512 Comparison of seven paediatric early warning scores, including the proposed National PEWS for England, to predict critical deterioration events in hospitalised children: a retrospective cohort study. (17th August 2022)

• Record Type: Journal Article
• Authors: Bracken, Abbey; Sefton, Gerri; Lane, Steven; Team, Detect Study; Haines, Caroline; Clerihew, Linda; Fitzsimons, John; Lim, Emma; Carrol, Enitan
• Is Part Of: Archives of disease in childhood. Volume 107(2022)Supplement 2
• Page Start: A364
• Page End: A365
• Publication Date: 2022-08-17
• Subjects: Children -- Diseases -- Periodicals; Infants -- Diseases -- Periodicals; 618.920005
• DOI: 10.1136/archdischild-2022-rcpch.588
• Languages: English
• ISSNs: 0003-9888

Abstract

Abstract : Aims: Objectives: The development of the National Early Warning Score has seen widespread uptake in adult care, with over 90% of acute trusts utilising NEWS-2. A National Paediatric Early Warning Score (PEWS) is in development for England, and this study is the first to validate its performance in hospitalised inpatients. We compared the performance of the proposed National PEWS to detect a critical deterioration event (CDE) requiring emergency transfer from an inpatient ward to the Paediatric Intensive Care or High Dependency Units, with six other PEWS currently in use in the UK (listed alphabetically: Alder Hey, Bedside, Bristol, Irish, National, Newcastle and Scotland). Figure 1 : Parameters used in each PEWS. Methods: This was a single centre cohort study in a tertiary children's hospital in England. We screened the observations of all young people, aged under 18, who were hospital inpatients between March 2018 and February 2019. Observations taken on day-case, ED and PICU wards were excluded, as were patients aged 18 and over and those missing the core PEWS components (heart rate, respiratory rate, work of breathing, SpO2, use of supplemental oxygen, nursing concern). Any other missing data were assumed to be normal. Observations were excluded from the study if they were less than one hour prior, to or following a CDE. PEWS scores were either calculated at the point of care or retrospectively. The primary outcome was the occurrence of a CDE within 4, 6, 12 and 24 hours from the observations being taken. Performance was compared using Area under the Receiver Operating Curve (AUC). Each patient who had a CDE (n=250) was matched to two non-CDE patients of the same age range (0-1, 1-2, 2-7, 7-13, 13+ years) and admitting specialty who did not (n=500). If multiple controls were available, they were matched on length of stay (time to discharge compared to time to critical deterioration), age in months and date of admission to hospital in that order. Results: 423, 321 observations were independently analysed, representing 11, 601 patients and 17, 519 patient visits. Median age was 57 months (IQR 16 – 128). The most common admitting specialties were General Paediatrics (5569 patient visits, 31.7%), Haematology-Oncology (2682 visits, 15.3%) and General Surgery (2369 visits, 13.5%). CDEs occurred on 250 of these visits, involving 217 patients (42.8% female). Median age was 8 months (IQR 1 – 61.25). CDEs most frequently occurred in patients admitted under General Paediatrics, (89 CDEs, 35.6%), Cardiology (59 CDEs, 23.6%) and Respiratory (38 CDEs, 15.2%). Figure 2 : Performance of maximum PEWS to predict CDE at 24, 12, 6 and 4 hours before event. Optimal cut-offs for sensitivity and specificity for the maximum proposed National PEWS at 24 and 12 hours prior to CDE was ≥6, and at 6 and 4 hours prior, was ≥5. Conclusion: All PEWS evaluated, including the National PEWS demonstrated excellent discrimination for CDEs. This single centre study supports the widespread roll out of proposed National PEWS, for predicting the occurrence of CDEs in hospitalised children, but further validation is required in other settings.