key: cord-1005165-2cdoara3
authors: Kirsch, Fabian; Lindemann, Ann-Kathrin; Geppert, Johanna; Borzekowski, Dan; Lohmann, Mark; Böl, Gaby-Fleur
title: Personal Protective Measures During the Coronavirus Pandemic in Germany
date: 2022-05-18
journal: Int J Infect Dis
DOI: 10.1016/j.ijid.2022.05.036
sha: af1b54f4d8d5399ea5f36fc5f74f4663109f2857
doc_id: 1005165
cord_uid: 2cdoara3
Objectives The coronavirus SARS-CoV-2 is highly contagious, and can only be contained if the majority of the population takes measures to protect themselves against infection. The present study aimed at investigating personal protective measures, their development over the course of the pandemic in Germany, and potential differences in behavior regarding gender, age, and education. Methods Data from 20 waves of the serial cross-sectional study “BfR-Corona-Monitor” were analyzed. The total sample consisted of N = 20,317 respondents (about 1,000 per wave). Data were collected via telephone surveys between June 2020 and March 2021. Results To protect themselves from infection, participants primarily relied on wearing covers for mouth and nose, keeping their distance from other people and washing their hands thoroughly. Analyses over time showed a strong positive correlation between the number of measures taken and the national incidence rate. Sociodemographic differences also emerged, with women, higher educated as well as younger respondents taking a higher number of protective measures. Conclusions Results indicated that in times of greater infection risks, people adapted accordingly and took more protective measures. However, based on sociodemographic differences, campaigns should especially focus on older, male and lower educated persons to enhance their protective behavior.
1 HIGHLIGHTS Protective behavior strongly correlated with national incidence rate over time Most taken measures: covering mouth and nose, keeping distance, washing hands Women took more protective measures than men Higher educated people took more protective measures than lower educated people There was a diverse result pattern for age groups in the uptake of measures
The first case of COVID-19 in Germany was reported on January 27, 2020 (Bavarian State Ministry of Health and Care, 2020). The virus SARS-CoV-2 spread fast and within the first two months, more than 42,000 cases had been notified (Robert Koch Institute, 2021) . Containment of the virus was not only complicated by the fact that it was found to be contagious even before the onset of symptoms, but also because some people can live through an infection without showing any symptoms and still infect others at the same time (Almadhi et al., 2021; Buitrago-Garcia et al., 2020) . In a study by Johannson et al. (2021) , it was estimated that about half of the transmissions happened this way. Protective measures were soon applied in all areas of life and the entire population was urged to implement them (e.g., Federal Ministry of Health, 2021). Such measures and recommendations included keeping a distance of 1.5 meters between people, staying at home, limiting the number of people one is allowed to meet, and washing one's hands thoroughly.
However, previous studies suggested that individual health behavior can vary in regard to sociodemographic aspects. The review by Hiller et al. (2017) shows that gender plays an important role in this, as women are usually more health-conscious. Men, on the other hand, are often found to engage less in preventive health behavior. This difference is also evident in preventive behavior during a pandemic (Bish & Michie, 2010) . Another factor that is associated with the conduction of health behavior is age. Most studies showed that in a pandemic situation older people are generally more likely to carry out preventive behavior (Bish & Michie, 2010; for an exception, see Pasion et al., 2020) . Behavioral effects of age and gender are particularly important with regard to COVID-19, as they are risk factors for a severe course of the disease. In fact, older people and men have been shown to be at a higher risk for a severe progression of the disease (Gallo Marin et al., 2020) .
In addition, the level of education may also influence the protective behavior. The review by Bish and Michie (2010) concluded that in most studies more educated people are generally more likely to adopt preventive behavior during a pandemic. A lower level of education is also often associated with an increased risk of numerous diseases and an unhealthier lifestyle (Mielck, 2012) . These findings were further supported by Lampert et al. (2005) , indicating that persons with low education have a higher risk for diseases than those with a higher level of education. However, they also found that preventive behavior was either not linked to the participants' education level or was even more likely to be taken up by persons with a lower level of education. The latter result pattern was also found in a study on protective behavior during the COVID-19 pandemic (Betsch et al., 2021) . In sum, previous studies yield to ambiguous results regarding preventive behavior during the course of a pandemic.
COVID-19 is an example of a highly contagious disease that developed into a global health crisis. The containment of this and future pandemics depends on a population-wide adaption of adequate protective measures. Therefore, the aim of the present study was to investigate personal protective behavior of citizens in Germany during the coronavirus pandemic between June 2020 and March 2021. Due to the dynamic course of the pandemic, we also tested for development over time and a potential correlation between protective measures and infection rates. Since previous research suggests that sociodemographic factors might influence the implementation of preventive behavior, we further explicitly tested for differences in regard to gender, age, and education. The knowledge about the impact of these factors on behavior can help in managing a health crises, for example by developing information campaigns for specific target groups.
We used data from 20 waves of the serial cross-sectional study "BfR-Corona-Monitor" (German Federal Institute for Risk Assessment, 2021). The sample consisted of 997-1,037 respondents per wave, resulting in a total sample size across all waves of N = 20,317 respondents. Data were collected on two consecutive days every other week between June 23, 2020 and March 17, 2021 via telephone surveys in the Federal Republic of Germany (see Table 1 ; three consecutive days for wave 4). Data collection was carried out by the market research institute Kantar GmbH as part of their daily omnibus telephone interviews (computer assisted telephone interviewing, CATI; Choi, 2004) .
Respondents had to be at least 14 years old to participate in the study. Samples were drawn via a random digit dialing procedure that included mobile as well as landline telephone numbers. When a mobile phone number was called, the person who answered the call was selected for the interview. This was different for landline telephone numbers, where the respondent within the household was randomly selected using the Kish selection method (Kish, 1949) . Data were statistically weighted to ensure representativeness (Gabler et al., 2016) : First, data were weighted regarding the number of mobile phones and landline numbers a person could be reached by. This ensured that each person had the same chance to be selected for an interview. Second, data were weighted according to various sociodemographic variables (gender, education, age, employment, size of city and German federal state). This process was carried out for each survey wave individually. As a result, the weighted sample is representative for the population in Germany with a balanced gender ratio (51% female) and a mean age of M = 49.1 years (SD = 19.9).
7
Protective behavior of the respondents was assessed using the question "Which of the following measures have you taken to protect yourself or your family from the novel coronavirus?". The question consisted of several items that referred to protective measures from different areas of everyday life, including the respondent's individual hygiene behavior (e.g., washing hands more thoroughly), changes in their social behavior (e.g., meeting friends and family less frequently), or their consumer behavior (e.g., having food delivered more frequently). The list of measures was predominantly developed based on previous waves of the BfR-Corona-Monitor, in which protective behavior was explored using an open-ended question (Kirsch et al., 2021) . In the present study, participants indicated for each measure on a binary scale if they applied it or not. Within the scope of this article we analyzed the nine protective measures that were consistently asked across the 20 waves (see Table 3 ).
Respondents could choose not to answer to the question of protective measures. These respondents (n = 60 across all waves; 0.03% of the total sample) were not considered in the statistical analyses presented below.
A sum score across these nine protective measures was computed, resulting in a score ranging from 0 (if none of the nine protective measures were taken) to 9 (if all of the nine protective measures were taken). As the question used a binary response format, we followed the recommendations by Gadermann et al. (2012) to compute an ordinal alpha as a parameter for reliability that is based on the polychoric correlation matrix. Reliability for the sum score was good with α = .77.
Sociodemographic variables. Among the sociodemographic variables, gender, age and education were used for further analyses. Gender was assessed binary (male vs. female). Age was assessed as a continuous variable and split into three groups for analyses: up to 39 years 8 of age, between 40 and 59 years, 60 years and over. These age limits are based on common definitions of age groups (e.g., Lachman, 2001; Klimczuk, 2016) . Further, from an epidemiological point of view, 60 years is also a critical threshold, as the mortality of COVID-19 increases considerably from this age onwards (Bonanad, 2020) . This has led the Standing Committee on Vaccination (STIKO) in Germany to prioritize this age group for vaccination (Vygen-Bonnet et al., 2021) . Education was assessed by five education levels.
The levels of "pupil", "secondary general school (Volks-/Hauptschule)" and "secondary school without Abitur" were combined to represent lower education, while the levels of "Abitur, university/polytechnic entrance qualification" and "academic degree (university, academy, polytechnic)" were combined to represent higher education.
7-day incidence rate. During the coronavirus pandemic, every day the Robert Koch Institute (2021) reported an incidence value for Germany based on the notified infection rates of the past seven days ("7-day incidence rate"). For each wave, we computed the mean of reported incidence values across the corresponding survey days. These incidence values were used for comparison with the development of the protective behavior.
Data were processed and analyzed using SPSS (Version 26). All analyses are based on the weighted sample (see "Participants and procedure") and, thus, Rao-Scott adjustment (Rao & Scott, 1984) for complex survey data was applied to all tests of significance. As measures of effect sizes, we computed the following coefficients: Cramer's V for differences in the uptake of single protective measures based on gender, education and age groups; Cohen's d for differences in the mean sum score of protective measures for gender and education; and partial η² for the difference in the mean sum score for the age groups. For interpretation of effect sizes, common thresholds were used (Cohen, 1988) , as shown in Table 2 .
9 Table 3 shows descriptive statistics for the total sample of the nine protective measures and their corresponding sum score across all 20 survey waves. On average, respondents reported taking M = 5.64 protective measures (SD = 1.59). The most frequently reported measures were using covers for mouth and nose, keeping more distance to other people as well as washing one's hands more thoroughly (each > 90%). In contrast, having food delivered more frequently was the least practiced protective measure (< 10%). Table 3 further shows the results of sociodemographic analyses across all waves. For most comparisons, sociodemographic differences were significant with small effect sizes.
Regarding gender comparisons, female respondents on average took a higher number of protective measures than male respondents. Considering single protective measures, female respondents implemented almost all measures more frequently than male respondents, with the greatest effect sizes for changes in their social behavior (e.g., keeping distance to other people). However, no gender difference was found regarding the use of the Corona-Warn-App, and male respondents were slightly more inclined to have food delivered more frequently.
Higher educated respondents also reported taking a higher number of protective measures than lower educated respondents. On level of single measures, the uptake rate was higher among higher compared to lower educated respondents in particular for changes in their social behavior (e.g., meeting friends and family less frequently). In addition, the greatest effect size was found for the use of the Corona-Warn-App. No educational difference was found for items referring to hygiene behavior (e.g., washing hands more thoroughly) as well as building up larger stocks and food delivery.
For age, we found the youngest age group (up to 39 years) taking a significantly higher number of protective measures than older respondents. On the level of single measures, however, we found a more diverse result pattern. For some measures (e.g., meeting friends and family less frequently), we found younger age groups reporting a higher uptake rate than older, with greatest effect size for using the Corona-Warn-App. For other measures (e.g., food delivery), we found the youngest and the oldest age group reporting a higher uptake rate than the middle age group. And for washing hands more thoroughly, the oldest age group reporting a higher uptake rate than both younger groups. There was no age difference for the usage of covers for mouth and nose as well as distance to other people. Figure 1 shows the development of the mean sum score of protective measures as well as the 7-day incidence rate over the period of the 20 waves. Both values were strongly correlated (r = .86, p < .001), that is, with increasing incidence rate respondents showed a higher number of protective measures. The development over time shows that in early waves (June to September 2020) both values were relatively low. Beginning October 2020, the incidence started to show a substantial increase, as did the sum score of protective measures, with a peak for both values in December 2020. In January and February 2021, incidence dropped and so did the sum score.
This study aimed to investigate personal protective behavior of the population in Germany during the coronavirus pandemic between June 2020 and March 2021 by analyzing 20 waves of the serial cross-sectional study "BfR-Corona-Monitor" (German Federal Institute for Risk Assessment, 2021). Overall, the most frequently reported measures were using covers for 11 mouth and nose, keeping more distance to other people as well as washing one's hands more thoroughly. These most common measures are in line with official orders, recommendations and campaigns by the federal government at that time (e.g., the "AHA-Formel" recommended to keep distance, pay attention to hygiene, and cover mouth and nose; Press and Information Office of the Federal Government of Germany, 2020).
Analyses over time showed a strong correlation between the number of protective measures taken by the respondents and the 7-day incidence rate in Germany that was reported every day by the Robert Koch Institute (2021). That is, when the incidence rate in Germany was low (e.g., in summer 2020), the population showed a relatively low number of protective measures. However, with rising incidence values (e.g., in October 2020) the population reacted by taking more protective measures. These results indicate that in times of greater infection risks (i.e., high incidence rate), people adapt their protection behavior accordingly.
At first glance, this result seems to be in contrast to a study by Rosman et al. (2021) showing that the acceptance of protective measures decreased over the course of the pandemic, leading to the assumption that the uptake of such measures would also decrease over time. However, Rosman et al. (2021) examined the period between March and July 2020, an early period where incidence values were decreasing or at a comparable low level, while in the present study, a time period with a more dynamic development of incidence values is covered (June 2020 to March 2021)including a sharp increase in reported infection numbers in autumn and winter 2020. In addition, various psychological theories outline that attitudes are not the only predictor for actual behavior (e.g., the theory of planned behavior; Ajzen, 1991) . In the context of COVID-19, for example, a study by Schillings et al. (2021) showed thatbesides the attitudeperceived expectations from the social environment (i.e., subjective norms) also play an important role for the intention to comply with social restrictions such as reducing social contacts or staying at home. Regarding sociodemographic differences, we found that women as well as younger and higher educated respondents on average take a higher number of protective measures. Effect sizes for these differences were small, but the results for gender are in line with previous research and could in part be explained by different attitudes influencing individual behavior. For example, women often feel more susceptible to diseases than men, which may then lead to a greater focus on preventive measures (for a review, see Bish & Michie, 2010) .
For education, our results are in line with previous research on other pandemics (e.g., Bish & Michie, 2010) , but there is a marked difference with research in context of the COVID-19 pandemic. For example, while our study seems to indicate an increasing uptake rate of protective measures with age, the study by Betsch et al. (2021) found a reverse effect. To better understand this discrepancy, it helps to look at the methodological differencesespecially with regard to the protective measures included in the questionnaire. For example, the study by Betsch et al. (2021) included more hygiene measures (e.g., covering the mouth when coughing, not touching one's face), while our study also considered changes in the social behavior (e.g., meeting friends or family less frequently, leaving one's home less frequently) and the use of the Corona Warn App. As shown in Table 3 , these particular items accounted for the largest differences in uptake regarding the education level of the respondents. Also, the paper by Betsch et al. (2021) analyzed data exclusively from March 2020 (though the COSMO study itself covers a larger timeframe, see Betsch et al., 2020; University of Erfurt, 2022). In addition, it is important to remember that not all people have the same resources or opportunities to take protective measures against an infectioneither because of their current job, their housing conditions or other limiting factors. Against this background, it is not surprising that studies have found that people with a lower educational status are more likely to be infected with SARS-CoV-2 (Hoebel et al., 2022) as well as perceiving the coronavirus as more dangerous (Wachtler et al., 2021 ). An even more diverse 13 pattern emerged with respect to the relationship between protective measures and age. Even though the number of taken protective measures was higher for the youngest age group (up to 39 years), the oldest age group (60 years and over) also showed a high uptake rate for some single measures, for example when it comes to washing hands more thoroughly or having food delivered more frequently. These mixed results are in line with the varied findings in previous studies. While a number of studies indicate that older people are more likely to engage in protective behavior (e.g., Atchinson et al., 2021) , others show no correlation with age (e.g., Helsingen et al., 2020) , or even an inverse correlation (e.g., Pasion et al., 2020) . The fact that there is no clear pattern indicates that other factors might also influence the link between age and protective behavior. For instance, the study by Pasion et al. (2020) showed that middle-aged adults had a higher risk perception regarding the coronavirus compared to older and younger adults. Further research on how age affects protective behavior should therefore focus more on the attitudes and risk perceptions of individuals in different age groups, as well as how their living conditions or other environmental aspects might influence their behavior (e.g., available resources, household size).
Consequently, the sociodemographic differences identified in this study indicate that some parts of the population may be of greater risk of becoming infected during a pandemic due to their less pronounced protective behaviorin this particular instance, men, people with a lower level of education and persons who are 40 years or older. This is especially concerning given the fact that particularly men and older people face a greater risk for a severe course of COVID-19 (Peckham et al., 2020; Katzenschlager et al., 2021) and people with a lower education level show higher infection rates (Hoebel et al., 2022) . This highlights the importance of communication formats or campaigns that address and provide information precisely to these target groups. However, the comparison with other studies also show the complexity of protective behavior during a pandemic. The possibility to implement many 14 measures is often also dependent on factors that one cannot directly influence oneself, such as the place of work or housing conditions. In order to better understand the protective behavior of the population and to derive better forecasts, future studies should also pay attention to the influence of individual circumstances on personal protective measures.
There are some limitations to this study. First, as the length of questionnaires is limited within omnibus telephone surveys, we assessed the protective behavior using a binary response scale (yes/no). A more granular response scale could have provided more information (e.g., frequency of behavior). Second, our list of protective measures is not exhaustive. That is, in addition to the nine measures we considered in our study, we cannot exclude that respondents may have shown other behaviors that could be classified as protective measures. As a final limitation, the serial cross-sectional study design cannot take into account the development of a single individual over time (intra-individual development), which only can be investigated using a longitudinal study design.
In conclusion, our results show that the population in Germany has shown a variety of behaviors to protect themselves or their family from an infection during the coronavirus pandemic. In times of greater infection risks, people adapted accordingly with a higher number of protective measures. However, based on sociodemographic differences, campaigns could be essential to educate older, male and lower educated persons on the necessity of protective behavior. Notes. a df numerator = 1, df dominator = 20.316; b df numerator = 2, df dominator ranging between 39.207 and 39.718, *** p < .001, ** p < .01, * p < .05.
The theory of planned behavior
The high prevalence of asymptomatic SARS-CoV-2 infection reveals the silent spread of COVID
Early perceptions and behavioural responses during the COVID-19 pandemic: A cross-sectional survey of UK adults
Bavarian State Ministry of Health and Care
Bestätigter Coronavirus-Fall in Bayern -Infektionsschutzmaßnahmen laufen
The four weeks before lockdown during the COVID-19 pandemic in Germany: A weekly serial crosssectional survey on risk perceptions, knowledge, public trust and behavior
Monitoring behavioural insights related to COVID-19
Demographic and attitudinal determinants of protective behaviours during a pandemic: A review
The effect of age on mortality in patients with COVID-19: A meta-analysis with 611,583 subjects
Occurrence and transmission potential of asymptomatic and presymptomatic SARS-CoV-2 infections: A living systematic review and meta-analysis
Computer assisted telephone interviewing (CATI) for health surveys in public health surveillance: Methodological issues and challenges ahead
Federal Ministry of Health
Chronik bisheriger Maßnahmen und Ereignisse [Coronavirus pandemic (SARS-CoV-2): Chronicle of measures and events to date]
Weighting, GESIS survey guidelines. Mannheim: GESIS -Leibniz Institute for the Social Sciences
Estimating ordinal reliability for likert-type and ordinal item response data: A conceptual, empirical, and practical guide
Predictors of COVID-19 severity: A literature review
German Federal Institute for Risk Assessment
The COVID-19 pandemic in Norway and Sweden -threats, trust, and impact on daily life: A comparative survey
Gender influence on health and risk behavior in primary prevention: A systematic review
Socioeconomic position and SARS-CoV-2 infections: Seroepidemiological findings from a German 19 nationwide dynamic cohort
SARS-CoV-2 transmission from people without COVID-19 symptoms
Can we predict the severe course of COVID-19 -A systematic review and meta-analysis of indicators of clinical outcome?
Serial cross-sectional data on the public's perception on the coronavirus during the first months of the pandemic in Germany
A procedure for objective respondent selection within the household
Adult Development, Psychology of
The SAGE Encyclopedia of Theory in Psychology
Beiträge zur Gesundheitsberichterstattung des Bundes: Armut, soziale Ungleichheit und Gesundheit [Contributions to federal health reporting: Poverty, social inequality and health
Empirische Belege für die zentrale Rolle der schulischen und beruflichen Bildung [Social inequality and health: Empirical evidence for the central role of schooling and vocational education
Voraussetzungen für eine moderne Gesellschaft [Healthy and educated: Prerequisites for a modern society
The AGE effect on protective behaviors during the COVID-19 outbreak: Sociodemographic, perceptions and psychological accounts
Male sex identified by global COVID-19 meta-analysis as a risk factor for death and ITU admission
Press and Information Office of the Federal Government of Germany
AHA
im neuen Alltag [The "AHA rules
On chi-squared tests for multiway contingency tables with cell proportions estimated from survey data
Werte) sowie Gesamtübersicht der pro Tag ans RKI übermittelten Fälle und Todesfälle [7-day incidences by federal states and counties (fixed values) as well as total overview of cases and deaths transmitted per day to the RKI]
Will COVID-19-related economic worries superimpose health worries, reducing nonpharmaceutical intervention acceptance in Germany?
New effect size rules of thumb
No way out? Intention zur Einhaltung von Corona-Ausgangsbeschränkungen im Kontext einer "extended theory of reasoned action
COSMO -COVID-19 Snapshot Monitoring
Socieoeconiomic inequalities in the prevalence and percieved dangerousness of SARS-CoV-2 infections in two early German hotspots: Findings from a seroepidemiological study
The authors would like to thank Kantar GmbH for managing data collection and the respondents for their participation in our study.
The authors declare that they have no conflict of interest.
None.
Ethical approval was not required for this study based on the following considerations: The study did not include medical aspects, person-identifiable data or sensitive or confidential data. No experimental manipulation or psychological tests were used. It was always possible for respondents to drop out of the survey before completion or to not answer one or more questions in the survey. In addition, data collection was carried out in line with the standards established by the Association of German Market Research Institutes (ADM; see https://www.adm-ev.de/en/standards-guidelines/). Respondents expressed their consent to participate in the surveys. All data were recorded and processed anonymously.
The authors declare that they have no conflict of interest.16