• An 18-month follow-up of 314 randomly selected psychiatric emergency room patients showed that, although there were 15 possible diagnostic groups the diagnoses made at initial interviews proved to be largely accurate. Follow-up results showed the prediction based on those initial diagnoses was 84% correct in the 299 patients (95%) that could be followed up. Follow-up results also showed that 88% of the individual diagnoses (as opposed to diagnosed patients) were correct. The kappa coefficients were high, all of them being above 0.60, some between 0.80 and 0.90. In addition to the relatively high diagnostic validity, the results also demonstrated a high item reliability for the more than 400 separate items for which kappa coefficients were calculated. Less than 15% of these items had a kappa coefficient of less than 0.60. This study thus demonstrated a high diagnostic reliability and a high diagnostic validity.
The basis of these diagnoses in a heterogeneous psychiatric population was the application of precisely defined and applied diagnostic criteria. The diagnostic process was blind in that the follow-up diagnosis was made in the absence of knowledge of the initial diagnosis. The use of precisely defined and systematically applied criteria may prove to be superior to multivariate analysis, cluster analysis, or other statistical techniques.
Robins E, Gentry KA, Munoz RA, Marten S. A Contrast of the Three More Common Illnesses With the Ten Less Common in a Study and 18-Month Follow-up of 314 Psychiatric Emergency Room PatientsIII. Findings at Follow-up. Arch Gen Psychiatry. 1977;34(3):285-291. doi:10.1001/archpsyc.1977.01770150043004