This cross-sectional, web-based survey study took place between March 19 and April 17, 2020. To ensure that our survey was highly visible to potential respondents, it was distributed via social media, email, direct communication methods, and advertisements in various digital communication channels. The recruitment of participants mainly took place at the Department of Psychology of Witten/Herdecke University. All participants were residents of Germany who were aged ≥16 years. All procedures in this study were performed in accordance with the ethical standards of the institutional review board of the Department of Psychology and Psychotherapy of Witten/Herdecke University and those of the American Psychology Association [18,19]. A letter of approval can be obtained from the first author.

Participants who fully completed the questionnaires were included in the statistical analysis. Descriptive statistical analyses were performed to describe the sample’s characteristics in terms of the variables that were included in this study. In addition, bivariate correlation values were computed to examine associations among the variables. A saturated path model [27] with manifest variables was used to test whether OCEAN personality traits, anxiety, and risk-taking behaviors were related to worries about hospital safety and hygiene, worries about medical practices, and anticipated behaviors toward nonurgent surgeries (Figure 1). To assess whether people’s risk of contracting COVID-19 and information-seeking behaviors (ie, those related to COVID-19) had an impact on their worries and anticipated behaviors, these variables were included in the analysis.

Age, gender, and educational level (ie, a dichotomous variable that accounted for primary and secondary education) were introduced in the model as covariates that needed to be controlled. All variables in the model were allowed to covary. Standardized regression coefficients (ie, β_i) for the path model (ie, the model for predicting affective reactions) and anticipated behaviors were calculated with the decomposition equation of correlations (ie, r_i), which is used to determine the direct and indirect effects that predictor variables (ie, X_i) have on dependent variables (ie, Y_j). In addition, the 95% CIs were calculated based on the 2.5 and 97.5 percentiles of the estimated SEs from bootstrapping [27]. As our study had four main dependent variables, Bonferroni correction was performed to lower the Cronbach α level for interpreting the results (ie, from .05 to .0125). To evaluate model fitness, the Chi-square test was used. According to Bollen and Long [28], Chi-square values should not be 2-5 times larger than the degrees of freedom. In addition, comparative fit index, Tucker-Lewis index, root mean square error of approximation, and standardized root mean square residual values were calculated as fit indices. Descriptive statistical analyses were conducted with SPSS, version 26 (IBM Corporation). The path analysis was conducted with Mplus, version 8.1 (Muthén & Muthén) [29].

Of the 1059 participants who took part in our survey, 929 (87.7%) had complete data sets. Thus, these 929 participants were included in the analyses. As indicated in Table 1, most of the participants (792/929, 85.3%) were female. The mean age of participants was 35.3 years (SD 12.9 years). Of the 929 participants, 890 (95.8%) stated that they were not infected with SARS-CoV-2, and only 7 (0.8%) stated that they were infected with SARS-CoV-2 (ie, at the time of the survey or before the survey). With respect to risk profiles, 683 (73.5%) participants reported that they did not exhibit any of the risk factors for contracting COVID-19, while 112 (12.1%) stated that they had a chronic lung disease. Almost all participants (884/929, 95.2%) subjectively felt restricted due to COVID-19–related regulations and measures. Details on participants’ sociodemographic characteristics are provided in Table 1.

Descriptive statistics and correlations among the variables in the path model are reported in Tables 2 and 3. Worries about proper hospital hygiene and medical practices positively correlated with neuroticism (r=0.12) and anxiety (r=0.21). Further, all four dependent variables intercorrelated with each other. For example, worries about hygiene and worries about safety significantly correlated with each other (r=−0.40; P=.001).

The path model predicted the associations between independent variables (ie, personality, risk-taking behaviors, and anxiety) and dependent variables (ie, feelings about security, worries about hospital hygiene and medical practices, and anticipated behaviors that relate to people’s decisions to postpone their own surgery or advise a person close to them against surgery). Figure 1 presents the parameter estimates of the model (ie, standardized solutions).

The following model-data fit indices were obtained: Chi-square value (χ²₅₄=942.94; N=929; P<.001), comparative fit index (1.00), Tucker-Lewis Index (1.00), root mean square error of approximation (<.01), and standardized root mean square residual (<.01). These values indicated a moderate model fitness. Table 4 provides the standardized regression coefficients of the path model, which was used to predict affective reactions and anticipated behaviors.

As outlined in Figure 1, the feeling of security with regard to hospitals and medical practices was significantly negatively related to anxiety (β=−.12; P=.001), which is in line with our hypothesis. Further, affective reactions to hospital hygiene and medical practices resulting from a bottleneck of appropriate personal protective equipment for health care workers were significantly positively related to anxiety (β=.16; P<.001) and nonsignificantly negatively related to extraversion (β=−.07; P=.054). Although anticipated behaviors that relate to advising a close person against surgery did not correlate with any of our hypothesized variables, anticipated behaviors that relate to one’s own surgery were negatively associated with neuroticism (β=−.08; P=.04) and risk-taking behaviors (β=−.09; P=.03). Such anticipated behaviors were also positively associated with anxiety (β=.08; P=.03). All of these associations however were not statistically significant after the Bonferroni correction. No other associations between the independent and dependent variables were found. However, women and older participants reported that they experienced higher levels of negative affective reactions and anticipated behaviors compared to men and younger participants, respectively.

To the best of our knowledge, our study is the first to investigate predictors of affective reactions that relate to hospital safety, hospital hygiene, and medical practices during the COVID-19 pandemic. We are also the first to investigate anticipated behaviors that relate to people’s decisions to postpone their surgery or advise a person close to them against surgery during the pandemic. Our findings are in line with those of a German-Austrian survey [30], which found that anxiety was positively related to security actions. Our results suggest that state anxiety is the most influential factor of anticipated health-related behaviors and concerns about safety or hygiene. Apart from state anxiety, none of the other hypothesized predictors (eg, risk-taking behaviors) or personality factors (eg, agreeableness or openness) had any significant association with affective reactions or anticipated behaviors. This is contradictory to the recent findings of Martin [31], who found that agreeableness was related to the perceived severity of the COVID-19 pandemic, and openness was related to low levels of anxiety with regard to contracting COVID-19. Although previous studies have suggested that individuals with high levels of neuroticism exhibit pronounced negative reactions to stressful events [32], our findings show that neuroticism was not associated with anticipated behaviors during the COVID-19 pandemic. In our study, we found that people with high levels of neuroticism were less likely to postpone their own surgery. This finding is comparable to that of an early US survey, which found that neuroticism was associated with high levels of concern [33]. The similarities in these results could be explained by the age of our participants. It is possible that our relatively young participants were not able to accurately imagine a scenario in which they are hospitalized. This might have influenced participants’ responses to our survey. With regard to the relationship between health policy formation and public responses [34], the most important finding of our study was that anxiety was related to both the affective reactions and anticipated behaviors of the participants. Allgleton and Kippax [35], who conducted an analysis on Australian HIV/AIDS policies, argued that suppressed anxiety can be used as a depressive position for eliciting a desired response in the general public [35]. Other studies [36-38] have also found that anxiety is an important predictive factor of taking preventive measures and exhibiting compliant behavioral responses during the 2009 influenza pandemic. Our findings support these empirical results. Health authorities should be aware that anxiety may not only affect individuals’ behaviors but also the behaviors of organizations and systems (eg, splitting and blaming) [39]. In addition, anxiety resulting from the COVID-19 pandemic might also encourage individuals to consult a physician later than necessary (ie, to present their complaints). This has already occurred [40]. Such behavior may result in harms to health, the development of depression [41], or the chronification of disease.

Aside from the strengths of our study (eg, its large sample size), several limitations also need to be mentioned. First, due to the dynamic nature of the pandemic, we decided to use a random sample. However, due to our survey dissemination methods, our sample may not be representative of the German population. The generalizability of our results is open to empirical debate, as our sample mostly consisted of middle-aged and well-educated women. Research has shown that compared to men, women are more likely to actively seek health-related information and pay more attention to potential worldwide pandemics [42]. Second, our sample mainly consisted of middle-aged individuals. Therefore, it is reasonable to assume that older people are more likely to postpone surgeries and operations due to the COVID-19 pandemic, as they are more susceptible to the disease than middle-aged people. Older people also have stronger health care needs than middle-aged people [17]. Further research on the COVID-19–related concerns of older individuals is needed. Third, data collection took place during the beginning of the pandemic in Germany. Therefore, it remains unclear whether individuals would have the same affective reactions and anticipated behaviors later into the pandemic. Furthermore, the affective reactions and anticipated behaviors of people from urban areas should be distinguished from those of people from rural areas, as COVID-19 spreads at different rates in different geographical areas [43]. However, during the first phase of the pandemic in Germany, no considerable differences were found in infection and death rates [43]. This finding is also supported by the results of a recent survey study [44], wherein the authors did not find any substantial differences in behavioral intentions between participants from rural and urban regions in China. Fourth, our dependent variables were only measured with one item that used a visual analog scale. This was done to keep the survey concise and specific. Unfortunately, validated measures such as the COVID-19–Induced Anxiety Scale or the Protective Behaviors Towards COVID-19 Scale [45] were not available at the time of our survey. According to Heller et al [46] and Price et al [47], visual analog scales have sufficient psychometric measurement properties. Thus, they can be used when no validated instrument is available. Fifth, although the fitness of our path model was acceptable, it could have been better. However, it should be noted that as the sample size increases and the degrees of freedom remain constant, the Chi-square value increases. This leads to the problem of plausible models being rejected due to a significant Chi-square value. Therefore, too much emphasis should not be placed on the significance of the Chi-square statistic [48]. Furthermore, it should be noted that our data are cross-sectional in nature. As such, causal conclusions cannot be drawn from our data. Future studies should be longitudinal in nature.

Our results provide further insight into affective reactions and anticipated health-related behaviors during the COVID-19 pandemic. Our findings indicate that OCEAN personality traits are not associated with affective reactions and anticipated behaviors. Therefore, specific distinctions do not seem necessary when designing messages about health risks and health protection measures (ie, those related to hospital and medical practices during the COVID-19 pandemic). Even though future research is needed to confirm our results, health care workers should address the issues of patients with anxiety seriously and directly. Clear communication is necessary when providing information on the specific actions that hospitals and medical organizations perform to protect patients and health care workers. This could also help with preventing the cancellation of nonurgent surgeries in hospitals.