Availability and Quality of Anthropometric Data in Swiss Childrens Hospitals: The SwissPedGrowth Project

OBJECTIVE: Anthropometric data are critical in paediatric care, routinely assessed during clinical visits, and available in electronic health records (EHRs). We describe the feasibility of extracting anthropometric data from heterogeneous EHR systems of Swiss childrens hospitals, evaluate their availability and quality, and assess the cohorts representativeness of the general population. METHODS: In this multicentre study (SwissPedGrowth), we retrospectively collected EHRs from patients <20 years who visited hospitals in Basel, Bern, Geneva, Lausanne, Luzern, St. Gallen, or Zurich between 2017-2023. Sociodemographic, administrative, and clinical information from EHRs were provided in a standardized way by a paediatric national data stream (SwissPedHealth), including the Swiss Neighbourhood Index of Socioeconomic Position (Swiss-SEP). We counted anthropometric recordings per visit to describe availability and used a self-developed and an existing (growthcleanr) algorithm to investigate data quality. To assess representativeness, we compared sociodemographic characteristics between SwissPedGrowth and the general paediatric population in Switzerland, computed standardized differences (effect size: 0.2 small, 0.5 medium, 0.8 large), and weighted the study population to reduce differences. RESULTS: We included 477,531 patients and 2,171,633 hospital visits; 54% boys, 71% Swiss, mean Swiss-SEP 65 (SD: 11), and median age at visit 6.3 [IQR: 2.3, 11.8] years. Height recordings were available for 20% of the visits, weights for 43%, and head circumferences for 5%, with better availability for inpatient stays than outpatient or emergency visits. Combining the self-developed and existing algorithm, 4% of heights and 3% of weights were flagged as outliers and 29% of heights and 31% of weights as carried forward from previous visits or same day duplicates. Sociodemographic differences between SwissPedGrowth and the general population were small or small-to-medium and disappeared after weighting. CONCLUSION: SwissPedGrowth demonstrates feasibility of extracting high-quality anthropometric data for paediatric growth research, but challenges regarding completeness and harmonization of EHR data across Swiss hospitals remain.