Logistic regression models were estimated for the binary dependent variables measuring self-reported HBV screening/testing and vaccination behavior to ascertain how important sources of information, and trust in those sources, were when controlling for socio-demographic and health coverage characteristics, and HBV transmission knowledge. The dependent variables (whether respondents were screened for HBV and whether respondents were vaccinated for HBV) only included those respondents who had ever been told that they should be screened or vaccinated for HBV. In other words, those respondents who had never been told they should be screened or vaccinated for HBV (40%) or who had never heard of HBV (26%) prior to completing the survey were not included in this analysis. Though the resulting sample was smaller, it did provide the opportunity to assess the possible connections between screening/testing and vaccination self-reported behavior and other variables.
To construct the logistic regression models, chi-square tests were conducted of the dependent variables (respondent had been screened/tested for HBV, respondent had been vaccinated for HBV) using each set of variables (socio-demographic characteristics, health coverage, sources of health information, trust in health information sources, HBV transmission knowledge). Only those variables that were statistically significant at the P < 0.05 level or better were retained for additional analysis. Pairwise Pearson correlation coefficients were calculated to ensure that the independent variables were not highly correlated. Using this variable screening process, the variables associated with physicians as sources of health information and trust associated with this health source were not retained (i.e., neither of these variables were significant at least at the P < 0.05 level for either of the two dependent variables).
This suggested that while a very large proportion of respondents reported that they received health information from their physicians, and that they trusted this source, this had little explanatory power with respect to HBV screening/testing or vaccination for those respondents who had been informed that they should be screened or vaccinated for HBV. As the number of observations in each of the models was relatively small (N = 63 for the model for HBV screening/testing, and N = 78 for HBV vaccination), the results should be viewed as suggestive rather than conclusive.
Table shows the results for a logistic regression model of HBV screening/testing for those respondents who had been informed that they should be screened for HBV. Two sets of variables were retained for the model after chi-square and Pearson correlation tests were conducted: socio-demographic and medical coverage characteristics, and information sources. Two variables were statistically significant at least at the P < 0.05 level in explaining the variation in having been screened/tested for HBV: being currently employed and getting health and health care information from friends. The results indicated that full-time employed respondents who have been informed that they should be screened/tested for HBV were over 6 times more likely to have been screened or tested for HBV compared to respondents who had been informed that they should be screened/tested for HBV but who did not report being full time workers (e.g., part time workers, retired, unemployed). In terms of health information sources, respondents who reported that they received health and health care information from their friends and who had been informed that they should be screened/tested for HBV were over 7 times more likely to also report that they had been screened or tested for HBV compared to respondents who had been informed that they should be screened/tested for HBV, but who did not report that they had received information about health or health care from their friends.
Logistic regression results for Hepatitis B screening/testing (N = 63)
Table shows the results of the multivariate logistic regression model for the dependent variable measuring self-reports of vaccination for HBV (only including those respondents who reported that they had been informed that they should be vaccinated). The only demographic or medical coverage variable that was a significant predictor of whether respondents reported that they had been vaccinated for HBV was whether the respondent reported being covered by Medicare; those who reported being covered by Medicare were 0.02 times as likely as respondents who did not report that they were covered by Medicare to also report that they had been vaccinated for HBV.3
Logistic regression results for Hepatitis B vaccination (N = 78)
There is one HBV transmission knowledge question that was significant at least at the P < 0.05 level; responding that Hepatitis B could be transmitted by eating unclean food (an incorrect response) was associated with also reporting being vaccinated for HBV. In other words, contrary to most interventions that might presume that more and better knowledge leads to protective and health promoting behavior, this result suggested that not understanding how HBV is transmitted (i.e., confusing HBV with Hepatitis A) was related to higher likelihood of being vaccinated for HBV. Those who answered this question incorrectly were over 8 times more likely to report that they had been vaccinated for HBV compared to respondents who answered this question correctly.
Similar to the model presented previously on self-reported HBV screening/testing behavior, being employed again appeared to be important in explaining the variation in self-reported vaccination for HBV, however, in this model, employment had a negative association. Obtaining information about health and health care from employee assistance programs was negatively associated with HBV vaccination for those respondents who had been informed that they should be vaccinated for HBV.