Dr Chantal Simon looks at the various risk scores that can be used to diagnose serious illnesses in children
All GPs are very nervous about missing a possible serious bacterial infection in a child. This is particularly true for small children as symptoms and signs may be more non-specific in this age group. Therefore, clinical decision tools to assist with identification of sick children are attractive for GPs.
I was recently asked to comment on the usefulness of a new score to identify serious bacterial infection in children [1]. The new score uses eight parameters: developmental delay, risk factors for infection, state variation (i.e. how alert and responsive the child is), temperature, capillary refill time, hydration status, respiratory rate and level of hypoxia. Based on these parameters, a score is derived. A score greater than eight suggests the possibility of severe bacterial infection.
My first observation is that the data used to validate this new score are entirely from children presenting to emergency departments. There is no evidence for the validity of this score in a primary care population, where children perhaps present earlier, and many children are safely managed with less severe infections without referral. For this reason I would personally be reluctant to use this score before further validation using a primary care population had been undertaken.
My second observation is that in the sample of 1447 children, 66 had proven serious bacterial infections. Only 10 had scores over eight (15%) whilst 56 (85%) had scores under eight. Furthermore, 21 had scores under five (31%). This highlights the inherent danger of clinical risk scores. No score is perfect, and reliance on a score to make a clinical decision may result in missed diagnoses with serious consequences.
There are a number of such clinical risk scores around for diagnosis of sick children. The Yale Observation Scale and the Young Infant Observation Scale are widely used in various guises all over the world. Both these scales have extensive validation data for their target population of small children. The 'SICK' score [2] has the advantage that it incorporates a computerised programme to calculate and interpret the score, and is validated for children up to the age of 12, but it has limited validation data using samples from emergency departments only.
In the UK, the most widely used scoring system is the NICE traffic lights system [3]. Most GPs use the NICE traffic lights despite some criticisms that they are not sufficiently evidence based, that they do not allow for clinical judgement enough, and possibly overdiagnose serious illness. The traffic lights are incorporated into many GP clinical computer systems.
There is considerable overlap between the NICE criteria for serious illness in children under five and the criteria for this new score. The only factors in the score that are not also in the 'red' panel of the NICE traffic light scoring system (for which immediate admission is recommended) are the weightings for children with developmental delay or risk factors for infection.
Although the new score has the advantage that it is validated in children and young people up to the age of 17, the median age of the children assessed was 19 months, with relatively few children over the age of five in the sample population (20%). I suspect that most GPs would lower their threshold for admission for children with risk factors for serious infection and developmental delay anyway and so I do not think that this score improves significantly on the traffic light system.
Dr Chantal Simon is a GP in Bournemouth and editor of the Oxford Handbook of General Practice
Dr Chantal Simon
