Charalampos S Siristatidis, Ahmed Gibreel, George Basios, Abha Maheshwari, Siladitya Bhattacharya

Gonadotrophin-releasing hormone agonists (GnRHa) are commonly used in assisted reproduction technology (ART) cycles to prevent a luteinising hormone surge during controlled ovarian hyperstimulation (COH) prior to planned oocyte retrieval, thus optimising the chances of live birth. To evaluate the effectiveness of the different GnRHa protocols as adjuncts to COH in women undergoing ART cycles. We searched the following databases from inception to April 2015: the Cochrane Menstrual Disorders and Subfertility Group Specialised Register, the Cochrane Central Register of Controlled Trials (CENTRAL) in the Cochrane Library (2015, Issue 3), MEDLINE, EMBASE, CINAHL, PsycINFO, and registries of ongoing trials. Reference lists of relevant articles were also searched. We included randomised controlled trials (RCTs) comparing any two protocols of GnRHa used in in vitro fertilisation (IVF) or intracytoplasmic sperm injection (ICSI) cycles in subfertile women. Two review authors independently selected studies, assessed trial eligibility and risk of bias, and extracted the data. The primary outcome measure was number of live births or ongoing pregnancies per woman/couple randomised. Secondary outcome measures were number of clinical pregnancies, number of oocytes retrieved, dose of gonadotrophins used, adverse effects (pregnancy losses, ovarian hyperstimulation, cycle cancellation, and premature luteinising hormone (LH) surges), and cost and acceptability of the regimens. We combined data to calculate odds ratios (OR) for dichotomous variables and mean differences (MD) for continuous variables, with 95% confidence intervals (CIs). We assessed statistical heterogeneity using the I² statistic. We assessed the overall quality of the evidence for the main comparisons using 'Grading of Recommendations Assessment, Development and Evaluation' (GRADE) methods. We included 37 RCTs (3872 women), one ongoing trial, and one trial awaiting classification. These trials made nine different comparisons between protocols. Twenty of the RCTs compared long protocols and short protocols. Only 19/37 RCTs reported live birth or ongoing pregnancy.There was no conclusive evidence of a difference between a long protocol and a short protocol in live birth and ongoing pregnancy rates (OR 1.30, 95% CI 0.94 to 1.81; 12 RCTs, n = 976 women, I² = 15%, low quality evidence). Our findings suggest that in a population in which 14% of women achieve live birth or ongoing pregnancy using a short protocol, between 13% and 23% will achieve live birth or ongoing pregnancy using a long protocol. There was evidence of an increase in clinical pregnancy rates (OR 1.50, 95% CI 1.18 to 1.92; 20 RCTs, n = 1643 women, I² = 27%, moderate quality evidence) associated with the use of a long protocol.There was no evidence of a difference between the groups in terms of live birth and ongoing pregnancy rates when the following GnRHa protocols were compared: long versus ultrashort protocol (OR 1.78, 95% CI 0.72 to 4.36; one RCT, n = 150 women, low quality evidence), long luteal versus long follicular phase protocol (OR 1.89, 95% CI 0.87 to 4.10; one RCT, n = 223 women, low quality evidence), when GnRHa was stopped versus when it was continued (OR 0.75, 95% CI 0.42 to 1.33; three RCTs, n = 290 women, I² = 0%, low quality evidence), when the dose of GnRHa was reduced versus when the same dose was continued (OR 1.02, 95% CI 0.68 to 1.52; four RCTs, n = 407 women, I² = 0%, low quality evidence), when GnRHa was discontinued versus continued after human chorionic gonadotrophin (HCG) administration in the long protocol (OR 0.89, 95% CI 0.49 to 1.64; one RCT, n = 181 women, low quality evidence), and when administration of GnRHa lasted for two versus three weeks before stimulation (OR 1.14, 95% CI 0.49 to 2.68; one RCT, n = 85 women, low quality evidence). Our primary outcomes were not reported for any other comparisons.Regarding adverse events, there were insufficient data to enable us to reach any conclusions except about the cycle cancellation rate. There was no conclusive evidence of a difference in cycle cancellation rate (OR 0.95, 95% CI 0.59 to 1.55; 11 RCTs, n = 1026 women, I² = 42%, low quality evidence) when a long protocol was compared with a short protocol. This suggests that in a population in which 9% of women would have their cycles cancelled using a short protocol, between 5.5% and 14% will have cancelled cycles when using a long protocol.The quality of the evidence ranged from moderate to low. The main limitations in the evidence were failure to report live birth or ongoing pregnancy, poor reporting of methods in the primary studies, and imprecise findings due to lack of data. Only 10 of the 37 included studies were conducted within the last 10 years. When long GnRHa protocols and short GnRHa protocols were compared, we found no conclusive evidence of a difference in live birth and ongoing pregnancy rates, but there was moderate quality evidence of higher clinical pregnancy rates in the long protocol group. None of the other analyses showed any evidence of a difference in birth or pregnancy outcomes between the protocols compared. There was insufficient evidence to make any conclusions regarding adverse effects.