Is evidence in the eyes of the beholder?


Moyer et al. conclude that 75% of primary interventions in hospitalized children are supported by high-level evidence.1 Will others reach the same conclusions?

Evidence was categorized as level I if at least one randomized trial clearly supported the use of the intervention and no trial showed clear evidence of harm; as level II if the intervention was supported by convincing non-experimental or observational evidence; and as level III if no substantial supporting evidence for the intervention was found in the literature. The discussion alludes to the potential assessor bias in attributing the level of evidence being used by clinicians. For example, the use of a nebulized beta agonist in bronchiolitis is interpreted as an example of high-level evidence supporting treatment. Although some Australian paediatricians use beta agonists in bronchiolitis, very few would believe that there is a strong case for their use in a 2-month-old infant with bronchiolitis.2 Furthermore, there is published evidence of harmful effects, as judged by worsening oxygen saturation.3

A second limitation of the report is that only the primary intervention was considered. Thus, every child with asthma was judged to have level I evidence for treatment because all received a nebulized beta agonist and oral or intravenous corticosteroids. Given that the study was conducted in early 1998, we would accept the use of nebulizers rather than spacer devices. However, many unanswered questions remain, such as what evidence was used to determine whether the child required admission to hospital,4 whether ipratropium bromide should be used as adjunctive therapy, whether supplemental oxygen was administered, what criteria were used to decrease intensity of therapy, what criteria were used to determine the timing of discharge from hospital,5 whether appropriate therapy was prescribed at discharge, whether a written asthma plan and arrangements for following up were provided, and whether unnecessary investigations were ordered (such as routine chest X-rays), or inappropriate drugs administered (such as anti­biotics and cough suppressants).

A more detailed analysis of these aspects might have led to the much more pessimistic views of hospital-based management that have been reported by others. We also need to consider whether or not to extrapolate results from adult studies. For example, a Cochrane Review has demonstrated that the addition of intravenous aminophylline to beta-2 agonists in adults with acute asthma is not of additional benefit, but has increased adverse effects.6 This review recommends against the use of aminophylline, because of side effects. How should we interpret the level of evidence about the use of aminophylline in children? Clearly, there is room for considerable variation in opinion.

Similarly, the use of oral glucose electrolyte solution in gastroenteritis is welcome. What about inappropriate use of intravenous fluids, antibiotics, anti-emetric agents, anti­diarrhoeal agents and paracetamol that has been noted previously?7,8 Many children received antibiotics for a variety of infections − therapy that was categorized as level II evidence. Consideration of the antibiotic used and the duration of therapy would have provided an opportunity to evaluate whether evidence had been followed (such as a 10-day course of penicillin for tonsillitis), or whether there was no evidence regarding appropriate duration of therapy. Was this the right antibiotic at the right time for the right duration?

Nearly one-quarter of the children in Professor Moyer's study were admitted for observation or evaluation only, and had no primary interventions. As implied above, many of us would regard the children in this category, those whose bronchiolitis was observed or who received supplemental oxygen, as having care that was more evidence-based than those called level I, because they received beta agonists. Similarly, the children who had toxic ingestion of substances for which no specific treatment is recommended benefited from masterly inactivity. Unfortunately, evidence-based medicine that bases its results on placebo controlled trials has difficulty in coping with situations in which the optimal approach is to withhold any intervention. On the other hand, some admissions to hospital may have been of no potential value, and, indeed, may have caused harm (such as psychological harm or physical harm in the form of nosocomial infection).

Moyer et al. have raised important questions and have highlighted the need to assess practice from an evidence base. In their correspondence with the Journal, they wrote ‘This is at best a very rough description of the extent to which evidence is used in paediatric practice, but is useful in comparing our specialty with others, and in pointing out the need for research to support all levels of paediatric practice’. Let us heed those words and not resort to simplistic summaries that tell us that X per cent of paediatrics is evidence-based. These are pilot data, which are not conclusive. The challenge is to build on this work.