What does the SART success rate for a given IVF program mean?

Abstract

Objective: The Society of Assisted Reproductive Technology (SART) database is used by many to assess the effectiveness of in-vitro fertilization (IVF) programs and the quality of the embryology laboratories. Factors other than just the embryology laboratory, influence IVF success. These include patient demographics, clinician management, and embryo transfer techniques. These factors are not reflected in the SART statistics, which only report overall center pregnancy rates for each age group. Our purpose is to determine if variations in individual physician success rates exist in a large IVF program, with a uniform laboratory and treatment protocols. Design: Retrospective analysis of the 1999 SART data for Boston IVF (BIVF) and its 13 physicians. The clinical pregnancy rate and live birth rate were analyzed for patients <38 years of age and for patients 38–40 years of age who underwent non-donor, fresh embryo transfer. An ‘ideal’ patient group <38 yrs of age, 3 or fewer IVF cycles, 8 or greater number of oocytes retrieved, total gonadotropin used between 1000–4500 units, and inseminated with the partners fresh sperm, was analyzed. Materials/Methods: The clinical pregnancy and live birth rates were obtained for BIVF as reported to SART. The same data was obtained for each individual physician and the results were compared between physicians using Chi Square analysis. Patient demographic characteristics were also analyzed. The technique of embryo transfer was not considered as a variable in the outcome of this study, because all BIVF physicians were randomly assigned to perform the procedure. Results: In 1999, 2215 IVF cycles were performed at BIVF. In the <38 yr age group, the clinical pregnancy and live birth rates were 28.3% and 24.8%, respectively. In the 38–40 yr age group, the clinical pregnancy and live birth rates were 20.5% and 15.5%.The mean number of cycles initiated per physician in the <38 yr age group was 125.07. In the 38–40 yr age group the mean number of cycles was 45.3. Among the 13 physicians the clinical pregnancy rate in the <38 yr group, ranged from 20.5%–35.1% (p = 0.019), and the live birth rates between 17.8%–31.1% (p = 0.045). For the 38–40 yr age group, the clinical pregnancy rate ranged from 10.6%–29.8% (p = 0.043) and the live birth rates between 7.0%–25.5% (p = 0.028). However, when the ideal patient group was analyzed there was no statistical difference among the pregnancy rates among the different physicians (p = 0.8). The physicians with the lowest overall pregnancy rates had higher pregnancy rates in the ideal patient group (p = 0.0026). These lower pregnancy rates might be a reflection of the proportion of difficult cases managed by these physicians. Additional demographic differences were present which will be discussed. Conclusions: Even with standard protocols, use of the same laboratory, and randomization of physicians performing the embryo transfer, significant differences remain in overall success rates among physicians in the same IVF program. These differences were however not present in the ideal patient group. The data suggest that patient demographics play a large role in IVF success. Hence, the assumption that success rates indicate the quality of the IVF program and its embryology laboratory is not be valid without taking into account variation caused by difficulty of patient care.