ATYPICAL ANTIPSYCHOTIC MEDICATIONS AND RISK OF FALLS

had physiological effects, most commonly cortisol suppression, suggesting that rapid absorption may not confer substantial benefit. Thus, head-to-head research comparisons of clinical effectiveness of the new preparation and the standard oral preparation are warranted. Third, they raise a concern that the assessment of appetite may have been inadequate. At the time of the study, there were no validated scales available. The instrument cited by the authors of the letter is currently in press at the time of this response. They next mention that improvement in prealbumin as a result of megestrol acetate has been previously demonstrated, but the literature cited was based on a retrospective chart review using a pre/post design rather than a controlled trial, so the benefit attributable to the drug cannot be determined. Finally, the authors of the letter cite our data to support the absence of clinical adrenal insufficiency as a result of adrenal suppression. The number of participants in our study was small, and it would probably be premature to draw this conclusion. Dr. Suneja notes the small percentage of assessed subjects who were actually recruited. The vast majority met exclusion criteria, which excluded participants for whom megestrol acetate had already been evaluated (e.g., cancer), those who had contraindications, and those who were demented. For this Phase II study, we elected to exclude persons who could not provide consent for a medication that has unknown toxicity in the population being studied. Nevertheless, the difficulty in recruiting eligible participants in this study certainly could affect the generalizability of the findings. Moreover, the recruitment strategies approved by our institutional review board precluded obtaining information to determine how much the recruited sample differed from the sample considered. Dr. Suneja raises the possibility of confounding by discharge site. Although there were some differences (none statistically significant) in percentage recruited from home versus nursing home across treatment arm, it would be difficult to attribute confounding to this distribution. Moreover, with such a small sample, imbalances are common, and it is uncertain that block randomization on site would have eliminated them. Larger studies should consider block randomization. Dr. Suneja asks whether baseline prealbumin was adjusted for, and it was. Finally, we agree with Dr. Suneja about the need for larger studies of megestrol acetate that examine clinical outcomes, including mortality and infectious complications. In summary, based on this study, we believe that further research on megestrol acetate needs to be conducted before it can be recommended for older persons who have poor appetite after a recent hospitalization.