Logo

70. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V.

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS)
07.-11.09.2025
Jena


Meeting Abstract

Prediction of parental confidence in the management of fever in their children – current analyses of FeverApp Registry data over 4 years with advanced modelling methods

Juliane Hardt 1
Ricarda Möhler 2
Christian Siegemeyer 1
Ludwig Cölln 1
Moritz Gwiasda 2
Larisa Rathjens 2
Christopher Germann 1
David Martin 2,3
Ekkehart Jenetzky 2,4
1Witten/Herdecke University, Faculty of Health, Department of Human Medicine, Institute of Integrative Medicine, Witten, Germany
2Witten/Herdecke University, Faculty of Health, Department of Human Medicine, Institute of Integrative Medicine, GKLS, Witten/Herdecke, Germany
3Eberhard-Karls-University Tübingen, University Hospital Tübingen, Department for Pediatrics, Tübingen, Germany
4University Medical Center of the Johannes-Gutenberg-University, Department of Child and Adolescent Psychiatry and Psychotherapy, Mainz, Germany

Text

Introduction: The German Fever App Registry [1], [2] is one of six BMBF-funded patient registries of a national initiative for health services research [3]. Besides using the information library, parents can document fever episodes of their children and relevant covariables in the FeverApp. Objective of the registry is the investigation of guideline-adherent management of fever in children (incl. reducing visits to pediatric practices and hospital’s emergency rooms and medication compared to currently usual in healthcare). Also the parent’s confidence in the management of feverish episodes of their children can be documented, so potential predictors of this confidence and effects on healthcare measures and -usage are investigated [4].

Methods: FeverApp data from September 2019 to August 2023 are selected. Residents of Germany, parents as caregivers and children and adolescents aged <18 yrs have been included in this analysis. Standard data preprocessing of the registry data was applied. Due to the data distributions, analyses are focused on the 1st documented fever episode of the first child per family. Descriptive analyses and orienting binary logistic regression analyses are applied as preparation for the prediction of parental confidence. Building upon that, ordinal logistic regression analyses and machine learning (ML) methods optimizing the prediction error are applied.

Results: We included 11,364 children (52.7% male; mean age/SD: 24.2 months/26 mo.; 9.3% with migration background) in the analysis population. At least 1 fever episode in the last 12 months was reported for 58.2% of children. Of the documenting parents 87.6% were mothers and > 50% of parents had a higher education. During the current fever episode 51.6% of parents reported a high or very high confidence. During that episode, 12% of children were given antipyretics. Preliminary analyses show female gender of child, age <12 months, no fever episode during last 12 months, current febrile temperature, warning signs, dehydration reported and well-being of the child rated as very low to moderate as significant predictors of a very low to moderate parental confidence in the fever management in multivariable logistic regression models. More comprehensive results of statistical and data science (ML) methods will be presented at the conference.

Conclusion: The registry data are one of the first large app-based datasets to investigate occurrence of feverish episodes in children with accompanying symptoms and fever management methods by parents and help to understand predictors of the current situation of healthcare and guideline adherence in children’s fever. The citizen-science approach with a health app (EMA tool) is related with possible selection bias in participation, active app use for documentation and documentation quality. However, a mobile app presumably performs better in EMA data collection than classic paper or eCRFs. We expected some of the results, but also some new predictors could be identified. Ongoing analyses may discover more predictors and optimize prediction models. Providing health apps with an information library and low-threshold documentation opportunities are valuable in extending evidence on healthcare realities in regular family home environments.

The authors declare that they have no competing interests.

The authors declare that a positive ethics committee vote has been obtained.


Literatur

[1] Martin D, Jenetzky E. Die FieberApp-Register - (Ecological Momentary Assessment (EMA)) Ökologische Momentaufnahme des Fiebermanagements in Familien hinsichtlich der Übereinstimmung mit aktuellen Empfehlungen [The FeverApp Register Study – Ecological Momentary Assessment (EMA) of fever management in families with regard to concordance with current recommendations]. DRKS Register entry no. DRKS00016591 with study publications. Available from: https://drks.de/search/de/trial/DRKS00016591
[2] Schwarz S, Martin D, Büssing A, Kulikova O, Krafft H, Gwiasda M, Hamideh Kerdar S, et al. Sociodemographic Characteristics and Interests of FeverApp Users. Int J Environ Res Public Health. 2021;18: 3121. DOI: 10.3390/ijerph18063121
[3] Harkener S, Jenetzky E, Rupp R, Dell J, Engel C, von Bargen MF, et al. Recommended data elements for health registries: a survey from a German funding initiative. BMC Med Inform Decis Mak. 2024 May 27;24(1):136.  DOI: 10.1186/s12911-024-02535-x
[4] Möhler R, Jenetzky E, Schwarz S, Gwiasda M, Rathjens L, Szὃke H, Martin D. Parental Confidence in Relation to Antipyretic Use, Warning Signs, Symptoms and Well-Being in Fever Management - Results from an App-Based Registry. Int J Environ Res Public Health. 2022;19:14502. DOI: 10.3390/ijerph192114502