To the Editor:

In the August 2006 article, "Study looks at weight impact on OC efficacy," the apparent conclusion that weight is not associated with differential pregnancy rates is not supported by the data presented.1

Heavier women (>155 pounds) did experience a 42% higher total pregnancy rate than women who weighed less, a difference that was not statistically significant. The article interprets this finding as supportive of a lack of difference in pregnancy rates according to weight of users. However, the small number of pregnancies, 20, limits conclusions that can be drawn and do not support this conclusion.

This number of pregnancies means that if a true difference of 42% exists, then this study has less than a 20% chance of demonstrating it and more than an 80% probability of erroneously concluding no difference. To have reasonable statistical power (80%) to demonstrate a difference of this magnitude would require about 270 pregnancies, which translates to a sample size of 25,750 women in each comparison arm.

Although the study's prospective nature does have the advantage of absence of recall bias, an important factor here, the infrequency of outcome measures is part of what makes contraceptive studies challenging. The weight issue really is a case where meta-analysis, a combined look at several studies, can be a powerful tool.

Michael Rosenberg, MD, MPH, President and CEO, HealthDecisions, Chapel Hill, NC

Reference

1. Zhang HF, LaGuardia KD, Creanga DL. Higher body weight and BMI are not associated with reduced efficacy in Ortho Tri-Cyclen Lo users. Presented at the Annual Clinical Meeting of the American College of Obstetricians and Gynecologists. Washington, DC; May 2006.