Suggested Improvements to the Overt Aggression Scale-Modified

To the Editor: The lack of a standard efficacy measure for outpatient studies of impulsive aggression led Coccaro et al. 1 to modify the Overt Aggression Scale (OAS) 2 to create the OAS-Modified (OAS-M). The one large (N=246) study to date which used the OAS-M 3 found that the total aggression score had a large variance and was highly skewed, which prevented the use of parametric statistics including covariance analyses. The large variance and skewness seemed due to the scoring system—to calculate the aggression scores, frequencies are multiplied by severity ratings, but there is no upper limit on frequencies. Therefore a patient who has many episodes of relatively mild irritability can have a very high total aggression score, much higher than a patient who, less frequently, has much more serious aggression (this also impairs face validity). Also, the calculated aggression scores have not yielded better drug-placebo discrimination than the much simpler global ratings. 4 Given the above, and my prior experience with the OAS-M, I modified the OAS-M as follows:

This revised OAS-M was then used in two fairly small placebo-controlled studies in outpatients with impulsive aggression, one with oxcarbazepine (N=48) 4 and one with levetiracetam (N=40). 5 Using the revised OAS-M resulted in clear and consistent evidence of benefit in the oxcarbazepine study (reliability can be inferred if a rating significantly discriminates an active medication from placebo). In the levetiracetam study, there was not a significant drug-placebo difference, but interrater reliability of the revised OAS-M was evaluated with 20 patients evaluated by two raters: Intraclass correlation coefficients were high (e.g., 0.918 for total aggression and 0.878 for global overt aggression, both significant at p<0.001). Supporting validity, correlations among the OAS-M measures and other aggression ratings were high in the levetiracetam study, and patient-rated global improvement and the hostility rating of the Brief Psychiatric Rating Scale also showed significant drug-placebo differences in the oxcarbazepine study.

Thus, the revised OAS-M seems more likely to discriminate between effective medication and placebo. Details about the rationale for the revised OAS-M (and a copy) can be obtained from the author ([email protected]).