Page 57 - Preventing pertussis in early infancy - Visser

Chapter 4
Quantitative study parents
(LimeSurvey) or a paper questionnaire. Data was processed in SPSS for MS Windows (version 20). All finished online questionnaires had no missing items, as this was a requirement for completion. None of the returned paper questionnaires were excluded. Uncompleted items in the questionnaires were entered as ‘missing’.
Data analysis
First, descriptive analyses on intention, determinants of intention, and determinants of attitude were performed (frequencies). Those measures for which more than 90% of the respondents filled in the same answering option were excluded from further analysis, because of low variability. Since intention showed a non-normal distribution (skewness - 1.43, SE 0.15, kurtosis 1.66, SE 0.29), the intention measure was dichotomized (≥ 5.0 positive intention, < 5.0 negative intention). Univariate and multivariate logistic regression analyses were used to assess the association between intention and its personal and psychosocial determinants. The Nagelkerke index was chosen to express a pseudo R2, in order to approximate the explained variance of the multivariate logistic model.
To assess the contribution of the attitudinal beliefs in explaining attitude, a univariate and multivariate linear regression analysis was used. For the multivariate linear model the R2 was used to express the explained variance. Those determinants that showed a significant bivariate association (p < 0.3) with the outcome variable (intention or attitude) in the univariate analyses were entered simultaneously in the multivariate model to assess their unique contribution to the explanation of the outcome variable. A final multivariate regression model was built by backwards eliminating the non-significant determinants, until only significant determinants (p < 0.05) remained.
Ethical review
The Medical Ethics Committee of district Arnhem – Nijmegen assessed the study and concluded that the study was exempted from their approval (reference number: 2010/475).
Results Population
The questionnaire was returned by 282 parents, of whom 185 were mothers (65.6 %) and 97 were partners (34.4 %; 94 males and two females) (Table 1). Most of the respondents completed and returned the paper questionnaire (243), only some used the possibility to fill in the questionnaire online (39). Respondents to the paper and online questionnaire did not differ on gender, education, income or intention to accept a pertussis cocooning vaccination. About half of the parents were highly educated (51.8%) (compared to 45% for the total Dutch population between 25 and 35 years of age (Statistics Netherlands 2016)), and 45.6% had a household income above average (compared to 47% for all Dutch households (Statistics Netherlands 2015)). A specific philosophical background (e.g. religion, homeopathy, natural cure or anthroposophy) was reported by 6.5% of the parents. Most
56
55