Eleven studies were identified (1,055 children, range 15 to 212) including six RCTs. Follow-up ranged from one to 50 months in seven studies. One RCT was multi-centre and double-blinded.
Coronary artery lesions: There were no significant differences in coronary artery lesions for corticosteroids versus conventional treatment in primary treatment (Ι²=55%; random-effects model; seven studies) or as additional treatment (Ι²=0%; fixed-effect model; four studies). Similarly, there were no significant differences in coronary artery aneurisms for corticosteroids versus conventional treatment for primary treatment (three studies), additional treatment (one study). or for all studies (Ι²=0%; fixed-effect model; four studies).
Duration of fever: There was no significant difference in duration of fever for corticosteroids versus conventional treatment in primary treatment (Ι²=68%; random-effects model; four studies), but the number of febrile days was significantly shorter in the corticosteroid group (MD -1.30 days).
Inflammatory markers: Due to high heterogeneity and incomplete data, a meta-analysis of the change in C-reactive protein after treatment was only possible for three studies. There was a significantly lower level after two intervention therapy regimes versus control (MD -0.16).
Rate of non-response to initial treatment: The numbers of non-responsive patients to initial treatment (with persistent fever for longer than 36 hours - temperature above 38°C or recrudescent fever associated with Kawasaki disease symptoms after an afebrile period of two to seven days) was significantly lower for corticosteroids versus conventional therapy (MD 0.50, 95% CI 0.32 to 0.79; Ι²=0%; fixed-effect model; five studies).
Results were reported for individual studies for other clinical outcomes including hospitalisation days and cost, serum cytokine levels, and adverse events.