The strength of evidence supporting luteal phase progestogen after assisted reproduction: A systematic review with reference to trial registration and pre-specified endpoints.

Objective: To measure the potential for outcome switching and selective reporting, in trials of luteal phase progestogen in assisted reproduction.

Study Design: Trials identified through Medline and Embase in August 2017 using the MeSH term "assisted reproductive technology, luteal phase support" and associated text words. Randomised controlled trials (RCTs) comparing progestogen of any type, dose, and route of administration, with placebo or no treatment as luteal phase support in subfertile women undergoing in vitro fertilization (IVF) or intrauterine insemination (IUI). Eight trials after IVF and eleven after IUI, involving 1040 and 2764 participants respectively, were included.

Results: None of the eight trials of progestogen therapy after IVF had been registered. Only 5/11 trials of progestogen after IUI had been registered, and only two of these prospectively. One of these had a registered primary outcome of "pregnancy sac plus heartbeat", but reported "pregnancy sac alone"; we judged this as an altered primary outcome. Three other trial had a registered primary outcome of "clinical pregnancy undefined" and reported "intra or extra-uterine pregnancy with a heartbeat"; we judged this alteration as minimal. That trial was negative. Overall, 26 different outcomes had been reported by the various trials. The three outcomes reported most often were pregnancy undefined (9/19), miscarriage (11/19) and clinical pregnancy (9/19). This suggests considerable potential for selective outcome reporting or outcome switching.

Conclusion: Apart from one negative trial, none of the evidence on luteal phase progestogen after assisted reproduction comes from prospectively registered trials: a slender reed indeed.