COVID-19 related distress in the Swedish population: Validation of the Swedish version of the COVID Stress Scales (CSS)

Abstract

Background: The COVID Stress Scales (CSS) assess health- and contamination-related distress in the face of a medical outbreak like the ongoing COVID-19 pandemic. Though the CSS is translated into 21 languages, it has not been validated in a Swedish national sample.

Aim: Our general objective is to provide a translation, replication, and validation of the CSS and test its convergent- and discriminant validity in relation to anxiety, health anxiety, depression, and stress in the general Swedish population. We also present latent psychometric properties by modelling based on item response theory.

Methods: Participants consisted of 3044 Swedish adults (> 18 years) from a pre-stratified (gender, age, and education) sample from The Swedish Citizen Panel. Mental health status was assessed by validated instruments, including the CSS, PHQ-4, SHAI-14, and PSS-10.

Results: Results indicate that our Swedish translation of CSS has good psychometric properties and consists of 5 correlated factors.

Discussion: The CSS is useful either as a unidimensional or multidimensional construct using the CSS scales to measure key facets of pandemic-related stress.

Conclusions: The findings support the cross-cultural validity of the CSS and its potential utility in understanding many of the emotional challenges posed by the current and future pandemics.

Fig 1. Item Information Function (IIF) of CSS.

The Item Information Function (IIF) indicates how much information each CSS-question (css1-css36) can provide from the latent modelled construct of Theta (in this case corresponding to a general COVID-related stress). Some specific questions may provide more information at a relatively higher, or lower, level of Theta depending on for example question specificity and phrasing. The level Theta is standardized and in this particular case levels higher than zero on the x-axis may be assumed to assess a higher degree of COVID-related stress compared to values lower than zero where the COVID-related stress may be less salient. The y-axis represents level of information and indicates where on the Theta scale each item is most sensitive. In Fig 1 we can see that most items have a curve that is flat at the top which entails that they can provide the same amount of information at a relatively wide range of Theta. For example, item 21 provides the most information compared to the other items, and it demonstrates approximately the same level of sensitivity from a level of nearly -2 to 2 of Theta.

Fig 2. Test Information Function (TIF) of CSS.

The Test Information Function (TIF) is an aggregate measure of the Item Information Function (IIF) presented in Fig 1. The TIF shows how the entire instrument (css1-css36 combined) performs at different levels of Theta and at what level or magnitude of the latent construct that we can retrieve as much information as possible. Fig 2 shows that the CSS instrument provides the most information approximately at a Theta level of 2 and, conversely, that the margin of error (SE) is at the lowest point at this particular level, meaning that the precision of the CSS-questions is likely higher at a higher level of COVID-related stress.