Low accuracy and interobserver/intraobserver agreement for colorectal sessile serrated lesions: A multicenter diagnostic study

Table 2 Diagnostic accuracy and interobserver agreement across reading rounds

Subtype	Round 1			Round 2			Round 3			P accuracy (round 1 vs round 3)	P accuracy (vs overall in round 1)
	Accuracy	κ	95%CI	Accuracy	κ	95%CI	Accuracy	κ	95%CI
Overall	66%	0.393	0.385-0.400	69%	0.441	0.434-0.449	72%	0.415	0.407-0.423	< 0.001a	reference
SSL	62%	0.409	0.394-0.423	60%	0.441	0.426-0.455	58%	0.301	0.286-0.316	0.032a	0.014
SSL-D	33%	0.114	0.099-0.129	33%	0.132	0.117-0.147	49%	0.127	0.112-0.141	0.007a	< 0.001
GCHP	71%	0.369	0.354-0.384	72%	0.402	0.387-0.417	69%	0.369	0.355-0.384	0.102	< 0.001
MVHP	62%	0.332	0.318-0.347	68%	0.425	0.411-0.440	65%	0.380	0.365-0.394	0.161	0.023
TSA	74%	0.587	0.572-0.602	75%	0.702	0.688-0.717	77%	0.703	0.688-0.718	0.238	0.004
TA	66%	0.429	0.415-0.444	71%	0.476	0.461-0.491	80%	0.487	0.472-0.52	< 0.001a	0.987
VA/TVA	65%	0.470	0.456-0.485	67%	0.472	0.457-0.487	74%	0.542	0.527-0.556	0.016a	0.815

^aP < 0.05 (round 3 vs round 1, uncorrected).

P < 0.0071 vs overall accuracy in round 1 (Bonferroni correction for 7 prespecified comparisons). Data from 1176 polyps assessed by 6 pathologists. κ: < 0.20, slight; 0.21-0.40, fair; 0.41-0.60, moderate; 0.61-0.80, substantial; > 0.80, almost perfect. GCHP: Goblet cell-rich hyperplastic polyp; MVHP: Microvesicular hyperplastic polyp; SSL: Sessile serrated lesion; SSL-D: Sessile serrated lesion with dysplasia; TA: Tubular adenoma; TSA: Traditional serrated adenoma; VA/TVA: Villous adenoma/tubulovillous adenoma.