Epidemic risk perceptions in Italy and Sweden driven by authority responses to COVID-19 | Scientific Reports
Study population
We included data from an anonymous survey on public risk perception carried out in Italy and Sweden in two different periods of the COVID-19 pandemic. Detailed information on the study has been published elsewhere31. In brief, the survey explores the public risk perception for nine threats (epidemics, floods, droughts, earthquakes, wildfires, terror attacks, domestic violence, economic crises, and climate change). Data were collected throughout a one-week period in August and November 2020. The samples were independent, and derived from two existing survey panels of 100,000 individuals in each country, set up by Kantar Sifo marketing research company32, and should be considered representative of the Swedish and Italian population for sex and age. Around 8000 individuals in the pool were invited to participate, if they did not reply, up to two reminders were sent. The capital regions were overrepresented: with a 1/9 sampling ratio in Italy and 4/6 sampling ratio in Sweden) (Supplementary Fig. 1). The missing data was quite low, < 5% for almost all variables included in the study, with the exception of political orientation in the Italian context where 20% of individuals preferred not to answer. The total sample included 8322 individuals. 4154 individuals participated in the survey in August (N = 2033, mean age 50.3 years, 53.0% of females in Italy and N = 2121, mean age 49.3 years, 49.9% of females in Sweden) and 4168 in November (N = 2004, mean age 49.4 years, 50.7% of females in Italy and N = 2164, mean age 47.9 years, 51.4% of females in Sweden).
Individuals that lived in the capital region were overrepresented, specific weights were applied in the analysis to take this into account. The present study was approved by the Italian Research Ethics and Bioethics Committee (Dnr 0043071/2019) and the Swedish Ethical Review Authority (Dnr 2019-03242). The study was carried out in accordance with the ethical standards set by the European Union under Horizon 2020 (EU General Data Protection Regulation and FAIR Data Management). Participants were informed that the participation was voluntary and they give their inform consent to participate in this study when completing the survey.
Risk perception of epidemics
The present study considered the public risk perception of epidemics considering seven domains: the likelihood of epidemics, epidemic impact on the individual and on the population, individual and authority preparedness, individual and authority knowledge of epidemics with a Likert-type scale ranging from 1, minimum to 5, maximum.
Predictors of the risk perception
Information on direct experience of an epidemic and socio-economic factors such as age, gender, employment (yes vs. no), relative income (from 1 to 5), university education (yes vs. no) were collected in the survey and included in the present study as possible predictors of risk perception.
Excess mortality
Excess mortality at regional level in Italy and Sweden during the first wave of the COVID-19 pandemic (15th February–15th May for Italy and 1st March–31st May for Sweden) was considered in the study. Regional level was defined according to Nomenclature of Territorial Units for Statistics (NUTS) 2 classification of the European Union33. We retrieved data on excess mortality among the Italian regions from Scortichini et al.17. To estimate the excess mortality in Sweden, we compared the COVID-19 outbreak with the pre-outbreak period. A two-stage interrupted time-series approach, that relied on a Poisson model with a function that constrains the excess risk as null at the beginning of March 2020, was used to calculate the excess mortality at the Swedish regional level34. The model was adjusted for time-varying confounders such as (i) seasonality using a natural spline term with 3 knots, (ii) indicators for the day of the week, (iii) air temperature using a term for the mean daily temperature. The information on temperature was retrieved from the ERA-5 reanalysis data set on the Copernicus climate data store35. We performed mixed-effects Poisson regression models with a random term for NUTS3 administrative units to calculate the excess mortality at regional (NUTS2) level taking into account the heterogeneity among NUTS3 administrative units.
National policy response
The Stringency index18 is a national response index and is used to quantify the measures implemented in response to the COVID-19 pandemic. The Stringency Index is a daily measure at country level that considers nine domains: school closures; workplace closures; cancellation of public events; restrictions on public gatherings; closures of public transport; stay-at-home requirements; public information campaigns; restrictions on internal movements; and international travel controls. In this paper, the level of national policy response was used as an ecological variable with four levels (Sweden up to August, Italy up to August, Sweden up to November and Italy up to November) and was defined as the area under the curve of the Stringency Index for each country, between two successive days up to the 5th August 2020 (first survey) and the 4th November 2020 (second survey). This measure was standardized on the value of Sweden in August (considered as the reference).
Statistical analysis
Possible differences in means and confidence intervals for seven items of risk perception between countries and over time were graphically presented using forest plots and stratified by country and period. Effect modification by country and period was examined using ordinal logistic regression models with risk perception (independent variables) and country and period as dependent variables. Results were presented as (i) Odds Ratios (ORs) for each country and period strata, (ii) ORs for country within strata of period and for period within strata of country, and (iii) interaction measures on additive and multiplicative scales36.
Second, multivariable ordinal logistic regression models were performed to evaluate the association of gender, age, employment, relative income, university education and experience of epidemics as possible predictors with the seven domains of risk perception (independent variables). The analysis was stratified by country and period.
Third, we examined if the risk perception varied according to which extent an area was affected by the first wave of the COVID-19 pandemic. We compared the means and confidence intervals for seven items of risk perception between the most affected region in terms of excess mortality (Stockholm region in Sweden about 60% excess mortality and Lombardy region in Italy about 100% excess mortality) and comparison with the rest of the country. Then, ordinal logistic regression models were performed to examine if the excess mortality at regional level (dependent variable) was associated with domains of risk perception (independent variables) stratifying for country and adjusting for gender, age, and relative income. Finally, the association between the level of implemented measures and risk perception was explored using adjusted ordinal logistic regression models.
The use of ordinal logistic regression models relied on the assumptions that the effect was linear on the log scale and that each independent variable had an identical effect for one unit increase of the ordinal dependent variable (proportional odds). Along with this, the goodness of fit of the ordinal logistic models was tested using the Deviance goodness of fit test. No multicollinearity among independent variables and correlation among errors from the models were detected.
As has been suggested in the literature, there are considerable risks in misinterpreting p-values37. Therefore, we opted to interpret the estimates in terms of possible direction of the effects and using ORs and 95% Confidence Intervals (CIs), which contain information on significance. Specifically, the width of the confidence interval and the size of the p-value are related: the narrower the interval is, the smaller the p-value is. Moreover, the confidence interval gives additional information that is related to the magnitude of the effect being investigated.
Statistical analyses were performed using Stata version 15.0 (StataCorp, College Station, TX, USA) and R version 4.0.338.
This article has been archived for your research. The original version from Nature.com can be found here.
