Combining natural language processing with patient-reported outcome measures scores to investigate the association of pandemic regulations with anxiety in children

The COVID-19 pandemic and the governmental regulations taken during the pandemic have been associated with the experienced anxiety of children. However, there is a gap in research that explains how and which regulations were associated with higher anxiety. In this study we combine quantitative and qualitative data through structural topic modelling (STM) and sentiment analysis.

Children and adolescents (N = 3,821, aged 8–18 years old) completed an open-ended question; “How are the corona-regulations for you?”, and the PROMIS Anxiety questionnaire, bi-annually from April 2020 until March 2023 (7 measurement occasions). This yielded 7,605 open-ended responses, of which 6,672 contained anxiety, which was classified as (sub)clinical for PROMIS T-scores > 50.8. We applied STM to the open-ended responses to identify relevant topics, using dichotomized Anxiety as covariate.

We used sentiment analysis to assess the affective state of responses and obtain polarity scores for each response (ranging from − 1 (negative) to + 1 (positive)). We identified the topics of importance to children during the pandemic and assessed the affective state of these responses per topic in the whole sample and split by presence of (sub)clinical anxiety. Eight topics emerged: (1) Adaptation/Resilience, (2) Distress, (3) Adherence to social distancing, (4) Limited social and family contact (5) Restrictions in activities/boredom, (6) Future perspectives, (7) Homebound (school closures), and (8) Lack of celebratory events.

Compared to their peers, children with (sub)clinical anxiety reported significantly more negative sentiment, particularly regarding adaptation/resilience (-0.19 vs. -0.10, Cohen’s D = 0.24) and distress due to lockdown (-0.06 vs. 0.02, Cohen’s D = 0.20). They also expressed stronger emotional language when discussing these topics. We successfully identified topics relating to the governmental regulations that are associated with (sub)clinical anxiety during the COVID-19 pandemic.

This study shows that children with (sub)clinical anxiety scores described the pandemic period and regulations slightly more negatively, which may indicate that they have more problems with coping and adapting to lockdown measures. This work was funded by the Netherlands Organisation for Health Research and Development (ZonMW) (grant: 10430372310005). Child and Adolescent Psychiatry & Psychosocial Care, Amsterdam UMC Location University of Amsterdam, Emma Children’s Hospital, Meibergdreef 9, Postbox 22660, 1100 AD, Amsterdam, The Netherlands Michiel A.

Luijten, Hedy A. van Oers, Josjan Zijlmans, Lotte Haverman & Tinca J. Polderman Amsterdam Public Health, Mental Health, Amsterdam, The Netherlands Amsterdam Reproduction and Development, Child Development, Amsterdam, The Netherlands Michiel A. Luijten, Hedy A. van Oers & Lotte Haverman Amsterdam University Medical Center, Vrije Universiteit Amsterdam, Epidemiology and Data Science, Amsterdam, The Netherlands Health Data Intelligence and Life Sciences, Oracle, Austin, TX, USA Nuffield Department of Orthopaedics, Rheumatology and Musculoskeletal Sciences, University of Oxford, Oxford, UK Amsterdam Public Health, Quality of Care, Amsterdam, The Netherlands LUMC Curium – Child and Adolescent Psychiatry, Leiden University Medical Center, Leiden, The Netherlands Jacintha M.