This review concluded that there was no substantial evidence for an association between statin use and risk of pancreatic cancer when statins are taken at daily doses for cardiovascular event prevention. Although aspects of this review were well conducted, the reliability of the conclusions are unknown given the limitations of the study data to comprehensively address the review question.

To evaluate the risk of pancreatic cancer as an outcome of statin use.

PubMed, BIOSIS Previews, and The Cochrane Library were searched from inception to August 2011 for publications in any language; search terms were provided. A manual search of references of relevant articles and reviews was undertaken.

Eligible studies were randomised controlled trials (RCTs), cohort studies, or case-control studies that compared statin therapy with placebo or no treatment. Trials had to report sufficient information on pancreatic cancer to reconstruct a 2x2 table of prevalence by statin intake. Trials that evaluated other lipid lowering drugs, such as fibrates, bile acid binding resins, and nicotinic acids, were excluded from the review.

One of the RCTs included postmenopausal women without evidence of cardiovascular disease; to other two RCTs included patients with heart disease. In the observational studies, patients included those with coronary heart disease, cancer, patients taking anti-hyperlipidaemic drugs, or people representative of the general population. In some studies, data were taken from various hospital or insurance databases.

The authors did not state how many reviewers selected studies for inclusion.

The quality of the RCTs was assessed based on the Cochrane handbook. Criteria evaluated included randomisation, allocation concealment, blinding, incomplete outcome data, selective reporting, and other bias. Each criteria was scored as "low", "unclear", or "high" risk of bias. Cohort and case-control studies were assessed using The Newcastle-Ottawa Scale which assessed eight items for selection, comparability, and exposure/outcome.

The authors did not explicitly state how many reviewers assessed quality.

For the RCTs, data on pancreatic cancer were extracted to calculate relative risk (RR) with 95% confidence intervals (CI). For the observational studies, risk ratios, odds ratios, hazard ratios, and standardised incidence ratios were extracted. The authors stated that these measures of effect provided similar estimates of relative risk because the incidence of pancreatic cancer was very low. Data were also extracted on confounding factors in the observational studies.

Three reviewers independently extracted the data, with any disagreements resolved by consensus.

The studies were pooled using a random-effects model to calculate an overall relative risk with 95% confidence intervals. Heterogeneity was assessed using Cochran's Q test and Ι².

Subgroup analysis was conducted by study design, statin type, period of follow-up, and duration of statin use. There were two cohort studies that reported data separately for gender, or statin use duration. These four sets of results were reported as separate studies in the meta-analysis. Sensitivity analysis was carried out by excluding studies one by one.

Publication bias was assessed using the Begg and Mazumdar's test, and the Egger's test.

Sixteen studies were included in the review (1,692,863 patients): three RCTs (7,118 patients), five cohort studies (1,143,697 patients), and eight case-control (542,048 patients) studies. For the RCTs, most of the quality criteria were considered to have been adequately addressed (at low risk of bias), although none of the trials reported allocation concealment. Mean or median duration of follow-up in the RCTs ranged from 3.9 to 10.4 years. The quality scores for the observational studies ranged from 3 to 9. Mean follow-up for the observational studies ranged from less than one year to 10.6 years.

The overall pooled result for all studies showed a non-significant difference between patients who received statins and those who received placebo for pancreatic cancer incidence (RR 0.89, 95% CI 0.74 to 1.07: Ι²=81%, high heterogeneity).

Subgroup analysis by study design found a null association between pancreatic cancer risk and statin use for RCTs (RR 0.99, 95% CI 0.44 to 2.21; three RCTs; there was no statistical heterogeneity), a non-significant increase in cancer risk for cohort studies (RR 1.05, 95% CI 0.93 to 1.19; five studies, seven comparisons; Ι²=40%), and a non-significant decrease in risk for case-control studies (RR 0.74, 95% CI 0.51 to 1.07; eight studies; Ι²=86%, high heterogeneity). Subgroup analysis of studies with long-term follow-up and long-term statin use did not make a difference to the results.

Sensitivity analyses by omission of any of the included studies did not alter the magnitude of the effect.

There was no evidence of publication bias.

There was no substantial evidence for an association between statin use and risk of pancreatic cancer when statins were taken at daily doses for cardiovascular event prevention.

The review question was clear. The inclusion and exclusion criteria were well defined for study type, intervention and outcome of interest, but studies were not restricted by population group. A search of relevant sources was undertaken with no language restrictions. It was not clear whether unpublished data were sought, although the authors did attempt to assess publication bias. It appeared that steps were taken to minimise reviewer error and bias, although this was not reported for all aspects of the review process.

The quality of the included studies was assessed using appropriate methods. It was not clear if the authors double counted the placebo group population to present four sets of comparisons from two studies. It may be also be argued that pooling of RCTs with observational studies was not appropriate. The authors acknowledged that the RCTs may not be powerful enough to investigate cancer outcomes.

Although aspects of this review were well conducted, the reliability of the conclusions are unknown given the limitations of the study data (duration, statistical power) to comprehensively address the review question.

The authors did not state any implications for practice and research.

Not reported.

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.