Key Questions and Analytic Framework

Using established methods,⁴ the USPSTF determined the scope and Key Questions for this review. Investigators created an analytic framework with the Key Questions and the patient populations, interventions, and outcomes reviewed (Figure 1).

Figure 1

Analytic Framework. Abbreviations: CVD=cardiovascular disease; CHD=coronary heart disease; CVA=cerebrovascular accident (stroke); KQ=key question.

Search Strategies

We searched the Cochrane Central Register of Controlled Trials, the Cochrane Database of Systematic Reviews, and Ovid MEDLINE to June 2016 for relevant studies and systematic reviews, with no start date limitations. Search strategies are available in Appendix A1. We also reviewed reference lists of relevant articles.

Study Selection

At least two reviewers independently evaluated each study to determine inclusion eligibility. We selected studies on the basis of inclusion and exclusion criteria developed for each Key Question (Appendix A2). The population for all Key Questions was adults age 40 years or older without prior CVD events (e.g., MI, angina, revascularization, stroke, or transient ischemic attack) or studies in which the proportion of patients with prior CVD events is less than 10 percent. We included studies that compared treatment versus no treatment or usual care without a statin and assessed effects on all-cause mortality, CHD- or stroke-related morbidity or mortality, or harms (including muscle injury, cognitive loss, diabetes, and hepatic injury), including studies that compared effects in subgroups defined by demographic (e.g., age, sex, or race/ethnicity) or clinical characteristics (e.g., specific cardiovascular risk factors, lipid parameters, or 10-year or lifetime cardiovascular risk). We also included studies that compared treatment strategies with statins to target LDL-C levels versus other treatment strategies and that evaluated how benefits and harms vary according to potency of statin treatment. For all Key Questions, we included randomized, controlled trials (RCTs) of statin therapy versus placebo or no statin. For Key Question 2, we included controlled observational studies reporting harms of statin use compared with nonuse. We included one meta-analysis of individual patient data that evaluated the association between degree of LDL-C reduction and clinical outcomes,⁴⁸ as the data were not available for us to perform this analysis. Otherwise, we reviewed reference lists of systematic reviews to identify potentially relevant studies. The selection of literature is summarized in the literature flow diagram (Appendix A3). Appendix A4 lists excluded studies with reasons for exclusion.

Data Abstraction and Quality Rating

We abstracted details about the study design, patient population, setting, screening method, interventions, analysis, followup, and results. Two investigators independently applied criteria developed by the USPSTF⁴ to rate the quality of each study as good, fair, or poor (Appendix A5). Discrepancies were resolved through a consensus process. When risk estimates were not reported for individual studies, we calculated the relative risk (RR) and 95 percent confidence interval (CI) if adequate data (number of events and sample size) were provided.

Data Synthesis

We conducted meta-analyses to calculate risk ratios for effects of statins on clinical outcomes using the DerSimonian and Laird random-effects model with Review Manager Version 5.2 software (The Cochrane Collaboration Nordic Cochrane Centre, Copenhagen, Denmark). Statistical heterogeneity was assessed using the I² statistic.⁴⁹ For stroke, we excluded hemorrhagic strokes when data permitted. When statistical heterogeneity was present, we performed sensitivity analysis with the profile likelihood method using Stata Version 10.1 (StataCorp, College Station, TX), as the DerSimonian and Laird model can result in overly narrow CIs in this situation.⁵⁰ We performed additional sensitivity and stratified analyses based on study quality, exclusion of trials that enrolled patients with prior CVD events, duration of followup, intensity of statin therapy (based on the ACC/AHA guideline),³⁰ mean TC and LDL-C levels at baseline, and whether the trial was stopped early. We constructed funnel plots to detect small sample effects (a marker for potential publication bias) for analyses with greater than 10 trials.⁵¹

We assessed the aggregate internal validity (quality) of the body of evidence for each Key Question (good, fair, poor) using methods developed by the USPSTF, based on the number, quality, and size of studies, consistency of results between studies, and directness of evidence.⁴

External Review

The draft report was reviewed by content experts (Appendix A6), USPSTF members, AHRQ Project Officers, and collaborative partners and was posted for public comment.

Response to Public Comments

The draft report was posted for public comment on the USPSTF Web site from December 22, 2015 to January 25, 2016, and few comments were received. No comments identified missing studies or errors in the evidence reviewed, resulting in no changes to the findings or conclusions of the report.