Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

NSAIDs for primary dysmenorrhoea

Systematic review
Results
Efficacy
Adverse effects
Comment

What extra can we find out about a topic when another systematic review comes along, as a Cochrane review [1] on NSAIDs for dysmenorrhoea? A previous review showed specific NSAIDs to be effective [2], but this newer review of NSAIDs takes a very critical view.

Systematic review

The Cochrane Menstrual Disorders and Sub-fertility Group trials register, Cochrane Central Controlled Trials Register, MEDLINE, and EMBASE were searched for randomised double blind trials of NSAIDs in primary dysmenorrhoea. Active and placebo controlled trials were included. Outcomes analysed were pain relief, absence from work or school, restriction of daily activities, and adverse effects. Studies in which fewer than 80% of randomised patients were available for a particular outcome were excluded.

Results

Sixty-three randomised controlled trials with 4,006 women were included, of which 44 used a crossover design. All were described as double-blind and eleven had adequate allocation of concealment. Information from crossover designs was used only if dichotomous data on the first treatment phase was available. Fourteen placebo-controlled trials had information on pain relief, and were all very small, with treatment group sizes between 15 and 35 women.

Women were aged between 12-47 years. Doses of NSAIDs varied across trials but were within the recommended daily doses. The duration of treatment was generally 3-5 days. Few trials compared the same treatments so there was sparse information available for any particular NSAID and most analyses were based on results from a single study. Findings of the meta-analysis were supported by direct comparisons of different NSAIDs, and crossover trials.

Efficacy

Most information was available for pain relief. Of 36 trials comparing NSAIDs with placebo, 14 (599 women) reported the number of women with at least moderate pain relief and were included in the meta-analysis. Results varied widely across trials (Figures 1 and 2) with between 5-40% (mean 20%) of patients achieving at least moderate pain relief with placebo and 17-95% (mean 67%) with NSAIDs. Overall there was a significant benefit with NSAIDs compared with placebo, relative benefit 3.4 (2.5 to 5.4) and NNT 2.1 (1.9 to 2.5) for at least moderate pain relief over 3-5 days. Most information was for naproxen (Figure 1) with an NNT of 2.5 (2.0 to 3.3). Figure 2 shows results by particular NSAIDs, including some cross-over trial data.


Figure 1: Percent of patients with at least moderate pain relief in placebo controlled trials included in the meta-analysis (naproxen in open circles, other NSAIDs in filled circles)





















Figure 2: Percent of patients with at least moderate pain relief with particular NSAIDs (red=active controls). All trials included in the meta-analysis





















Requirement for additional medication and absence from work or school were reported less often. Women taking NSAIDs were less restricted in their daily activities than women taking placebo (216 women, 39 % restricted vs 68% placebo), and were less likely to be absent from work or school (229 women; 36% versus 72%).

Adverse effects

Overall NSAIDs caused significantly more adverse effects than placebo (1030 women), but there was no significant difference between NSAIDs and placebo for gastrointestinal effects (432 women) or nervous system adverse effects (229 women).

The median withdrawal rate of was 10%. Women withdrew for reasons including lack of efficacy or adverse effects, but numbers for these were not available.

Comment

Despite the inclusion of a large number of trials, they generally had uninterpretable designs, they were small, with too few patients to make much sense for different NSAIDs. That so many crossover trials failed to report results for the first treatment phase meant that much of the available information could not sensibly be included in the meta-analysis. We know that NSAIDs are effective, but we do not know which dose of which NSAID is the most effective and least harmful. Figure 2 is the closest we can come to a comparison of efficacy, and that is not much help because there is so little data. While pain is a useful outcome, and while reduction in pain suggests a reduction in other things like restriction of activities and absenteeism from work or school, we have no idea what women think is the most important outcome for them. One might have hoped for more information than this.

Frankly, in 2004, this is a disgrace. Half the population are women, dysmenorrhoea affects half of them, and one in 10 is incapacitated by it every month. Yet all we have is information on fewer than 600 women in trivial trials. It is just not good enough.

Reviews all too often moan on about the need for more research in a rather unthinking way, but here is a barn door case for more and better research. We might have some from pharmaceutical companies with new analgesics to sell, but the comprehensive research needed will only come when health services take an interest. Bandolier's feminine side is writing to her parliamentary representatives.

References:

  1. J Marjoribanks et al. Nonsteroidal anti-inflammatory drugs for primary dysmenorrhoea (Cochrane Review). In: the Cochrane Library, Issue 4, 2003.
  2. WY Zhang , A Li Wan Po. Efficacy of minor analgesics in primary dysmenorrhoea: a systematic review. British Journal of Obstetrics & Gynaecology 1998 105:780-789.

previous or next story