Seasonality of acute kidney injury phenotypes in England: an unsupervised machine learning classification study of electronic health records

Abstract

AbstractBackgroundAcute Kidney Injury (AKI) is a multifactorial condition which presents a substantial burden to healthcare systems. There is limited evidence on whether it is seasonal. We sought to investigate the seasonality of AKI hospitalisations in England and use unsupervised machine learning to explore clustering of underlying comorbidities, to gain insights for future intervention.MethodsWe used Hospital Episodes Statistics linked to the Clinical Practice Research Datalink to describe the overall incidence of AKI admissions between 2015-2019 weekly by demographic and admission characteristics. We carried out dimension reduction on 850 diagnosis codes using multiple correspondence analysis and applied k-means clustering to classify patients. We phenotype each group based on the dominant characteristics and describe the seasonality of AKI admissions by these different phenotypes.FindingsBetween 2015-2019, weekly AKI admissions peaked in winter, with additional summer peaks related to periods of extreme heat. Winter seasonality was more evident in those diagnosed with AKI on admission. From the cluster classification we describe six phenotypes of people admitted to hospital with AKI. Among these, seasonality of AKI admissions was observed among people who we described as having a multimorbid phenotype, established risk factor phenotype, and general AKI phenotype.InterpretationWe demonstrate winter seasonality of AKI admissions in England, particularly among those with AKI diagnosed on admission, suggestive of community triggers. Differences in seasonality between phenotypes suggests some groups may be more likely to develop AKI as a result of these factors. This may be driven by underlying comorbidity profiles or reflect differences in uptake of seasonal interventions such as vaccines.FundingThis study was funded by the National Institute for Health and Care Research (NIHR) Health Protection Research Unit (HPRU) in Modelling and Health Economics, a partnership between UK Health Security Agency (UKHSA), Imperial College London, and London School of Hygiene and Tropical Medicine. The views expressed are those of the authors and not necessarily those of the National Health Service, NIHR, UK Department of Health or UKHSA.Research in contextEvidence before this studyWe searched for articles in Medline using the terms (“Seasons/” OR “Seasons”) AND (“Acute Kidney Injury/” OR “Acute Kidney Injury” OR “AKI” OR “ARF”). We also search Embase using the terms (“Seasonal variation/” OR “Seasonal variation” OR “Season/” OR “Season”) AND (“Acute kidney failure/” OR “Acute kidney failure” OR “AKI” OR “ARF”. Articles published until 20/01/2023 in any language were included. Only two studies investigated seasonality of AKI in the UK and indicated winter increases in admissions. However, both studies aggregate AKI hospitalisations into quarterly counts and therefore were unable to show acute weekly changes in AKI admissions and timings of peaks. Studies outside of the UK varied in their conclusions of summer or winter increases in AKI admissions and the profile of patients driving this variation.Added value of this studyThis is the largest and most granular investigation of AKI seasonality in England, investigating 198,754 admissions in a weekly time series detecting acute changes in incidence and differences in peaks year to year. We demonstrate consistent peaks in the winter as well as acute peaks in the summer. Most records indicated AKI was diagnosed on admission therefore suggestive of community triggers of AKI. We included more data on the profile of patients than previously published studies. Our novel approach to investigate the profile of seasonal admissions using unsupervised machine learning suggests some groups may be more affected by seasonal triggers than others.Implications of all the available evidenceAKI is a common syndrome which leads to hospitalisation with a significant burden on the health system. We demonstrate a conclusive seasonal pattern to AKI admissions which has important implications on healthcare provision planning, public health, and clinical practice in England. Future research on AKI should take into account seasonality; uncertainty remains on the main drivers and aetiology of the seasonal patterns observed.