How accurately can the NICE Traffic Light System predict serious illness in acutely unwell children under five? A retrospective cohort study.

Talk Code: 
Amy Clark
Kathryn Hughes, Rebecca Cannings-John
Author institutions: 
Cardiff University, PRIME Centre Wales, Cardiff University Centre for Trials Research


The NICE Traffic Light System was created to facilitate the assessment of acutely unwell children under the age of five in primary care. It characterises their clinical features into ‘green’, ‘amber’, or ‘red’; corresponding to a low, intermediate, or high risk of serious illness respectively. Studies in international and emergency settings have demonstrated that this tool has good sensitivity but poor specificity in the detection of serious illness. However, no studies have evaluated the predictive value of this clinical decision rule for identifying seriously ill children in UK general practice, a setting with a different presenting population and lower incidence of serious illness. The aim of this study is to evaluate the predictive value of the Traffic Light System in identifying serious illness in acutely unwell children under five presenting to UK general practice.


This retrospective cohort study performed secondary analysis on a dataset of acutely unwell children in general practice, collected for the Diagnosis of Urinary Tract Infection in Young Children study (DUTY). This dataset was linked to routinely collected hospital data to identify the children who were admitted to hospital after consulting with their GP. Their clinical features captured in general practice were categorised according to the Traffic Light System, as part of a separate unpublished study. ICD-10 clinical diagnosis codes were used to identify the outcome of serious illness in those admitted to hospital. Logistic regression will be used to evaluate the predictive value of the Traffic Light categories.


This study is still in progress, with completion by May 2021. Current findings have revealed that 32% of 6,791 children presenting to general practices with an acute illness were categorised as ‘red’, 63% as ‘amber’, and 6% as ‘green’. GPs referred 1.6% of the ‘red’ children for same day hospital assessment (30/2149); 41 children from ‘red’ or ‘amber’ categories were referred overall (0.6%).Over 98% of children have been linked to hospital admission records (6703/6791). A total of 139 children seen in GP were admitted to hospital within 7 days; serious illness was diagnosed in 17 of these.


This study is important because no previous research has determined the predictive value of this clinical decision tool in UK general practices, where it is widely used and recommended by NICE. NICE guidelines advise that ‘red’ children require immediate hospital assessment. However, our initial findings suggest that high numbers of children fit the ‘red’ criteria, and that GPs are only referring a small percentage of these to hospital. The results of this study may highlight the need for amendments to the tool; to improve its ability to correctly identify the most unwell children, while preventing unnecessary hospital admissions for children who are more likely to have a self-limiting viral illness.

Submitted by: 
Amy Clark
Funding acknowledgement: 
I would like to thank the Wolfson Foundation for their funding of this project as part of their intercalated award bursary.