A careless shoe-string, in whose tieI see a wild civility:Do more bewitch me, than when artIs too precise in every part. “Delight in Disorder,” Robert Herrick (1591–1674)
Herrick's 17th-century sonnet seems the antithesis of the scientific movement toward evidence-based medicine. And, of course, it mostly is: there is no place for the “careless” or the “wild.” Yet those of us who still exalt in the “art” of medicine are less “bewitched” when guidance becomes “too precise in every part.” This is a subversive sentiment and, if applied to our patients, must be done so thoughtfully and responsibly. However, it may lead us to a useful insight: evidence-based medicine is not incompatible with significant practice variation.
In this month's issue of Pediatrics, Goldman et al1 describe (with a perceptible sense of dismay) practice variation across Canadian pediatric emergency departments in the management of febrile infants younger...
I agree with Dr. Hampers that a likely explanation for the variation in the management of febrile infants is the fact that pediatricians vary in their tolerance of risk, but I think there is more to it than that. I think that most pediatricians think that not all febrile infants carry the same risk of having an SBI. Despite the "dismay" of Goldman and colleagues and probably many febrile infant guideline developers, there is no convincing evidence that variations in management are associated with differences in outcomes.
Wenneberg's important pioneering studies of variations in the rates of tonsillectomy or hysterectomy in different, but apparently similar, geographic areas gave variation a bad name--and justifiably so. When variation in practice leads to substantial differences in clinical outcomes, then figuring out ways to encourage clinician to follow a guideline makes sense. However clinicians faced with a heterogeneous group of infants whose one common feature is fever and in whom a tiny fraction have an SBI, it's hard to see how one can sell adherence to a guideline.
Goldman and colleagues use the curious term "outcome variable" to describe "variation in treatment decision." Treatment decisions are, of course, process variables. The outcome of interest is whether the infant had an SBI or not. Presumably the authors have the data to determine if there was any association between the independent variable of "followed the guidelines or not" and the dependent or outcome variable of "had an SBI." They indicate that they interviewed parents 7-10 days later so presumably they have the data to answer this question. The fact that they didn't present the data or test their implied hypothesis that not following guidelines is bad (not because it causes guideline makers "dismay" but that it results in infants with SBI's being missed and suffering as a result) makes me think that there was no such association. Thankfully the impact of HIB and pneumococcal conjugate vaccines have had a dramatic effect on decreasing the prevalence of meningitis and bacteremia (though presumably not on UTI's). Thus it seems unlikely that their sample size, even in a multicenter study, was sufficient to exclude a possible association between practice and outcome.
Although catheterizing a well-appearing febrile infant with no risk factors is still traumatic for the parents and invasive for the infant, it's probably worth doing when no source is found, but what's the evidence that either going or not going beyond that results in differences in outcomes?
Conflict of Interest:
None declared