Fibrates for primary prevention of cardiovascular disease events

Tobias Jakob; Alain J Nordmann; Stefan Schandelmaier; Ignacio Ferreira‐González; Matthias Briel

doi:10.1002/14651858.CD009753.pub2

. 2016 Nov 16;2016(11):CD009753. doi: 10.1002/14651858.CD009753.pub2

Fibrates for primary prevention of cardiovascular disease events

Tobias Jakob ¹, Alain J Nordmann ¹, Stefan Schandelmaier ¹, Ignacio Ferreira‐González ², Matthias Briel ^1,^✉

Editor: Cochrane Heart Group

PMCID: PMC6464497 PMID: 27849333

Abstract

Background

Fibrates are effective for modifying atherogenic dyslipidaemia, and particularly for lowering serum triglycerides. However, evidence that fibrates reduce mortality and morbidity associated with cardiovascular disease (CVD), or overall mortality and morbidity, in the primary prevention of CVD is lacking.

Objectives

This Cochrane Review and meta‐analysis aimed to evaluate the clinical benefits and harms of fibrates versus placebo or usual care or fibrates plus other lipid‐modifying drugs versus other lipid‐modifying drugs alone for the primary prevention of cardiovascular disease (CVD) morbidity and mortality.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE (Ovid), Embase (Ovid), CINAHL (EBSCO), and Web of Science (all from inception to 19 May 2016). We searched four clinical trial registers (last searched on 3 August 2016) with the help of an experienced professional librarian. We searched the databases to identify randomised controlled trials (RCTs) evaluating the clinical effects of fibrate therapy in the primary prevention of CVD events. We did not impose any language restrictions.

Selection criteria

We aimed to include all RCTs comparing the effects of fibrate monotherapy versus placebo or usual care, or fibrates plus other lipid‐modifying drugs versus other lipid‐modifying drugs alone. Included studies had a follow‐up of at least six months for the primary prevention of CVD events. We excluded trials with clofibrate, because it was withdrawn from the market in 2002.

Data collection and analysis

Two review authors independently screened titles and abstracts for potential study inclusion. Two review authors independently retrieved the full‐text papers and extracted data. Disagreements were resolved by consensus. We calculated risk ratios (RRs) and accompanying 95% confidence intervals (CIs) for aggregate data on primary and secondary outcomes. We tested for heterogeneity with the Cochrane Q‐test and used the I² statistic to measure inconsistency of treatment effects across studies. Using the GRADE approach, we assessed the quality of the evidence and used the GRADE profiler software (GRADEpro GDT) to import data from Review Manager 5 to create 'Summary of findings' tables.

Main results

We identified six eligible trials including 16,135 individuals. The mean age of trial populations varied across trials; between 47.3 and 62.3 years. Four trials included individuals with diabetes mellitus type 2 only. The mean treatment duration and follow‐up of participants across trials was 4.8 years. We judged the risks of selection and performance bias to be low; risks of detection bias, attrition bias, and reporting bias were unclear. Reporting of adverse effects by included trials was very limited; that is why we used discontinuation of therapy due to adverse effects as a proxy for adverse effects. Patients treated with fibrates had a reduced risk for the combined primary outcome of CVD death, non‐fatal myocardial infarction, or non‐fatal stroke compared to patients on placebo (risk ratio (RR) 0.84, 95% confidence interval (CI) 0.74 to 0.96; participants = 16,135; studies = 6; moderate‐quality of evidence). For secondary outcomes we found RRs for fibrate therapy compared with placebo of 0.79 for combined coronary heart disease death or non‐fatal myocardial infarction (95% CI 0.68 to 0.92; participants = 16,135; studies = 6; moderate‐quality of evidence); 1.01 for overall mortality (95% CI 0.81 to 1.26; participants = 8471; studies = 5; low‐quality of evidence); 1.01 for non‐CVD mortality (95% CI 0.76 to 1.35; participants = 8471; studies = 5; low‐quality of evidence); and 1.38 for discontinuation of therapy due to adverse effects (95% CI 0.71 to 2.68; participants = 4805; studies = 3; I² = 74%; very low‐quality of evidence). Data on quality of life were not available from any trial. Trials that evaluated fibrates in the background of statins (2 studies) showed no benefits in preventing cardiovascular events.

Authors' conclusions

Moderate‐quality evidence suggests that fibrates lower the risk for cardiovascular and coronary events in primary prevention, but the absolute treatment effects in the primary prevention setting are modest (absolute risk reductions < 1%). There is low‐quality evidence that fibrates have no effect on overall or non‐CVD mortality. Very low‐quality evidence suggests that fibrates are not associated with increased risk for adverse effects.

Plain language summary

Fibrates for patients without established cardiovascular disease

Review question

What are the benefits and harms of using fibrate treatment compared to placebo or usual care for preventing cardiovascular disease in people at increased risk of developing cardiovascular disease?

Background

Cardiovascular disease is the most common cause of death, illness, disability, and reduced quality of life in industrialised countries. One of the major risk factors for cardiovascular disease is elevated low‐density lipoprotein cholesterol (LDL‐C, 'bad' cholesterol). In addition, persons with elevated serum triglycerides and low levels of high‐density lipoprotein cholesterol (HDL‐C, 'good' cholesterol) are also at increased risk for cardiovascular disease events such as heart attacks or strokes. Fibrates lower serum triglycerides, modestly raise HDL‐C, and modestly lower LDL‐C. Therefore, long‐term therapy with fibrates may help prevent cardiovascular disease events, in particular in combination with statins, for which it has been shown that they substantially lower LDL‐C and reduce the risk of heart attack, stroke, and overall mortality.

Study characteristics

The evidence is current to May 2016. We identified six eligible primary prevention trials including 16,135 individuals without established cardiovascular disease that compared fibrate therapy with placebo or usual care. The mean age of the trial populations varied between 47.3 and 62.3 years; the majority of included individuals had diabetes mellitus type 2. The mean treatment duration and follow‐up of participants across trials was 4.8 years.

Key results and quality of the evidence

Moderate‐quality evidence suggests a risk reduction of 16% with fibrate therapy for the combined outcome of death due to cardiovascular disease, heart attack, or stroke. In absolute terms, the risk for this combined outcome in patients with cardiovascular risk factors but without established cardiovascular disease was on average reduced from 5.0% to 4.3% over five years. Moderate‐quality evidence also suggests a risk reduction for fatal and non‐fatal heart attacks with fibrates, but there is low‐quality evidence for no risk reduction for overall mortality or death from non‐CVD with fibrates. Very‐low quality evidence suggests that there is no increased risk for adverse effects with fibrate treatment. The reporting of adverse effects by identified trials was very limited. Data on quality of life were not available from any included study. Trials that evaluated fibrates in the background of statin treatment showed no benefits in preventing cardiovascular events.

Summary of findings

Summary of findings for the main comparison. Summary of findings table.

Fibrates compared to placebo for patients in primary prevention of cardiovascular disease events
Patient or population: patients in primary prevention of cardiovascular disease events  Settings: outpatients  Intervention: fibrates  Comparison: placebo
Outcomes	*Illustrative comparative risks (95% CI)**		Relative effect  (95% CI)	No of Participants  (studies)	Quality of the evidence  (GRADE)	Comments
	Assumed risk	Corresponding risk
	Placebo	Fibrates
Combined CVD death, non‐fatal MI, or non‐fatal stroke   Follow‐up: 2 years to 5 years	62 per 1000	52 per 1000   (49 to 59)	RR 0.84   (0.74 to 0.96)	16,135  (6 studies)	⊕⊕⊕⊝  moderate¹
Combined coronary heart disease death or non‐fatal MI   Follow‐up: 2 years to 5 years	46 per 1000	36 per 1000   (31 to 42)	RR 0.79   (0.68 to 0.92)	16,135  (6 studies)	⊕⊕⊕⊝  moderate¹
Overall mortality   Follow‐up: 1 year to 5 years	36 per 1000	36 per 1000   (30 to 45)	RR 1.01   (0.81 to 1.26)	8471  (5 studies)	⊕⊕⊝⊝  low^{1, 2}
Non‐CVD mortality   Follow‐up: 1 year to 5 years	22 per 1000	22 per 1000   (16 to 29)	RR 1.01   (0.76 to 1.35)	8471  (5 studies)	⊕⊕⊝⊝  low^{1, 2}
Discontinuation of therapy due to adverse effects   Follow‐up: 2 years to 5 years	261 per 1000	360 per 1000   (185 to 699)	RR 1.38   (0.71 to 2.68)	4805  (3 studies)	⊕⊝⊝⊝  very low^{1, 3, 4}
The basis for the assumed risk* (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).  CI: confidence interval; CVD: cardiovascular death; MI: myocardial infarction; RR: risk ratio.
GRADE Working Group grades of evidence  High quality: Further research is very unlikely to change our confidence in the estimate of effect.  Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.  Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.  Very low quality: We are very uncertain about the estimate.

Open in a new tab

¹ High risk of bias for incomplete outcome data in FIRST 2014 and SENDCAP 1998 (attrition bias).  ² The estimate is imprecise because the 95% CI includes clinically‐relevant harm as well as clinically‐relevant benefit; in addition there is some evidence for publication bias.  ³ There is considerable inconsistency across studies (I² of 74%) without plausible explanation.  ⁴ The estimate is imprecise, because the 95% CI includes clinically‐important benefit as well as clinically‐important harm.

Background

Description of the condition

Cardiovascular disease (CVD) is the most common cause of death, illness, disability, and reduced quality of life in industrialised countries (Thom 2006; WHO 2014). With the increasing incidence of obesity, metabolic syndrome, and type 2 diabetes mellitus, the disease burden of CVD may escalate further in the years to come (Shaw 2010; WHO 2015). One of the major risk factors for CVD is elevated low‐density lipoprotein cholesterol (LDL‐C). In individuals with elevated LDL‐C, 3‐hydroxy‐3‐methylglutaryl coenzyme A (HMG‐CoA) reductase inhibitors (statins) are considered to be the first choice of pharmacological therapy, since they reduce CVD events and total mortality independently of baseline LDL‐C levels in primary and secondary prevention of CVD (4S 1994; Baigent 2005; Graham 2007; Heart Protection Study 2002; Hooper 2001; Lestra 2005; Mills 2010). The finding of elevated serum triglycerides and low levels of high‐density lipoprotein cholesterol (HDL‐C) can identify persons who are at increased risk for CVD, and who might benefit from further lipid‐modifying therapy (Graham 2007; Khoury 2011). Several causes underlie elevated triglycerides and low levels of HDL‐C in the general population: being overweight and obese, physical inactivity, cigarette smoking, excess alcohol intake, very high‐carbohydrate diets (> 60% of total energy), type 2 diabetes, chronic renal failure, nephrotic syndrome, certain drugs (corticosteroids, protease inhibitors for HIV, beta‐adrenergic blocking agents, estrogens), and genetic factors (Chait 1990; Stone 1994). When triglyceride levels are ≥ 200 mg/dL, the presence of increased quantities of atherogenic remnant lipoproteins can increase coronary heart disease risk substantially beyond that predicted by LDL‐C alone (Steiner 1987). In patients with type 2 diabetes mellitus, the United Kingdom Prospective Diabetes Study (UKPDS) identified low levels of HDL‐C (< 40 mg/dL in men, < 45 mg/dL in women) as the second most important coronary risk factor after LDL‐C (Turner 1998). However, it remains unclear if therapeutic interventions to raise levels of HDL‐C indeed translate into CVD risk reduction (Briel 2009).

Description of the intervention

Currently available fibrates in North America and/or Europe include: gemfibrozil, fenofibrate, fenofibric acid, bezafibrate, etofibrate, and ciprofibrate. Clofibrate is no longer in use due to excess mortality (United Nations 2003; WHO 1978). Fibrates are effective for modifying atherogenic dyslipidaemia; i.e. they prominently lower serum triglyceride levels, modestly raise levels of HDL‐C, and modestly lower levels of LDL‐cholesterol (Fazio 2004; Abourbih 2009; Jun 2010). Therefore, they appear to be particularly suitable for prevention of CVD events in people with low levels of HDL‐C and high triglycerides (Khoury 2011).

Potential side effects or adverse effects from fibrate therapy are increased venous thrombotic events, pancreatitis, reversible rise in creatinine (described with all fibrates except gemfibrozil), rise in homocysteine, and elevations in transaminases, gallbladder disease (since fibrates increase the cholesterol content of bile), and myositis/rhabdomyolysis, in particular for combinations of gemfibrozil with statins (Davidson 2007; Wierzbicki 2010).

How the intervention might work

Improvements in atherogenic dyslipidaemia mainly through reduction of serum triglycerides and modest increase of HDL‐C may result in a reduction of CVD events (Graham 2007; Lee 2011). The underlying mechanisms are probably fibrate induced changes in gene transcriptions controlling parts of lipid metabolism (Staels 1998). Fibrates seem to activate transcription factors, specifically peroxisome proliferator‐activated receptors (PPARs). PPAR activation leads to an increased production of major HDL apolipoproteins and lipoprotein lipase. In addition, fibrates enhance the catabolism of triglycerides through stimulation of cellular fatty acid uptake and conversion to acyl‐CoA derivatives.

Why it is important to do this review

Evidence from systematic reviews that fibrates reduce overall mortality or CVD mortality is lacking, although fibrates reduced non‐fatal coronary events in several previously published meta‐analyses (Abourbih 2009; Jun 2010; Keene 2014). In the Jun 2010 meta‐analysis that included 18 trials with a total of 45,058 participants, fibrate therapy was associated with a 10% relative risk reduction for major CVD events and a 13% relative risk reduction for coronary events, but was not associated with a risk reduction of stroke. There was no effect of fibrate therapy on the risk of overall mortality or CVD mortality. None of the systematic reviews and meta‐analyses reported separate outcomes for the effect of fibrates in primary or secondary prevention.

The main evidence for clinical benefit comes from placebo controlled trials with older fibrates such as gemfibrozil (for which some safety concerns were raised primarily when used in combination with statin therapy) and clofibrate (which is no longer available due to safety concerns) (Graham 2004; Jones 2005; Khoury 2011; Rubins 1999). The large Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial failed to show a reduction in CVD events when patients with type 2 diabetes were treated with fenofibrate plus statins rather than with statins alone (Ginsberg 2010). In the Keene 2014 meta‐analysis, fibrate therapy reduced the risk for non‐fatal myocardial infarction, but this effect was only statistically significant in the trials without statin treatment and not in those with statin treatment. The difference between these subgroups was, however, not significant. Similarly, the Ip 2015 meta‐analysis showed there was not more clinical benefit from statin‐fibrate combination therapy compared with statin monotherapy. Therefore, there remains uncertainty as to whether fibrates alone, or as add‐on therapy to statins, are effective in reducing CVD events in people at increased risk of developing CVD.

The controversy about the benefit of fibrates is also to some extent reflected in the results of a recent population level, observational cohort study using Intercontinental Marketing Services (IMS) Health data of patients prescribed fibrates in the USA and Canada (Jackevicius 2011). In this study, authors observed an increase in prescriptions for fibrates (particularly fenofibrate) in the USA during the past decade, while prescriptions for fibrates in Canada remained stable.

Limited healthcare resources demand an unbiased appreciation of the available evidence about the benefit and harm of fibrate therapy. In the USA, annual sales of fibrates total more than USD 1 billion (Bloomberg 2011). Since current evidence about the effectiveness of fibrate therapy in the primary prevention of CVD is missing, the results of this Cochrane Review and meta‐analysis will serve as a basis for decision‐making in health care.

In contrast to another Cochrane Review that focuses on the secondary prevention of CVD and stroke (Wang 2015), and a currently withdrawn Cochrane Review protocol on the combined therapy of statins and fibrates for individuals with dyslipidaemia (Ciapponi 2014), our Cochrane Review and meta‐analysis focuses on the effects of fibrates as monotherapy or as add‐on therapy to other lipid‐modifying drugs for the primary prevention of CVD morbidity and mortality in patients at risk for cardiovascular events.

Objectives

This Cochrane Review and meta‐analysis aims to evaluate the clinical benefits and harms of fibrates versus placebo or usual care or fibrates plus other lipid‐modifying drugs versus other lipid‐modifying drugs alone for the primary prevention of cardiovascular disease (CVD) morbidity and mortality.

Methods

Criteria for considering studies for this review

Types of studies

We included published and unpublished randomised controlled trials (RCTs) that documented relevant clinical outcome data and had a length of treatment of at least six months. In order to be included as a primary prevention trial, fewer than 10% of participants were required to have established cardiovascular disease (CVD).

Types of participants

Adults (i.e. individuals older than 18 years of age) at increased risk of CVD events (e.g. individuals with type 2 diabetes mellitus, or dyslipidaemia, or both) but not established CVD.

Types of interventions

Fibrate therapy versus placebo or usual care (combination therapy of a fibrate plus statins versus statins alone was considered in a sensitivity analysis).

We excluded clofibrate as an intervention, because it is no longer in use. In 1979, clofibrate was withdrawn in some countries and its approval indications were severely restricted in many others due to adverse effects (United Nations 2003; WHO 1978).

Types of outcome measures

Outcome measures were patient‐relevant clinical outcomes and adverse effects of fibrate therapy.

Primary outcomes

Combined outcome of major CVD events (CVD death, non‐fatal myocardial infarction, or non‐fatal stroke).

Secondary outcomes

Overall mortality.
Combined coronary heart disease death (mortality) or non‐fatal myocardial infarction as a composite endpoint.
Non‐CVD mortality.
Diabetic retinopathy (progression of pre‐existing diabetic retinopathy, development of new diabetic retinopathy according to individual trial definitions).
Incidence of albuminuria (incidence of macroalbuminuria, incidence of micro‐ or macroalbuminuria).
Patient perceived quality of life.

In addition, we investigated adverse effects of fibrate therapy (especially increases in serum creatinine or liver values (transaminases), myositis/rhabdomyolysis, pancreatitis, venous thrombotic events, gallbladder diseases, and the number of individuals discontinuing therapy due to adverse effects). Because adverse effects were often not specified as outcomes in included trials, we used discontinuation of therapy due to adverse effects as a proxy. We defined the presence or absence of increases in serum creatinine and liver values according to individual trial definitions.

Search methods for identification of studies

Electronic searches

We searched the Cochrane Central Register of Controlled Trials (CENTRAL; 2016, Issue 2) in the Cochrane Library, MEDLINE (Ovid, 1946 to May week 2, 2016), Embase Classic and Embase (Ovid, 1947 to 2016 May 18), CINAHL (EBSCO, 1937 to 19 May 2016), Conference Proceedings Citation Index‐Science (CPCI‐S), and Science Citation Index Expanded (SCI‐EXPANDED) on Web of Science (Thomson Reuters, 1990/1900 to 19 May 2016) on 19 May 2016.

We used the Cochrane sensitivity‐maximising RCT filter for MEDLINE and terms as recommended in the Cochrane Handbook for Systematic Reviews of Interventions for Embase (Lefebvre 2011). We did not impose any language restrictions. The search strategy that was developed to search MEDLINE (Ovid), together with the Cochrane Heart Group, was adapted where necessary to search the other databases listed. The search strategies can be found in Appendix 1.

We searched the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP)) (www.apps.who.int/trialsearch) and individual trial registers for protocols of ongoing trials on 3 August 2016 (www.clinicaltrials.gov; www.controlled‐trials.com; www.isrctn.org). For any potentially relevant ongoing trials that were identified, we contacted authors by email or fax to get more precise information on trial methods, recruitment status, and outcome data in case of unpublished trial protocols.

Searching other resources

We searched for reference lists of identified articles, editorials, and reviews on the topic for further eligible studies.

Data collection and analysis

Selection of studies

TJ, AJN, IFG, and MB, working in teams of two, independently reviewed potentially eligible titles and abstracts. If either review author believed the study could have been eligible, we obtained the full report. After obtaining full reports of the studies (either in full peer‐reviewed publication or press article), TJ and AJN independently assessed eligibility from full‐text papers. Discrepancies were resolved by reviewers’ consensus or, if needed, by third party arbitration (MB). Primary prevention trials were defined as trials including fewer than 10% patients with already established cardiovascular disease (CVD).

Data extraction and management

Working in teams of two, TJ, AJN, and SS used pre‐piloted forms to independently extract all relevant data on baseline characteristics of trials, patient populations, outcomes, and additional data provided by original trial investigators. Any disagreements between review authors were resolved by discussion and consensus.

For the ACCORD 2010 trial, we obtained the full dataset from the National Institutes of Health, USA. In a first step, we reproduced the reported results; in a second step, we stratified the analyses by primary and secondary prevention and included the results for primary prevention in the meta‐analysis. We recalculated the following outcomes using complete cases only: combined outcome of CVD death, non‐fatal myocardial infarction, or non‐fatal stroke; combined outcome of fatal myocardial infarction (or coronary heart disease death) or non‐fatal myocardial infarction; overall mortality; non‐CVD mortality; micro‐ and macroalbuminuria; and elevated serum creatinine.

Assessment of risk of bias in included studies

Working in teams of two, TJ, AJN, and MB independently assessed the risk of bias of each included trial with respect to concealment of treatment allocation, random sequence generation, blinding of patients, blinding of caregivers, blinding of assessors of clinical outcomes, completeness of follow‐up, and selective reporting of outcomes (Higgins 2011). Any disagreements were resolved by discussion and consensus or by third party arbitration, if needed. We explored the influence of individual 'Risk of bias' criteria on summary estimates in sensitivity analyses.

Measures of treatment effect

We used risk ratios (RRs) and accompanying 95% confidence intervals (CIs).

Unit of analysis issues

Except for the ACCORD 2010 trial we used aggregated data for analysis.

Dealing with missing data

In case of unpublished or incomplete data, or published data that could not be directly integrated into our analysis, we contacted the investigators of the original trials and asked for the relevant information.

Assessment of heterogeneity

We tested for heterogeneity using the Cochrane Q‐test and used the I² statistic to measure inconsistency of treatment effects across different lipid‐lowering interventions (Higgins 2002; Higgins 2003; Higgins 2011).

Assessment of reporting biases

We checked for outcome reporting bias by comparing reported outcomes to outcomes mentioned in corresponding trial protocols, outcomes mentioned in trial registries, or outcomes mentioned in the Methods section of publications. We investigated the presence of publication bias by means of funnel plots (Sterne 2001).

Data synthesis

We pooled treatment effects across studies and calculated a weighted average RR for all outcomes in the treatment and control groups by using a fixed‐effect model (default) if there was no heterogeneity among pooled studies. In cases where we detected heterogeneity, we applied a random‐effects model, as described in Section 9.5.4 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). The random‐effects model and the fixed‐effect model typically give identical results when there is no heterogeneity among pooled studies. All analyses were conducted using Review Manager 5.3 (RevMan5 2014).

We created a 'Summary of findings' table using the following outcomes: combined CVD death, non‐fatal myocardial infarction, or non‐fatal stroke; combined coronary heart disease death or non‐fatal myocardial infarction; overall mortality; non‐CVD mortality; and discontinuation of therapy due to adverse effects. We used the five Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) considerations (study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of a body of evidence as it relates to the studies which contribute data to the meta‐analyses for the prespecified outcomes (GRADE Handbook 2013). We used methods and recommendations described in Section 8.5 and Chapter 12 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011), using GRADEpro software (GRADEpro 2015). We justified all decisions to down‐ or upgrade the quality of studies using footnotes and we made comments to aid reader's understanding of the review where necessary.

Subgroup analysis and investigation of heterogeneity

Due to the small number of included trials we did not conduct any subgroup analyses.

Sensitivity analysis

We examined treatment effects according to risk of bias (reported concealed treatment allocation versus no reported concealed treatment allocation, reported blinding of patients and caregivers versus no reported blinding of patients and caregivers, and reported blinded outcome assessment versus no reported blinded outcome assessment ). In addition, we conducted sensitivity analyses with respect to fibrate monotherapy versus trials evaluating fibrates in addition to other lipid‐modifying drugs (e.g. statins), and prompted by editors of the Heart Group, we conducted sensitivity analyses with respect to follow‐up duration (4 years and more versus less than 4 years).

Results

Description of studies

Results of the search

Our literature search for articles, editorials, and reviews on the topic yielded 7735 deduplicated hits (Figure 1). After exclusion of articles based on titles and abstracts, we reviewed 50 publications in full‐text. We identified a total of eight primary prevention trials which fulfilled our inclusion criteria, but excluded two trials that used clofibrate, which is no longer in use (DIS 1991; WHO 1978). Thus, we were able to analyse data from six primary prevention trials in total (ACCORD 2010; Emmerich 2009; FIELD 2005; FIRST 2014; Frick 1987; SENDCAP 1998). We were able to obtain separate primary prevention data from three of these trials that included patients with established cardiovascular disease (CVD) (ACCORD 2010; Emmerich 2009; FIRST 2014). For FIELD 2005, we could only retrieve primary prevention data for our combined primary outcome (CVD death, non‐fatal myocardial infarction, or non‐fatal stroke). For ACCORD 2010 we were able to obtain the complete dataset from the National Institutes of Health, USA, which enabled us to separate primary prevention from secondary prevention participants in the analysis for all outcomes except for adverse effects. Primary prevention trials included a total of 16,135 patients (8087 in the intervention group and 8048 in the control group).

Included studies

Design

All included trials were randomised controlled multicentre trials (undertaken in the USA, Canada, Germany, Finland, France, Sweden, Australia, New Zealand, UK). Individual follow‐up‐periods of included patients varied from two to five years (Table 2). Five of the six trials compared outcomes by intention‐to‐treat‐analysis (ACCORD 2010; Emmerich 2009; FIELD 2005; Frick 1987; SENDCAP 1998), while FIRST 2014 did not report the method of outcome analysis (Characteristics of included studies).

1. Baseline characteristics of primary prevention trials.

Trial name	Patients total, n	Follow‐up, years	Average age, years (SD)	Men (%)	Diabetes   mellitus type 2 (%)	Arterial hypertension (%)	Smokers (%)	Establised cardiovascular disease at baseline (%)
ACCORD 2010	3502	4.7	62.3 (NA)	NA	100	NA	NA	0
Emmerich 2009	244	2	58.6 (NA)	28.7	100	NA	NA	0
FIELD 2005	7664	5	61.5 (6.8)	61.5	100	53.3	8.9	0
FIRST 2014	480	2	NA	NA	47.5	75	NA	0
Frick 1987	4081	5	47.3 (4.6)	100	2.6	14	36	0
SENDCAP 1998	164	3	51 (8)	71	100	16	18	0

Open in a new tab

SD: standard deviation

NA: not available

Sample sizes and description of participants

The number of included participants in individual trials varied from 164 to 7664 (Table 2; Characteristics of included studies). Three trials included mostly men (FIELD 2005; Frick 1987; SENDCAP 1998). The mean ages of study participants varied in primary prevention trials between 47.3 and 62.3 years. Four trials included individuals with type 2 diabetes only (ACCORD 2010; Emmerich 2009; FIELD 2005; SENDCAP 1998); of the patients included in FIRST 2014, 47.5% had diabetes type 2 and in Frick 1987, 2.6%. Further information on included trials are summarised in the section Characteristics of included studies.

Interventions

Three trials investigated the effect of fenofibrate (ACCORD 2010, 160 mg/d; FIELD 2005, 200 mg/d; FIRST 2014, 135 mg/d). One trial evaluated the effects of etofibrate (Emmerich 2009, 1000 mg/d), one the effects of bezafibrate (SENDCAP 1998, 400 mg/d), and another the effect of gemfibrozil (Frick 1987, 1200 mg/d). Two trials used fibrate therapy on top of a statin therapy (FIRST 2014, atorvastatin 10 mg/d to 40 mg/d in both study groups; ACCORD 2010, simvastatin 20 mg/d in both study groups), whereby participants in the intervention groups additionally received fibrates while participants in the control group additionally received placebo in these two trials. In FIRST 2014, study investigators had the option of adding ezetimibe 10 mg to the treatment regimen in both study groups, if low‐density lipoprotein cholesterol (LDL‐C) was still ≥ 3.34 mmol/l (130 mg/dl) despite application of the maximum atorvastatin dose (40 mg). All other trials used fibrates in the intervention groups as monotherapy and placebo in the control groups. Frick 1987 recommended a cholesterol‐lowering diet, increase in physical activity, and a reduction in smoking to all patients (both study groups). Further information on included trials are summarised in Characteristics of included studies.

Outcomes

The primary outcome of FIRST 2014 was the rate of change of the left and right mean posterior wall (carotid intima‐media thickness) measured by ultrasound. ACCORD 2010 defined as a primary outcome the composite endpoint until the first occurrence of non‐fatal myocardial infarction, non‐fatal stroke, or death from cardiovascular causes, while the primary outcome of the ACCORD eye substudy was the rate of progression of diabetic retinopathy by at least three steps on the Early Treatment of Diabetic Retinopathy Study (ETDRS) severity scale or development of proliferative diabetic retinopathy necessitating photocoagulation therapy or vitrectomy. Emmerich 2009 defined the improvement or worsening of vision assessed by Consenus Clinical Global Impression of Change (CCGI‐C) as a primary outcome (comparison of fundus results at study entry and after one year of treatment). The primary outcomes of FIELD 2005 were coronary events (coronary heart disease death or non‐fatal myocardial infarction), while the FIELD 2007 eye subtrial investigated primarily the progression of diabetic retinopathy of at least two‐steps in the ETDRS‐severity scale. The primary endpoint of SENDCAP 1998 was the change in the carotid artery intima‐media thickness. The primary outcomes of Frick 1987 were fatal and non‐fatal myocardial infarctions and cardiac death. Further information on included trials are summarised in Characteristics of included studies.

Overview of data availability for our primary and secondary outcomes

Concering our combined primary outcome of CVD death, non‐fatal myocardial infarction, or non‐fatal stroke, and for the combined outcome of coronary heart disease mortality or non‐fatal myocardial infarction we were able to pool data from all six trials. For the secondary outcomes of overall mortality and non‐CVD mortality we were able to pool data from all trials with the exception of FIELD 2005. For the outcome of new diabetic retinopathy or the progression of pre‐existing diabetic retinopathy, we were not able to obtain data from primary prevention populations: ACCORD 2010 (Accord eye subtrial), FIELD 2005 (Field eye subtrial), and Emmerich 2009 reported only on patients with and without established CVD (mixed prevention). ACCORD 2010 was the only trial providing data on progression to albuminuria. None of the included trials reported on patient perceived quality of life.

Overview of data availability for adverse effects

Two of six trials reported increases in serum creatinine and transaminases levels (Emmerich 2009; FIRST 2014). None of the six trials reported the occurrence of myositis/rhabdomyolysis. FIRST 2014 reported the occurrence of pancreatitis and FIRST 2014 and Emmerich 2009 reported venous thrombotic events. Three trials reported gallbladder diseases and individuals discontinuing therapy due to adverse effects (Emmerich 2009; FIRST 2014; Frick 1987). Primary prevention data on adverse effects in ACCORD 2010 were not available.

Excluded studies

Reasons for exclusion are presented in the Characteristics of excluded studies table and in Figure 1. We excluded two trials with clofibrate because the drug is no longer in use (DIS 1991; WHO 1978). We excluded 18 trials because more than 10% of trial participants had established CVD at baseline (Acheson 1972; BECAIT 1996; Begg 1971; Benderly 2009; BIP 2000; CDP fibrates 1975; Cullen 1974; DAIS 2001; Frick 1993; Haim 2007; Harrold 1969; LEADER 2002; Liamina 2011; LOCAT 1997; Newcastle 1971; SSP 1971; VA Cooperative 1973; VAHIT 1999). Eleven trials did not report relevant outcome data to be included in this meta‐analysis and review (Chew 2008; Domagala 2013; Farnier 2012; Kawano 2013; Pioneer 2013; Pruski 2009; Rajamani 2011; Rosenson 2007; Scott 2010; Tkacheva 2010; Valensi 2010). We excluded five trials because of the lack of a placebo/usual care group (Bays 2008, fibrate + statin versus fibrate only; Derosa 2009, fibrate only versus statin only versus fibrate and statin in combination; Dohmen 2013, fenofibrate versus bezafibrate; Lee 2012, fibrate and statin versus statin only); Yang 2013, multiple interventions). Two trials were retrospective trial designs and not RCTs (Klempfner 2014; Webb 2009). In six trials the follow‐up time was too short (< 6 months) to be included (Aguilar‐Salinas 2001; Betterridge 1994; Chen 2013; Khan 2014; Krysiak 2010; Massin 2014).

Risk of bias in included studies

For details see Characteristics of included studies, Table 1, Figure 2, and Figure 3.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Two trials did not report the method of random sequence generation (Emmerich 2009; Frick 1987), while four trials did (ACCORD 2010; FIELD 2005; FIRST 2014; SENDCAP 1998); therefore we judged the risk of bias in this domain to be low. All but one trial reported concealment of treatment allocation (Emmerich 2009), and therefore we rated the risk of bias for this domain as low too.

Blinding

Study participants and personnel were blinded to treatment groups in all six trials; therefore we judged the risk of performance bias as low.

Blinding of outcome assessment (data collectors and data adjudicators) was reported in three trials (ACCORD 2010; FIELD 2005; Frick 1987). SENDCAP 1998 did not report whether data collectors were blinded, but reported that data adjudicators were blinded. In two trials, blinding of outcome assessment was not reported (Emmerich 2009; FIRST 2014). Therefore, we rated the risk for detection bias as unclear for all outcomes other than mortality.

Incomplete outcome data

We judged two trials to be at high risk of attrition bias. FIRST 2014 had a loss to follow‐up of 31% of patients without reporting the reasons, but numbers of patients lost to follow‐up between the intervention and control groups were similar. SENDCAP 1998 lost 22% of patients to follow‐up without reporting reasons. The remaining trials had negligible losses to follow‐up. Therefore, we rated the risk for attrition bias in these remaining trials as low.

Selective reporting

We searched for study protocols on PubMed (www.pubmed), ClinicalTrials.gov (www.clinicaltrials.gov), and the WHO International Clinical Trials Registry Platform (ICTRP)) (www.apps.who.int/trialsearch). We found study protocols for three trials (ACCORD 2010; FIELD 2005; FIRST 2014). All reports on randomisation procedures, blinding of patients and caregivers, blinding of outcome assessment, inclusion and exclusion criteria, and primary and secondary endpoints in trial protocols were in congruence with the corresponding information reported in these individual trials. These trials reported alI clinical outcomes and all adverse effects. For the remaining three trials, we did not find study protocols (Emmerich 2009; Frick 1987; SENDCAP 1998). SENDCAP 1998 did not report any adverse effects. Emmerich 2009 did not report the number of patients discontinuing therapy due to adverse effects.

Publication bias

The visual inspection of the funnel plot for the combined primary outcome of CVD death, non‐fatal myocardial infarction, or non‐fatal stroke did not provide compelling evidence for a potential publication bias (Figure 4). However, the funnel plots for overall mortality (Figure 5), and non‐CVD mortality (Figure 6), showed an asymmetry in a way that the risk for harm from fibrate therapy could be underestimated.

Funnel plot of comparison: 1 Outcomes/adverse effects, outcome: 1.1 Combined CVD death, non‐fatal MI, or non‐fatal stroke.

Funnel plot of comparison: 1 Outcomes/adverse effects primary prevention trials, outcome: 1.3 Overall mortality.

Funnel plot of comparison: 1 Outcomes/adverse effects primary prevention trials, outcome: 1.4 Non‐CVD mortality.

Other potential sources of bias

Another potential source of bias is the involvement of pharmaceutical companies in the conduct, analysis, or publication of clinical trials. Design, study conduct, analysis, and financial support of FIRST 2014 were provided by the pharmaceutical company, AbbVie. AbbVie participated in the interpretation of data, review, and approval of the article. The trial by Emmerich 2009 was sponsored and conducted by Merz‐Pharmaceuticals. SENDCAP 1998 was supported by grants from Boehringer Mannheim, but this company seemed to have had no influence on the performance and data analysis of the trial. In ACCORD 2010 some of the investigators received fees from different pharmaceutical companies, but the study was co‐ordinated independently. The two remaining trials did not seem to have conflicted interests (FIELD 2005; Frick 1987).

Effects of interventions

See: Table 1

Primary outcome

Combined outcome of major cardiovascular disease (CVD) events

Six trials with a total of 16,135 participants reported data for the combined outcome of CVD death, non‐fatal myocardial infarction, or non‐fatal stroke (ACCORD 2010; Emmerich 2009; FIELD 2005; FIRST 2014; Frick 1987; SENDCAP 1998). Patients in fibrate groups had a reduced risk for the primary combined outcome (risk ratio (RR) 0.84, 95% confidence interval (CI) 0.74 to 0.96; I² = 0%; moderate quality of evidence) compared with placebo (Analysis 1.1.1).

1.1 — Comparison 1 Fibrates versus placebo, Outcome 1 Combined CVD death, non‐fatal MI, or non‐fatal stroke.

Secondary outcomes

Overall mortality

We found no effect of fibrate treatment on overall mortality (RR 1.01, 95% CI 0.81 to 1.26; participants = 8471; studies = 5; I² = 0%; low‐quality evidence; Analysis 1.2.1).

1.2 — Comparison 1 Fibrates versus placebo, Outcome 2 Overall mortality.

Combined coronary heart disease death (mortality) or non‐fatal myocardial infarction as a composite endpoint

We found a decreased risk for coronary heart disease death or non‐fatal myocardial infarction with fibrate therapy compared to placebo (RR 0.79, 95% CI 0.68 to 0.92; participants = 16,135; studies = 6; I² = 0%; moderate‐quality evidence; Analysis 1.3).

1.3 — Comparison 1 Fibrates versus placebo, Outcome 3 Combined coronary heart disease death or non‐fatal MI.

Non‐CVD mortality

We found no effect of fibrate treatment on non‐CVD mortality (RR 1.00, 95% CI 0.76 to 1.33; participants = 8471; studies = 5; I² = 0%; low‐quality evidence; Analysis 1.4).

1.4 — Comparison 1 Fibrates versus placebo, Outcome 4 Non‐CVD mortality.

Diabetic retinopathy (progression of pre‐existing diabetic retinopathy, development of new diabetic retinopathy according to individual trial definitions)

No primary prevention trial reported data on diabetic retinopathy (development of new diabetic retinopathy or progression of pre‐existing diabetic retinopathy). However, using data from mixed prevention populations (patients with and without established CVD) of three trials (ACCORD 2010; Emmerich 2009; FIELD 2005), the risk for progression of pre‐existing diabetic retinopathy was decreased in the fibrate group compared to the placebo group (RR 0.67, 95% CI 0.56 to 0.81; participants = 2901; studies = 3; I² = 0%; Analysis 1.5). Using data from two trials reporting results on the development of new diabetic retinopathy (Emmerich 2009; FIELD 2005), we found a RR of 0.60 (95% CI 0.17 to 2.09; participants = 1308; studies = 2; I² = 73%; Analysis 1.6).

1.5 — Comparison 1 Fibrates versus placebo, Outcome 5 Progression of pre‐existing diabetic retinopathy.

1.6 — Comparison 1 Fibrates versus placebo, Outcome 6 Development of new diabetic retinopathy.

Incidence of albuminuria (incidence of macroalbuminuria, incidence of micro‐ or macroalbuminuria)

The incidence of albuminuria was only reported in one trial (ACCORD 2010). For the incidence of macroalbuminuria (>= 300 ratio mg albumine / g creatinine) we found a RR of 0.85 (95% CI 0.67 to 1.09; participants = 3246) for fibrate treatment compared with placebo (Analysis 1.7). For the incidence of micro‐ or macroalbuminuria (>= 30 ratio mg albumine / g creatinine) we found a RR of 0.92 (95% CI 0.83 to 1.01; participants = 3246) for fibrate treatment compared with placebo (Analysis 1.8).

1.7 — Comparison 1 Fibrates versus placebo, Outcome 7 Incidence of macroalbuminuria (>= 300 ratio mg albumine/g creatinine).

1.8 — Comparison 1 Fibrates versus placebo, Outcome 8 Incidence of micro‐ or macroalbuminuria (>= 30 ratio mg albumine/g creatinine).

Patient perceived quality of life

No trial reported on patient perceived quality of life.

Adverse effects

We found an increased risk for raised serum creatinine levels with fibrate treatment compared with placebo (RR 1.88, 95% CI 1.65 to 2.15; participants = 4173; studies = 3; I² = 0%; Analysis 1.9). There were only five patients with raised serum transaminase levels (4 in fibrate groups versus 1 in placebo groups) in two trials (Emmerich 2009; FIRST 2014), resulting in imprecise evidence (RR 2.94, 95% CI 0.47 to 18.21; participants = 724; I² = 0%; Analysis 1.10). No primary prevention trial reported on the risk for myositis/rhabdomyolysis. There was only one patient with pancreatitis (fibrate group) in one trial reporting on pancreatitis (FIRST 2014; RR 2.74, 95% CI 0.11 to 66.88; participants = 480; Analysis 1.11). There were only few venous thrombotic events (3 in fibrate groups, 2 in placebo groups) resulting in imprecise evidence (RR 1.46, 95% CI 0.24 to 8.78; participants = 724; studies = 2; I² = 0%; Analysis 1.12). Three trials reported on the incidence of gallbladder disease (Emmerich 2009; FIRST 2014; Frick 1987), with a summary RR of 1.33 (95% CI 0.68 to 2.62; participants = 4805; studies = 3; I² = 0%; Analysis 1.13). Because adverse effects were mostly not specified as an outcome in included trials, we used discontinuation of therapy due to adverse or side effects as a proxy; there were 712 patients (29.4%) who discontinued therapy due to adverse or side effects out of 2420 patients in fibrate groups compared with 622 (26.1%) out of 2385 patients in placebo groups (RR 1.38, 95% CI 0.71 to 2.68; participants = 4805; studies = 3; I² = 74%; very low‐quality evidence; Analysis 1.14).

1.9 — Comparison 1 Fibrates versus placebo, Outcome 9 Raised serum creatinine.

1.10 — Comparison 1 Fibrates versus placebo, Outcome 10 Raised serum transaminases.

1.11 — Comparison 1 Fibrates versus placebo, Outcome 11 Pancreatitis.

1.12 — Comparison 1 Fibrates versus placebo, Outcome 12 Venous thrombotic events.

1.13 — Comparison 1 Fibrates versus placebo, Outcome 13 Gallbladder disease.

1.14 — Comparison 1 Fibrates versus placebo, Outcome 14 Discontinuation of therapy due to side effects.

Sensitivity analyses

We performed sensitivity analyses for the primary outcome of combined CVD death, non‐fatal myocardial infarction, or non‐fatal stroke, and for overall mortality with respect to reported allocation concealment, reported blinding of patients and caregivers, reported blinded outcome assessment, trials using fibrate monotherapy versus trials using fibrates in addition to statins, and trials with at least four years of follow‐up).

Concealed treatment allocation versus unclear concealed treatment allocation

There were four trials that reported concealed treatment allocation (ACCORD 2010; FIELD 2005; FIRST 2014; SENDCAP 1998), with a RR for the combined primary outcome (CVD death, non‐fatal myocardial infarction, or non‐fatal stroke) of 0.88 (95% CI 0.76 to 1.01; participants = 11810; studies = 4; I² = 0%; Analyses 1.1.1 and 1.1.2). The sensitivity analysis on overall mortality focusing on trials that reported concealed allocation yielded a RR of 1.00 (95% CI 0.77 to 1.30; participants = 4146; studies = 3; I² = 0%) between patients treated with fibrates compared to patients on placebo (Analyses 1.2.1 and 1.2.2).

Blinding of patients and caregivers versus unclear blinding of patients and caregivers

All six included trials reported blinding of patients and caregivers, making this sensitivity analysis obsolete for the combined primary outcome and overall mortality (Analyses 1.1.3, 1.1.4, 1.2.3, and 1.2.4).

Blinded outcome assessment versus unclear blinded outcome assessment

There were four trials that reported blinding of outcome assessors (ACCORD 2010; FIELD 2005; Frick 1987; SENDCAP 1998), with a RR for the combined primary outcome (CVD death, non‐fatal myocardial infarction, or non‐fatal stroke) of 0.85 (95% CI 0.72 to 0.99; participants = 15,411; I² = 0%) when comparing fibrate groups with placebo groups (Analyses 1.1.5 and 1.1.6). The sensitivity analysis on overall mortality focusing on trials which reported blinded outcome assessment yielded a RR of 1.02 (95% CI 0.82 to 1.27; participants = 7747; studies = 3; I² = 0%; Analysis 1.2.5 and 1.2.6).

Trials using fibrate monotherapy versus trials using fibrates in addition to statins

Four included trials used fibrates as monotherapy (Emmerich 2009; FIELD 2005; Frick 1987; SENDCAP 1998), with a RR for the combined primary outcome (CVD death, non‐fatal myocardial infarction, or non‐fatal stroke) of 0.79 (95% CI 0.68 to 0.92; participants = 12,153; studies = 4; I² = 0%; Analysis 1.1.7). Two trials used fibrates in addition to statins (ACCORD 2010; FIRST 2014), with a RR for the combined primary outcome of 0.99 (95% CI 0.78 to 1.26; participants = 3982; studies = 2; I² = 0%; Analysis 1.1.8). The corresponding analyses for overall mortality yielded a RR of 1.02 (95% CI 0.68 to 1.52; participants = 4489; studies = 3; I² = 0% for trials using fibrate monotherapy (Analysis 1.2.7), and a RR of 1.01 (95% CI 0.78 to 1.31; participants = 3982; studies = 2; I² = 0%) in trials using fibrates in addition to statins (Analysis 1.2.8). There was some evidence for a difference in treatment effects for the primary combined outcome between groups of trials with fibrate monotherapy and groups of trials using fibrates in addition to statins (I² between groups of 61%), suggesting no relevant risk reduction with fibrates when using them together with statins.

Trials with at least four years of follow‐up versus trials with less than four years of follow‐up

There were three trials with at least four years of follow‐up (ACCORD 2010; FIELD 2005; Frick 1987), with a RR for the combined primary outcome (CVD death, non‐fatal myocardial infarction, or non‐fatal stroke) of 0.85 (95% CI 0.72 to 1.00; participants = 15,237; studies = 3; I² = 36%; Analyses 1.1.9 and 1.1.10). There were two trials with at least four years of follow‐up that reported on overall mortality among primary prevention patients with a RR of 1.03 (95% CI 0.82 to 1.28; participants = 7583; studies = 2; I² = 0%) between patients treated with fibrates compared to patients on placebo (Analyses 1.2.9 and 1.2.10).

Discussion

Summary of main results

In trials of primary prevention of cardiovascular disease (CVD) that compared fibrate therapy to placebo or usual care, this Cochrane Review and meta‐analysis found moderate‐quality evidence for a reduced risk for CVD events (CVD death, non‐fatal myocardial infarction, or non‐fatal strokes) of 16%, and for coronary events (coronary heart disease death or non‐fatal myocardial infarction) of 21%. These effects are modest (≦ 1%) on an absolute scale, taking into account a low baseline risk of 5% or lower over five years for CVD events. In terms of numbers needed to treat (NNTs), we calculated that, assuming a baseline risk of 6%, 112 individuals have to be treated over five years to prevent one CVD event. Furthermore, assuming a baseline risk of 4%, we calculated that 125 individuals have to be treated for five years to prevent one coronary event (coronary heart disease death or non‐fatal myocardial infarction). When fibrates were applied in addition to a statin (2 primary prevention trials) compared with statin therapy alone, the treatment effect tended towards no effect. Low‐quality evidence suggests that fibrates have no effect on overall mortality or non‐CVD mortality. There was very low‐quality evidence for no difference between fibrate treated patients and patients receiving placebo or usual care, with respect to discontinuation of therapy due to adverse effects.

Overall completeness and applicability of evidence

This Cochrane Review included all primary prevention trials of fibrate therapy. No primary prevention trial reported results on diabetic retinopathy or quality of life, and only one trial provided data on the development of albuminuria (ACCORD 2010). The mean age of trial participants ranged from 47.3 to 62.3 years. Thus, the results of this meta‐analysis cannot necessarily be generalised to the elderly population. The applicability of our results to female individuals is limited since a large majority of trial participants were male. In addition, the results of our study may not easily be generalisable to individuals without type 2 diabetes mellitus since the number of non‐diabetic participants was too small to conduct separate subgroup analyses. Finally, data in this review were primarily from trials without background statin therapy, which is now considered usual care. Only two more recent trials evaluated the effect of combined statin and fibrate therapy (ACCORD 2010; FIRST 2014), and did not find a reduction in the primary outcome of combined CVD death, non‐fatal myocardial infarction, or non‐fatal stroke compared to statin therapy alone (Analysis 2.4).

Quality of the evidence

We employed the GRADE approach (GRADE Handbook 2013) to interpret result findings and the GRADE profiler software (GRADEpro 2015) allowed us to import data from Review Manager 5 to create 'Summary of findings' tables. According to the GRADE approach, the quality of the evidence was moderate for the primary composite outcome of CVD death, non‐fatal myocardial infarction, or non‐fatal stroke and for the secondary composite outcome of coronary heart disease death or non‐fatal myocardial infarction due to an increased risk of attrition bias for SENDCAP 1998 and FIRST 2014 (Table 1). For overall mortality and non‐CVD mortality we judged the evidence to be of low‐quality, because of wide confidence intervals (imprecise estimates) and an increased risk of attrition bias for SENDCAP 1998 and FIRST 2014; in addition we found some evidence for potential publication bias (Figure 5; Figure 6). For discontinuation of therapy due to adverse effects, we rated the quality of evidence as very low due to substantial heterogeneity of results across trials that we cannot explain in stratified analyses, an increased risk of attrition bias for FIRST 2014, and imprecision (95% confidence interval (CI) of summary effect includes clinically‐important benefit as well as clinically‐important harm).

Potential biases in the review process

With the aim to identify and include all eligible randomised trials on the topic, we conducted a highly sensitive literature search, including trial registers and contacts with experts in the field for unpublished studies. Still, funnel plots for overall mortality and non‐CVD mortality indicate some asymmetry that could be due to publication bias (Figure 5; Figure 6). The present results may, therefore, underestimate the risk for harm from fibrate therapy.

Agreements and disagreements with other studies or reviews

Previously published systematic reviews did not report separate effects of fibrate therapy for primary prevention (Abourbih 2009; Jun 2010; Keene 2014), except for an older systematic review of ours (Studer 2005). Our results are consistent with two previous systematic reviews reporting the effects of fibrate therapy in a combined primary and secondary prevention population. Jun 2010 reported a 10% risk reduction for major CVD events and a 13% risk reduction for coronary events in individuals treated with fibrate therapy, but did not report any effects on overall mortality, CVD mortality, or non‐CVD mortality. They reported a decreased risk for the new development of diabetic retinopathy, including two trials in a pooled primary and secondary CVD prevention population (risk ratio (RR) 0.63, 95% CI 0.49 to 0.81; I² = 41.5%) (Emmerich 2009; FIELD 2005). Our analysis, including results from individuals from the ACCORD 2010 trial, confirms these findings.

Keene 2014 reported an odds ratio (OR) of 0.80 (95% CI 0.74 to 0.87) for non‐fatal myocardial infarction in individuals treated with fibrates, but no difference in overall mortality, coronary heart disease mortality, nor stroke. More recently, Wang 2015 reported a RR of 0.88 (95% CI 0.83 to 0.94) for a composite outcome of non‐fatal stroke, non‐fatal myocardial infarction, and vascular death in individuals treated with fibrate therapy for the secondary prevention of CVD based on a meta‐analysis of 13 trials, involving 16,112 participants. However, there was no difference in the primary composite outcome after excluding clofibrate trials. There was no difference in overall mortality, vascular death, strokes, nor adverse effects.

In conclusion, our meta‐analysis confirms results of previous meta‐analyses in mixed or secondary CVD prevention trials, reporting a moderate reduction in a composite vascular outcome, as well as for the primary prevention of CVD events.

Authors' conclusions

Implications for practice.

There is moderate‐quality evidence that currently used fibrates lower the risk of a combined outcome of cardiovascular (CVD) death, non‐fatal myocardial infarction, or non‐fatal stroke by 16%, and the risk of a combined outcome of coronary heart disease death or non‐fatal myocardial infarction by 21% in the primary prevention of CVD, without increasing or decreasing overall mortality or non‐CVD mortality (low‐quality evidence). Very low‐quality evidence suggested no difference in discontinuation of treatment due to adverse effects in the fibrate groups compared to placebo groups. Our sensitivity analysis stratifying studies by presence/absence of additional therapy with statins suggests that the benefits of fibrate therapy may no longer be relevant in the context of background therapy with statins (Analysis 2.4). Nevertheless, the beneficial effects of fibrates in terms of a decreased risk for major CVD events without background therapy with statins appear modest on an absolute scale (≦ 1%). There was insufficient data to analyse fibrate therapy for the prevention of diabetic retinopathy in primary prevention individuals.

The recent guidelines from the National Institute for Health and Care Excellence (NICE) in the UK did not recommend fibrates for primary nor secondary prevention of CVD (Rabar 2014). However, the International Atherosclerosis Society (IAS) states that if non‐high‐density lipoprotein cholesterol (non‐HDL‐C) and triglycerides remain elevated when the low‐density lipoprotein cholesterol (LDL‐C) goal is achieved, the addition of a fibrate should be considered (IAS Recommendations 2014), although there is no compelling evidence from clinical trials to support this consideration.

Implications for research.

Collected outcome data on CVD events should be reported separately for stroke, fatal stroke, fatal myocardial infarction, non‐fatal myocardial infarction an revascularisation procedures, such as bypass grafts and angioplasty. Further randomised controlled trials (RCTs) would be needed to better estimate a potential benefit of fibrate therapy for primary prevention of CVD events on top of statin therapy, and better estimate benefits and harms in non‐diabetic patients, in women, and in elderly individuals. In future primary prevention trials of fibrates, data on diabetic retinopathy (development of new diabetic retinopathy and progression of pre‐existing diabetic retinopathy), and development of albuminuria should be systematically captured, as well as typical known adverse effects of fibrates such as raised serum creatinine, raised serum transaminases, myositis/rhabdomyolysis, pancreatitis, venous thrombotic events, gallbladder diseases, and the number of patients discontinuing therapy due to side effects, to achieve more precise estimates for these outcomes. Special attention should be given to the systematic reporting and analysis on most relevant adverse effects, which were rather limited in this review due to inadequate reporting.

What's new

Date	Event	Description
9 March 2017	Amended	Minor adjustments to analyses 1.1 and 1.2 ‐ totals suppressed

Open in a new tab

Acknowledgements

We acknowledge Cynthia M Holas and Kristin A Costello from AbbVie pharmaceutics, North Chicago, USA, for sharing non‐published data from the FIRST 2014 study. We would like to thank MERZ Pharma, A Gebauer and Roman Görtelmeyer, for additional data on primary prevention patients, adverse effects, and mortality from Emmerich 2009. We acknowledge the support of the National Institutes of Health for providing us with the ACCORD 2010 dataset which allowed us to analyse primary prevention patients separately. Thanks to Neera Bhatnager from McMaster University, Hamilton, Canada, for her help with the first literature search. Finally, we would like to thank Nicole Martin from the Cochrane Heart Group for refining our search strategy and conducting several updated searches.

Appendices

Appendix 1. Search strategies

CENTRAL

#1 MeSH descriptor Fibric Acids explode all trees

#2 (fibrate* or fibric acid*)

#3 MeSH descriptor Clofibric Acid explode all trees

#4 (gemfibrozil or gemfibrocil)

#5 MeSH descriptor Gemfibrozil, this term only

#6 (ciprofibrat*)

#7 MeSH descriptor Bezafibrate, this term only

#8 MeSH descriptor Clofibrate, this term only

#9 (fenofibrate)

#10 MeSH descriptor Fenofibrate, this term only

#11 (bezafibrate or clofibrate)

#12 (bezafibrate or 41859‐67‐0 or asufibrat or azufibrat or (bm‐15075 or bm15075))

#13 (bezafibrate)

#14 (befibrat or befizal or beza or bezabeta or bezacur or bezafibratum)

#15 (bezafisal or bezagen or beza‐lande or bezalande or bezalex or bezalip)

#16 (bezamerck or bezamil or beza‐puren or bezapuren or bezastad or bezatol)

#17 (bionolip or cedur or colser or difaterol or durabezur or eulitop)

#18 (hadiel or klestran or (lo‐44 or lo44) or liparol or lipitrol or lipocin)

#19 (lipox or norlip or pms‐bezafibrate or polyzalip or redalip)

#20 (reducterol or regadrin b or sklerofibrat or solibay or wayfrato)

#21 (zafibral or zimbacol or ciprofibrate or 52214‐84‐3 or bi‐lipanor)

#22 (hyperlipen or cirpol or lipanor or modalim or oroxadin) or (win‐35833 or win35833)

#23 (clofibrate or 637‐07‐0 or 882‐09‐7 or abitrate or amotril)

#24 (anparton or anti‐lipide‐ratiopharm or apolan or arterioflexin)

#25 (arteriosan or artevil or ateculon or ateriosan or atheropront)

#26 (atromidin or atromid‐s) or (ay‐61123 or ay61123)

#27 (bioscleran or cartagyl or claripex or clobren or clofibral)

#28 (clofibrat or clofibratum or clofirem or clofi or clofibrin or clofibrinic acid or clofinit or clofenapate or methylclofenapate)

#29 (colebron or col or geromid or hyclorate). or (ici‐28257 or ici28257 or ici55695 or ici‐55695 or nsc1149 or nsc‐1149)

#30 (liaptene or lipavil or lipavlon or lipilim or lipofacton or lipomid)

#31 (liprinal or miscleron or misclerone or neo‐atromid or athromidin or normet)

#32 (regulipid or sinteriod or serotinex or sklerepmexe or skleromexe)

#33 (s trat os or stratos or sklero‐tablinene or sklerolip or ticlobran)

#34 (xyduril or gemfibrozil) or 25812‐30‐0 or apo‐gemfibrozil or ausgem

#35 (bolutol or (ci‐719 or ci719) or decrelip or deopid or dom‐gemfibrozil)

#36 (dropid or elmogan or emfib or fibrocit or fibros or gemcor)

#37 (gemfibril or gemfibromax or gemhexal or gemizol or gemlipid)

#38 (gemnpid or gen‐gemfibrozil or nugemfibrozil or genlip or genozil or genozil)

#39 (gozid or hidil or ipolipid or jezil or lipazil or lipidys or lipox gemfi or gemfi 1a pharma or lipidil or lofibra or lipison)

#40 (litarek or locholes or lopid or lopizid or mariston or med‐gemfibrozil)

#41 (norpid or novo‐gemfibrozil or pilder or pms‐gemfibrozil or poli‐fibrozil)

#42 (tentroc or tiba or trialmin or fenofibrate or procetofen)

#43 (49562‐28‐9 or apofen or ankebin or apo‐fenofibrate or apo‐feno‐micro)

#44 (catalip or cil or controlip or dom‐fenofibrate or durafenat)

#45 (elasterin or fegenor or fenobeta or fenobrate or fenogal or fenofanton or phenofibrate)

#46 (fenolip or fenotard or feno‐micro‐200 or fulcro or gen‐fenofibrate)

#47 (lexemin or (1f‐179 or 1fl79) or (1f‐178 or 1fl78) or lipanon or lipanthyl)

#48 (lipantil or liparison or lipcor or lipidax or lipidil or lipil)

#49 (lipirex or lipoclar or lipofene or lipofen or lipo red or liposit)

#50 (lipovas or lipsin or livesan or nolipax or normalip or procetofene)

#51 (qualecon or scleril or procetoken or protolipan or secalip)

#52 (supralip or tilene or tricor or volutine or tocofibrate) or 50465‐39‐9

#53 (transferal or theofibrate or etofylline clofibrate) or 54504‐70‐0

#54 (duolip or (ml‐1024 or ml1024) or theofol)

#55 (#1 OR #2 OR #3 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16 OR #17 OR #18 OR #19 OR #20 OR #21 OR #22 OR #23 OR #24 OR #25 OR #26 OR #27 OR #28 OR #29 OR #30 OR #31 OR #32 OR #33 OR #34 OR #35 OR #36 OR #37 OR #38 OR #39 OR #40 OR #41 OR #42 OR #43 OR #44 OR #45 OR #46 OR #47 OR #48 OR #49 OR #50 OR #51 OR #52 OR #53 OR #54)

#56 MeSH descriptor Cardiovascular Diseases explode all trees

#57 MeSH descriptor Cerebrovascular Disorders explode all trees

#58 (coronary)

#59 (heart next (disease* or attack*))

#60 (coronary next (disease* or event*))

#61 (cerebrovascular* or cardiovasc*)

#62 (myocardial next (infarct* or re?vascular* or ischemi*))

#63 (stroke* or mortal*)

#64 (morbid near/5 (heart* or cardiovascula* or coronary* or isch?em* or myocard*))

#65 (cardio* or cardia* or heart* or coronary* or angina* or ventric* or myocard* or pericard* or isch?em*)

#66 (apoplexy or (brain near/2 accident*))

#67 ((brain* or cerebral or lacunar) near/2 infarct*)

#68 MeSH descriptor Hypertension explode all trees

#69 (hypertensi* or (peripheral arter* disease*))

#70 ((high or increased or elevated) near/2 blood pressure)

#71 MeSH descriptor Hyperlipidemias explode all trees

#72 (hyperlipid* or hyperlip?emia* or hypercholesterol* or hypercholester?emia* or hyperlipoprotein?emia* or hypertriglycerid?emia*)

#73 (emboli* or arrhythmi* or thrombo* or atrial fibrillat* or tachycardi* or endocardi* or (sick next sinus))

#74 (#56 OR #57 OR #58 OR #59 OR #60 OR #61 OR #62 OR #63 OR #64 OR #65 OR #66 OR #67 OR #68 OR #69 OR #70 OR #71 OR #72 OR #73)

#75 (#55 AND #74)

MEDLINE

Cochrane sensitivity‐maximising RCT filter applied (Lefebvre 2011)

1 exp Fibric Acids/ or fibrate$.mp. or fibric acid$.tw. or exp Clofibric Acid/

2 gemfibrozil.mp. or Gemfibrozil/ or gemfibrocil.tw.

3 bezafibrate/ or clofibrate/ or ciprofibrat$.mp. [mp=title, abstract, original title, name of substance word, subject heading word, protocol supplementary concept, rare disease supplementary concept, unique identifier]

4 fenofibrate.mp. or Procetofen/

5 (bezafibrate or clofibrate).mp.

6 bezafibrate/ or bezafibrate.mp. or 41859‐67‐0.rn. or asufibrat.mp. or azufibrat.mp. or (bm‐15075 or bm15075).rn,mp.

7 (befibrat or befizal or beza or bezabeta or bezacur or bezafibratum).mp.

8 (bezafisal or bezagen or beza‐lande or bezalande or bezalex or bezalip).mp.

9 (bezamerck or bezamil or beza‐puren or bezapuren or bezastad or bezatol).mp.

10 (bionolip or cedur or colser or difaterol or durabezur or eulitop).mp.

11 (hadiel or klestran or (lo‐44 or lo44) or liparol or lipitrol or lipocin).mp.

12 (lipox or norlip or pms‐bezafibrate or polyzalip or redalip).mp.

13 (reducterol or regadrin b or sklerofibrat or solibay or wayfrato).mp.

14 (zafibral or zimbacol or ciprofibrate or 52214‐84‐3 or bi‐lipanor).mp,rn.

15 (hyperlipen or cirpol or lipanor or modalim or oroxadin).mp. or (win‐35833 or win35833).mp,rn.

16 clofibrate.mp. or 637‐07‐0.rn. or 882‐09‐7.rn. or abitrate.mp. or amotril.mp.

17 (anparton or anti‐lipide‐ratiopharm or apolan or arterioflexin).mp.

18 (arteriosan or artevil or ateculon or ateriosan or atheropront).mp.

19 (atromidin or atromid‐s).mp. or (ay‐61123 or ay61123).mp,rn. or azionly.mp.

20 (bioscleran or cartagyl or claripex or clobren or clofibral).mp.

21 (clofibrat or clofibratum or clofirem or clofi or clofibrin or clofibrinic acid or clofinit or clofenapate or methylclofenapate).mp.

22 (colebron or col or geromid or hyclorate).mp. or (ici‐28257 or ici28257 or ici55695 or ici‐55695 or nsc1149 or nsc‐1149).mp,rn.

23 (liaptene or lipavil or lipavlon or lipilim or lipofacton or lipomid).mp.

24 (liprinal or miscleron or misclerone or neo‐atromid or athromidin or normet).mp.

25 (regulipid or sinteriod or serotinex or sklerepmexe or skleromexe).mp.

26 (s trat os or stratos or sklero‐tablinene or sklerolip or ticlobran).mp.

27 (xyduril or gemfibrozil).mp. or 25812‐30‐0.rn. or apo‐gemfibrozil.mp. or ausgem.mp.

28 bolutol.mp. or (ci‐719 or ci719).mp,rn. or decrelip.mp. or deopid.mp. or dom‐gemfibrozil.mp.

29 (dropid or elmogan or emfib or fibrocit or fibros or gemcor).mp.

30 (gemfibril or gemfibromax or gemhexal or gemizol or gemlipid).mp.

31 (gemnpid or gen‐gemfibrozil or nugemfibrozil or genlip or genozil or genozil).mp.

32 (gozid or hidil or ipolipid or jezil or lipazil or lipidys or lipox gemfi or gemfi 1a pharma or lipidil or lofibra or lipison).mp.

33 (lipofor or lipogen or lipoite or lipozid or lipozil or lipur).mp.

34 (litarek or locholes or lopid or lopizid or mariston or med‐gemfibrozil).mp.

35 (norpid or novo‐gemfibrozil or pilder or pms‐gemfibrozil or poli‐fibrozil).mp.

36 (polyxit or recozil or riva‐gemfibrozil or synbrozil or taborcil).mp.

37 (tentroc or tiba or trialmin or fenofibrate or procetofen).mp.

38 49562‐28‐9.rn. or apofen.mp. or ankebin.mp. or apo‐fenofibrate.mp. or apo‐feno‐micro.mp.

39 (catalip or cil or controlip or dom‐fenofibrate or durafenat).mp.

40 (elasterin or fegenor or fenobeta or fenobrate or fenogal or fenofanton or phenofibrate).mp.

41 (fenolip or fenotard or feno‐micro‐200 or fulcro or gen‐fenofibrate).mp.

42 lexemin.mp. or (1f‐179 or 1fl79).mp,rn. or (1f‐178 or 1fl78).mp,rn. or lipanon.mp. or lipanthyl.mp.

43 (lipantil or liparison or lipcor or lipidax or lipidil or lipil).mp.

44 (lipirex or lipoclar or lipofene or lipofen or lipo red or liposit).mp.

45 (lipovas or lipsin or livesan or nolipax or normalip or procetofene).mp.

46 (qualecon or scleril or procetoken or protolipan or secalip).mp.

47 (supralip or tilene or tricor or volutine or tocofibrate).mp. or 50465‐39‐9.rn.

48 (transferal or theofibrate or etofylline clofibrate).mp. or 54504‐70‐0.rn.

49 duolip.mp. or (ml‐1024 or ml1024).mp,rn. or theofol.mp.

50 or/1‐49

51 exp Cardiovascular Diseases/

52 exp Cerebrovascular Disorders/

53 coronary.mp.

54 (heart adj (disease$ or attack$)).mp.

55 (coronary adj (disease$ or event$)).mp.

56 (cerebrovascular$ or cardiovasc$).mp.

57 (myocardial adj (infarct$ or re?vascular$ or ischemi$)).mp.

58 (stroke$ or mortal$).mp.

59 (morbid adj5 (heart$ or cardiovascula$ or coronary$ or isch?em$ or myocard$)).mp.

60 (cardio$ or cardia$ or heart$ or coronary$ or angina$ or ventric$ or myocard$ or pericard$ or isch?em$).tw.

61 (apoplexy or (brain adj2 accident$)).tw.

62 ((brain$ or cerebral or lacunar) adj2 infarct$).tw.

63 exp Hypertension/

64 (hypertensi$ or peripheral arter$ disease$).tw.

65 peripheral arter$ disease$.tw.

66 ((high or increased or elevated) adj2 blood pressure).tw.

67 exp Hyperlipidemias/

68 (hyperlipid$ or hyperlip?emia$ or hypercholesterol$ or hypercholester?emia$ or hyperlipoprotein?emia$ or hypertriglycerid?emia$).tw.

69 (emboli$ or arrhythmi$ or thrombo$ or atrial fibrillat$ or tachycardi$ or endocardi$ or (sick adj sinus)).tw.

70 or/51‐69

71 randomized controlled trial.pt.

72 controlled clinical trial.pt.

73 randomized.ab.

74 placebo.ab.

75 drug therapy.fs.

76 randomly.ab.

77 trial.ab.

78 groups.ab.

79 71 or 72 or 73 or 74 or 75 or 76 or 77 or 78

80 exp animals/ not humans.sh.

81 79 not 80

82 50 and 70 and 81

EMBASE

Cochrane RCT filter applied (Lefebvre 2011)

1 exp Fibric Acids/ or fibrate$.mp. or fibric acid$.tw. or exp Clofibric Acid/

2 gemfibrozil.mp. or Gemfibrozil/ or gemfibrocil.tw.

3 bezafibrate/ or clofibrate/ or ciprofibrat$.mp. [mp=title, abstract, subject headings, heading word, drug trade name, original title, device manufacturer, drug manufacturer, device trade name, keyword]