Intended for healthcare professionals


Parenteral penicillin before admission to hospital for meningitis

BMJ 2006; 332 doi: (Published 01 June 2006) Cite this as: BMJ 2006;332:1283
  1. Duncan Keeley (duncan.keeley{at}, general practitioner
  1. The Health Centre, Thame OX9 3JZ

    The benefit remains uncertain despite new evidence

    General practitioners worry about seeing—and even more about missing—meningococcal disease. We know that affected children can deteriorate very rapidly and that the disease has a high mortality. And we know that if we suspect this diagnosis we should give parenteral penicillin while arranging urgent transfer to hospital. This is the advice in the British National Formulary, the Drug and Therapeutics Bulletin, and the guidelines for general practitioners from the Meningitis Research Foundation.1

    It stands to reason that in a rapidly progressive bacterial infection the earliest possible administration of antibiotics should have a beneficial effect on outcomes. We carry the penicillin in our bags. But not everything that stands to reason proves to be the case—and the uncertainty of the evidence base for prehospital administration of penicillin in suspected meningococcal disease is added to by a paper published in this week's BMJ.2

    This study (p 1295) is one of a series reporting a large national case-control study of the clinical management of 448 children aged 0-16 presenting between 1997 and 1999 with meningococcal disease. It compares fatal and non-fatal cases. Previous papers have identified early symptoms of sepsis that medical staff should be aware of (leg pains, cold hands and feet, abnormal skin colour)w1 and possible deficiencies in hospital management associated with poorer outcomes.w2

    The present study is confined to the subgroup of 158 children in whom a diagnosis of meningococcal disease was suspected by the general practitioner at the initial contact, two thirds of whom were given prehospital penicillin. Children who were given penicillin were seven times more likely to die (adjusted odds ratio 7.45, 95% confidence interval 1.47 to 37.67). One likely explanation for this finding is that the children who were given penicillin were already more severely ill. This is intuitively probable—and, up to a point, they were. Information on severity of illness at initial assessment by the general practitioner was limited (an indication of incomplete documentation of vital signs), but on arrival at hospital the children given penicillin scored, on average, 2.5 points higher on a validated illness severity score with a range of 0-15 (the Glasgow meningococcal septicaemia prognostic score).

    But parenteral antibiotics given in advance of other supportive measures could lead to worse outcomes than if they had not been given. The management of children with meningococcal disease includes giving oxygen, correcting hypovolaemia with plasma expanders, and in some cases measures to provide inotropic or ventilatory support.w3 Rapid bacteriolysis before access to these supportive measures might cause clinical deterioration.

    Guidelines for ambulance staff include giving oxygen and intravenous fluids in transit to children with evidence of shock,w4 but gaining intravenous access in shocked children is difficult. Use of the intraosseous route is increasingly taught to paramedical and emergency department staff and could be learnt by GPs. This study found no association between time taken to reach hospital after assessment by the general practitioner (in the observed range of up to three hours) and severity score on arrival, whether or not the children received antibiotics before admission.

    The paper discusses the mixed evidence from previous studies, with British studies from the early 1990s suggesting benefit and more recent Danish studies pointing the other way.w5-w8 The authors of the Clinical Evidence chapter on meningococcal disease,3 and a systematic review of earlier observational studies also published in this issue (p 1299), find the evidence for pre-admission parenteral penicillin inconclusive.4 In the systematic review, a beneficial effect of prehospital penicillin was more likely where a higher proportion of patients received this treatment, but this finding, as the authors acknowledge, could also be accounted for by severity confounding.

    The strength of the study by Harnden et al lies in restricting the analysis to children in whom the diagnosis was made or suspected by the general practitioner—the patients for whom the decision to give penicillin arises. The other observational studies do not differentiate between such patients and those in whom the diagnosis was not made at first contact and who might therefore have reached hospital and started definitive treatment with more delay. In the overall case-control study group such children arrived at hospital a median of 12 hours later than those given penicillin.2 The apparent benefit of parenteral penicillin in the earlier British studies might have been accounted for by earlier hospital admission in the children receiving penicillin. A randomised clinical trial, as called for by the authors, would be ethically justifiable given these uncertainties, but very difficult to organise. Without such a study, further analysis of improved observational data—preferably collected in an international collaboration5—may be the next best thing.

    Should our practice with respect to prehospital parenteral penicillin change? On the basis of this study alone, probably not. But the Meningitis Research Foundation will undoubtedly be considering the implications of these findings and forming a view, which frontline practitioners will be keen to hear. Single case-control studies, highly prone to bias and confounding, are not the most reliable form of evidence. But accumulated evidence from case-control studies can build up to justify a counterintuitive policy change which proves beneficial: witness the story of prone sleeping position and sudden infant death.

    Meanwhile we should remember the undoubted contributions general practitioners can make: educating parents; early diagnosis, with better awareness of early sepsis symptoms and signs; and rapid transfer to hospital with optimum supportive care, with the general practitioner waiting for and, if necessary, travelling with the ambulance. Measuring, documenting, and monitoring pulse, respiratory rate, and capillary return—not something that general practitioners routinely do—is important for detecting hypovolaemic shock, which is the main immediate life threatening feature of septicaemia. Doing this and getting the child to hospital fast—preferably while instituting intravenous or intraosseous bolus fluid replacement if shock is present—may offer the best opportunity for reducing morbidity and mortality. These measures may be more important for improving survival than administering parenteral penicillin in the community.


    • Competing interests None declared.

      Embedded Image References w1-w8 are on

      Research p 1295


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    View Abstract

    Log in

    Log in through your institution


    * For online subscription