The association between protection motivation and hepatitis b vaccination intention among migrant workers in Tianjin, China: a cross-sectional study

We surveyed migrant workers, who had lived for 6 months in the Binhai New District in Tianjin, but did not possess a Binhai New District household registration hukou. Given the high mobility and the lack of local registration of migrants, sampling “hidden populations”, such as migrant workers, is difficult [24]. Employing the convenience sampling method, used, for example, in a previous study of migrant children’s immunization in Beijing [25], we surveyed three migrant worker-dense industries between July and October in 2017: (1) manufacturing (including workers from food and electronic manufacturing companies); (2) retail (including workers in supermarket and electrical appliances selling markets); and (3) services (including taxi drivers, hotel and restaurant workers). Following previous studies on sample size selection [26, 27], we calculated that at least 385 migrant workers should be surveyed based on the expected willingness to participate of 50% and a tolerance error of 0.1. Considering 10% dropped samples and balancing the sample size in each industry, questionnaires were administered to 150 workers in each industry who met the inclusion criteria of living at least 6 months in the Binhai New District in Tianjin, but did not possess a Binhai New District household registration hukou, and self-identified as having no HBV vaccination history. After excluding unqualified and incomplete questionnaires, our sample consisted of 406 migrant workers, 133 from the retail, 119 from the manufacturing and 154 from the services industry, with a response rate of 90.2%. We divided migrants into Tianjin migrants, with a non-Binhai New District hukou, but a Tianjin hukou, and non-Tianjin migrants, with a non-Binhai New District and non-Tianjin hukou.

A pilot survey was performed both to train postgraduate students from Tianjin University of Traditional Chinese Medicine to conduct the survey and to test the questionnaire for its comprehension. Supported by the Binhai local health authority, we contacted the principals in selected companies and factories to facilitate the survey and to ensure a high response rate. Data on participants’ socio-demographic characteristics, migrant-industry characteristics, PMT measurement items and HBV vaccination intentions were obtained in face-to-face interviews, where all participants were volunteers and informed about their right not to answer questions. The intention to take the HBV vaccine was measured by a self-report question: “Will you take the HBV vaccine in the future?” with a binary yes or no response. The vaccination intention rate was 67.2%.

Based on previous studies and migrant workers’ characteristics, we designed items for each PMT construct using a 5-point Likert scale ranging from 1 (strongly disagree) to 5 (strongly agree). The severity of HBV was measured by the migrant workers’ cognition on clinical symptoms, the costs for the individual and their family, and social discrimination. Vulnerability refers to the migrant’s perceived infection probability compared with local workers. Similar to previous studies, response efficacy was measured by the migrant workers’ perception on safety and effectiveness of HBV vaccine. Self-efficacy was measured by migrants’ attitude to the advice of nearby people, the work organization, family, friends or doctors on vaccinations [28]. Response costs are the migrant workers’ realized barriers facing the individual to adopt the protective behavior, measured by the vaccination information and worries about the side effects of the vaccine, and the price of protective behavior, measured by the cost of the HBV vaccine.

As set out in Table 1, factor analysis and Cronbach’s alpha were employed to test the validity and reliability of the PMT measurement tool. As PMT assumes a close relationship between some of the constructs, an oblique rotation specially designed to maximize the correlation among factors was selected to extract the factors [20]. Using the minimum eigenvalue criterion of 1.0, Table 1 displays the PMT factors generated by producing a factor score on five dimensions, comprising severity, vulnerability, response efficacy, self-efficacy and response costs. Each respondent’s PMT factor scores were dichotomized by the cutoff point of their mean value: high (min ≤ factor score ≤ mean) and high (mean<factor score ≤ max).

The covariates included socio-demographic and migrant-industry characteristics of the migrant workers. Socio-demographic characteristics consisted of age, sex, education level, marital status, self-rated health and income group. Education level was divided into three groups: low (below senior high school), medium (senior high school or equivalent) and high (above senior high school). Marital status included two categories: married and unmarried (including the divorced and widow). Self-rated health was measured by the respondents’ self-assessment of their current health compared with those who were in the same age cohort and was divided into three groups: poor, medium and good. The income group variable was based on tertiles of the respondent’s average monthly income. Migrant-industry characteristics included migrant location (non-Tianjin migrant or Tianjin migrant); whether the migrant was accompanied by family or not; and industry groups (manufacturing, retail and services).

We estimated the frequency of migrant workers’ intention to obtain the HBV vaccination. Severity factor, vulnerability factor, response efficacy factor, self-efficacy factor and response costs factor were produced by principal component factor analysis. Bivariate analysis of PMT factors and covariates with the dependent variable were conducted by logistic regression modeling. As set out in Models 1–3 below, socio-demographic characteristics variables, migrant-industry variables and PMT factors were added into the binary logistic regression model, where the dependent variable was the intention to HBV vaccinate.

Model 1: f¹(P) = α¹ + β¹₁X_{Socio−demographic characteristics} + ε¹

Model 2: f²(P) = α² + β²₁X_{Socio−demographic characteristics} + β²₂X_{Migrant−industry characteristics} + ε²

Model 3: f³(P) = α³ + β³₁X_{Socio−demographic characteristics} + β³₂X_{Migrant−industry characteristics} + β³₃X_PMT factors + ε³

Maladaptive response rewards were not included in our PMT model. In some previous vaccination studies, maladaptive response rewards were measured by saving money or time, avoiding the side effects of the vaccine and acquiring natural immunity to subsequent infection [31, 32]. Since HBV is incurable, acquiring natural immunity does not exist, and saving expenses and worries about the side effects have been included in response costs, so maladaptive response rewards may not be a serious omission in our PMT model. However, maladaptive response rewards warrants empirical study in future HBV research. Second, our study did not assess vaccination behaviour directly. While it has been shown that vaccination behavior in empirical studies can be predicted by previous intention in a wide range of contexts [33, 34], future studies should include vaccination behavior directly. Third, as a result of convenience sampling, our results and findings need confirmation through studies of other migrant worker populations, regions and industries. Finally, data on Tianjin’s migrant workers was collected through a convenience sample, which may limit the external generalisation of our results.