Design: Cochrane systematic review and meta-analysis of 87 randomised trials, GRADE certainty very low to high
Population: about 36,000 reproductive-age women seeking emergency contraception within five days of unprotected intercourse
Primary outcome: number of pregnancies
Effect: low-dose mifepristone vs levonorgestrel RR 0.73 (95% CI 0.59-0.90); mid-dose RR 0.67 (0.49-0.91)

A Cochrane review pooled 87 randomised trials and about 36,000 participants seeking emergency contraception within five days of unprotected intercourse, comparing mifepristone with levonorgestrel, the Yuzpe regimen and the copper intrauterine device. The critical outcome was the number of pregnancies.

Low-dose mifepristone, under 25 mg, reduced pregnancy compared with levonorgestrel (RR 0.73, 95% CI 0.59-0.90, no heterogeneity across 14 studies and 8,752 participants) at high certainty, and cut side effects substantially (RR 0.26, 0.17-0.38). Mid-dose mifepristone, 25 to 50 mg, gave RR 0.67 (0.49-0.91) against levonorgestrel at moderate certainty. Mid-dose beat low-dose only slightly (RR 0.77, 0.59-1.00) at the cost of more delayed menses. Against the Yuzpe regimen, mifepristone at any dose gave RR 0.14 (0.05-0.41).

Delayed menses is the trade-off, and it scales with dose: RR 1.52 (1.11-2.08) for low-dose against levonorgestrel, and 1.32 (1.15-1.50) for mid-dose against low-dose. Early menses, by contrast, became less common. Seventy-nine of the 87 trials were done in China, which the reviewers flag as limiting how far the side-effect and satisfaction findings travel. Ulipristal was an eligible comparator but no trial supplied that comparison.

In Indian practice mifepristone is available and familiar, but it is a scheduled drug with restrictions on sale, and emergency contraception here is supplied overwhelmingly as over-the-counter levonorgestrel. This review is a reason to know the comparative numbers when counselling a woman who has a choice, not a reason to treat mifepristone as the routine first-line supply.

When counselling on emergency contraception, be able to say that mifepristone prevented more pregnancies than levonorgestrel in randomised comparisons
Warn about delayed menses specifically — it is the commonest adverse effect and rises with mifepristone dose
Tell women that a later period after mifepristone is expected, and set a threshold for a pregnancy test rather than leaving it open
A copper intrauterine device remains the most effective option and the evidence against mifepristone here was very low certainty
Check local regulatory and supply status before changing what your service stocks

Why it matters

Most services stock one emergency contraceptive and never revisit the choice; this puts a number on what that choice costs.

The statistics, in plain English

A risk ratio of 0.73 with an interval of 0.59 to 0.90 means roughly a quarter fewer pregnancies, and because the interval stays below 1.0 the direction is secure. I-squared of 0% means the 14 trials agreed closely, which is why this outcome was graded high certainty. The mid-dose versus low-dose comparison, 0.77 with an upper bound of exactly 1.00, is the weakest of the effectiveness findings — it touches no difference. The side-effect comparison carried I-squared of 72% in one analysis, meaning trials disagreed, which is part of why it was downgraded.