Dr. van Harten and Colleagues Reply

To the Editor: Jay D. Sherr, Pharm.D., and colleagues state that every strategy for treating tardive dyskinesia should include the cessation of antipsychotic drugs if clinically feasible. We strongly support this idea, and it would be a misconception if the opposite was concluded from our article. The central point of our report was that if treatment with neuroleptics is required, the drugs should preferably not be administered intermittently because this may increase the risk of tardive dyskinesia. The authors suggest that we interpreted correlation as causation and that causation may be reversed in such a way that extrapyramidal syndromes are the cause of drug interruptions. However, it is very unlikely that tardive dyskinesia was the main reason for drug interruptions. Had it been, then one would expect this information to have been noted in the records. Furthermore, our finding is also supported by the results of animal studies (1). We cannot rule out the possibility that acute extrapyramidal syndromes like parkinsonism, akathisia, and acute dystonia were a main reason for drug interruptions; we did not assess the history of acute extrapyramidal syndromes. Because acute extrapyramidal syndromes may be a risk factor for tardive dyskinesia, this idea deserves attention (1). However, it is not likely that in our study this explanation would be valid: according to the patient records, the main reason for those interruptions was discharge from the psychiatric hospital.

Dr. Sherr and colleagues further suggest that 1) the cumulative anticholinergic dose must be entered in the logistic regression analysis before examining the predictive value of drug interruptions and that 2) the total time of neuroleptic exposure should be used instead of the cumulative amount of neuroleptic exposure. Reanalyzing the data on the basis of these two suggestions did not change the results.

The fact that we were unable to find a relationship between the cumulative amount of neuroleptics and tardive dyskinesia in our cross-sectional study may be because of the long mean duration of neuroleptic treatment in our population. A high mean cumulative neuroleptic dose will obscure this relationship (ceiling effect), particularly if the relationship can only be found during the first years of antipsychotic treatment. Incidence studies clearly show that such a relationship does exist (1, 2).

In short, our conclusion that drug interruptions may be a risk factor for tardive dyskinesia seems the most likely explanation from the data.