This review found no association between statin therapy and the incidence of all cancers or site-specific cancers in the short term, independent of length of follow-up and type of statin. The authors' conclusions appear appropriate and are likely to be reliable. However, some studies with relevant data might have been missed.

To evaluate the association between statin therapy and the risk of cancer.

Bibliographic databases (MEDLINE, EMBASE, Web of Science, ISI Proceedings and BIOSIS Previews) and electronic trials registers were searched from inception to November 2005; the search terms were reported. The bibliographies of all included studies were reviewed to identify additional relevant studies. Only peer-reviewed papers published in full in the English language were eligible.

Study designs of evaluations included in the review

Randomised controlled trials (RCTs) and observational studies were eligible. The median length of follow-up was 3.6 years (range: 1.0 to 10.4) in the RCTs and 6.2 years (range: 3.3 to 7.2) in the observational studies.

Specific interventions included in the review

Trials comparing statins with placebo, or observational studies comparing statins with no statins, were eligible. Observational studies comparing a statin treatment with other lipid-lowering agents were excluded. The statins included were lovastatin, simvastatin, fluvastatin, atorvastatin and pravastatin. No details of the doses or regimens used were reported.

Participants included in the review

Inclusion criteria were not explicitly stated for the participants. Studies evaluating highly specific populations, such as renal transplant patients and those with familial hypercholesterolaemia, were excluded. In the included studies, the mean age of the participants ranged from 18 to above 80 years and the proportion of women ranged from 0 to 100%.

Outcomes assessed in the review

Studies that reported on the strength of association (risk ratio odds ratio, RR and OR, respectively) between statin use and all-cancer or site-specific cancer incidence or fatality, or data to enable the calculation of the strength of association, were eligible. The site-specific cancers were breast, prostate, colorectum, lung, genitourinary or gastric cancer, and melanoma.

How were decisions on the relevance of primary studies made?

The authors stated that titles and abstracts of retrieved studies were systematically reviewed against the inclusion criteria. However, it was unclear how many reviewers performed the selection.

The authors stated that the quality of the included studies was not assessed on the grounds that such an assessment in meta-analyses is controversial and the results are potentially misleading.

One reviewer extracted the data and a second reviewer checked their accuracy. The RR or OR, or data to enable their calculation, were extracted from each included study. Data from the longest follow-up were used where appropriate. Authors were contacted for additional information.

How were the studies combined?

The studies were combined using a fixed-effect meta-analysis. The analysis was conducted separately for RCTs and for observational studies. A pooled RR (OR for case-control studies) and corresponding 95% confidence interval (CI) were calculated. Small study bias was assessed by visual inspection of funnel plots and the use of Egger and Begg tests.

How were differences between studies investigated?

Statistical heterogeneity was evaluated using the I-squared statistic. Meta-regression analysis was used to evaluate length of follow-up (less than or equal to 5 years or greater than 5 years), statin type (lipophilic versus hydrophilic) and statin potency (low, medium or high). The analyses were repeated using a random-effects model. Methodological aspects of individual studies were discussed where there was evidence of statistical heterogeneity.

Twenty-six RCTs (103,573 patients) and 12 observational studies (826,854 patients) were included.

Pooled data from RCTs showed no significant association between statin therapy and the incidence of all cancers (RR 1.00, 95% CI: 0.95, 1.05, p=0.98; 26 trials) or site-specific cancers:

breast (RR 1.01, 95% CI: 0.79, 1.30, p=0.92; 7 trials),

prostate (RR 1.00, 95%: 0.85, 1.17, p=0.99; 4 trials),

colon-rectum (RR 1.02, 95% CI: 0.89, 1.16, p=0.83; 9 trials),

lung (RR 0.96, 95% CI: 0.84, 1.09, p=0.49; 9 trials),

genitourinary (RR 0.95, 95% CI: 0.83, 1.09, p=0.46; 5 trials),

melanoma (RR 0.86, 95% CI: 0.62, 1.20, p=0.38; 4 trials), or

gastric (RR 1.00, 95% CI: 0.35, 2.85, p=0.99; 1 trial).

There was no evidence of statistical heterogeneity in any of the analyses, with the exception of breast cancer (I-squared 43%), which was no longer evident after excluding one study.

Statin therapy was not associated with all-cancer fatality (16 trials; RR 1.01, 95% CI: 0.93, 1.09, p=0.91) and seemed not to be related to higher fatality in prostate (1 study), colorectal (1 study) and lung (1 study) cancers.

There was no evidence of differential effects for subgroups defined a priori. The tests of Begg and Egger suggested no evidence of publication bias.

The results of observational studies results were generally consistent with those from RCTs.

There were relatively too few studies to explore subgroups by meta-regression analysis.

There was no evidence that statin therapy is associated with an increased risk of cancer in the short term. A small protective effect of statin therapy on all cancer and colorectal cancer was suggested by a few observational studies, although it did not reach statistical significance. There was no evidence to support differences based on statin type or length of follow-up.

This review had clearly stated inclusion criteria with respect to the study design, outcomes and intervention. A broad definition of participants was used, which was appropriate given the aim of the review. Several relevant sources were searched for relevant studies and the potential influence of publication bias was evaluated. Although no evidence of publication bias was detected, it might have been introduced by the apparent exclusion of non-English articles and the inclusion of full-paper peer-reviewed publications only. This may be of particular importance in this review as the authors highlighted that the primary outcomes of included studies were cardiovascular events and it is possible that studies with relevant cancer data might not have been published. It was unclear if the studies were selected by two independent reviewers, therefore reviewer error and bias might have been introduced into the review process. The authors stated that study quality assessment was not performed on the grounds that its use is controversial and misleading in meta-analyses. However, additional information on quality components of the included studies could have been provided to give an indication of the reliability of the evidence presented.

Details presented on the included studies suggest that the decision to statistically combine the studies was appropriate. Statistical heterogeneity was assessed and the authors stated that there was no significant heterogeneity for the main outcomes. Subgroup analysis was conducted to investigate subgroups specified a priori.

The review was generally well conducted. The authors' conclusions seem reliable based on the evidence presented, although it is possible that some studies providing potentially relevant data might have been missed, or errors made in study selection.

Practice: The authors did not state any implication for practice.

Research: The authors stated that it would be important to have long-term follow-up studies to evaluate any late effects of statins. They also suggested that further investigations would be needed to explore possible differences between statins, or to assess the association between statin use and less common cancers.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.