Abstract
BACKGROUND: Gonadotropin releasing hormone agonists (GnRHa) are used in assisted reproduction cycles to reversibly block pituitary function and prevent a luteinizing hormone surge. In the short and ultrashort protocols of GnRHa administration, injection of gonadotropins is commenced a few days after the start of GnRHa. In the long protocols (with GnRHa started either in the midluteal phase or in the early follicular phase) gonadotropin administration is delayed until pituitary desensitization has been achieved, usually 2-3 weeks. OBJECTIVES: To conduct a systematic overview of available data comparing short or ultrashort and long GnRHa protocols for pituitary desensitization in in vitro fertilization (IVF) and gamete intra-fallopian transfer (GIFT) treatment cycles. SEARCH STRATEGY: Search strategies included on-line searching of the MEDLINE and EMBASE data bases and the Cochrane Menstrual Disorders and Subfertility Group's Specialised Register from 1982 to 1998, and hand searching of bibliographies of relevant publications and reviews, and abstracts of scientific meetings. SELECTION CRITERIA: Randomized trials of short or ultrashort versus long (follicular or luteal phase start) GnRHa protocols in IVF or GIFT treatment cycles. DATA COLLECTION AND ANALYSIS: Data were extracted into 2 x 2 tables. For the primary outcome, clinical pregnancy per cycle started, the overall common odds ratio (OR) and the risk difference with 95% confidence interval (CI) were calculated after verifying the presence of homogeneity of treatment effect across all trials. The following subgroup comparisons were performed: ultrashort versus long protocols, short versus long protocols and, within each of these comparisons, subgroups of studies which used the long protocol with follicular phase start or the long protocol with luteal phase start. Secondary outcomes considered were clinical pregnancy per oocyte retrieval and per embryo transfer, spontaneous abortion, ongoing/delivered pregnancy per cycle started, number of ampoules of gonadotropin used, number of oocytes retrieved, and fertilization rate. MAIN RESULTS: Twenty-six trials met the inclusion criteria. The common OR for clinical pregnancy per cycle started was 1.32 (95% CI, 1.10 - 1.57) in favour of the long GnRHa protocol. The studies were subgrouped, depending on whether, in the long protocol, the GnRHa was commenced in the follicular phase (8 trials) or luteal phase (16 trials). The respective ORs were 1.54 (95% CI, 1.11 - 2. 13) and 1.21 (95% CI, 0.98 - 1.51). After excluding the four trials using the ultrashort protocol, the OR for long versus short protocols (22 trials) was 1.27 (95% CI, 1.04 - 1.56). A comparison of long versus ultrashort protocols (4 trials) produced an OR of 1. 47 (95% CI, 1.02 - 2.12). REVIEWER'S CONCLUSIONS: On the basis of clinical pregnancy rate per cycle started, this meta-analysis demonstrates the superiority of the long protocol over the short and ultrashort protocols for GnRHa use in IVF and GIFT cycles.