Efficacy of vitamin B-6 in the treatment of premenstrual syndrome:systematic reviewBMJ 1999; 318 doi: https://doi.org/10.1136/bmj.318.7195.1375 (Published 22 May 1999) Cite this as: BMJ 1999;318:1375
All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Wyatt et al.'s systematic review of the efficacy of Vitamin B-6 for
premenstrual syndrome (PMS) is to be welcomed in tackling an important
clinical problem (1). The methods used in this review, however, may have
led to an over-estimate of the treatment's efficacy.
This arises from the way the outcome data appear to have been
extracted from primary studies for meta-analysis. The range of outcomes
reported in primary studies was broad, and by implication so too was the
range of assessment methods for these outcomes. Many were in the form of
continuous measures (e.g. PMS symptom scores, emotional symptoms), whilst
others were binary and reported the proportion of subjects showing an
improvement in symptoms. The continuous outcome data appear to have been
converted into such binary format to facilitate meta-analysis and
calculation of odds
ratio data of efficacy compared to placebo.
This is not invalid, but evidence from another 'diffuse' topic
systematic review we have conducted (2) suggests that effect sizes in
binary outcomes are on average greater than for continuous outcomes. In
that review 23 studies reported continuous outcomes and 59 reported binary
outcomes (e.g. the proportion of subjects accepting treatment or modifying
health behaviour). The mean effect size of binary outcomes was 0.36 and
the mean effect size of continuous outcomes was
0.18 (statistically significant difference, p=0.0062). Our review also
found the effect sizes of randomised controlled trials to be smaller (mean
0.24) than for other study designs (mean 0.4; statistically significant
difference, p=0.0159). This latter finding is in keeping with other
literature (3,4), suggesting that the pattern of effect size variations
has some generalisability for other reviews.
The implication is therefore that converting continuous outcome data
into binary format for meta-analysis purposes may lead to an over-estimate
of efficacy. It may be more appropriate to analyse the outcomes from
different studies in effect size format, which allows comparison between
outcomes measured on different scales (5). A true mean (to indicate
efficacy) can be produced by adjusting the
data to eliminate the systematic excess in effect size evident in binary
data. Then health care professionals may get a more realistic impression
of the treatment's efficacy relative to placebo. If the absolute changes
(or equivalent 'numbers-needed-to-treat' data) were presented as well as
these relative changes or comparative data, then professionals would also
have information with which to assess the true or likely effect that might
be expected in practice.
Dr Adrian Edwards & Dr Kerenza Hood
Department of General Practice
University of Wales College of Medicine
Llanedeyrn Health Centre
Cardiff CF3 7PN
1 Wyatt KM, Dimmock PW, Jones PW, O'Brien PMS. Efficacy of Vitamin
B-6 in the treatment of premenstrual syndrome: a
systematic review. BMJ 1999;318:1375-81.
2 Edwards AGK, Barker J, Bloor M, Burnard P, Covey J, Hood K et al. A
systematic review of risk communication - improving effective clinical
practice and research in primary care. Cardiff. University of Wales
College of Medicine, Department of Gener Practice. Report to
NHS Executive. 1998.
3 Colditz GA, Miller JN, Mosteller F. How study design affects outcomes
in comparisons of therapy. 1.Medical. Statistics in Medicine 1989;8:441-
4 Sacks H, Chalmers TC, Smith H. Randomised versus historical controls
for clinical trials. American Journal of Medicine 1982;72:233-8.
5 Slavin RE. Best evidence synthesis: an intelligent
alternative to meta-analysis. Journal of Clinical Epidemiology 1995;48:9-
Competing interests: No competing interests
1) The term PMS was defined in the introduction, but was this definition
used in the trial? Without a strict definition IN your inclusion
criteria, this trial is putting apples in with the oranges. (How do we
know these patients truly have PMS?)
2) The use of odds ratios and "better"/"not better" in trial allow a
minimal analysis of qualitative benefits, but whether the benefit is
CLINICALLY significant CANNOT be determined, nor can one say how it helps
patients? (What does "better" really mean in PMS?). I am sure, however,
that a drug company somewhere that makes B6 will be using these results in
their advertising... :)
I do hope that a good, powerful RCT is undertaken, but my bet is that
it will show no effect.
Competing interests: No competing interests