Background: In March 2022, a concerning rise in cases of unexplained pediatric hepatitis was reported in multiple countries. Cases were defined as acute hepatitis with serum transaminases >500 U/L (aspartate transaminase [AST] or alanine transaminase [ALT]) in children aged 16 years or younger. We explored a simple federated data analytics method to search for evidence of unreported cases using routinely held data. We conducted a pragmatic survey to analyze changes in the proportion of hospitalized children with elevated AST or ALT over time. In addition, we studied the feasibility of using routinely collected clinical laboratory results to detect or follow-up the outbreak of an infectious disease.
Objective: We explored a simple federated data analytics method to search for evidence of unreported cases using routinely held data.
Methods: We provided hospitals with a simple computational tool to enable laboratories to share nondisclosive summary-level data. Summary statistics for AST and ALT measurements were collected from the last 10 years across all age groups. Measurements were considered elevated if ALT or AST was >200 U/L. The rate of elevated AST or ALT test for 3-week- to 5-year-olds was compared between a period of interest in which cases of hepatitis were reported (December 1, 2021, to August 31, 2022) and a prepandemic baseline period (January 1, 2012, to December 31, 2019). We calculated a z score, which measures the extent to which the rate for elevated ALT or AST was higher or lower in the period of interest compared to a baseline period, for the 3-week- to 5-year-olds.
Results: Our approach of sharing a simple software tool for local use enabled rapid, federated data analysis. A total of 34 hospitals in the United Kingdom, the Netherlands, Ireland, and Curaçao were asked to contribute summary data, and 30 (88%) submitted their data. For all locations combined, the rate of elevated AST or ALT measurements in the period of interest was not elevated (z score=-0.46; P=.64). Results from individual regions were discordant, with a higher rate of elevated AST or ALT values in the Netherlands (z score=4.48; P<.001), driven by results from a single center in Utrecht. We did not observe any clear indication of changes in primary care activity or test results in the same period.
Conclusions: Hospital laboratories collect large amounts of data on a daily basis that can potentially be of use for disease surveillance, but these are currently not optimally used. Federated analytics using nondisclosive, summary-level laboratory data sharing was successful, safe, and efficient. The approach holds potential as a tool for pandemic surveillance in future outbreaks. Our findings do not indicate the presence of a broader outbreak of mild hepatitis cases among young children, although there was an increase in elevated AST or ALT values locally in the Netherlands.
Keywords: Ireland Curacao; Netherlands; United Kingdom; acute hepatitis; all age groups; children; data analytics; disease surveillance; federated analytics; hospital; laboratory; outbreak detection; outbreaks; pandemic preparedness; pandemic surveillance; pediatric hepatitis; pre-pandemic; public health; single center.
© Maaike C Swets, Steven R Kerr, Brian MacKenna, Louis Fisher, Merel van Wijnen, Diederik Brandwagt, Paul W Schenk, Pieter Fraaij, Leonardus G Visser, Sebastian Bacon, Amir Mehrkar, Alistair Nichol, Patrick Twomey, Philippa C Matthews, ISARIC4C Hepatitis Study Group, Malcolm G Semple, Geert H Groeneveld, Ben Goldacre, Iain Jones, J Kenneth Baillie. Originally published in JMIR Public Health and Surveillance (https://publichealth.jmir.org).