Kawasaki Disease (KD) has been the bane of my clinical practice for decades: an illness that mimics many others with no definitive diagnostic test, with the majority of cases self-resolving without sequelae but a few with devastating consequences. On the other hand, KD has been a boon for teaching of evidence-based medicine, with the current article a great example.
Source: Chen S, Dong Y, Kiuchi MG, et al. Coronary artery complication in Kawasaki Disease and the importance of early intervention . JAMA Pediatr. 2016; 170(12):1156-1163; doi:10.1001/jamapediatrics.2016.2055. See AAP Grand Rounds commentary by Dr. David Spar (subscription required).
The current study reports findings of a systematic review and meta-analysis of studies comparing IVIG therapy to IVIG plus steroids (either as initial or rescue therapy) for KD. They found that the patients receiving steroids, especially if they were classified as high risk for coronary disease development, had a lower rate of coronary artery problems in followup. Fortuitously, for this blog anyway, the Cochrane Library published a full meta-analysis last month on essentially the same issue, reaching a similar, perhaps even stronger, conclusion: "... treatment with a long course of steroids should be considered for all children diagnosed with KD until further studies are performed." What is especially instructive for me is how different these 2 meta-analyses are.
The 2 studies differed greatly in the literature search strategies. Chen utilized only 3 search terms and, surprisingly for a meta-analysis, included non-randomized trials in the review. Nine of the 17 included studies weren't randomized. The Cochrane review utilized a pre-existing trials search in the Cochrane system, plus a 15-step search strategy for the CENTRAL database, overall much a much more thorough search than Chen.
My main concern with the Chen study is inclusion of non-randomized trials, which opens up the study's conclusions to all sorts of bias difficult to assess. The main concern with this approach is that sicker patients may be treated differently from those less ill, a form of selection bias. So, were the sickest patients more likely to be treated with steroids Probably so, in some of the articles in Chen's study. The degree to which this might have affected the results is difficult to ascertain, but I'm surprised this research group chose to include non-randomized trials for a condition where most untreated patients will not experience the outcome of greatest importance, coronary artery aneurysms.
This brings to mind an article on steroid treatment for KD, published during my first year of pediatric residency, that came to the opposite conclusion: "... steroid might act adversely to cause a progression of coronary lesions in [KD]." I've used this study many times in teaching EBM, as an example of what happens without randomization. Here, clinicians at 5 different institutions could choose which of 5 different protocols they could use to treat KD. All 5 groups received oral cephalexin, 3 received steroids, and others received varying combinations of aspirin, warfarin, or nothing else. Talk about confusing! I still remember how this study impacted KD teaching for several years thereafter: corticosteroids were said to be contraindicated for KD, and statements to that effect appeared in the Red Book from 1982 through 1994. (You'll need to take my word on that, I went through my collection of Red Books to verify it, but sadly you won't find old Red Books online.)
So, beware of non-randomized trials, and hope that we get some clarity in the future about treatment for KD. Clinicaltrials.gov lists only 8 treatment trials for KD, and only 1 is randomized. I'm also aware of 1 other randomized trial for refractory KD that is set to open enrollment soon.
