Empirical weighting of the Standardized Mini Mental State Examination items among nursing home residents

Missing items in the Mini Mental State examination are dealt with in different ways. The main aims of this study were to calculate a weighted item score for organic disorder on the basis of the item score of the Standardized Mini Mental State Examination (SMMSE) test regardless of the completeness of the test and to compare the criterion validity of the weighted scores with the ordinal scores for the SMMSE test. With a participation of 91%, the study included 100 nursing residents. All residents were tested with the SMMSE test and concurrently diagnosed in accordance with the ICD-10 by a consultant psychiatrist. The two assessments were mutually blinded. Multiple conditional forward logistic regression was used to select the items that most strongly predicted organic disorder as assessed by the psychiatrist. The weighted score had significantly better validity parameters, performed better on a receiver operating curve (ROC), and was better at dichotomizing the population into organic disorder than the commonly used ordinal score. We propose that missing items in the SMMSE should be scored as missing and included in empirical weighting on SMMSE items, which will then be substantially more valid than the SMMSE score itself.