Fidelity of Delivery in a Multisite Randomized Clinical Trial of Intervention Efficacy for Infants With Unilateral Cerebral Palsy.

The checklist was written for this trial alone, so its scores cannot be compared with fidelity measured elsewhere or picked up and used in a clinic as it stands. Rater agreement was reported as percentage agreement, which does not account for agreement that would happen by chance, so true agreement is lower than the 84.3% between raters. The worst item agreed only 37% of the time. Only about 10% of home visits were scored, and the authors say scoring a larger proportion might have given different results. Just 8 of 72 sessions fell below the high-fidelity mark, which was too few to test what influences fidelity. The comparisons by intervention group, timing and therapist experience are descriptions, not statistical findings. Almost everything scored high, so there was little spread in the scores to show what makes delivery slip. Thirteen of the 21 checklist items were removed after looking at how they performed, so the final measure was shaped by these same data, and the authors note that weighting items differently could have shown more variation. The study records whether therapists delivered the therapy as intended, not whether the therapy works. The authors state that the impact of different levels of fidelity on child outcomes is unknown.

The authors declare no conflicts of interest. No funding statement appears in the article text, and no funder's role in designing, running or writing the study is described; the paper is indexed as receiving NIH extramural and non-US government research support.

Do not read this as evidence that Baby-CIMT or Baby-BIM improves an infant's hand function. The study scored 72 video-recorded sessions and measured no child outcomes at all; it reports only whether therapists delivered the therapy as the manual described, and the authors state that the link between fidelity level and child outcome is unknown.