Abstract
The recent study by Montejo, Richesson, Padilla, Zychowicz, and Hambley (2017) concluded that a multipronged occupational health education program to promote influenza vaccinations led to higher immunization rates among employees. However, this conclusion was not well supported by study data for two reasons: year-to-year differences in the experimental group and the nature of comparisons with the control group.
The study used a nonexperimental design in which two workplace sites within the same corporation were compared. One site received an intervention of a comprehensive education program with free on-site influenza vaccinations plus gift bags promoted by posters, flyers, badge stickers, and employee group health education. The control site did not receive this intervention. The intervention site had higher immunization rates (37%) than the nonintervention site (27.2%) during the 2015-2016 influenza season. Although this finding was reported as statistically significant, we calculated the effect size for the intervention and found it was only r = 10.4% (p = .015). Effect sizes (r) can range from 0% to 100%, and by Cohen’s criteria, 10% is considered a low effect size (Cohen, 1988). The argument could be made that the intervention had a small effect that did not justify its time and expense.
A second finding in the study was an increase in the influenza vaccination rates after the intervention program was implemented. Employees were immunized at a rate of 37% after the program was implemented during the 2015-2016 season compared with 35% before the program was implemented in the 2014-2015 season. We computed the statistical significance for this difference assuming that the number of employees was the same in the two consecutive seasons and found that the difference was not statistically significant (p = .756). By this measure, we offer the alternative finding that the immunization rate was not significantly higher in the postintervention period compared with the preintervention period.
It should be noted that the above study had many strengths, including an appropriate experimental design for the hypothesis, a large sample size for all groups, an innovative health promotion and education program, focus on a significant occupational health problem, excellent coverage of previous studies, and a well-written report. Nonetheless, due to the statistical considerations described above, we suggest that it is premature at this time to accept the conclusion that the intervention was sufficiently effective to warrant adoption elsewhere.