The ICD-10 clinical field trial for mental and behavioral disorders: Results in Canada and the United States

Division of Epidemiology and Services Research, NIMH, National Institutes of Health, Rockville, MD 20857.
American Journal of Psychiatry (Impact Factor: 12.3). 10/1994; 151(9):1340-50. DOI: 10.1176/ajp.151.9.1340
Source: PubMed


To help evaluate the impact of proposed revisions to the chapter on mental and behavioral disorders for ICD-10, the World Health Organization (WHO) Division of Mental Health organized an international clinical field trial to evaluate draft clinical descriptions and diagnostic guidelines. The authors compare interrater diagnostic reliability results from this field trial for clinicians in Canada and the United States of American with those from all other clinicians worldwide, as well as with those from field trials conducted to evaluate drafts of DSM-III.
Two or more clinicians at each clinical center independently evaluated each patient, following a study protocol that allowed clinicians to list up to six diagnoses. In Canada and the United States, 96 clinicians completed 1,781 assessments among 491 patients, and elsewhere in the world 472 clinicians completed 7,495 assessments among 1,969 patients.
Summary kappa coefficients at two-, three-, and four-character ICD-10 code levels were 0.76, 0.65, and 0.52, respectively, for Canadian and U.S. clinicians and 0.83, 0.75, and 0.62 for clinicians elsewhere. The mean number of diagnoses per assessment for Canadian and U.S. clinicians was 2.1; for clinicians elsewhere it was 1.7. More multiple coding of diagnoses for substance use disorders, mood (affective) disorders, and personality disorders by Canadian and U.S. clinicians accounted for much of the difference in diagnostic coding and in interrater reliability between them and clinicians elsewhere.
Interrater diagnostic reliability in Canada and the United States was similar to that of clinicians worldwide and also to results from the DSM-III field tests. Use of more multiple coding of selected disorders by Canadian and U.S. clinicians may reflect the influence of DSM-III and DSM-III-R, which encourage multiple diagnostic entries and the use of separate multiaxial coding for personality disorders, and may have reduced interrater concurrence for some categories. Further, collaborative development of ICD-10 with DSM-IV has aligned these two systems more closely.

6 Reads
  • Source
    • "Within the sample originally chosen to represent approximate equal numbers of FHP and FHN subjects at age 20%, 35.9% of the men developed an AUD, including 23.9% with dependence. While rates vary across studies, in general, the lifetime risk for alcohol dependence in the general population of males is estimated at approximately 15% (Kessler et al., 1997, Regier et al., 1994), and the risk for dependence in sons of alcoholics is estimated at approximately 40% (Cotton, 1979; Goodwin, 1985). Using these figures, one might predict an approximate 30% of dependence in the current population, a figure not markedly different from the lifetime risk observed here. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The low level of response (LR) to alcohol is related to a family history (FH) of alcohol use disorders (AUDs), and each predicts alcohol-related outcomes. Few studies have evaluated the interrelationships between the number of alcoholic relatives, LR, and a range of alcohol-related outcomes. This study tests the hypotheses that there will be an inverse relationship between LR and FH and that LR will be a better predictor of the maximum quantity of alcohol consumed. Data were extracted from personal interviews with 376 males from 20 years of follow-up in the San Diego Prospective Study. Level of response had been established at about age 20 through alcohol challenges in this population, about half of whom had at least one alcoholic relative. Face-to-face follow-ups with both the subjects and additional informants were carried out 10, 15, and 20 years later. These analyses used correlations and regressions to evaluate the relationship between the 2 major predictors (FH and LR) and 5 alcohol-related outcomes over 20 years of follow-up. As predicted, the alcohol challenge-based LR correlated significantly with the number of alcoholic relatives (up to -0.17 for subjects with clearly high and low LR scores). Each of the 2 predictors correlated with the 5 outcomes, including the maximum quantity of alcohol consumed since original testing, maximum frequency, nondiagnostic alcohol-related problems, the number of 11 DSM-IV (Fourth Diagnostic and Statistical Manual of Mental Disorders) abuse and dependence items, and having developed alcohol abuse or dependence. In hierarchical regression analyses, LR contributed significantly to the prediction of all 5 outcomes, even when considered in the context of FH, with only LR predicting drinking quantity and frequency, but both items adding to the prediction of alcohol-related problems and diagnoses. These results were not affected by the intensity of usual drinking when LR had been measured at age 20. Both FH and LR contributed to a range of alcohol-related outcomes, with LR alone significantly predicting maximum quantity and frequency in regression analyses.
    Preview · Article · Sep 2006 · Alcoholism Clinical and Experimental Research
  • Source
    • "At present there are two established classification systems for mental disorders: the International Classification of Diseases (ICD) that is published by the World Health Organization (WHO), and the classification system recommended by the American Psychiatric Association (APA), the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). The classification systems converged to a great extent in their last revisions, and diagnoses therefore seem to be largely comparable, although differences exist (Regier et al., 1994). "

    Preview · Article ·
  • Source

    Full-text · Article ·
Show more