Introduction Paediatric standards are widely used by clinicians to understand how values of an important variable for an individual child compare to the typical range for a population of similar children, allowing them to identify and diagnose disease or deterioration. While there are published age-related standards for variables such as height and weight among healthy children, these are often unavailable for other uncommon variables, such as creatinine levels or gait, and in rare-disease populations. Our research explored analytical methods so that GOSH-specific standards can be computed from electronic health record (EHR) data and used to assess patients.

Methods We developed an analytical method within the GOSH clinical informatics platform (PICTURE) to build standards from EHR data. PICTURE allows one to define phenotype-based patient cohorts, extracts their data from the EHR, and applies generalised analytical methods to synthesise information derived from individual trajectories. We fit Generalized Additive Models for Location, Scale and Shape (GAMLSS) to construct growth standards. Using non-parametric (typically splines) flexible models, GAMLSS can fit standards to many types of data stored in the EHR such as vital signs and laboratory test results.

Results We constructed standards for two example applications in cardiology patients (n=3068). We produced age- and sex-specific weight standards and visualised how GOSH-cardiology patient weight gain differs from that in a healthy population. We also produced sex- and age-specific standards for serum creatinine and show how they contrast with the usual discretised normal ranges.

Conclusions This investigation shows how standards can be generated based on EHR data as part of the GOSH clinical informatics platform. Easy access to GOSH-specific standards has the potential to support clinicians identifying how patients are deviating from what would be expected for healthy or similar patients. Further work is needed to validate the accuracy and utility of information from single-centre standards.

Acknowledgements for funding or support This work is supported by the NIHR GOSH BRC. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. This work is supported by the Great Ormond Street Hospital Children’s Charity.