Search Strategy, Study Selection, and Data Abstraction

We searched the following databases for primary studies: MEDLINE^®, Embase^®, and the Cochrane Central Register of Controlled Trials (CENTRAL) from May 2008 through July 2013. We also reviewed relevant review articles. In addition, we requested and reviewed Scientific Information Packets provided by the pharmaceutical manufacturers.

Abstract and full-text screening was performed by two independent reviewers using prespecified eligibility criteria (Table A). All articles included in the prior review were reviewed during the full-text screen. Conflicts were resolved by consensus adjudication.

Table A

Study inclusion and exclusion criteria.

Data abstraction was conducted with a senior reviewer (faculty-level project investigator) abstracting data from articles while having access to the first reviewer's data abstraction. Differences in opinion were resolved through consensus adjudication and, for difficult cases, during team meetings.

Risk-of-Bias Assessment

Risk of bias was assessed independently by two reviewers using the Cochrane Collaboration's tool. For studies included from the prior review, we used the quality assessments from that report, which used the Jadad Score.

Data Synthesis

We compared lower potency statins in combination therapy with higher potency statin monotherapy, which enabled us to synthesize data across statin type and statin dose. We used specific criteria to determine statin potency (Table B).

Table B

Different dosing of specific statins based on potency to reduce LDL-c.

We calculated and displayed the mean differences with 95-percent confidence intervals for the individual studies grouped by combination therapy agent, statin potency, and population for all comparisons. We considered meta-analysis where there were three or more similar studies. We report qualitative synthesis of data for most outcomes because of the lack of outcomes meeting our criteria for meta-analysis and significant heterogeneity detected when meta-analyses were conducted (I² >50%).

Strength of the Body of Evidence

We graded the quantity, quality, and consistency of the evidence for the following outcomes: mortality, acute coronary events, revascularization procedures, serious adverse events, LDL-c, and HDL-c. We used an evidence grading scheme recommended by the “Methods Guide for Effectiveness and Comparative Effectiveness Reviews.”¹⁵ We created evidence grades for each comparison and outcome by combination agent, statin potency, and population. We used four domains to yield a final evidence grade: risk of bias, consistency, directness and precision.

The final strength-of-evidence (SOE) grades were: (1) “high” grade (indicating high confidence that the evidence reflects the true effect and further research is very unlikely to change our confidence in the estimate of the effect); (2) “moderate” grade (indicating moderate confidence that the evidence reflects the true effect and further research may change our confidence in the estimate of the effect and may change the estimate); (3) “low” grade (indicating low confidence that the evidence reflects the true effect and further research is likely to change our confidence in the estimate of the effect and is likely to change the estimate); and (4) “insufficient” grade (no evidence identified). A comparison-outcome pair with high SOE was one with low risk of bias, directness, consistency, and precision. Moderate SOE indicated that a high risk of bias was noted or that two of the following were observed: a moderate risk of bias, inconsistency, indirectness, or imprecision. Low SOE indicated a high risk of bias and two or more of the following or a moderate risk of bias and three of the following: inconsistency, indirectness, and imprecision.

Investigators writing each section completed the SOE grading, which was then reviewed by the team.

Applicability

We describe the applicability of studies in terms of the degree to which the study population, interventions, outcomes, and settings were relevant to individuals at high CHD risk requiring aggressive lipid-modifying therapy and features that may affect the effectiveness of the intervention.

Results of Literature Searches

Figure B summarizes the search results. The literature search identified 4,293 unique citations. During the title and abstract screening we excluded 3,396 citations; during the article screening we excluded 380 citations (see Appendix D in the full report). Fifty-five studies, reported in 59 articles, were included. All trials were randomized controlled trials.

Figure B

Summary of search (number of articles). CHD = coronary heart disease; LDL = low-density lipoprotein; RCT = randomized controlled trial; SIP = Scientific Information Packet ^a Total exceeds the number of citations in the exclusion box because citations (more...)

Overview of Included Trials by Potency and Agent

The SOE was variable across comparisons evaluating the effectiveness and safety of combination therapy versus intensification of statin monotherapy. Evidence for all the clinical outcomes of mortality, acute coronary events, and revascularization procedures was graded as insufficient across all potency comparisons for all combination therapy regimens.

Seven comparisons had moderate SOE for LDL-c and HDL-c outcomes. However, all other comparisons and outcomes had low or insufficient evidence. The interventions and approaches that effectively lowered LDL-c or raised HDL-c are described by combination therapy regimen below. The SOE for the body of evidence is provided in Table C for general populations and Table D for subgroups.

Table C

Summary of the strength of evidence for general populations.

Table D

Summary of the strength of evidence for subgroups.

Combination Therapy Versus Intensification of Statin Monotherapy

Key Findings

The evidence suggests that some combination therapy regimens may confer benefits with respect to lowering LDL-c, including bile acid sequestrants (up to 14 percent greater LDL-c reduction) and ezetimibe (up to 21 percent greater LDL-c reduction). LDL-c is an important factor in the development of atherosclerotic cardiovascular disease, and higher levels of LDL-c have been associated with greater risk of this disease.^4,5 However, there is insufficient evidence to address whether the LDL-c–lowering benefits achieved with these medications translate into decreased rates of atherosclerotic cardiovascular disease. Prior trials comparing combination regimens to statin monotherapy, such as ENHANCE, AIM-HIGH, and ACCORD-Lipid, have demonstrated that combination therapy can lead to superior lipid outcomes but fail to reduce clinical outcomes such as cardiovascular death, MI, revascularization, or stroke.^7,13,14

We also found that some combination therapy regimens may confer benefits with respect to raising HDL-c, including ezetimibe (up to 6 percent) and niacin (up to 27 percent). In particular, given that only one prior study has demonstrated the benefit of pharmacologically raising HDL-c with respect to prevention of CVD events,¹⁶ the potential long-term clinical benefits of these combination regimens with respect to their HDL-c effects are unclear.

The strength of evidence is provided for all observed comparisons in general populations in Table C and for subgroups in Table D. Most trials included in this report were of relatively short duration (<3 months). In this limited timeframe, investigators are unlikely to capture any changes in a chronic condition such as atherosclerotic cardiovascular disease, which typically develops and progresses over a number of years. Powering such studies is especially difficult, given that both arms are taking statins, which would reduce the baseline incidence of cardiovascular events. Therefore, currently it is not possible to draw conclusions about the clinical implications of the surrogate marker changes identified. However, until additional data are available, these results may help health care providers tailor lipid-modifying regimens based on individual patient needs and concerns for adverse events.¹⁷

Applicability

Many trials that met our inclusion criteria were implemented in patients with hyperlipidemia, and most were designed to evaluate effects on lipid measures and short-term harms. The results of most trials generalize to patients with hyperlipidemia uncomplicated by other major comorbid conditions. Interestingly, we identified fewer trials that were conducted among patients at high risk for CHD, such as those with diabetes or preexisting cardiovascular disease. These patients could benefit the most from improvement in their lipid profiles and are the most likely to be receiving more aggressive lipid-modifying regimens in clinical practice.

Limitations of the Evidence Base and Review Process

The SOE was insufficient for many comparisons and outcomes because of a paucity of studies and poor quality of existing studies. Trials were frequently downgraded in risk-of-bias assessment for lack of blinding by participant and study personnel (performance bias), for not reporting the blinding of outcome assessors (detection bias), or for not accounting for losses to followup or handling of incomplete data (attrition bias). Few studies reported variance estimates for the between-group differences in any outcomes over time. In some instances, the studies did not report a mean difference or point estimate, stating only that there was no significant difference between the groups. In addition, some studies did not report an intention-to-treat analysis and others did not specify the number analyzed in each arm. All of these factors limited our ability to conduct meta-analyses. Where we conducted meta-analyses, substantial heterogeneity was present in most cases.

The evidence base was also limited due to the short duration of most studies. Most trials we identified were of relatively short duration, despite the fact that these medications are currently used in clinical practice as chronic long-term medications. Studies were of insufficient duration to adequately assess long-term clinical outcomes, including mortality, acute coronary events, and revascularization procedures. In addition, losses to followup and medication adherence were often not reported by intervention arm in trials, which may bias our results. While our findings may suggest that one therapeutic option provides a benefit over another, we cannot comment on the tolerability of or persistence with the regimen, given the lack of data and short trial duration. Additional long-term trials are needed to compare the tolerability, side effects, and harms with prolonged use of these combinations.

The review process imposed limitations as well. First, the review focused narrowly on combination therapy compared with statin intensification. As a result, many studies of add-on combination therapy versus the same statin dose or nonstatin monotherapy were excluded because they did not address the Key Questions. Given several previous reviews on dietary modification and reduction of lipids and CVD risk, we did not include these therapies in this review.^18,19 Further, we did not examine differences in statin response based on genetic variations.^20,21 Second, we excluded non–English-language publications, although we do not believe this introduced significant bias. Third, because this review was conducted prior to the release of the 2013 cholesterol treatment guidelines from the American College of Cardiology/American Heart Association Task Force, we could not define our population eligibility criteria to match their four “statin benefit groups” and our potency categorizations differ slightly from those in the guidelines.¹⁰

Future Research Needs

We suggest that most comparisons and outcomes that have low or insufficient evidence are future research needs. In order to answer whether there are long-term benefits with respect to mortality, acute coronary events, and revascularization procedures, future investigators need to make these endpoints the primary outcomes of their trials and ensure that trials are of sufficient duration to actually capture these events (at least 12 months and preferably longer). Short-term trials using surrogate endpoints are of diminishing value at this point.

We further suggest that future studies focus on high-risk CHD populations and populations with greater burden of cardiovascular disease to determine which strategy provides better short-term improvements in lipid profile and long-term clinical benefits. These populations include patients with diabetes and preexisting cardiovascular disease, as well as Black and Native American populations.²² It may be worthwhile to explore differences between men and women, as the ACCORD trial showed benefit of combination therapy with fibrate in men and potential harms with this combination therapy in women.¹⁴ Such studies would have tremendous impact on clinical practice, as these patients with greater burden of cardiovascular disease are the most likely to need a more aggressive lipid-modifying regimen.

While head-to-head comparisons of a combination regimen to intensification of statin therapy may answer important clinical questions, these trials do not help clinicians decide between different combination therapy options. Once the effectiveness of each combination regimen on long-term clinical outcomes is established, the next step to inform clinical decisionmaking would be to help clinicians determine how to select the most appropriate lipid-modifying regimen from all available options. We suggest that future studies conduct head-to-head comparisons of multiple combination regimens against each other as well as against intensification of statin monotherapy to address this need. Additionally, it would be useful to examine whether it is possible to achieve LDL-c reductions consistent with those from potent statins (50–60%) in patients who are unable to tolerate full-dose statin therapy and what the clinical effects of these reductions would be. Furthermore, it would be useful to determine if LDL-c lowering of 50 percent achieved with a statin and a bile acid sequestrant is as efficacious as similar LDL-c lowering with a statin and ezetimibe, and whether both used together are as efficacious as a potent statin alone. Finally, alternative study designs, such as observational studies using registry data from electronic medical records, may also provide useful data on clinical outcomes.

Conclusions

Although many studies looked at intermediate outcomes, few studies addressed the question of which approach produces better clinical outcomes. Combination of statin with ezetimibe or bile acid sequestrant lowered LDL-c better than intensification of statin monotherapy, but evidence for clinical outcomes (mortality, acute coronary events, and revascularization procedures) was insufficient across all potency comparisons for all combination therapy regimens. Additional studies evaluating long-term clinical benefits and harms are needed to better inform clinical decisionmaking, patient choice, and clinical practice guidelines.