Objectives: In trials of acute severe infections or inflammations frequent administration of non-randomised treatment (ie, intercurrent event) in response to clinical events is expected. These events may affect the interpretation of trial findings. Swissped-RECOVERY was set up as one of the first randomised controlled trials worldwide, investigating the comparative effectiveness of anti-inflammatory treatment with intravenous methylprednisolone or intravenous immunoglobulins in children and adolescents with Paediatric Inflammatory Multisystem Syndrome Temporally Associated with SARS-CoV-2 (PIMS-TS). We present one approach towards improving the interpretation of non-randomised treatment in a randomised controlled trial.

Design: This is a pre-planned ancillary analysis of the Swissped-RECOVERY trial, a randomised multicentre open-label two-arm trial.

Setting: 10 Swiss paediatric hospitals (secondary and tertiary care) participated.

Participants: Paediatric patients hospitalised with PIMS-TS.

Interventions: All patient-first intercurrent events, if applicable, were presented to an independent adjudication committee consisting of four international paediatric COVID-19 experts to provide independent clinical adjudication to a set of standardised questions relating to whether additional non-randomised treatments were clinically indicated and disease classification at the time of the intercurrent event.

Results: Of 41 treatments in 75 participants (24/41 (59%) and 17/41 (41%) in the intravenous methylprednisolone and immunoglobulin arms of the trial, respectively), two-thirds were considered indicated. The most common treatment (oral glucocorticoids, 14/41, 35%) was mostly considered not indicated (11/14, 79%), although in line with local guidelines. Intercurrent events among patients with Shock-like PIMS-TS at baseline were mostly considered indicated. A significant proportion of patients with undifferentiated PIMS-TS at baseline were not attributed to the same group at the time of the intercurrent event (6/12 unchanged, 4/12 Kawasaki disease-like, 2/12 Shock-like).

Conclusion: The masked adjudication of intercurrent events contributes to the interpretation of results in open-label trials and should be incorporated in the future.

Trial registration numbers: SNCTP000004720 and NCT04826588.