic characteristics of respondents and their children were collected using an iPad. A pilot was conducted among 15 parents in Beijing and Jinan in July 2019 to examine the acceptability, comprehensibility and validity. A few modifications were implemented based on the feedback from the pilot. An example of a final choice set was shown in figure 1.

Figure 1

An example of discrete choice question (translated version).

Study population and data collection

This DCE, as well as a related DCE on parental preference on vaccination for children in general,33 were embedded in a nationwide project on Strategies of Influenza Vaccination in China study.34 A multistage sampling method was adopted to elicit parental values and preferences for influenza vaccines across the country, the details of which has also been reported elsewhere.33 Initially, 10 provinces/municipalities were selected according to geographical location and the level of economic development, including the eastern region (Shandong and Shanghai), western region (Gansu and Chongqing), southern region (Yunnan and Guangdong), northern region (Beijing and Jilin), middle region (Henan and Jiangxi), which can be seen in figure 2. Next, except for three municipalities (Beijing, Shanghai and Chongqing), in each of the other seven provinces, one provincial capital and one non-provincial-capital city were chosen. A district and a county were randomly selected from each city. Finally, 30 parents with at least one child aged between 6 months and 5 years old were randomly recruited from each community healthcare centre or station.

Figure 2

Provinces/municipalities selected in China.

According to a rule of thumb suggested by Orme,35 a sample size of 75 (500×3/2×10=75) would be desirable for the main effects model based on the number of analysis cells, alternatives and choice sets. We aimed to recruit a minimum of 100 respondents in each region.26 36 Hence, we intended to survey 60 parents in each province and 120 parents in each region.

The anonymous survey was administered between August and October 2019. Data were collected through one-by-one face-to-face interviews with parents waiting for routine vaccination for their children or remaining for observation after routine vaccination. The vaccination rates for routine vaccines, such as DTaP and HepB, were more than 95% in China.37 so the sample bias for participants recruited from the vaccination sites was very limited. Before enrolling in the survey, respondents were informed about the purpose and content of the survey by interviewers who have been trained by the research team. Electronic written consent was obtained from all respondents.

Statistical analysis

Responses to the hardcopy DCE questionnaire were double-entered into a database set up by the EpiData V.3.1 software and then matched with other sociodemographic characteristics obtained from the iPad for statistical analyses. In cases where the number of missing DCE responses was more than two tasks or the majority of sociodemographic data was missing, respondents were excluded from the final analysis.

A mixed logit model was employed to analyse DCE data which takes into account potential preference heterogeneity.38 The utility function can be written as follows:

Embedded Image

Where Uijt is the utility that respondent i derives from choosing alternative j in the choice set t, Xijt is a vector representing the levels of the attributes, β is a vector of coefficients corresponding to attribute levels and εijt is a random error term. The cost attribute was treated as a continuous variable, while other attributes were dummy coded. In a mixed logit model, coefficients of attribute levels are commonly assumed to follow a normal distribution to account for preference heterogeneity, that is, β is composed of a mean coefficient as well as a SD. A significant positive (negative) coefficient represents a positive (negative) preference for an attribute level. The importance of an attribute can be calculated through the difference of level coefficients in the same attribute. Therefore, the relative importance of attributes can be estimated by comparing the utility range of each attribute.39

We further examined whether the elicited preferences varied by particular sociodemographic characteristics. Finally, vaccination update rates were predicted to facilitate the interpretation of DCE results to decision-makers. Descriptive analyses including Student’s t-test, χ2 test and Wilcoxon rank-sum test were adopted to compare means and proportions between subgroups, respectively. All statistical analyses were conducted using Stata V.12.1 software. The mixed models were estimated by simulated maximum likelihood using the Stata command developed by Hole40 and 2000 random draws were used to achieve stability.

Patient and public involvement

The study did not involve the patients. The public was involved at the stage of questionnaire design, pretesting and feedback from respondents was incorporated into questionnaire revisions.


A total of 600 parents consented and participated in the survey. Among them, 3 and 18 parents were excluded from the analysis due to missing sociodemographic information and failure in completing the majority of DCE questions, respectively. Among the remaining 579 parents, they had a mean age of 31 years old, most (79%) of them are mothers of children and the mean age of their children was 2 years old. At the time of the survey, 355 (61%) parents were working and 337 (58%) had at least two children. Among DCE responses, 449 (78%) respondents passed the consistent test (ie, duplicated task) and they were treated as the main study sample. There was no significant difference in sociodemographic characteristics between those who passed and who failed the consistent test except for the region (urban vs rural). More details on respondents’ socio-demographic characteristics are presented in table 2.

Table 2

Sociodemographic characteristics of the study population

Importance rating

Figure 3 showed the relative importance of five DCE attributes ranked by respondents prior to the pairwise choice tasks. The most important attribute was the protection rate followed by the risk of severe side effect events, whereas the out-of-pocket cost of the vaccine and duration of vaccine-induced protection were less important.

Figure 3

Importance rating of attributes.

Discrete choice experiment results

The DCE results incorporating the second-stage choices and based on the main study sample are reported in table 3. As a sensitivity analysis, the full sample analysis results are shown in online supplemental table S1, while the analyses on forced-choice responses from the main study sample are presented in online supplemental table S2. All attributes were statistically significant. Overall, similar patterns can be seen in the supplementary material.

Table 3

Mixed logit model results with only main effects

Focusing on table 3, the mixed logit model estimates suggested that the higher the protection rate, the longer the duration of vaccine-induced protection, the lower the risk of severe side effects, the lower the cost, the more likely that parents would be willing to vaccine for their children. Contrary to our initial hypothesis, respondents prefer domestic rather than imported vaccination. Most estimated SD were significant, indicating the existence of preference heterogeneity among parents.

The vaccine with the lowest risk of severe side effects had the highest preference weight when compared with a relatively high risk of severe side effects, followed by the highest protection rate. And the duration of vaccine-induced protection was less important. Reducing the risk of severe side effects from high to low could yield 4.4 (2.626/0.596) times as much as utilities increasing the duration of vaccine-induced protection from 6 to 12 months.

The coefficient of non-vaccination was significantly negative, indicating that on average the parents were more likely to vaccinate their children against influenza regardless of the vaccine profile described by attributes and levels.

To evaluate whether there was a significant difference between parents with various characteristics, a series of interaction terms between respondents’ characteristics and attribute levels were explored and the result was reported in table 4. We found that parents who were beyond 30 years old or lived in urban were more likely to choose vaccination. Highly educated, those beyond 30 years old and those who lived in rural areas placed a higher weight on the highest protection rate. Those who lived in rural areas also had a stronger preference for the lowest risk of severe side effects. Other than what has been reported, we found no significant influence between attribute levels and the working status of parents and the gender of children.

Table 4

Results of mixed logit model with main effects and interactions

Predicted uptake rates for different scenarios

Figure 4 showed the results of predicted probability when changing a particular attribute level based on results reported in table 3. Corresponding to the reference within DCE’s main effect analysis, the scenario was selected as the baseline presented by 70% protection rate, 6-month duration, high risk of severe side effects, domestic and costing CNY150. For the change within an attribute, the decrease in the risk of serious adverse effects from high to low had the largest effect on preference for influenza vaccines, in which the probability of taking that vaccination increased by 86%. For the changes with multiple attributes, the vaccine with an 80% protection rate was preferred to the free one with a 12-month duration. On the other hand, the impact of cost and duration change was small. The most attractive vaccine was ‘⑤+⑥’ one, which has the lowest risk of severe side effects and the highest protection rate.

Figure 4

Simulated probabilities for influenza vaccination under change of a single attribute. notes: the baseline was presented by a 70% protection rate, 6-month duration, high risk of severe side effect, domestic and costing CNY150.


This study has estimated parental preference for vaccinating against influenza for their children. To the best of our knowledge, this is the first nationwide study to explore parental preference for influenza vaccine delivery using DCEs in mainland China. A previous DCE study conducted in Hong Kong Special Administrative Region surveyed the adult to assess the relative effects of different factors on influenza vaccination choices.41

We found that on average respondents from this study preferred vaccination against influenza for their children from the hypothetical vaccination scenarios, which is consistent with other DCE study findings.31 42 The relatively high acceptance was also documented in another survey that aimed to study the knowledge, attitudes, and practices towards the influenza vaccine among young workers in China.43

In general, all the attributes included in our study were statistically significant and preference heterogeneity existed among both observable and non-observable personal characteristics. Among all the attributes, the risk of severe side effects and the protection rate of the vaccine were the top two most important characteristics perceived by parents. Their important roles in the choice for vaccination are in line with other influenza vaccine DCE studies.20 31 Similar findings have also been reported in other vaccines. A DCE study surveying girls’ preference for HPV vaccination reported that respondents preferred low severe side effects44 and other studies found willingness to vaccinate was closely related to vaccine safety and efficacy.42 45 The above findings could suggest that reducing the risk of severe side effects and increasing vaccine effectiveness could be regarded as two universal procedures to effectively achieve higher vaccination coverage.

Somewhat surprising, given the recent Changchun Changsheng vaccine incident, this study found that parents preferred the domestic vaccine to the imported vaccine. In 2017 and 2018, Changchun Changsheng Biotechnology Co., had two consecutive cases of serious violations of the drug production quality management specification, such as fraud in the vaccine production process. It has had a very bad impact on society. However, the same finding was also reported in one recent DCE study conducted in Shanghai, even though there are substantial differences, for example, study population.23 One potential reason for which domestic vaccine was preferred may be that it is thought to be more effective46 and more accessible. And the other is that the regulatory environment is more stringent. Indeed, the government facilitated a public consultation after the incident in 2018,47 and the Standing Committee of the National People’s Congress voted to adopt the first Vaccine Administration Act in 2019, which aimed to tighten vaccine regulation.48

The out-of-pocket cost was found to be less important compared with the other attributes. Based on the calculation of uptake rates, the probability of vaccination was affected slightly by a change in cost. This differs from some previous studies in which cost was found to be an important factor driving preferences.21 39 49 The above results were incomparable for our study due to differences in targeted vaccines. In reality, the out-of-pocket cost of the influenza vaccine is affordable when compared with the household income. For example, the highest out-of-pocket cost of the influenza vaccine made up about 1% of the monthly income in our study. Furthermore, most families in China are willing to spend more for their children,50 and the cost is not a key factor.

When studying the preference heterogeneity, the protection rate has again stood out as a key attribute that those who were older, lived in a rural area or got higher education all placed a higher weight on a higher protection rate. By far influenza vaccine has not been included in the national immunisation programme schedule in China and to improve the vaccination rate in particular for people mentioned above, providing more information about as well as improving the safety and effectiveness of vaccines will be the most important factor.

Consistent with the results of our study, vaccine safety and serious adverse events are repeatedly shown to be a top concern for parents.51 Not only the provision of information to parents or education interventions, but also communication strategies should be focused on for healthcare communicators/practitioners. Communication processes that build rapport and trust are needed. Healthcare providers play a vital part and are often the most trusted sources of vaccine information.52 For the relevant regulatory department, the strict supervision of domestic vaccines should be strengthened to increase parents’ trust in influenza vaccine, to improve the vaccination rate of influenza vaccine for children. Vaccine providers should conduct self-examination and establish good credit. On the premise of improving the safety and effectiveness of influenza vaccines, vaccine manufacturers should pay more attention to publicity and brand building.

The present study had several limitations. First, our study includes 600 respondents recruited from 10 provinces (and among them, 449 of them were included for the main analysis) which maybe not large enough to represent the whole of China. However, we did not find significant regional preference heterogeneity in the analysis. Second, though attributes included in our study were identified and selected through previous literature, interview with experts and focus group discussions, following the recommended procedure, we cannot guarantee that all attributes concerned with parental vaccination choice were included. Third, we did not differentiate barriers and facilitators among factors associated with the vaccine, it may be more useful to distinguish between barriers and facilitators. Finally, similar to most DCE studies, the external validity of DCE results cannot be testified. Nevertheless, the consistency test and importance rating were implemented to confirm DCE’s internal validity.


Vaccinating influenza vaccines is the most effective measure to prevent the prevalence of influenza. Although WHO and the Chinese CDC have recommended the influenza vaccine to the whole population, especially the youth, the vaccination rate is extremely low. This study aimed to investigate national parents’ preference for vaccinating against influenza for their children based on a nationwide sample. Based on a DCE, the study showed that on average parents were more willing to vaccinate their children. Among the five attributes been examined, the risk of severe side effects and protection rate were key drivers of preference among parents in China and preference heterogeneity was found among parents. The findings from this study will shed light on future policy implementation to improve the influenza vaccination rate in China.

Data availability statement

Data are available upon reasonable request. Data are available upon request. To obtain data, please email Shunping Li (

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants. The study received ethical approval from the Peking University Institutional Review Board (IRB00001052-19076). Participants gave informed consent to participate in the study before taking part.


We would like to thank all participants for their contribution to this study.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Contributors Guarantor: SL. Conceptualisation and supervision: SL and HF. Data curation: SL and TG. Formal analysis: TG and PL. Funding acquisition: HF. Methodology: GC, TG and PL. Project administration: SL and GC. Write-original draft: TG. Writing-review and editing: SL, TG, GC, PL, XL, HR, XM, ZH, HF.

  • Funding This work was supported by the National Natural Science Foundation of China (grant number 71774006).

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting or dissemination plans of this research. Refer to the Methods section for further details.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.