Author Affiliations: Department of Noncommunicable Disease Epidemiology, Cochrane Heart Group, London School of Hygiene and Tropical Medicine, London, England.
Evidence to support the use of statins for preventing recurrent cardiovascular disease (CVD) is strong, derived from individual patient data meta-analysis1 and from systematic reviews of randomized trials.2 This evidence underpins the use of statins in clinical guidelines internationally as an uncontroversial treatment for secondary prevention that is both cost-effective and safe.3- 5
In this issue of The Archives, Gutierrez and colleagues6 present a meta-analysis of statin trials to determine whether the effects are different in men and women. Their literature search was confined to keyword terms in PubMed and English language, and data were extracted by a single investigator—all potential sources of bias. Indeed the basic search for the terms statins/cholesterol-lowering medications and CVD yields only 5000 articles, of which 878 survive the additional limits imposed. So is it possible that Gutierrez and colleagues have missed some relevant studies?
There have been several recent systematic reviews of statin trials to compare with the analysis by Gutierrez et al,6 whose limited search and restrictive inclusion criteria resulted in most of the relevant data being lost. Of the 21 trials included in the recent Cholesterol Treatment Trialists (CTT) publication,7 only 5 were included by Gutierrez and colleagues. Several of the omitted trials would have been eligible for inclusion if the authors had accepted usual-care comparison groups, used in some of the larger randomized controlled trials, and participants without prior cardiovascular disease.
But is it really likely that there are sex differences in treatment effects? Evidence accrued in 2004 from a systematic review of 13 randomized trials (N = 990 events in 11 435 randomized women),8 cited by the authors, showed convincing evidence of benefit in women (relative risk [RR] for total coronary events, 0.80 [95% CI, 0.71-0.91]). The CTT7 published individual patient data analyses from 21 randomized trials (N = 3395 events in 39 000 randomized women) showing similar benefits for major vascular events in women (RR, 0.84 [95% CI, 0.77-0.91]) and men (RR, 0.78 [95% CI, 0.75-0.81]). Furthermore, statins decrease low-density lipoprotein (LDL) cholesterol levels in women by about the same amount as in men. Since the association between cholesterol levels and cardiovascular events is similar in men and women,9 it would be expected that reductions in cholesterol level caused by statins would also reduce clinical events equally in both sexes (but it is possible that some of the effect of statins is not in their reduction of LDL cholesterol level).
While Gutierrez and colleagues6 reported similar findings for coronary events in men and women, the stroke and all-cause mortality findings cause them to doubt the efficacy of statins for these outcomes. However, a recent meta-analysis using more inclusive criteria reports similar benefits in women and men for stroke outcomes.10 The biggest relevant trial, the Heart Protection Study,11 contributes substantially more deaths than all the trials pooled by Gutierrez and colleagues.6 Adding the findings of this trial to the meta-analysis would result in narrower CIs, and the pooled RR for all-cause mortality among women would be 0.89 (95% CI, 0.78-1.01), remarkably similar to the estimate in men (RR, 0.85 [95% CI, 0.80-0.90]) (Figure).
Figure. Effects of statins on all-cause mortality. HPS indicates Heart Protection Study11; RR, relative risk.
Focusing on a lack of statistical significance in the findings for women is misleading. The real issue is not significance but whether the effect size in women is materially different from the effect size in men. Overinterpretation of imprecisely estimated effects is a serious problem in meta-analyses and in primary studies.12 In the study by Gutierrez et al,6 the effect on stroke and all-cause mortality in women is consistent with the effect in men. If a statistical test is wanted, the appropriate P value is for the sex interaction for the outcome by sex. We suggest that statins work just as well in women as in men.
Correspondence: Dr Taylor, Department of Noncommunicable Disease Epidemiology, Cochrane Heart Group, London School of Hygiene and Tropical Medicine, Keppel St, London WC1E 7HT, England (Fiona.Taylor@lshtm.ac.uk).
Financial Disclosure: None reported.
Taylor F, Ebrahim S. Statins Work Just as Well in Women as in MenComment on “Statin Therapy in the Prevention of Recurrent Cardiovascular Events”. Arch Intern Med. 2012;172(12):919-920. doi:10.1001/archinternmed.2012.2434