Excluding numeric side-effect information lowers vaccine intentions
Shoots-Reinhard, B., Lawrence, E. R., Schulkin, J., & Peters, E. (2022). Excluding numeric side-effect information produces lower vaccine intentions. Vaccine. doi: https://doi.org/10.1016/j.vaccine.2022.06.001
If you use the following link before July 30, 2022, you can read a copy of the article for free: https://authors.elsevier.com/a/1fDkP,60n7kMYm
In a public health crisis like COVID-19, vaccine uptake can have striking effects on disease outcomes. As one U.S. example, during the summer of 2021, unvaccinated Americans died at 11 times the rate of vaccinated Americans.
Despite these striking statistics, many Americans remain concerned about vaccine safety. Despite this concern, communication about side effects does not often inform people about how likely the side effects are. However, research from our lab and others has demonstrated that providing likelihoods for medications (e.g., “10%)—instead of verbal likelihood labels (e.g., “common”)—reduces overestimation of side effect likelihood and increases medication uptake. We suspected similar effects may occur with vaccines.
In a diverse online convenience sample (N = 595), we told people to imagine their doctor recommended a vaccine and gave them a list of side effects of the hypothetical vaccine. Some participants only saw the list of side effects (right column below), some saw the list + verbal likelihood labels (right two columns), some saw the list + numeric likelihood information (left and right columns), and some saw the list with both verbal and numeric information (all three columns):
As we suspected, providing numeric information increased vaccine intentions—70% of those who received numeric information were predicted to be moderately or extremely likely to vaccinate compared to only 54% of those who did not receive numeric information (p<.001), controlling for age, gender, race, education, and political ideology.
We found even more striking effects for people who indicated in an earlier survey that they were hesitant to get a vaccine a doctor recommended. For these hesitant participants, there appeared to be a benefit to giving both numeric and verbal risk information (Figure 1). Among the vaccine hesitant, 43% were predicted to be moderately or extremely likely to get the vaccine when provided numeric information and verbal labels compared to only 24% in the list only (i.e., standard-of-care) group, controlling for covariates.
Figure 1. Vaccine-hesitant participants (on the right) were more willing to receive the recommended vaccine when provide numeric and verbal likelihood information (scale: 1=not likely to get vaccine to 6=extremely likely to get vaccine).
Additional analyses suggested that: 1) the numeric information reduced the number of side effect likelihoods that were overestimated and 2) the combination of numeric and verbal likelihood information helped people realize that most side effects were not serious; it also reduced concern about rare, very serious blood clotting.
We suspect that a switch from the standard list of side effects to more detailed information including both numeric information and verbal labels could help increase COVID-19 vaccination rates. For example, in October 2020, about 120 million people were hesitant to vaccinate. If we assume an effect on actual vaccination half the size we found on intentions, then we estimate that about 11 million more Americans (3% more of the population) would have been convinced to vaccinate.
This project was supported by the United States National Science Foundation (SES-2022478, SES-2017651, and SES-2029857), United States Health Resources and Services Administration (UA6MC19010), and a First Year Research Experience Grant from the University of Oregon.