To the Editor:-We appreciated the article by Song et al. who assessed the potential benefit of subhypnotic propofol dosages applied at the end of anesthesia to reduce postoperative nausea and vomiting (PONV) after maintenance with sevoflurane or desflurane. Although the study was not double-blinded, we would like to express our respect for the nice study design, including the power analysis, with a sufficient number of patients and the consequent stratification and randomization.
However, we see problems with some studies focusing on PONV, and we would like to take this article as an opportunity to raise two questions.
First, we agree that it is reasonable to select "high-risk patients" for antiemetic trials. However, a selection of females or a specific type of operation, or both, can be questioned. Recent studies identifying the clinically most significant factors [2,3]and developing a risk score for the prediction of PONV or postoperative vomiting only (PV) [4,5]were able to show that the incidence of PONV or PV after inhalational anesthesia is mainly related to the patient specific characteristics: female gender, nonsmoking history, history of motion sickness or PONV, young age, and the duration of anesthesia. Furthermore, there is some evidence that a score to predict PV is applicable to other types of surgery because incidences of PV in various types of surgery were mainly related to the distribution of individual risk factors. Thus, it is not surprising that females undergoing laparoscopic cholecystectomies had a PONV incidence of approximately 60%. Thus, bearing the importance of individual risk factors in mind, it is difficult to understand why especially patients with a previous history of motion sickness or PONV were excluded from this study. 
Second, data presentation are usually not standardized so that studies cannot be compared or used for meta-analyses. We therefore would like to suggest a minimal standard for outcome presentation for nausea, vomiting, PONV, and rescue treatment for the time intervals of 0-2, 2-6, 6-24 and 0-24 h postoperatively. In contrast, the reported 12-h assessment appears quite artificial and clinically of little relevance, because this would often require disturbances during the night or inaccurate time scaling in the data acquisition. An informative adjunct are Kaplan-Meyer curves that, in contrast to this paper, are not used very often. Again, it would be nice to have them separately for nausea, vomiting, PONV, and rescue treatment.
All in all, we do not intend to criticize this specific article because it bears interesting aspects, but to raise an awareness of difficulties concerning patient selection and data presentation in PONV studies. A more standardized presentation might facilitate future quantitative systematic reviews.
Christian C. Apfel, M.D.
Norbert Roewer, M.D.
Professor and Department Head; Department of Anesthesiology; University of Wuerzburg; Wuerzburg, D-97080, Germany; Apfel@anaesthesie.uni-wuerzburg.de
(Accepted for publication January 21, 1999.)