Intended for healthcare professionals


Recognising serious illness in feverish young children in primary care

BMJ 2007; 335 doi: (Published 30 August 2007) Cite this as: BMJ 2007;335:409
  1. Anthony Harnden, university lecturer in general practice
  1. Department of Primary Health Care, University of Oxford, Oxford OX3 7LF
  1. anthony.harnden{at}

    Increasing the opportunity for prompt clinical assessment is the priority

    The diagnosis and management of children with fever is an important part of primary care. Parents and children put their trust in general practitioners, who rightly worry about making a correct judgment. Although general practitioners have substantial clinical experience of assessing febrile children, half of children with meningococcal disease are sent home at first primary care consultation.1 How can this be and what can we do to improve our assessment of febrile children?

    Several diagnostic difficulties face us. Firstly, serious bacterial infection in children is now rare. The successful introduction of Haemophilus influenzae type B and meningococcal C vaccines into the UK childhood immunisation schedule has led to a dramatic reduction in the prevalence of invasive infections caused by these organisms.2 3 Early reports suggest that the recently introduced conjugate pneumococcal vaccine will do the same.4 Studies have reported a large variation in estimated incidences of invasive bacterial disease in preschool children, but the quoted 1%5 is almost certainly now a considerable overestimate. In England the only serious febrile illness in childhood whose incidence is increasing is Kawasaki disease, and that may be the result of improved recognition.6

    The time point in the illness that the child is seen is crucial. Early symptoms of meningococcal disease—coryza, sore throat, nausea—mimic those of common and self limiting viral infections.1 Reference to the precise time that the illness started may be helpful, but it is the velocity that is of key importance.7 The problem is that the velocity may not be linear or even exponential but may have sudden unexpected changes in acceleration resulting in marked and unexpected clinical deterioration. Indeed the established clinical paradigm of a dichotomous split between “viral” and “bacterial” infections is too simplistic—the velocity and clinical course reflects a complex interaction between viral infection, bacterial pathogens, and host response.

    Evidence is lacking about the utility of vital signs in primary care. No established reference values exist. For example, we know that fever increases heart rate and that the rate varies according to the age of the child, whether the child is crying, and when antipyretic medication was last given. Children of the same age have different baseline heart rates and these rates may respond differently to fever irrespective of the severity of illness. Moreover, precise measurement of heart rate is difficult, especially in very young children with higher rates. Measurement error is a problem even when machinery is used.

    Changes in NHS policy have led to the primary care of febrile children presenting outside office hours being delivered by an increasing number of professional groups. Doctors, nurses, staff working for the telephone helpline NHS Direct, out of hours centres, and accident and emergency departments may all have different levels of skill and experience. This is a major concern because the most solid evidence for recognising clinical severity in febrile children in primary care is a global assessment by an experienced clinician.8 9 The global assessment involves eliciting a clear history and careful observation of signs, including alertness, activity, colour, and respiratory effort.

    Concerns from paediatricians about the late diagnosis of children with serious infection and the fragmentation of primary care led to the recent publication by the National Institute for Health and Clinical Excellence (NICE) of guidelines for the assessment and initial management of young children with feverish illness.10 The authors' objectives are laudable and the literature review is comprehensive, but their recommendations rely too much on consensus techniques and widespread consultation rather than being a rigorous interpretation of the evidence.

    Although it may be practical in triage settings, it is premature to recommend that every febrile child visiting a general practitioner should routinely have a measurement of temperature, heart rate, respiratory rate, and capillary refill. A careful global assessment, examination, and medical record are rightly considered good practice. But over-reliance on vital signs with a low positive predictive value may result in the inappropriate referral of large numbers of children while children who develop a serious illness are sent away. General practitioners must not be persuaded to disregard their intuition. Nor does the “traffic light” system proposed by NICE add much value: children with “green” features are self evidently currently well; children with “red” features are very sick; and those with the “amber” features described—wakes only with prolonged stimulation and has nasal flaring and swelling of a limb or joint—would cause general practitioners to actively intervene in most circumstances.

    To improve the care of children with feverish illness in primary care we should be offering less telephone advice and more opportunities for a prompt clinical assessment. We should recognise that we are seeing only a brief snapshot of a dynamic illness and should always empower and make it easy for the parent to consult again—even a few hours later—if symptoms deteriorate. We should trust our clinical intuition and refer and re-refer if concerned. Meanwhile a pressing need exists for more primary care research into the time course of febrile illnesses and the utility of a combination and sequential record of vital sign measurements before concrete recommendations can be made.


    • Competing interests: None declared.

    • Provenance and peer review: Commissioned; not externally peer reviewed.


    View Abstract