|Home | About | Journals | Submit | Contact Us | Français|
Aneja and Carcillo's review of adrenal insufficiency and the rationale for steroid use in sepsis is both concise and timely.1 Although they acknowledge that further studies are needed in paediatrics before recommending a 7 day course of hydrocortisone in catecholamine resistant septic shock, they suggest that the logic for this therapy is “compelling”, recommending a bolus of hydrocortisone followed by an infusion of up to 50 mg/kg/day in states of sepsis‐induced relative adrenal insufficiency.
This compelling logic is incorporated within their statement that, among children with septic shock, survivors exhibit dramatically increased plasma levels of cortisol compared to non‐survivors (Aneja and Carcillo, fig 1). Furthermore, these levels are roughly consistent with those achieved following doses of hydrocortisone of 50 mg/kg (Aneja and Carcillo, fig 2).1 Unfortunately, figure 1 of Aneja and Carcillo's paper is very misleading, for two reasons. First, the data used for the septic categories within this graph come from one paediatric paper only, published in 1999.2 Second, the apparent dramatic difference between survivors and non‐survivors (186 vs 74 μg/dl, shown as dots on the graph) represents the upper limit of the range; thus this large discrepancy could have been due to one or two outliers. Indeed comparison of the median values between survivors and non‐survivors (the solid lines) provides a more modest difference (45 vs 39 μg/dl), and falls far short of that attained following administration of 50 mg/kg of hydrocortisone. There have been many paediatric papers published since then, with the majority showing no consistent differences between survivors and non‐survivors3,4; indeed the authors' own paper in 2005 reached the same conclusion.5
The waters are further muddied by the use of etomidate, which produces an iatrogenic state of adrenal insufficiency.6 This is illustrated beautifully by three papers from the Netherlands. In 2000 and 2002, Joosten et al reported lower cortisol levels among non‐survivors of meningococcal sepsis (although reported as two papers, the patient group in the earlier paper appeared to be largely a subset of the latter).7,8 The same authors re‐analysed these and more contemporary data, and showed that the apparent differences in both cortisol levels, and also the ratio of ACTH to cortisol, are dramatically attenuated when etomidate use is factored in.9
Interpretation of previous papers on this subject is further obscured by the fact that all have measured total, rather than biologically active free, cortisol. Hamrahian et al showed that the concept of relative adrenal insufficiency in the critically ill disappears when free cortisol is measured.10 This is further supported by work from Morel and colleagues, who were unable to differentiate between apparent “responders” and “non‐ responders” to steroids on the basis of any tests of adrenal function.11
The question of steroid benefit in catecholamine resistant septic shock remains largely unanswered. Annane's trial needs to be repeated in paediatrics, but this time we should not use etomidate and we should measure free cortisol. Until such a time, I will probably continue to use steroids, largely because I'm too nervous not to!
Competing interests: None declared.