Naproxen Sodium for Pain Control With Intrauterine Device Insertion: A Randomized Controlled Trial

    loading  Checking for direct PDF access through Ovid


I was very interested in the unexpected findings reported by Ngo et al.1 The double-blind, randomized, placebo-controlled trial was designed to evaluate the effect of naproxen sodium on their primary outcome, pain with intrauterine device (IUD) placement. Although they found no difference in pain at the time of IUD placement, they report a significant difference at 5 and 15 minutes after IUD placement. I believe that the study actually demonstrates no relevant difference in pain at any time point with use of naproxen sodium compared with placebo 1–1.5 hours before IUD placement.
The investigators initially established sample size estimates for nulliparous (n=84) and parous (n=76) women but discontinued the parous cohort owing to slow enrollment. Instead of just enrolling 84 nulliparous women, they overenrolled to include 118 women in the final analysis. This larger sample allowed the investigators to calculate what they claim is a statistically significant finding at both 5 and 15 minutes after IUD placement. Median visual analog scale (VAS) pain scores at 5 minutes after placement were 16.5 and 26 (P=.01) for naproxen and placebo users, respectively. At 15 minutes, the median scores were 12.8 and 24 (P=.01), respectively.
The debate concerns what difference in VAS scores can be considered clinically significant. The authors state that changes in scores of 9–14 mm define a minimum change that is clinically significant. However, their references are articles dating from 1996 to 2001, primarily from emergency medicine literature. The level of clinical significance was evaluated more closely in the anesthesia literature in 1998 (during the same time frame) in a study that validated the correlation of a 100-mm VAS and an 11-point oral scale and that the VAS imprecision is ±20 mm.2 The 20-mm minimum has been used most commonly in recent IUD studies evaluating pain, including one in this journal.3–5
The differences reported in pain at 5 and 15 minutes do not meet the 20-mm threshold as clinically significantly different. The authors used 15 mm as their estimate of significance in their original sample size calculation; the findings at 5 and 15 minutes do not meet this criterion either. Thus, although the sample was large enough for the authors to claim statistical significance, I believe the findings fail to show any clinical difference.
    loading  Loading Related Articles