The steering committee (M.K. (chair), D.d.L., J.B., P.B. and P.R.) defined disease conditions (see ESM) and identified ten European panel members who were internationally established paediatric MV investigators with recent peer-reviewed publications (last 10 years). An electronic literature search in PubMed and EMBASE (inception to September 1, 2015) was performed using a combination of medical subject heading terms, text words related to MV and disease-specific terms. All panel members screened the references for eligibility, defined by (1) age <18 years, (2) describing non-invasive or invasive respiratory support, and (3) type of design (i.e. any type of clinical study except for case-series and reports). Publications were excluded if they described diseases exclusively linked to the perinatal period. The proposal by Chatburn (ESM, Table 2) was used for ventilator taxonomy [3 (link), 4 (link)].
Recommendations were drafted by all panel members, and subsequently discussed at a two-day meeting in Rome, Italy (September 2015). This resulted in a final set of recommendations, subjected to electronic voting (December 2015) using the Research and Development/University of California, Los Angeles (RAND/UCLA) appropriateness method scale [5 ]. Recommendations were scored from 1 (complete disagreement) to 9 (complete agreement). Median score (95% confidence interval) was calculated after eliminating one lowest and highest value. Recommendations were labelled “strong agreement” (median 7–9 and no score <7), “equipoise” (median 4–6) or “disagreement” (median 1–3). Recommendations without “strong agreement” were rephrased. Revised recommendations retaining “strong agreement” after the second electronic voting (February 2016) were labelled “weak agreement” and the percentage of agreement (number of individual scores ≥7 divided by 15) quantified the level of disagreement. As it was expected a priori that there would be very few RCTs or systematic reviews, it was decided by the steering committee to keep the consensus guideline descriptive and not use the GRADE system [6 (link)].
Recommendations were drafted by all panel members, and subsequently discussed at a two-day meeting in Rome, Italy (September 2015). This resulted in a final set of recommendations, subjected to electronic voting (December 2015) using the Research and Development/University of California, Los Angeles (RAND/UCLA) appropriateness method scale [5 ]. Recommendations were scored from 1 (complete disagreement) to 9 (complete agreement). Median score (95% confidence interval) was calculated after eliminating one lowest and highest value. Recommendations were labelled “strong agreement” (median 7–9 and no score <7), “equipoise” (median 4–6) or “disagreement” (median 1–3). Recommendations without “strong agreement” were rephrased. Revised recommendations retaining “strong agreement” after the second electronic voting (February 2016) were labelled “weak agreement” and the percentage of agreement (number of individual scores ≥7 divided by 15) quantified the level of disagreement. As it was expected a priori that there would be very few RCTs or systematic reviews, it was decided by the steering committee to keep the consensus guideline descriptive and not use the GRADE system [6 (link)].
