Systematic review on the association of COVID-19-related conspiracy belief with infection-preventive behavior and vaccination willingness

In times of a pandemic, not only infections but also conspiracy narratives spread among people. These have the potential to influence the course of the pandemic. Here we summarize and critically evaluate studies from the first year of the pandemic presenting findings on the association between COVID-19-related conspiracy belief and infection-preventive behavior and vaccination willingness. A systematic literature search was conducted using the databases COVID-19 Data Portal, APA PsycArticles, Psychology and Behavioral Sciences, Scopus, and PubMed. After removing duplicates, studies meeting the previously defined inclusion and exclusion criteria were subjected to title and abstract screening and content reviewed and analyzed subsequently. The systematic literature search yielded 17 studies meeting our pre-specified inclusion criteria. Twelve studies examined infection-preventive behavior (N = 16,485), and ten studies vaccination willingness (N = 20,210). In summary, belief in COVID-19-related conspiracy narratives was negatively associated with vaccination willingness and infection-preventive behavior. The results point to the importance of the content of the conspiracy narratives. Various explanatory approaches and possible moderator variables are discussed, referencing the state of research on conspiracy beliefs and health-related preventive behavior after the first year of the pandemic. We argue that future studies should strive for a consistent operationalization and use of the term conspiracy belief.

during the COVID-19 pandemic, conspiracy narratives circulated within a short period of time. They vary in detail and abstractness and differ in terms of content, for example addressing the origin, characteristics or spread of the virus, targeting specific actors and their interests, or doubt the existence of the coronavirus. Conspiracy narratives have the potential to influence the course of the crisis that initially fostered their occurrence. This is because effective control of the highly contagious virus requires widespread public adoption of preventive behavior [6] and vaccination of a high proportion of the population [7]. The negative association between conspiracy belief and health-related preventive measures has already been demonstrated in various research projects [8][9][10][11]. People who believe in conspiracies where not only found to be less likely to attend annual checkups, visit the dentist, and use sunscreen [12], according to Jolley and Douglas [13], they are also less likely to support public action on pandemics. Furthermore, conspiracy belief correlates with concerns towards vaccination and lowers vaccination willingness [12,14,15]. Similar effects might be expected with respect to COVID-19. To control the ongoing COVID-19 pandemic, effective vaccines are an essential tool because they can be critical in reducing transmission, hospitalizations, and the need for intensive care [16,17]. However, the hoped-for indirect community protection, known as herd immunity, can only occur if enough people are willing to be vaccinated [18]. Until herd immunity is achieved, infection prevention measures remain critical to contain the pandemic [19]. This paper therefore examines the relationship between belief in COVID-19-related conspiracy narratives and infection-preventive behavior and vaccination willingness. Since the studies dealing with this question were conducted at different points in time and the samples considered vary to a large extent and there is no uniform operationalization of the constructs, comparability is significantly hampered. Therefore, a systematic examination of similarities and differences is necessary to reliably determine the state of research.

Definition of the surveyed factors
In international psychological and social science research, there is conceptual disagreement in analyzing conspiracies. Commonly, the terms conspiracy theory and conspiracy belief are used. According to Douglas et al. [20, p. 4] conspiracy theories can be defined as "attempts to explain the ultimate causes of significant social and political events with claims of secret plots by two or more powerful actors". Instead of the misleading term conspiracy theory, which seems to suggest a scientific theoretical background, we decided to use the term conspiracy narrative in our review. Belief in conspiracy narratives is referred to as conspiracy belief. The term COVID-19-related conspiracy belief refers to belief in conspiracy narratives that relate to the COVID-19 pandemic. Any behavior that contributes to reducing the spread of COVID-19 and thus to pandemic containment is subsumed under the term infection-preventive behavior. On the one hand, this includes the implementation of hygiene measures such as hand washing or hand disinfection, adherence to cough and sneeze etiquette, avoiding touching one's face in public, and cleaning surfaces. On the other hand, this includes measures of social distancing or physical distancing such as staying at home, keeping a distance, limiting contact, avoiding crowds, and avoiding contact with sick or elderly people. Also included are specific measures such as wearing a mask or gloves, staying home with symptoms of illness, using testing services, avoiding travel and using a contact-tracking app. Vaccination willingness refers to the expressed intention to be vaccinated against COVID-19 if vaccination is available. The vaccination hesitancy variable is understood to be the opposite of vaccination willingness.

Literature search procedure
In the first step the search terms for the literature search were determined. To identify relevant articles related to the prevailing COVID-19 pandemic, the search terms COVID-19 and coronavirus were used. For coverage of studies dealing with beliefs in conspiracy narratives, the keywords conspiracy theories and conspiracy beliefs were used. To identify appropriate studies to answer the research question, the final search term was created by linking the search terms ((conspiracy theories OR conspiracy beliefs) AND (COVID-19 OR coronavirus)) and conducting a systematic literature search on February 3, 2021 using the databases COVID-19 Data Portal, APA PsycArticles, Psychology and Behavioral Sciences, Scopus, and PubMed. Duplicates were removed in the following step. The search resulted in a varying number of matches in the databases (see Additional file 1: Table A1). Using the previously defined inclusion and exclusion criteria, the remaining studies were subjected to title and abstract screening and finally reviewed for content. In total, 17 studies were extracted through the described search process and with the help of the inclusion and exclusion criteria (see Additional file 1: Table A2), which form the research basis of the present paper. Figure 1 shows the systematic search process, including the results obtained in each step.

Inclusion and exclusion criteria
Only articles with data collected and published during the COVID-19 pandemic were included (i.e., studies that collected and published their data between March 11, 2020 and February 2, 2021) [21]. To ensure the quality of the selected studies, a further requirement was publication in a peer-reviewed journal. Preprints were excluded. To increase the comparability and generalizability of the data, only quantitative, empirical studies that investigated an adult sample were included. The included studies measured COVID-19-related conspiracy beliefs as well as infection-preventive behavior and/or vaccination willingness, according to the definition established in advance (see Definition of the surveyed factors). Only studies in English were included and only those studies that reported statistical analyses about the association between COVID-19-related conspiracy beliefs and infection-preventive behavior and/or between COVID-19-related conspiracy beliefs and vaccination willingness.

Results
The total sample size of the twelve studies examining infection preventive-behavior (see Table 1) is N = 16,485, with the smallest sample counting N = 407 participants [22], the largest sample counting N = 2501 participants [23], and the average age ranging from 31 years [24] to 47 years [23]. The participants come from different nations. Two studies examined a Polish sample [25,26], and one sample each was collected in Serbia [22] and Turkey [24]. The majority of the studies was conducted in English-speaking countries: Three studies each examined U.S. samples [27][28][29], and UK samples [23,30,31]. In addition, there is a study with samples both from the US and UK [32], and an international sample with participants from 66 nations, but with the majority also from the US or UK [33]. Conspiracy belief was measured in a variety of ways. For example, Alper et al. [24] used two items that asked about a conspiratorial plan behind the spread of coronavirus, while Freeman, Waite, et al. [23] used 48 items to cover the belief in a wide variety of conspiracy narratives. Imhoff and Lamberty [32] used two sets of conspiracy narratives with three items each: COVID-19 is a hoax and SARS-CoV-2 is humanmade. Oleksy et al. [26] distinguished between general and government-related conspiracy narratives. To assess infection-preventive behavior, the studies reviewed varied in the number of examined behaviors that reduce the  [33], and the largest was − 0.524 [32]. The various regression analyses yielded odds ratios ranging from 0.37 [30] to 14.34 [23], standardized regression coefficients from 0.36 to 0.38 [31], and unstandardized regression coefficients from 0.04 [26,28] to 0.834 [32]. In the SEM model of Biddlestone et al. [33], conspiracy belief emerged as a negative predictor of social distancing: β = − 0.04, p < 0.001. In the crossed-lagged panel model of Bierwiaczonek et al. [27], conspiracy belief at an earlier measurement time point predicted social distancing at the following measured time point: β (− 0.062 to − 0.067), p = 0.002. Alper et al. [24] and Earnshaw et al. [28] found no significant association between conspiracy belief and infectionpreventive behavior. The difference in the strength of the association is striking, depending on the content of the conspiracy narratives on which the conspiracy belief variable is based. For example, general COVID-19-related conspiracy beliefs were not related to infection preventive-behavior (b (SE) = 0.02 (0.02)) in Oleksy et al. [26], whereas belief in government-related conspiracy narratives (b (SE) = − 0.04 (0.02)) were. Imhoff and Lamberty [32] found that belief in the conspiracy narrative that COVID-19 was a hoax had negative predictive power for pandemic-controlling behavior (S1 The largest negative association reported in the studies was between believing in a conspiracy narrative that views Jews as the authors of the virus and following the "stay home" directive: OR = 14.34, 95% CI [11.26, 18.25] [23]. Biddlestone et al. [33] concluded that conspiracy beliefs are negatively related to social distancing (r = − 0.16, p < 0.001; SEM model: β = − 0.04, p < 0.001) but not hygiene measures (r = − 0.02).
The total sample size of the ten studies that examined vaccination willingness is N = 20,210. The smallest sample counts N = 396 participants [34], the largest sample N = 5114 participants [35], and the average age ranges from 26 years [34] to 47 years [35]. Two studies tested individuals from the USA [28,29] three from the UK [23,31,35], and one sample each is from Serbia [22], Italy [36], and France [34]. In addition, there is a study with samples from both Turkey and the UK [37] and an Arabic sample with participants from 18 nations, with the majority coming from Jordan and Kuwait [38]. Conspiracy beliefs were measured in different ways. Prati [36] and Salali and Uysal [37] each used one item that asked about conspiracy belief in the non-natural origin of COVID-19, while Freeman, Waite, et al. [23] used 48 items to cover belief in a wide variety of conspiracy narratives. Bertin et al. [34] distinguished three groups of conspiracy narratives that addressed supposed machinations of different groups: a threatening foreign outgroup (China), unspecified outgroups (e.g., industrialists), and a national ingroup (French government). Vaccination willingness, or vaccination hesitancy [35], was assessed with one item each. Studies reported either correlation coefficients [22,23,34,35], odds ratios [23,28,[36][37][38], standardized regression coefficients [29,34], or unstandardized regression coefficients [22,28,31]. Freeman and Loe et al. [35] constructed a SEM model. Nine out of ten of the reviewed studies demonstrated a negative correlation of conspiracy belief and vaccination willingness. The smallest reported correlation was − 0.23 [34], and the largest was − 0.53 [22]. The various regression analyses yielded odds ratios ranging from 0.26 [28] to 2.73 [38], standardized regression coefficients from − 0.11 [34] to 0.35 [31], and unstandardized regression coefficients from 0.45 [31] to 1.63 [35]. In the SEM model of Freeman, Loe, et al. [35], conspiracy belief and vaccine hesitancy were positively associated: β = 0.38, p < 0.001. Prati [36] concluded in her study that the conspiracy belief in non-natural origins was not a significant negative predictor of willingness to vaccinate: OR = 0.88, 95% CI [0.22, 3.55]. There is a difference in the strength of the association depending on the content of the conspiracy narratives underlying the conspiracy belief variable. For example, the conspiracy belief in the artificial origin of COVID-19 was not significantly correlated with vaccination willingness in Prati [36]. Also in Salali and Uysal [37]

Discussion
The systematic literature review on the link between COVID-19-related conspiracy belief and infection-preventive behavior and vaccination willingness presented here provides clear evidence in favor of a negative association between conspiracy belief and infection-preventive behavior and vaccination willingness. Ten of twelve of the reviewed studies demonstrated a negative association of COVID-19-related conspiracy beliefs with infectionpreventive behavior, and nine of ten studies showed a negative association with vaccination willingness. Studies on vaccination willingness showed a largely consistent picture. The reported correlation coefficients showed a medium to high linear relationship of the variables, with values ranging from − 0.23 [34] to − 0.53 [22]. While simple correlations typically have a high face validity and can be straightforwardly interpreted also by laypersons, this may sometimes not do justice to the complexity of the subject matter, because they do not allow to control for other predictors, unlike regression coefficients. Another advantage of regression coefficients is that regression coefficients are robust even with small sample sizes. The various regression analyses demonstrated the negative relationship between conspiracy belief and willingness to get vaccinated, largely with small to very small effect sizes, and in some cases with a medium effect [39]. Only Prati [36] found no negative association between COVID-19-related conspiracy beliefs and vaccination willingness. This could be related to the fact that only a single item asking about the non-natural origin of COVID-19 was used. This is also true for the study by Salali and Uysal [37], who also asked about the belief in the non-natural origin of COVID-19 and demonstrated a negative association between conspiracy belief and vaccination willingness. One explanation for the differences in effect sizes could be that the different conspiracy narratives, and thus the content of each conspiracy belief, is crucial for the strength of the association. Not only Prati [36] and Salali and Uysal [37] looked at the belief in the artificial origin of COVID-19. Also Sallam et al. [38] found an association: OR = 0. 47 [38]. This could indicate a mediating factor, such as threat-or risk perception that moderates the association between conspiracy beliefs and vaccination willingness, and that varies in intensity depending on the narrative. One might assume that a person who believes that the virus was fabricated in a laboratory would nevertheless believe it to be dangerous and be willing to protect themself with a vaccine. However, if the vaccine itself is integral part of the conspiracy belief and is therefore thought to be more dangerous than the disease it protects against, low willingness to vaccinate is not surprising [22]. Here, not only risk or threat perception, but also the resulting COVID-19-related anxiety and the associated degree to which individuals perceive themselves as affected by the conspiracy could mediate the association with infectionpreventing behavior, which would be consistent with the findings of the Brewer et al. [40] meta-analysis. An indirect influence of conspiracy beliefs via feelings of powerlessness, as addressed by Jolley and Douglas [13,15] in their work, which could lead to reduction in infectionpreventive behavior or vaccination willingness, is also possible. Moreover, the present data suggest that conspiracy beliefs of any kind are negatively related to vaccination willingness. This could be explained by a core motive of conspiracy belief: People who believe in conspiracy narratives assume that important information, such as the truth about the harmfulness of vaccines or the origin of the pandemic, is being hidden from the public [41,42]. This may explain why they are suspicious of vaccination campaigns by the state and of vaccination itself, even if they believe the virus is dangerous. Discouragement of vaccination might as well be mediated through social media usage patterns. Chadwick et al. demonstrated that people with a high conspiracy mentality who primarily seek information on social media platforms discourage vaccination [43]. Harmful misinformation spread on social media and the risk of drifting into ideologically isolated echo chambers could therefore contribute to the perpetuation of the pandemic situation. A challenge also recognized by the WHO, who stated we are not only fighting a pandemic but also an infodemic [21]. Especially in the fast-paced pandemic situation, analysis of vaccinerelated content on social media provides important information on public opinion [44]. The varying strength of the association with infectionpreventive behavior could also be explained in terms of the content of the conspiracy narratives studied. In Imhoff and Lamberty [32], for example, individuals who believed that the pandemic was a hoax perceived the pandemic as less threatening, whereas there was no significant relationship between believing that the virus was human-made and threat perception. Belief in the conspiracy narrative that COVID-19 is a hoax was negatively related to infection-preventive behavior, whereas the relation with the belief that SARS-CoV-2 is humanmade was less clear. In both US-samples the negative correlation between believing COVID-19 is a hoax and infection-preventive behavior was significant, in the UK it was not. The correlations between believing SARS-CoV-2 is human-made varied in both countries regarding their direction and significance. These findings suggest that a conspiracy, which pictures the pandemic as less threatening, may lead to reduced adherence to preventive behavior. While Oleksy et al. [26] found no significant association between general COVID-19-related conspiracy beliefs and infection-preventive behavior, there was a negative association with beliefs in government-related conspiracy narratives. The population is most prominently urged by the government to engage in protective behavior, and the government imposes infection prevention measures to this end. The suggestion is that conspiracy narratives about hidden malicious intent or government misconduct therefore reduce infectionpreventive behavior among the population. However, it must be critically noted here that Oleksy et al. only found a small difference between the two sets of conspiracy narratives, which could possibly also have been caused by a differential measurement error. Also noteworthy is the largest negative association reported in the studies: Belief in the conspiracy narrative that Jews created the virus to collapse the economy and benefit financially from was strongly negative related to stay-at-home behavior in a UK sample: OR = 14.34, 95% CI [11.26, 18.25] [23]. According to a study by Staetsky [45] 13% of the population in the UK holds antisemitic prejudices which could be an explanation for this surprisingly large association. COVID-19-related conspiracy narratives may help to satisfy the need to make sense of events during a period, which is characterized by uncertainty, thereby regaining a sense of control [46,47]. Another explanation for why conspiracy beliefs and infection-preventive behavior are negatively connected is addressed by Bierwiaczonek et al. [27] who argue that belief in an external threat may lead people to seek social support, and thus prevent the minimization of social contact in the interest of infection prevention. This could also explain the findings of Biddlestone et al. [33], who found that conspiracy beliefs are negatively related to social distancing (r = − 0.16, p < 0.001; SEM model: β = − 0.04, p < 0.001) but not hygiene measures (r = − 0.02). Alper et al. [24] and Earnshaw et al. [28] found no significant association between conspiracy beliefs and infection-preventive behavior. However, in Alper et al. [24], the reliability of the measurement of conspiracy beliefs is questionable, as only a rather general item ("Coronavirus was developed and spread around the world by certain people for their own purposes") and an inversely coded version of the item were used to capture the construct. Earnshaw et al. [28] used a 6-item questionnaire, arguably a more reliable measurement tool, but again, as in most of the studies reviewed, the reliability and validity of the measurements are not assessed. This is not surprising, because due to the novelty of the studied construct, no established instruments were readily available. As a consequence, most of the authors of the reviewed studies decided to develop their own instruments, for example to assess infection-preventive behavior in the context of the COVID-19 pandemic. In many of the studies reviewed, the selection of the surveyed preventive behavior was based on either government guidelines or those of the WHO. Other authors developed questionnaires based on studies of other infectious diseases, such as HIV [28]. For those instruments that were adapted, it is unclear to which extent the reliability and validity of the measurements can be applied to the novel context of the COVID-19 pandemic. There is a broad consensus that it is of great importance to check the consistency of results across time or across different observers (i.e., the issue of reliability) and to carefully check whether the instruments used actually measure what they are supposed to measure (i.e., the issue of validity). However, these comprehensive quality controls come at the expense of the rapid dissemination and implementation of the findings, which has to be weighed carefully. Evidently, this issue is not limited to highly dynamic situations like a pandemic. In their systematic review, Goreis and Voracek [10] found substantial heterogeneity in the operationalization and hence a lack of scientific consensus in psychological research on conspiracy narratives in general. When interpreting the present findings, the aspect of the consistency and the accuracy of the measurement instruments should therefore always be taken into account. A further aspect that may be relevant is the heterogeneity in the samples studied and their thus highly varying representativeness (see Tables 1, 2). A limitation of the present work are the broad definitions of the variables COVID-19-related conspiracy belief and infection-preventive behavior. The inclusion criteria could have been further specified to reliably identify more specific effects of conspiracy narratives on particular behavior. Another limitation is the measure of vaccination willingness. It remains unclear to what extent expressed intention to be vaccinated is related to actual behavior. Since the literature search was conducted, other relevant papers have been published (e.g., [47][48][49]), which is not surprising given the dynamic nature and the global relevance of the pandemic. This review focuses on the literature published during the first year of the pandemic. Thus, developments that occurred during the second pandemic year such as broadly implemented government vaccination campaigns, media coverage on the effectiveness and side effects of vaccination, and the discourse on restricted social participation for unvaccinated people, are not reflected in it. The present review only summarizes the state of the literature and knowledge at the particular point in time at which the literature search was conducted, which represents a limitation. However, it may also serve as an important reference point to track the evolution of COVID-19-related conspiracy belief over time (e.g., after the second year of the pandemic).

Conclusion
Belief in COVID-19-related conspiracy narratives is negatively associated with vaccination willingness and infection-preventive behavior. These findings are consistent with previous research linking conspiracies and healthrelated preventive behavior and suggest that the content of conspiracy narratives is associated with the strength of the effect. Even very small negative effects on taking behavior to contain the pandemic translate into higher rates of infection, thus affecting the health of countless people long-term and resulting in a cost in human lives. Reduced vaccination willingness due to belief in conspiracies could significantly impede successful control of the pandemic, thus prolonging the current crisis. Because belief in COVID-19-related conspiracy narratives and resistance to preventive behavior and future vaccination are closely related, it could be critical to counter them to contain the spread of the virus. Key to successful combating is a better understanding of the phenomenon of conspiracy narrative. Further research projects should focus on the question of what specific conspiracy narratives, or what types of conspiracy narratives, are related to which specific behavior resulting in pandemic progression and ask what variables moderate this effect. Also, research on the question of how conspiracy narratives spread is advisable. Furthermore, a review of the published literature after the second year of the pandemic would be helpful to assess to what extend COVID-19-related conspiracy belief is influenced by factors such as new findings on vaccination or restrictions for unvaccinated individuals and the public debate about it. Moreover, a precise definition and consistent, psychological operationalization of the construct conspiracy belief with differentiated measurement instruments would be beneficial to increase the quality of the results and to ensure their comparability. Systematic reviews have the potential to generate insights that go beyond those of a single study, because studies differ in many different aspects such as the time and place of the data collection and sample characteristics, to name but a few. Especially in a dynamic and complex situation such as a pandemic, systematic reviews are needed to provide more people with a structured overview of scientifically sound information.