To the Editor: We thank Drs. Saraf, Chandra, and Ross for their comments regarding our article. Drs. Saraf and Chandra raise concerns regarding methodological shortcomings in our review. Their criticism focuses on the lack of screening of databases that contain nonindexed journals and the lack of specific criteria for the inclusion of studies that they feel were not assessed for their methodological quality. The authors conclude that because of the heterogeneity of studies, it was inappropriate to calculate incidence rates by combining results from these trials.

We agree with the notion that, unfortunately, too few studies are available that provide data on the development of tardive dyskinesia in patients treated with a second-generation antipsychotic for 1 year or longer. This is particularly true for randomized controlled trials. Given the dearth of long-term studies and the clinical importance of tardive dyskinesia, it seems unlikely that we would have found additional studies in journals that are not indexed by MEDLINE, although we cannot be certain of this. In addition to our comprehensive MEDLINE search, we screened proceedings and abstracts of major psychiatric meetings and contacted the manufacturers of all second-generation antipsychotics for unpublished data. In our Methods section, we clearly delineated the criteria used for inclusion of the reviewed studies. These were open or controlled treatment with any second-generation antipsychotic (i.e., amisulpride, aripiprazole, clozapine, olanzapine, quetiapine, risperidone, sertindole, sulpiride, ziprasidone, or zotepine) that involved at least 20 subjects, lasted 1 year or longer, and included data on newly identified cases of tardive dyskinesia or dyskinesia. Furthermore, we discussed the limitations of the reviewed studies regarding the open nature of most of the trials, differences in patient populations (e.g., age, ethnicity, gender, diagnosis, severity/chronicity of illness, etc.), employed rating scales and rating intervals, definitions of "caseness," doses of second-generation antipsychotics, the use of high-dose first-generation antipsychotic agents in the trials with an active comparator, the lack of data in drug-naive and pediatric populations, and shortcomings in some of the statistical analyses.

Given the heterogeneity of the available studies, it is surprising, however, that the reported 1-year incidence rates of tardive dyskinesia are relatively homogenous when stratified by age, which is the most robust risk factor for the development of tardive dyskinesia. In adults, the rates varied from 0% to 1.5%; in the elderly, the rates varied from 0% to 13.4%. These figures are about one-fifth of the widely reported 1-year incidence rates of tardive dyskinesia associated with first-generation antipsychotics, which are approximately 5% in adults (1–3) and 25%–30% in the elderly (4–7). Although the relatively small number of available studies and differences in their design precluded a formal meta-analysis based only on randomized controlled trials, explicit inclusion and exclusion criteria were used to justify a systematic review.

In his letter, Dr. Ross calls for caution in interpreting the presented data, pointing out methodological limitations of the studies that were included in the review. His criticism focuses on the lack of studies lasting for longer than 1 year, the relatively high dropout rates, and insufficient reporting on numbers of patients experiencing acute extrapyramidal symptoms or receiving co-medications that may have masked tardive dyskinesia. In particular, Dr. Ross argues that the dropping out of patients because of acute extrapyramidal symptoms may have introduced a selection bias toward longer-term exclusion of patients at particular risk for the development of tardive dyskinesia.

Clearly, studies lasting for longer than 12 months and with lower dropout rates are desirable. We also agree that a more comprehensive reporting of side effects leading to study withdrawal would have improved the quality of the studies. However, although patients experiencing acute extrapyramidal side effects appear to be at higher risk for the development of tardive dyskinesia and a skewed dropout of patients having these side effects could artificially lower the incidence rates of tardive dyskinesia, this would be even more the case for patients in the three double-blind randomized trials that used medium to high doses of haloperidol as a comparator. Despite this, haloperidol was still associated with approximately five times higher 1-year incidence rates of tardive dyskinesia (1) in adults. The same applies to the use of concomitant sedative-hypnotic or anxiolytic medications, which, in addition, have not been found to consistently reduce the symptoms of tardive dyskinesia (8, 9). However, if the study group size allows, this question should be tested in future studies by stratifying the analyses by use of concomitant medications and controlling for dropouts due to acute extrapyramidal symptoms.

The issues raised underscore the need for more methodologically rigorous, controlled long-term studies to further our understanding of the true risk for tardive dyskinesia associated with second-generation antipsychotic agents. Nevertheless, we feel that despite the limitations of the available database, clinicians should consider these results in their choice of medications.

Reprints are not available; however, Letters to the Editor can be downloaded at http://ajp.psychiatryonline.org.