How is anxiety measured in the Czech Republic? We took a closer look at the popular GAD-7 questionnaire

The INPSY research team, in collaboration with researchers from Charles University, validated the GAD-7 questionnaire—a widely used tool for screening generalized anxiety disorder. Since there are many different Czech translations in circulation, a large-scale methodological study was sorely needed. The research team utilized a large dataset of approximately 5,500 respondents across 20 samples and four translations and published their results in the highly prestigious Journal of Anxiety Disorders.

A significant portion of the population suffers from anxiety, and it appears that the prevalence of anxiety disorders has been increasing recently. It is therefore essential to have high-quality measurement tools available that can be used both for screening (i.e., identifying at-risk individuals for the purpose of more detailed diagnosis) and for researching the prevalence of anxiety disorders and other related factors.

There are a number of similar questionnaires, one of which is the seven-item GAD7 screening questionnaire (Generalized Anxiety Disorder 7-item questionnaire). In the past, the National Institute of Mental Health (NÚDZ), for example, used it in its widely publicized studies. You can actually try it out for yourself online on several different websites, such as here.

However, the GAD-7 is a foreign questionnaire, originally validated in English for the American population. Foreign questionnaires must always be “adapted” to the target language. This process involves not only linguistic translation but also research-based validation to ensure that the questionnaire functions as intended. And it is precisely this thorough adaptation and its research validation that were sorely lacking in our country.

Moreover, due to the questionnaire’s popularity, a paradoxical situation arose in which at least five different translations were available in Czech. However, different translations may function slightly differently, which then leads to reduced generalizability of studies. For example, if researchers reach certain conclusions in one study using one translation, we cannot automatically expect that identical results will be observed in another study using a different translation. Questions translated differently are also answered with varying degrees of ease. It can then easily happen that different translations will systematically differ in respondents’ average scores; in other words, the same respondent with the same level of anxiety may achieve a higher score in one translation than in another. This can ultimately affect the interpretation of so-called critical scores, which are used in practice to identify people at risk of anxiety—and upon which studies on the prevalence of anxiety in the population are also based.

A research team consisting of Hynek Cígler, Petra Pátková Daňsová, and Ondřej Prikner from INPSY, joined by Alena Javůrková and Jaroslava Raudenská from Charles University, attempted to address this issue. The researchers first identified 20 different research datasets in which the GAD-7 questionnaire had been used: not only various clinical and population studies, but also master’s and bachelor’s theses. These included all data available as of February 2020. Four different, independent translations of the questionnaire were used in these individual studies. Unfortunately, no data was available for the fifth translation, published on Pfizer’s website.

The researchers then examined various properties of the GAD-7 questionnaire both within each sample individually and collectively. First and foremost, the results from international studies were confirmed: the questionnaire measures anxiety very accurately (with high reliability), has a good factor structure (each question is unique, and the questionnaire captures a single psychological trait, not a “mixture” of them), and effectively distinguishes between individuals with and without anxiety. The most interesting result, however, emerged from a comparison of the individual translations, which, somewhat surprisingly, functioned very similarly, with the exception of a single question in one of the versions where a minor translation error occurred. The questionnaire therefore measures the same thing regardless of the translation, and the results of the variously translated versions can be directly compared. And this holds true not only across translations but also over time—for example, it is easy to compare patients before and after treatment and use the GAD7 to measure the effectiveness of therapy.

These results demonstrate that the GAD-7 questionnaire is robust to the translation process. The results of the published study may therefore be very useful for international researchers as well—it appears that the exact wording of the questionnaire items is not essential; the content itself is more important. This can be utilized in future translations of the questionnaire into other languages, or in cross-cultural studies that work with different language versions.

However, the article has a clear benefit, particularly for Czech researchers. On the one hand, it supports the comparability of individual language versions across previously conducted studies; on the other hand, it selected the most suitable translation from the four available (although, as we have already noted, the differences were negligible). Furthermore, it provides provisional norms for the general adult population (men and women separately) and the population of psychiatric patients, which can be used with caution. Other researchers may also appreciate that all source data and analytical scripts are available, which serve both to verify all calculations and may facilitate further research using the GAD7 questionnaire in the future.

Recommended citation:

Cígler, H., Daňsová, P. P., Javůrková, A., Raudenská, J., & Prikner, O. (2026). Validity and factor structure of the Czech GAD-7 across twenty samples and four independent translations. Journal of Anxiety Disorders, 103149. https://doi.org/10.1016/j.janxdis.2026.103149