Effects of agonist dose on in-vitro fertilizationDear Sir, Alvarez et al. (1997) have made a useful contribution to this field of research. However, there are possible alternative interpretations for their findings which we feel might be usefully considered. Their study appears flawed in that there are no data concerning concentrations of follicle stimulating hormone (FSH) on day 3 of the cycle, or on previous pregnancies and their outcomes -both features are important when assessing outcome variables. In our experience, we also find that obese patients require significantly longer duration to achieve down-regulation, using the same criteria as the authors, but this has no impact on pregnancy rate.Finally, and most importantly, the chosen cut-off point of 13 days duration of gonadotrophin-releasing hormone (GnRHa) administration was not justified in their manuscript. We find this biased, since in their small sample shifting two patients from GII (longer duration) to GI (shorter duration) by choosing a cut-off of 14 or 15 days can lead to no significance in the pregnancy rate between the two groups. To demonstrate this statistically we carried out the following exercise: the reported pregnancy rates in the study for GI and GII were 44 and 20% respectively which is 9/27 and 12/45. Using the c 2 test gives P ϭ 0.027. By increasing the cut-off point by 1 or 2 days the pregnancy rates may be: 12/43 and 9/29 (assuming no change in the number of the pregnant patients) which makes P ϭ 0.06, (not significant). Furthermore, the authors did not indicate how often they measured the patients serum oestradiol to verify the concentrations Ͻ50 pg/ml. This has to be on a daily basis so that the cut-off point may be in days. However, every other day or weekly measurement of oestradiol does not give accurate duration in days. On the other hand, as the authors indicated in their study, the patients were seen in the first few days of their period for ultrasound scan and oestradiol measurement. Since their measurements were not on a specific day, inadvertent over-estimation of the days of GnRHa administration could not be ruled out. In conclusion, we feel that a prospective study is necessary to address these issues effectively. Care must be taken in over-interpreting data collected from small numbers.Reference Alvarez, C., Cremades, N., Blasco, N. and Bernabeau, R. (1997) (Alvarez et al., 1998). In response to this comment we would like to clarity that patients were separated into two groups according to the number of days of agonist administration, and the value of the median (13 days) was chosen as a cut-off point. When we chose other cut-off points (12 or 14 days), the pregnancy rate per cycle and per transfer was significantly lower in the patients group suppressed for a longer period (Neyro et al., 1994) (Tables I and II).Ovarian quiescence was verified for the first or second day of cycle with no follicular growth observed by transvaginal ultrasound and an oestradiol serum concentration of Ͻ50 pg/ml on the same day. Cer...