RELIABILITY OF A COMORBIDITY MEASURE - THE INDEX OF COEXISTENT DISEASE (ICED)

The reliability of an established comorbidity index (the Index of Go-E xistent Disease) was tested using retrospective data from the case not es of elderly patients who had undergone total hip replacement. Inter- rater reliability was examined twice, first with two raters (n = 39) a nd then with three (n = 49). Intra-rater reliability was assessed usin g one rater (n = 45). Reasons for any lack of reliability were explore d. The inter-rater reliability of the ICED was moderate (kappa 0.5-0.6 ). While the Functional Severity index performed well (kappa 0.6-1.0), the Index of Disease Severity subindex was less reliable (kappa 0.4-0 .5). Differences between raters had an impact on the observed associat ion between comorbidity and serious post-operative complications. Intr a-rater reliability was excellent (kappa 0.9). Several reasons why int er-rarer reliability was only moderate were identified, mostly related to uncertainties in applying the ICED. The reliability of the ICED ne eds to be improved before it is used more widely with retrospective da ta. This might be achieved by further clarification of the instruction s for its use. (C) 1997 Elsevier Science Inc.