NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Sharma M, Ansari MT, Soares-Weiser K, et al. Comparative Effectiveness of Lipid-Modifying Agents [Internet]. Rockville (MD): Agency for Healthcare Research and Quality (US); 2009 Sep. (Comparative Effectiveness Reviews, No. 16.)

  • This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

Cover of Comparative Effectiveness of Lipid-Modifying Agents

Comparative Effectiveness of Lipid-Modifying Agents [Internet].

Show details

4Summary and Discussion

This report addresses the effectiveness and safety of adding lipid modifying agents to statin therapy. Few long term studies were available reporting on major clinical endpoints such as incidence of myocardial infarction, mortality, adverse events and adherence. Most of the available evidence focused on short term studies of surrogate markers linked to vascular disease. The largest number of trials was found for the ezetimibe plus statin combination, with fewer studies for other combinations.

In treated individuals whose lipid profile is suboptimal the clinician must decide whether to increase the dose of statin and continue monotherapy or to add another medication. However, the comparator for most trials was not a higher, but rather the same dose of statin monotherapy. Indeed, a number of publications specifically stated that the comparator was the starting dose of the particular ongoing statin. Of note, a recent meta-analysis comparing more intensive statin treatment with less intensive treatment demonstrated a significant reduction in LDL-c levels in high risk patients with more intensive therapy. No statistical difference was observed in discontinuation rates attributable to drug related harms.226 Further, as discussed below, multiple medications may decrease adherence to treatment, a critical factor in determining the outcomes of individuals on long term preventive therapies.

The choice of nonstatin medication to be added to therapy was more difficult to address. A single included study compared statin combination therapies using niacin or ezetimibe, with statin monotherapy.139 All treatments examined resulted in similar reductions in LDL-c, while the niacin combination therapy resulted in significantly greater increases in HDL-c. No other direct comparisons of various combinations were identified, so the effect of these strategies can only be compared indirectly.

Clinical Outcomes

All cause mortality and vascular death in individuals requiring intensive lipid lowering therapy was not specifically examined in trials of combination therapy and higher dose statin monotherapy, for any of the combinations studied.

We therefore examined all trials providing evaluable data on these mortality endpoints, for all statin doses, and found a neutral odds ratio for all-cause mortality with ezetimibe, bile acid sequestrants, fibrate, niacin, or omega-3 fatty acids in combination with statins, compared with statin monotherapy. It should be noted that there were few deaths in the included trials, which is likely a function of the relatively short periods of followup. Thus the statistical power to observe such differences was low. This finding is in keeping with that of Josan et al, who noted a neutral impact on all-cause mortality with intensive statin therapy compared with lower dose statin therapy, in a quantitative systematic review of seven trials.226 Similar findings were noted for vascular mortality. Among all reports of this outcome there were few participants with fatal myocardial infarctions, with no observed differences between combination and monotherapy treatments. A previous meta-analysis of mortality comparing classes of lipid modifying therapy to placebo suggested benefits from all therapies considered in the present review with the exception of fibrates, which were associated with an excess of noncardiovascular mortality.227 A subsequent report suggested that this association disappears if trials employing clofibrate are excluded from statistical pooling.228 Clofibrate is not approved for use in the United States.

No significant difference was noted in the occurrence of non-fatal myocardial infarction or acute coronary syndrome. Indeed, there was no evidence of additional benefit from combination therapy when compared to higher dose statin monotherapy for any clinical outcomes. Several caveats are important to note regarding clinical outcomes. First, as noted above, most of the evidence to date has focused on short duration studies aimed at intermediate outcomes and there is insufficient data for most outcomes of clinical importance. Second, the comparator arms rarely explored higher statin doses which may have advantages in terms of medication adherence and, for some medications, cost. While some data exists for the benefit of niacin and sequestrants alone or in combination with a statin in coronary heart disease, it remains unclear if the marginal benefit of adding these agents to a lower dose of a statin is a better strategy than increasing statin dose, particularly for individuals managed to ATP III targets.43

Stroke was a very rare event in this group of trials and no conclusions can be drawn regarding the differential impact of these interventions on its occurrence. While stroke is commonly considered to be an indication for lipid modification, note should be made that the NCEP ATP III guidelines specify symptomatic carotid disease as a coronary heart disease risk equivalent. This therefore excludes cardioembolic stroke, stroke due to small vessel disease and intracerebral hemorrhage as guideline supported indications for therapy. Ample evidence supports the beneficial effects of statins on stroke incidence in individuals with cardiac disease, but there is only a single trial demonstrating benefit from statin therapy in individuals treated after stroke.18,229 In addition, lower cholesterol levels are associated epidemiologically with higher rates of intracranial hemorrhage, and statin treatment may increase the likelihood of its occurrence.230,231 Thus the therapeutic window, balancing potential risks and benefits, for treatment following stroke may be narrower than for coronary heart disease. Further work is required to identify the characteristics of individuals with stroke whose potential for benefit with intensive lipid lowering exceeds any potential for harm. Specifically, trials are required with significant recruitment from secondary prevention stroke populations.

Serious Adverse Events

Our review of serious adverse events and cancer was not constrained to specific statin dose comparisons but rather included all trials comparing combination therapy with statin monotherapy. The ezetimibe combination had the largest number of trials reporting this outcome, but the majority of these were less than 24 weeks duration, with a small minority reporting up to 52 weeks. In these longer duration studies the serious adverse events rate was approximately 10 percent in the combination and monotherapy groups, with no significant difference in proportions of participants experiencing serious adverse events between groups. One large trial of omega-3 fatty acids added to statin therapy in a Japanese population did not demonstrate any increase in cancer compared with monotherapy.141 Data for the other interventions was sparse or had significant limitations.

Surrogate Outcomes

Ample evidence supports the selection of LDL-c as the primary target for lipid modifying therapy. A number of studies have established the correlation of LDL-c cholesterol and incident coronary heart disease or recurrent myocardial infarction in men and women.232–237 Law et al, in a review of 164 trials of statins noted that these interventions reduce LDL-c by an average of 70 mg/dL, with a range from 70 to 108 mg/dL. For each reduction of 40 mg/dL cardiac events were reduced by 11 percent in the first year, 24 percent in the second and over 30 percent subsequently.238 A lower incidence of major cardiovascular events is associated with more intensive statin therapy than with less intensive treatment.226

When compared to a higher dose of statin, no significant difference was found in LDL-c reduction for fibrate in combination with statins compared with statin monotherapy in populations requiring intensive lipid lowering therapy. However, in two trials, 10 to 20 percent significant additional mean percentage reductions in LDL-c were demonstrated in high CHD risk participants, in favor of lower dose simvastatin plus ezetimibe combination therapy compared with higher dose monotherapy. There were no trials with this comparison for niacin, bile acid sequestrants or omega-3 combinations with statins.

Overall, there was at best scant evidence to support a greater lowering of LDL-c with any of the five combinations reviewed than with higher dose statin therapy in participants requiring intensive lipid lowering therapy. However, when combinations were compared with similar doses of statin monotherapy in this population, statin-ezetimibe combinations caused additional reductions in LDL-c compared with statin monotherapy. In populations requiring intensive lipid lowering therapy, all 18 trials exceeding 6 weeks in duration of statin-ezetimibe combination therapy were associated with a greater reduction in LDL-c ranging from 4 to 27 percent. This compares to indeterminate or inconsistent results for Fibrates, BAS, Niacin and Omega-3, possibly due to small sample sizes, differences in statin dosages, or few to absent trial data in populations requiring intensive lipid lowering therapy. Both BAS and ezetimibe interfere with absorption from the intestines and would be expected to have an impact on LDL-c levels. When used as monotherapy, BAS have been shown to decrease LDL-c by 15 to 30 percent43,239 and ezetimibe by 18 percent,240 while the LDL-c reduction by fibrate has been considered to be marginal.241

Evidence was reviewed for the outcome of attainment of ATP III LDL-c goals for combination therapy compared with a higher dose statin. Ezetimibe in combination with lower dose simvastatin compared with higher dose simvastatin monotherapy was associated with a significantly greater odds of attaining the LDL-c target, with an odds ratio of 7.21 (95% CI 4.30, 12.08), on the basis of two pooled trials. No difference was noted for fibrate on the basis of a single small trial, and no evidence was available for niacin, BAS or omega-3 combinations. As treatment to a target LDL-c is both the major goal of therapy as well as a justification for using combinations, this represents an important issue to be addressed in future work.

HDL-c is identified in the ATP III report as inversely correlated with coronary heart disease risk, and while the relationship is continuous, a level below 40 mg/dL has been identified as low.43 A target for therapeutic intervention has not been set by these guidelines and it remains unclear whether raising HDL-c has an impact on coronary heart disease that is independent of LDL-c levels. For the direct comparison of combination therapy versus higher dose statin monotherapy, a single trial suggested no difference for combinations with ezetimibe or fibrate in participants requiring intensive lipid lowering therapy. In another trial, a significant increase in mean percentage change was noted for the combination of rosuvastatin 10 mg/day plus niacin 2 g/day compared with rosuvastatin 40 mg/day monotherapy in participants with combined dyslipidemia and low HDL-c (below 45 mg/dL), who were not necessarily in need of intensive lipid lowering therapy. Niacin has an effect on HDL-c levels at low doses, while higher doses are required to reduce LDL-c.43 Thus these findings are consistent with previous work with this agent.196

Some evidence suggests that treatment with niacin plus a statin may affect the progression of intermediate markers of atherosclerosis. Taylor et al examined the progression of CIMT in individuals with CAD and HDL-c below 45 mg/dL, treated with niacin or placebo added to ongoing statin therapy. Over a one year period combination therapy was associated with a nonsignificantly lower rate of progression than the comparator monotherapy group. The majority of the comparator group was on a statin, usually simvastatin, but the diversity of statin treatments in this group makes interpretation somewhat difficult.196 While the accumulated evidence suggests that raising HDL-c levels may be helpful in high risk populations, the target levels and optimal strategies remain unknown.

Two trials investigated this outcome measure in participants in need of intensive lipid lowering therapy. One compared niacin-statin combination with background statin monotherapy (ARBITER-2) while the other investigated simvastatin 80 mg/day plus ezetimibe 10 mg/day, compared with simvastatin 80 mg/day monotherapy (ENHANCE trial). No significant differences were found between the treatments. No evidence was found pertaining to the question of lower dose statin in combination therapy versus higher dose monotherapy for CIMT. The ENHANCE trial was similar in design to the two year ASAP trial that showed significant regression in CIMT with atorvastatin 80 mg/day compared with simvastatin 40 mg/day.242. Important differences can be recognized between the ASAP and ENHANCE trials, including a higher baseline CIMT and inclusion of statin naïve participants in ASAP. Further, as pointed out by Brown and Taylor, none of the intervention studies on CIMT of two years or less duration have demonstrated an effect.41The questions surrounding the findings of the ENHANCE trial will require further long term studies focused on clinical outcomes.

Adherence and Harms

Scant evidence exists, comparing short term harms and treatment adherence for lower dose statin in combination therapy with higher dose monotherapy across all populations. No significant treatment differences were noted.

A common adverse event with niacin is flushing, reported by as many as 88 percent of individuals initiating slow release niacin.243 Of note however, the absolute rates of withdrawal in niacin plus statin treatment groups in four trials were not more than 10 percent, even with significant odds in favor of monotherapy (2.38, 95% CI 1.63, 3.47). On average 5 percent withdrew from statin plus BAS combination therapy in contrast to 2 percent from statin monotherapy, but the pooled odds from nine trials was not significant (OR 1.80; 95% CI 0.68, 4.76).

No participant developed rhabdomyolysis across all 87 RCTs investigating five statin combination therapies. These results are recognized in and consistent with extant literature.244 However, this lack of evidence fails to shed light upon the relative safety of lower dose statin in combinations, compared with higher dose monotherapy.

Comparing statin combination therapies with monotherapy using similar statin doses in single trials, significantly fewer participants adhered to pravastatin plus cholestyramine combination therapy (OR 0.41, 95% CI 0.26, 0.65), rosuvastatin plus cholestyramine (OR 0.10, 95% CI 0.04, 0.25), and rosuvastatin plus niacin combination treatments (OR 0.42, 95% CI 0.21, 0.85).

Medication adherence is a significant issue in determining population benefit. In a population based cohort study, Sokol demonstrated an association between medication adherence and lower medical costs and reduced hospitalization rates in individuals with hypercholesterolemia.245 In general, medication nonadherence rates range from 20 to 50 percent.246,247 Chronic conditions and preventive therapies are associated with poorer adherence rates than acute conditions.247 The complexity of medication regimen and the number of medications may play a role in adherence.248,249 Thus there may be benefit from less complex regimens employing fewer separate agents. Medication adherence and persistence are related but distinct concepts. Adherence is defined as the extent to which an individual acts in accordance with the prescribed interval and dose of a dose regimen while persistence is the accumulation of time from initiation to discontinuation of therapy.250 The previous literature uses the terms interchangeably. While this report refers to adherence, this outcome was rarely reported and for most reports we could only extract data regarding the proportion of participants withdrawing from treatment.


There is dearth of evidence regarding lower dose statin in combination therapy versus higher dose monotherapy in subgroups. Absence of evidence or at best scant trial evidence precluded definitive conclusions regarding short term and longer term efficacy.

Seven trials were included which reported on surrogate outcomes in participants with diabetes mellitus, comparing ezetimibe combination therapy with statin monotherapy. Considerable heterogeneity precluded a summary point estimate, but the results favored combination therapy in all trials with the mean percentage change in LDL-c from baseline ranging from 4 to 26 percent. Only one trial compared higher dose monotherapy with combination therapy. Asymmetry of the funnel plot was noted in this group of trials. While such asymmetry may be the consequence of publication bias it can be observed for other reasons including heterogeneity. The small number of studies and the presence of heterogeneity in this group of trials makes it difficult to ascribe the observed lateralization to publication bias.251

There was no analyzable evidence comparing lower dose statin combination therapy with higher dose statin monotherapy for the mean percentage change from baseline or changes scores in participants with diabetes mellitus. Reductions in triglyceride levels and elevations in HDL-c are considered to be desirable, albeit with a lower level of evidence. Both non-HDL-c cholesterol and apolipoprotein b (Apo-B) correlate with cardiovascular risk. The ATP III guidelines recommend non-HDL-c as a secondary target in individuals with hypertriglyceridemia. A single trial compared combination therapy with fibrates versus a higher dose statin in individuals with diabetes. The combination was favored with a mean percentage change from baseline of −13.57 mg/dL (95% CI −24.16 mg/dL, −2.98 mg/dL).125 Evidence was available for non-HDL-c in diabetics for the comparison of combination therapy with ezetimibe or fibrates, with similar dose statin monotherapy. In six trials an additional reduction of 4 to 27 percent was seen with the addition of ezetimibe to statin. A single trial with fibrate combination therapy resulted in no significant benefit.

While current treatment guidelines in diabetes continue to support the primacy of statin therapy and LDL-c reduction in managing vascular risk,252a consensus panel assembled by the American Diabetes Association and the American College of Cardiology Foundation recommended targets for both non HDL-c and Apo-B in individuals with diabetes, established cardiac disease or combinations of risk factors.253 The panel acknowledged that further data was needed regarding these therapeutic targets and that there was a lack of robust data on the effects of combination therapies on outcomes. The optimal management of individuals with diabetes as well as the role of targets other than LDL-c will continue to evolve as the results of ongoing trials of niacin and fibrates in combination with statins become available.

With few exceptions, included trials were mixed with respect to gender, but limited subgroup data in women did not show a definitive difference in LDL-c reduction between lower dose statin plus ezetimibe combination therapy versus higher dose statin monotherapy. This finding may be the result of the small number of participants available for analysis. Most trials were comprised of a majority white population of European descent. Goff et al in a multicenter cohort study reported that the prevalence of dyslipidemia was similar in populations of African and Hispanic descent in the US, but that they were less likely to be treated or controlled.254 Mexican Americans are significantly less likely than non-Hispanic whites to be aware of and treated for dyslipidemia, despite having only slightly lower prevalence of the condition. While access and socioeconomic issues impact treatment and control, members of both populations fall into groups who require lipid modifying treatment and trials of these therapies should reflect that reality.

Demographic trends suggest that the elderly will grow more sharply than other segments of the population, with those over 65 increasing from 37 million in 2005 to 81 million in 2050.255 The average age of participants in included trials was in the fifties, limiting generalizability to older populations. Deedwania et al compared moderate statin therapy (pravastatin 40 mg/day) with intensive therapy (atorvastatin 80 mg/day) in individuals aged 65 to 85 years, with coronary artery disease. The atorvastatin group had fewer deaths and major cardiac events, though there was an increase in the proportion with elevated hepatic enzymes.256 Robinson et al257 examined 16 trials of ezetimibe plus statin or placebo combination therapy and statin monotherapy. All included trials were performed by Merck/Schering Plough and published by December 2006. A four week single blind placebo run in was followed by a treatment period of six to12 weeks. The analysis was not by intention to treat. Of the over 13,000 individuals randomized, approximately 4,400 were over 65 years, of whom 1,147 were over 75 years of age. Neither the treatment effect for surrogate outcomes nor the incidence of adverse events varied significantly by age group. Limitations, including trial selection and short durations, limit the applicability of this review. Further it must be acknowledged that individuals over the age of 65 are heterogeneous with respect to the probability of benefit as well as susceptibility to adverse events. Trials of combination therapy need to involve the larger population of Americans at older ages who require therapy, to confirm efficacy and tolerability.

Long-term nonrandomized studies directly investigating clinical effectiveness, serious adverse events and cancer were also lacking. Sparse event data in the three included studies could not guide any definitive conclusions.

C-reactive protein is an inflammatory biomarker which predicts vascular risk and may improve risk stratification beyond that afforded by LDL-c.258 Ridker et al examined the impact of rosuvastatin monotherapy in a group of apparently healthy individuals with LDL-c levels below current levels for drug therapy, but elevated levels of C-reactive protein.259 Significant reductions in the occurrence of cardiac events, stroke and vascular death were observed in the treated group, suggesting that this biomarker may identify a subgroup for treatment. The role of combination therapy in this population has not been studied to date.


Our review does not examine specifically the addition of a combination medication to maximal statin therapy. There are instances, such as familial dyslipidemia, in which maximal statin therapy may be insufficient, so combinations may be required to achieve primary or secondary treatment goals in some individuals.

The assessments of clinical outcomes, harms and treatment adherence were limited by the paucity of long term studies with a sufficient number of events to offer meaningful results. The search for specific harms was limited to prespecified important events rather than all potential adverse experiences. Thus studies investigating specific minor adverse experiences were not captured unless adverse events lead to nonadherence or withdrawal from treatment. Composite outcomes were rarely reported in the included trials and imputation was not attempted as the possibility of double counting could not be avoided. We used a conservative approach to pooling with a strict limit to allowable heterogeneity which precluded pooling of results in a number of instances.

A number of caveats apply to the evaluation of surrogate outcomes. First, the absolute benefit in measures of LDL-c and HDL-c may depend on the baseline status, including intensity of prior statin therapy and comorbidities. In conducting our review we used percentage change from baseline as that was most commonly reported, as well as change scores if data permitted. Percentage change from baseline has lower statistical power and may fail to protect against bias in the case of baseline imbalances.3 Triglyceride levels may be reported as medians rather than means when the distribution is skewed, but there are no widely accepted methods to pool data reported as medians. For this reason we specifically extracted means because there are techniques to pool these values. However, due to this methodological limitation the impact of therapies on triglyceride levels may be underestimated.

Indirect comparisons are hazardous given the potential differences in trial populations and design. Indirect comparisons may inflate estimates of differences and were not attempted.4

A large number of studies were funded by pharmaceutical companies. Evidence suggests that industry sponsorship of research is associated with a greater likelihood of results favoring the sponsored product.5–7 We did not detect many instances of possible publication bias based upon funnel plots in this review, but the power to detect was limited.

A number of concerns regarding trial quality were identified. Only 26 of 87 (30%) included randomized controlled trials reported allocation concealment and 21 (24 percent) reported an intention to treat analysis. Clinical end points were rarely adjudicated and blinding was not consistently reported.


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...