Effects of Korean-AGREE Ⅱ Scoring Guide on Improving the Reliability of the Scores Moo-Kyung Oh 1, Heuisug Jo1,2,3, Youkyoung Lee3,4,5 1Department of Preventive Medicine, Kangwon National University Hospital, Chuncheon, Korea 2 Department of Health Management and Policy, Kangwon National University School of Medicine, Chuncheon, Korea 3Executive Committee for clinical practice guideline, The Korean Academy of Medical Sciences 4Department of Laboratory Medicine and Genetics, Soonchunhyang University Bucheon Hospital, Bucheon, Korea 5Department of Laboratory Medicine and Genetics, Soonchunhyang University College of Medicine, Cheonan, Korea I. Background and Purpose • Current status of Clinical Practice Guidelines in Korea Reliability • Higher ICC in SG Observed in almost domain and stastically significant in overall assessment • Korean medical guideline information center (KoMGI) • A couple of CPGs development handbook • More than 100 of guidelines of the last decade Development Table 1. Inter-rater reliability of K-AGREE Ⅱ domain scores in CPG A • Korean AGREE II instrument • Web-based evaluation systems Approval Appraisal Differences of healthcare environments • KoMGI Dissemination A lack of experiences in evaluation of CPGs Scoring Guide Implementation Inter-rater disagreement Domain 1 Scoring Guide group ICC (95% C.I.) p-value 0.815 (-0.344~0.995) 0.046 Domain 2 0.430 (-3.137~0.986) 0.251 -0.762 (-11.791~0.955) 0.595 Domain 3 0.722 (0.175~0.938) 0.011 0.473 (-0.565~0.882) 0.121 Domain 4 0.718 (-1.048~0.993) 0.096 -0.296 (-8.411~0.967) 0.503 Domain 5 0.424 (-1.925~0.960) 0.229 0.273 (-2.693~0.950) 0.312 Domain 6 0.000 (-16.443~0.999) 0.391 0.000 (-16.443~0.999) 0.391 0.826 (0.671~0.918) <0.001 0.680 (0.395~0.850) <0.001 Domain Overall Non-Scoring Guide group ICC (95% C.I.) p-value 0.682 (-1.307~0.992) 0.116 Table 2. Inter-rater reliability of K-AGREE Ⅱ domain scores in CPG B This study aim to examine the effects of the K-AGREE II scoring guide to reduce inter-rater differences. II. Methods Study design Random assignment Appraisal Scoring Guide Group 14 Appraisers Non-Scoring Guide Group 2 CPG Reliability Statistics • To identify the distribution • To evaluate the reliability • To evaluate the consistency Domain 2 0.769 (-0.675~0.994) 0.068 0.769 (-0.679~0.994) 0.069 Domain 3 0.796 (0.394~0.954) 0.002 0.424 (-0.710~0.871) 0.155 Domain 4 -1.333 (-15.940~0.941) 0.670 0.000 (-6.260~0.975) 0.422 Domain 5 0.888 (0.431~0.992) 0.005 0.272 (-2.696~0.950) 0.312 Domain 6 0.667 (-4.814~1.000) 0.182 0.792 (-2634~1.000) 0.116 Overall 0.869 (0.753~0.939) <0.001 0.662 (0.362~0.841) <0.001 Consistency • Higher association in SG Association improve in SG, especially distinctive in CPG B Consistency (Korean AGREE II only) Non-Scoring Guide group ICC (95% C.I.) p-value -0.333 (-0.806~0.966) 0.512 Analysis Distribution (Korean AGREE II + Scoring Guide) Domain 1 Scoring Guide group ICC (95% C.I.) p-value 0.821 (-0.303~0.995) 0.043 Domain Descriptive analysis (Domain specific) Intra-class correlation (Domain specific) Association among appraisers (Each items) III. Results Table 3. Association of SG in CPG A Table 4. Association of Non-SG in CPG A Appraiser 1 2 3 4 Appraiser 1 2 3 4 1 1 .622 .348 .481 2 .622 1 .393 .596 3 .348 .393 1 -.052 4 .481 .596 -.052 1 1 1 -.225 -.434 -.459 2 -.225 1 .373 .502 3 -.434 .373 1 .833 4 -.459 .502 .833 1 Table 5. Association of SG in CPG B Table 6. Association of Non-SG in CPG B Appraiser 1 2 3 4 Appraiser 1 2 3 4 1 1 .853 .453 .491 2 .853 1 .651 .749 3 .453 .651 1 .641 4 .491 .749 .641 1 1 1 .556 .441 .127 2 .556 1 .479 .128 3 .441 .479 1 .372 4 .127 .128 .372 1 Distribution • Higher Scores and higher variability in Non-SG Distinctive in domain 2, 3, and 5 IV. Discussion • Scoring guide reduce the inter-rater disagreemet and improve the overall reliability of the K-AGREE II instrument. Domain Scores Those effects remarkable in low level CPGs development Inter-rater disagreement reflects the healthcare environment characteristics Domain Stakeholder involvement Scope Stake Rigour Clarity Applica- Editorial and bility Indepen involve- develop and -dence Purpose ment -ment Presentation Scope Stake Rigour Clarity Applica- Editorial and bility Indepen involve- develop and -dence Purpose ment -ment Presentation Fig. 1. Distribution of K-AGREE Ⅱ Fig. 2. Distribution of K-AGREE Ⅱ domain scores according Scoring domain scores according Scoring Guide users in CPG A Guide users in CPG B Limitation Lack of experience in stakeholder involvement Different understanding of stakeholder Low level of CPGs implementation Applicability Confuse implementation with dissemination Scoring Guide Providing clear standards regarding the stakeholder and the level of participation Providing clear definition of implementation Providing the methodologies and resources in detail