Delusion-proneness predicts COVID-19 vaccination behavior

Abstract

Introduction: Vaccination-related conspiracy ideation is related to reduced compliance with public health advice globally. Such beliefs have previously been linked to the delusion-proneness trait. However, it is not known how this extends to getting vaccinated.

Methods: Here, we examined how delusion-proneness, as assessed by Peters et al. Delusions Inventory (PDI), is associated with COVID-19 vaccination in a sample of 273 subjects. We also examined whether delusion-proneness predicted the time to get vaccinated, after the vaccine became available.

Results: Unvaccinated subjects were more delusion-prone than vaccinated subjects (W=2225.5, p<0.001, effect-size=0.27). Among vaccinated subjects, higher delusion-proneness was related to longer time to get vaccinated ( ${\text{r}}_{\text{s}}$ =0.27, p<0.001). These effects remained after adjusting for anxiety, ADHD, and ASD (Autism Spectrum Disorder) traits as well as for psychiatric diagnoses and sex. Path analyses indicated that the effect of delusion-proneness on vaccination rate was strongly mediated through COVID-19 conspiracy ideation, suggesting that delusion prone individuals first develop specific delusion-like ideas regarding vaccination, which then delays vaccination. An exploratory analysis of written text by subjects instructed to explain why they had vaccinated or not, revealed a difference in reasoning between the groups. Unvaccinated individuals were primarily motivated by concerns about personal safety and potential side effects, while vaccinated individuals stated a desire to protect themselves and others as the primary reasons to get vaccinated.

Discussion: Our results suggest that delusion-proneness is a key factor for attaining conspiracy beliefs, at least in relation to COVID-19 pandemic, and associated with lower vaccination rates as well as longer time to get vaccinated.

Keywords: COVID-19; conspiracy ideation; delusion proneness; psychosis; schizophrenia; vaccination.

Figure 1

Timeline showing dates of data sampling for the study where the relation between delusion proneness and later development of COVID-19 conspiracy believes was studied (7) and present study, including important dates pertaining to the COVID-19 pandemic.

Figure 2

(A) Boxplot showing the comparison of means of PDI between unvaccinated and vaccinated subjects. Each data point represents an individual. (B) Scatter plot showing the relationship between PDI and time to get vaccinate (in months) from when the vaccination program started. The cloud represents accumulated effect of overlapping data points, while black dots represent single data points with jitter. The darker shade of blue represents more data points (N =229), *** p <.001. PDI: Peters et al. Delusions Inventory. rw = Wilcoxon's effect size. rs = Spearman's correlation.

Figure 3

Path analyses (with standardized parameter values) of how delusion-proneness (measured with PDI) predicts vaccination with COVID-19 Conspiracy ideation as a mediator. PDI = Peters et al. Delusions Inventory; CCQ Total = total score on the COVID-19 Conspiracy Questionnaire. ** p <.01, *** p <.001.

Figure 4

(A) Word cloud showing most frequent words used by unvaccinated subjects (left) and for vaccinated subjects (right). The bigger words are more frequently used than the smaller ones. Results from the text analysis suggested that unvaccinated individuals used more negative words and themes regarding the vaccination while vaccinated individuals used more positive words about protecting themselves and other individuals. Thus, while both groups were concerned about safety issues, conceptualizations of safety differed between them. A group comparison was performed in order to test if the average number of words used was significantly different between vaccinated and unvaccinated subjects and found that on average, unvaccinated subjects used more words (mean=52.03, SD=83.67, median=20.5, range 2-422) to describe why they did not get vaccinated compared to the vaccinated subjects (average=14.74, SD=10.92, median=11, range 2-61) (W=989.9, p=.002). This result remained significant even after excluding two extreme cases in the non-vaccinated group (W=989.5, p=.009). We then conducted a correlation analysis and found no significant correlation between number of words and PDI (r=-.22, p=0.22) or CCQ (r=-.09, p=0.62). (B) Average frequency of themes found in subject’s responses on why they did or did not get vaccinated (unvaccinated n = 36, vaccinated n = 237). Note that every subject’s response was tested for all identified themes.