Helen Salisbury: Early warning scores and medicine by numbers

“Do you remember that child you saw two days ago?” All doctors dread what might come next: we fear that something bad has happened because we missed a clinical sign or made a poor judgment. This kind of worry keeps us all alert, even at the end of a busy day as the duty doctor.

Hospitals have developed scoring systems to help medical and nursing staff gauge how sick a child is, and the Royal College of Paediatrics and Child Health has suggested that these should be rolled out in the community.12 There could be advantages not only in improving detection of children at risk of deterioration but also in shared understanding and ease of handover.

However, research recently published in the British Journal of General Practice found that these scores weren’t useful in primary care.3 They lacked both sensitivity, meaning that they missed a significant proportion of children who needed hospital care, and specificity, advising escalation for many who could safely be looked after at home.

These findings don’t come as a surprise to practising GPs. Early warning systems used on a hospital ward were developed for a different baseline population, as the children there have already been judged to be significantly unwell. Most of the children we see in the community, fortunately, are not.

The hospital systems also incorporate data not routinely collected in a GP surgery: for instance, while we take account of pulse, temperature, and respiratory rate, we rarely measure blood pressure in children. Instead, we use many other pieces of information to decide how sick a child is. Changes in eating, drinking, and play are key: the toddler with an alarmingly high temperature who is energetically investigating my cupboards is unlikely to be significantly unwell, whereas the 18 month old who sits quietly on her father’s lap and doesn’t protest when I examine her ears and throat is far more concerning.

In a traditional GP setting, we have the advantage of prior knowledge of families and how they respond to their children’s illnesses. It can take a lot to make an experienced mother decide to bring her feverish child to the doctor, and I’d pay particular attention to her worry and be especially wary of being too quick to reassure. There may also be situations where, even though the child isn’t in immediate danger, concerns about the family’s capacity to respond to “safety netting” advice would change my threshold for referral.

Assessing acutely unwell children, and accurately picking out those few who need hospital care from the vast majority who don’t, isn’t something that can be done by algorithm or by following a protocol. It’s a skill built up over time, with baseline knowledge supplemented by repeated experience. And it’s yet another reason why acute and urgent care should remain an integral part of general practice rather than being separated into same day access hubs, staffed mostly by non-doctors.45 Seeing a succession of snotty toddlers may not be everyone’s idea of a fun day at the surgery, but if GPs don’t practise this skill they may lose it or never even acquire it.

Medicine by numbers, delegated to less trained healthcare staff, is not an option. This recent study confirms that it would result in overburdened paediatric emergency departments and an increased likelihood of children at risk of deterioration slipping through the algorithm.