48. Does it exist an interobserver variability in the interpretation of SSEPs in comatose patients after cardiac arrest? An answer from the Italian Multicentric study ProNeCA

Abstract

Evaluate inter-observer variability reading of SSEPs in post-anoxic comatose patients included in ProNeCa. 3 readers and a “Gold Standard”, central coordinatior of Italian study, decided to interpretate independently 13 SSEPs. SEPs evaluation was splited into “Absent” (A), “Present” (P) and Not Valutable (NV). The degree of interobserver agreement was calculated using Overall Agreement, Kappa Coefficients and Intraclass Correlation Coefficient (ICC). Overal Agreement between raters (69.2%; 95%CI 38.6–90.9) and agreement of each rater vs. GS was good (from Rater1 = 84.6% to Rater3 = 92.3%). Interrater Agreement was substantial, Exact kappa = 0.64. For category A was 0.58 (moderete agreement), NV 0.62 (substantial agreement) and Present 0.77 (good). Single raters were in good agreement vs. GS (two had a weighted κ⩾0.800), higher than K for bivariate comparison among raters (weighted κ from 0.68 to 0.87). ICC yielded a good 0.747 [0.49–0.91] P = 0.0287. Interobserver agreement for A in accord with lecterature was moderate, with quite wide confidence intervals. Each rater is from fair to good agreement with the GS. To reach a more reliable agreement and reduce CI we decided to increse the sample size: 32 SSEP have been reed but not yet included in this analysis.