This review assessed the effectiveness of a variety of lipid-lowering treatments in preventing stroke. The authors concluded that statins reduce the risk of stroke in people with and without coronary heart disease. No other lipid-lowering agents appeared to have this effect. The review was well conducted and the authors' conclusions are supported by the data presented.

To assess the effects of lipid-lowering agents in preventing stroke in people with or without coronary heart disease (CHD).

MEDLINE, EMBASE, Pascal, Index Medicus and the Cochrane Library were searched to August 2002. No language restrictions were applied. The reference lists of relevant reviews and identified studies were checked for any additional publications to June 2003.

Randomised controlled trials (RCTs) with a minimum follow-up of 6 months were eligible for inclusion.

Studies that compared any lipid-lowering treatment (pharmacological or dietary) with placebo or usual diet were eligible for inclusion. Studies of interventions with obvious adverse events, such as hormone replacement therapy or outdated interventions (e.g. ileal bypass surgery), were excluded. The treatments in the included studies were statins (pravastatin, lovastatin, simvastatin, fluvastatin, atorvastatin), fibrates (clofibrate, bezafibrate, gemfibrozil, fenofibrate), resins (colestipol, cholestyramine), n-3 fatty acids (fish oil, linolenic acid), diet (diet, soya bean oil), others (niacin, policosanol, garlic), or a combination of these.

Studies on people with or without CHD were eligible for inclusion. Studies restricted to people with previous strokes were excluded. The mean age of participants in the included studies ranged from 45 to 75 years, and in most studies the majority of the participants were men (between 20 and 100%). Where reported, just over half of the included studies contained only patients with CHD, the proportion of patients with diabetes ranged from 0 to 35%, and 0 to 100% had hypertension. Baseline total cholesterol levels ranged from 4.5 to 8.5 mmol/L.

To be included, the studies had to report on nonfatal and fatal stroke and total mortality. The primary outcomes of interest were ischaemic and haemorrhagic stroke (fatal or nonfatal). The secondary outcomes were nonfatal or fatal myocardial infarction (MI).

Two reviewers independently assessed studies for inclusion. Any disagreements were resolved by consensus.

The quality of studies was assessed on the following criteria: concealment of allocation, blinding and the proportion of participants with complete follow-up. Two reviewers independently assessed study quality. Any disagreements were resolved by consensus.

The authors did not state how the data were extracted for the review, or how many reviewers performed the data extraction.

Investigators of included studies were contacted for additional information, when required. Risk ratios were calculated for the individual studies. The percentage change in total cholesterol in each trial was calculated as the difference in the mean change in cholesterol level from baseline to follow-up between the intervention and control groups. Where no exact levels were available, information was extracted from graphs.

A random-effects model was used to estimate pooled weighted risk ratios (RRs). A funnel plot was used to assess for publication bias. The number-needed-to-treat (NNT) was calculated by multiplying the averaged weighted mean annual baseline risk with the mean relative risk reduction. Meta-regression was used to assess any association between nonfatal and fatal stroke and changes in cholesterol levels and type of cholesterol-lowering treatment.

Heterogeneity was assessed using the Breslow-Day test and the I-squared statistic. Differences between the studies were investigated using subgroup analyses based on drug class (statins, fibrates, resins, n-3 fatty acids and dietary interventions), the presence or absence of CHD, and studies that did or did not differentiate between stroke and transient ischaemic attack. Sensitivity analyses were performed to investigate the influence of study quality.

Sixty-five RCTs (200,607 participants) were included.

Lipid-lowering interventions were associated with a statistically significant reduction in nonfatal and fatal stroke (RR 0.89, 95% confidence interval, CI: 0.83, 0.96). There was no evidence of significant heterogeneity (P=0.24, I-squared 10%, 95% uncertainty interval, UI: 0%, 34%). Statins statistically significantly reduced the risk of fatal or nonfatal stroke (RR 0.82, 95% CI: 0.76, 0.90), whereas non-statin interventions did not (RR 0.98, 95% CI: 0.90, 1.07).

Neither all lipid-lowering interventions, nor statins alone, appeared to affect the incidence of haemorrhagic stroke (RR 1.21, 95% CI: 0.78, 1.89 and RR 1.03, 95% CI: 0.49, 2.16, respectively). There was no evidence of significant heterogeneity (P=0.3, I-squared 15%, 95% UI: 0%, 57% and P=0.13, I-squared 46%, 95% UI: 0%, 82%, respectively).

In the subgroup of trials of people without CHD, statins statistically significantly reduced the risk of nonfatal and fatal stroke (RR 0.77, 95% CI: 0.62, 0.95). There was no evidence of significant heterogeneity (P=0.6, I-squared 0%, 95% UI: 0%, 75%). The reduction was similar in trials of people with CHD (RR 0.75, 95% CI: 0.65, 0.87; heterogeneity, P=0.9, I-squared 0%, 95% UI: 0%, 51%). Subgroup analyses showed that trial quality did not significantly alter the results. The NNT per year with statins, to prevent fatal or nonfatal stroke, was 617 (95% CI: 463 to 1,111) in high-risk populations with established CHD and 2,778 (95% CI: 2,083 to 5,000) in low-risk populations without established CHD. All other pooled data suggested that no other lipid-lowering treatments were statistically significantly beneficial in reducing stroke in people with or without CHD.

The results suggested that statins reduce the risk of stroke in people with and without CHD.

This review had clearly stated aims and inclusion criteria. A number of relevant databases were searched without language restrictions. The study selection and quality assessment processes were carried out in duplicate, which helps reduce the potential for reviewer error or bias. The studies were quality assessed using appropriate criteria, and the results of the quality assessment exercise were used to perform subgroup analyses. Adequate details of the included studies were tabulated. The statistical methods for pooling data appeared appropriate, and the authors used subgroup analyses to investigate the effects of differences between the included studies. The review appears to have been well conducted and the authors' conclusions are likely to be reliable.

The authors did not state any implications for practice or further research.

This additional published commentary may also be of interest. Kingsbury KJ. Review: statins reduce non-fatal and fatal strokes in patients with and without coronary heart disease. Evid Based Nurs 2005;8:86.

This record is a structured abstract written by CRD reviewers. The original has met a set of quality criteria. Since September 1996 abstracts have been sent to authors for comment. Additional factual information is incorporated into the record. Noted as [A:....].