Association between response to ovarian stimulation and miscarriage following IVF: an analysis of 124 351 IVF pregnancies

Is there a relationship between ovarian reserve, quantified as ovarian response to stimulation, and miscarriage rate following IVF treatment? There is a strong association between the number of oocytes retrieved and miscarriage rate following IVF treatment, with the miscarriage rate decreasing with an increasing number of oocytes and then levelling off: poor responders have a higher miscarriage rate across all age groups. Poor ovarian response is a manifestation of a decline in the quantity of the primordial follicle pool. Whether poor ovarian response is associated with a decline in oocyte quality contributing to miscarriage is however debated. Anonymous data were obtained from the Human Fertilization and Embryology Authority (HFEA), the statutory regulator of assisted reproduction treatment (ART) in the UK. The HFEA has collected data on all ART performed in the UK since 1991. Data from 1991 to June 2008 involving 402 185 stimulated fresh IVF cycles and 124 351 pregnancy outcomes were analysed. Data on all women undergoing a stimulated fresh IVF treatment cycle with at least one oocyte retrieved during the period from 1991 to June 2008 were analysed for their early pregnancy outcomes. There was a strong association between the number of oocytes retrieved and the clinical miscarriage rate. The miscarriage rate fell from 20 to 13% with an increasing number of oocytes before levelling off. Stepwise logistic regression identified three cut-off points (4, 10 and 15 oocytes) at or beyond which the probability of clinical miscarriage fell. There was no increase in miscarriage rate with very high oocyte numbers (> 20 oocytes). The lowest risk of miscarriage (9.9%) was for women under 38 years of age, with primary infertility without a female cause and producing more than three oocytes. Although the analysis was performed only on stimulated IVF cycles (excluding unstimulated cycles), the data had the limitation that there was no information on the total gonadotrophin consumption. The model was adjusted for age and type of infertility, but the dataset contained no information on other confounders such as body mass index (BMI) of the women to allow adjustment. Analysis of this extensive dataset suggests that poor responders have a higher risk of clinical miscarriage, indicating that poor ovarian response is associated with a parallel decline in both oocyte quantity and quality. The miscarriage rate is also higher with advanced age, secondary infertility and a female cause of infertility compared with a younger age, male factor infertility and unexplained cause.