Research Methods Measurement and Statistics

A new method for estimating the effectiveness of emergency contraception that accounts for variation in timing of ovulation and previous cycle length

Mikolajczyk RT, Stanford JB

Published June 14, 2005 Fertility and Sterility, 83(6), 1764-1770
DOI 10.1016/j.fertnstert.2005.01.097 PMID 15950649

RRM Academy Synopsis

New method cuts bias in simulated emergency contraception trials

A 2005 simulation study built a new way to estimate how well emergency contraception works, using each woman's previous cycle length and the spread in ovulation timing. In simulated trials, the new method gave the least biased estimates of effectiveness. The cycles came from 125 women.

Study at a glance, simulation study: A new method gave the least biased estimates of effectiveness
Study at a glance, simulation study of 725 cycles from 125 women: A new method gave the least biased estimates of effectiveness. Source: Mikolajczyk RT et al., 2005, PMID 15950649.

Key Findings

  • The new method drew on 725 cycles from 125 women. Its fecund window ran 14 days, from day 10 to day 23 counted back from the last day before the next menses.
  • In simulated trials using the previous cycle's length, the unadjusted Trussell method overestimated a true effectiveness of 50% by a median of 18 percentage points. The new method overestimated it by 3.
  • In the same simulations at a true effectiveness of 75%, the unadjusted median overestimates were 9 points for the Trussell method, 7 for the Dixon method and 1 for the new method.
  • With a 28-day cycle assumed for everyone, the median overestimates grew for all methods. At a true 50%, the new method overestimated by 15 points and the Trussell method by 26.
  • In simulated 5-day exposure frames, every method overestimated effectiveness when intercourse fell early in the cycle and underestimated it when late. The new method gave the least biased results there.

Interpretation

This simulation study tested two assumed effectiveness levels, 50% and 75%. For each one, it ran 1,000 simulated trials of 120 cycles. In every scenario, the new method had less bias than the Trussell or Dixon methods. The authors call these results preliminary and ask for replication with other data sets. The findings support larger studies that test this method against real effectiveness data.

RRM Context

The daily pregnancy numbers behind the new method came from fertility awareness users. They charted their peak mucus day to mark ovulation. Restorative reproductive medicine treats this kind of cycle charting as a root-cause tool. It evaluates a woman's own fertile signs instead of an assumed cycle length.

Abstract

Objective

To develop a new method for estimating the effectiveness of emergency contraception (EC) by using information about previous menstrual cycle length, accounting for the variation in the day of ovulation within the menstrual cycle, and comparing the validity of the new and previous methods.

Methods

Secondary analysis of a data set with a biological marker of ovulation and its distribution in the cycle. Based on a sample of cycles with known length and a known biological marker of ovulation, we simulated trials of predetermined EC effectiveness and then calculated estimates of EC effectiveness based on old and new methods.

Results

Under some conditions, all methods produced biased estimates of effectiveness with simulated trials, especially when the actual effectiveness was low. The systematic bias was minimized with the new method. The new method was robust with regard to the distribution of the day of intercourse in women presenting for EC.

Conclusions

Future studies of EC effectiveness should consider both the uncertainty in predicting the day of ovulation and previous cycle length. Our estimates of daily fecundity should be replicated with other data sets.

Read the guide

  • Fertility Awareness Methods Compared

    Clinician-reviewed comparison of fertility awareness-based methods (Billings, Creighton, Sensiplan, Marquette, Standard Days, apps). Every effectiveness figure cited and labeled by metric.

    30 min read

Topics

By this author

Related research

Research Methods › Measurement and Statistics › Statistical Methods · Fertility Awareness › Effectiveness › Typical and Perfect Use
Rafael T Mikolajczyk, Joseph B Stanford
R Mikolajczyk, Joe Stanford, Joey Stanford, J Stanford
PMID 15950649 15950649 DOI 10.1016/j.fertnstert.2005.01.097 10.1016/j.fertnstert.2005.01.097 Mikolajczyk et al. 2005, Mikolajczyk 2005