Open AccessInter‐rater reliability of the Abnormal Involuntary Movements Scale (AIMS) in a multi‐centre trial: results from Department of Veterans Affairs Cooperative Study #394
Author(s) -
Edson Robert,
Lavori Philip,
Bloch Daniel,
Tracy Kathlene,
Adler Lenard A.,
Rotrosen John
Publication year - 2000
Publication title -
international journal of methods in psychiatric research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.275
H-Index - 73
eISSN - 1557-0657
pISSN - 1049-8931
DOI - 10.1002/mpr.89
Subject(s) - veterans affairs , inter rater reliability , tardive dyskinesia , rating scale , psychology , physical therapy , confidence interval , reliability (semiconductor) , bonferroni correction , medicine , statistics , psychiatry , developmental psychology , mathematics , power (physics) , schizophrenia (object oriented programming) , physics , quantum mechanics
The US Department of Veterans' Affairs (VA) conducted Cooperative Study #394 at nine sites to determine whether vitamin E was a safe and effective treatment for tardive dyskinesia (TD). The study used the Abnormal Involuntary Movement Scale (AIMS) to measure TD. To monitor inter‐rater reliability (IRR) on the AIMS, personnel at each site did initial AIMS assessments, on seven occasions, on a set of five unique subjects from videotapes. On four of these occasions raters re‐assessed sets of five subjects from seven to 13 months after their initial ratings. We analysed the initial ratings of 34 videotaped subjects and the second ratings of 19. To measure IRR, we used the intra‐class correlation coefficient (ICC) for each initial or second rating of the sets of interviews. The ICCs ranged from 0.50 to 0.86, and the ICC for the initial ratings over all 34 subjects was good (0.75; 95% confidence interval 0.64 to 0.83). To determine whether there was significant ‘rater drift’ during the study, we performed linear mixed effects regression on the data with fixed effects for rating (initial or second), rater type, and site and a random effect for rating. The results indicated that scores varied little between ratings (about 0.05 points). Copyright © 2000 Whurr Publishers Ltd.