Fibrates for secondary prevention of cardiovascular disease and stroke
Abstract
Background
Fibrates are a class of drugs characterised by mainly lowering high triglyceride, raising high‐density lipoprotein (HDL) cholesterol, and lowering the small dense fraction of low‐density lipoprotein (LDL) cholesterol. Their efficacy for secondary prevention of serious vascular events is unclear, and to date no systematic review focusing on secondary prevention has been undertaken.
Objectives
To assess the efficacy and safety of fibrates for the prevention of serious vascular events in people with previous cardiovascular disease (CVD), including coronary heart disease and stroke.
Search methods
We searched the Cochrane Central Register of Controlled Trials (CENTRAL; Issue 9, 2014) on the Cochrane Library, MEDLINE (OVID, 1946 to October week 1 2014), EMBASE (OVID, 1980 to 2014 week 41), the China Biological Medicine Database (CBM) (1978 to 2014), the Chinese National Knowledge Infrastructure (CNKI) (1979 to 2014), Chinese Science and Technique Journals Database (VIP) (1989 to 2014). We also searched other resources, such as ongoing trials registers and databases of conference abstracts, to identify further published, unpublished, and ongoing studies.
Selection criteria
We included randomised controlled trials (RCTs) in which a fibrate (for example gemfibrozil, fenofibrate) was compared with placebo or no treatment. We excluded RCTs with only laboratory outcomes. We also excluded trials comparing two different fibrates without a placebo or no‐treatment control.
Data collection and analysis
Two review authors independently selected trials for inclusion, assessed risk of bias, and extracted the data. We contacted authors of trials for missing data.
Main results
We included 13 trials involving a total of 16,112 participants. Eleven trials recruited participants with history of coronary heart disease, two trials recruited participants with history of stroke, and one trial recruited participants with a mix of people with CVD. We judged overall risk of bias to be moderate. The meta‐analysis (including all fibrate trials) showed evidence for a protective effect of fibrates primarily compared to placebo for the primary composite outcome of non‐fatal stroke, non‐fatal myocardial infarction (MI), and vascular death (risk ratio (RR) 0.88, 95% confidence interval (CI) 0.83 to 0.94; participants = 16,064; studies = 12; I2 = 45%, fixed effect). Fibrates were moderately effective for preventing MI occurrence (RR 0.86, 95% CI 0.80 to 0.93; participants = 13,942; studies = 10; I2 = 24%, fixed effect). Fibrates were not effective against all‐cause mortality (RR 0.98, 95% CI 0.91 to 1.06; participants = 13,653; studies = 10; I2 = 23%), death from vascular causes (RR 0.95, 95% CI 0.86 to 1.05; participants = 13,653; studies = 10; I2 = 11%, fixed effect), and stroke events (RR 1.03, 95% CI 0.91 to 1.16; participants = 11,719; studies = 6; I2 = 11%, fixed effect). Excluding clofibrate trials, as the use of clofibrate was discontinued in 2002 due to safety concerns, the remaining class of fibrates were no longer effective in preventing the primary composite outcome (RR 0.90, 95% CI 0.79 to 1.03; participants = 10,320; studies = 7; I2 = 50%, random effects). However, without clofibrate data, fibrates remained effective in preventing MI (RR 0.85, 95% CI 0.76 to 0.94; participants = 8304; studies = 6; I2 = 47%, fixed effect). There was no increase in adverse events with fibrates compared to control. Subgroup analyses showed the benefit of fibrates on the primary composite outcome to be consistent irrespective of age, gender, and diabetes mellitus.
Authors' conclusions
Moderate evidence showed that the fibrate class can be effective in the secondary prevention of composite outcome of non‐fatal stroke, non‐fatal MI, and vascular death. However, this beneficial effect relies on the inclusion of clofibrate data, a drug that was discontinued in 2002 due to its unacceptably large adverse effects. Further trials of the use of fibrates in populations with previous stroke and also against a background treatment with statins (standard of care) are required.
PICOs
Plain language summary
Fibrates for secondary prevention of cardiovascular disease and stroke
Review question
The aim of this study was to assess the effect of fibrates for the prevention of major events including heart attacks, strokes, and circulatory disease death in people with existing circulatory disease.
Background
Fibrates have been used for many years as a treatment to prevent further heart attacks and strokes in people who already have disease of their circulatory system. Fibrates are a class of drugs that work by positively influencing fats in the the blood, that is lowering triglyceride, raising high‐density lipoprotein ('good') cholesterol and lowering the small dense fraction of low‐density lipoprotein ('bad') cholesterol. The drug class includes clofibrate, gemfibrozil, fenofibrate, bezafibrate, and ciprofibrate. Successful adjustment of the blood fats in people with disease of their circulatory system could mean these individuals are less likely to have heart attacks, stroke, and die from their circulatory disease. There is currently no clear evidence for benefit of fibrates in such people.
Study characteristics
The duration of fibrates ranged from 12 months to 8 years.
Key results
We included 13 trials in this review with a total of 16,112 participants with a history of coronary heart disease or stroke. This review includes evidence identified up to October 2014.
Our analysis showed that when compared primarily to placebo, fibrates can be effective for prevention of composite outcome of non‐fatal stroke, non‐fatal heart attack (myocardial infarction), and death due to circulatory disease. However, this beneficial effect relies on the inclusion on data from clofibrates, a drug that was discontinued in 2002 because of safety concerns. In other words, there is no good evidence to support the use of currently available fibrates in the prevention of future heart attacks, strokes, and circulatory disease death in people with existing circulatory disease.
Quality of the evidence
In combination with clofibrate data, quality of evidence was moderate for the composite (non‐fatal stroke, non‐fatal myocardial infarction (MI), and vascular death) and MI (non‐fatal or fatal) outcomes and low for stroke (ischaemic or haemorrhagic, non‐fatal or fatal) and death from vascular or any cause during the treatment and scheduled follow‐up period. The quality of evidence without clofibrate data was moderate for MI (non‐fatal or fatal) outcome and low for the composite (non‐fatal stroke, non‐fatal MI, and vascular death), stroke (ischaemic or haemorrhagic, non‐fatal or fatal), and death from vascular or any cause outcomes during the treatment and scheduled follow‐up period.
Authors' conclusions
Summary of findings
| Fibrates compared with control for secondary prevention of cardiovascular disease and stroke | ||||||
| Patient or population: People with previous clinical manifestations of cardiovascular disease, transient ischaemic attack, or stroke | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Control | Fibrates | |||||
| Composite outcome of non‐fatal stroke, non‐fatal myocardial infarction, and vascular death | Study population | RR 0.88 | 16,064 | ⊕⊕⊕⊝ | ||
| 233 per 1000 | 205 per 1000 | |||||
| Moderate‐risk population | ||||||
| 261 per 1000 | 230 per 1000 | |||||
| Death from any cause during the treatment and scheduled follow‐up period | Study population | RR 0.98 | 13,653 | ⊕⊕⊝⊝ | ||
| 185 per 1000 | 182 per 1000 | |||||
| Moderate‐risk population | ||||||
| 219 per 1000 | 215 per 1000 | |||||
| Death from vascular causes during the treatment and scheduled follow‐up period | Study population | RR 0.95 | 13,653 | ⊕⊕⊝⊝ | ||
| 132 per 1000 | 126 per 1000 | |||||
| Moderate‐risk population | ||||||
| 142 per 1000 | 135 per 1000 | |||||
| Myocardial infarction (non‐fatal or fatal) during the treatment and scheduled follow‐up period | Study population | RR 0.86 | 13,942 | ⊕⊕⊕⊝ | ||
| 190 per 1000 | 163 per 1000 | |||||
| Moderate‐risk population | ||||||
| 227 per 1000 | 195 per 1000 | |||||
| Stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period | Study population | RR 1.03 | 11,719 | ⊕⊕⊝⊝ | ||
| 83 per 1000 | 86 per 1000 | |||||
| Moderate‐risk population | ||||||
| 61 per 1000 | 63 per 1000 | |||||
| *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). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1More than 75% of included studies did not report the details of randomisation, and four trials withdrawn or loss to follow‐up less than 20%. 2Confidence interval overlaps no effect. | ||||||
| Fibrates compared with control for secondary prevention of cardiovascular disease and stroke ‐ without clofibrate data | ||||||
| Patient or population: People with previous clinical manifestations of cardiovascular disease, transient ischaemic attack, or stroke | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Control in cardiovascular disease and stroke | Fibrates | |||||
| Composite outcome of non‐fatal stroke, non‐fatal myocardial infarction, and vascular death without clofibrate | Study population | RR 0.9 | 10,320 | ⊕⊕⊝⊝ | ||
| 204 per 1000 | 183 per 1000 | |||||
| Moderate‐risk population | ||||||
| 204 per 1000 | 184 per 1000 | |||||
| Death from any cause during the treatment and scheduled follow‐up period without clofibrate | Study population | RR 1.01 | 7909 | ⊕⊕⊝⊝ | ||
| 146 per 1000 | 148 per 1000 | |||||
| Moderate‐risk population | ||||||
| 99 per 1000 | 100 per 1000 | |||||
| Death from vascular causes during the treatment and scheduled follow‐up period without clofibrate | Study population | RR 0.98 | 7909 | ⊕⊕⊝⊝ | ||
| 73 per 1000 | 71 per 1000 | |||||
| Moderate‐risk population | ||||||
| 74 per 1000 | 73 per 1000 | |||||
| Myocardial infarction (non‐fatal or fatal) during the treatment and scheduled follow‐up period without clofibrate | Study population | RR 0.85 | 8304 | ⊕⊕⊕⊝ | ||
| 149 per 1000 | 126 per 1000 | |||||
| Moderate‐risk population | ||||||
| 132 per 1000 | 112 per 1000 | |||||
| Stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period without clofibrate | Study population | RR 0.94 | 7189 | ⊕⊕⊝⊝ | ||
| 56 per 1000 | 53 per 1000 | |||||
| Moderate‐risk population | ||||||
| 60 per 1000 | 56 per 1000 | |||||
| *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). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1Studies did not report the details of randomisation. | ||||||
Background
Description of the condition
Cardiovascular disease (CVD) is a group of disorders of the heart and blood vessels, including hypertension, coronary heart disease, cerebrovascular disease, peripheral vascular disease, heart failure, and rheumatic heart disease (WHO 2010). CVD remains the leading cause of death in the world, claiming an estimated 17.5 million lives a year in 2012, representing 31% of all global deaths (WHO 2014). Although mortality from CVD has shown a downward trend in high‐income countries, the overall burden continues to increase in high‐, middle‐, and low‐income countries (Roger 2011). In recent decades, over three‐quarters of CVD deaths have taken place in low‐ and middle‐income countries (WHO 2015). Out of the total of 17.5 million CVD deaths, about 7.4 million deaths are attributed to coronary heart disease, and 6.7 million deaths are attributed to stroke (WHO 2014). CVD is a major public health challenge worldwide. Effective strategies for the prevention of CVD are imperative.
Description of the intervention
Fibrates have been available for clinical use for many years (Brown 2007). They include clofibrate, gemfibrozil, fenofibrate, bezafibrate, and ciprofibrate. Clofibrate is not commercially available due to little benefit and excess in total mortality (Khoury 2011). Fibrates are a class of drugs characterised by mainly lowering high triglyceride (TG), raising high‐density lipoprotein (HDL) cholesterol, and lowering the small dense fraction of low‐density lipoprotein (LDL) cholesterol (Fazio 2004). Generally, their most prominent effect is lowering TG (Fazio 2004). Except for lowering the small dense fraction of LDL, their effects on LDL cholesterol vary, depending largely on the patient's degree of hypertriglyceridaemia (Caslake 1993; de Graaf 1993; Tikkanen 1998). Epidemiological studies and meta‐analyses have shown that low HDL cholesterol and high TG are significantly associated with increased risk of CVD (Castelli 1986; Castelli 1992; Gordon 1977; Hokanson 1996; Miller 1977; Stampfer 1991). In addition, CVD risk remains considerable in spite of statin therapy (Khoury 2011). Statins tend to have less effects on TG and HDL cholesterol. Therefore, fibrates seem to be effective for prevention of CVD, especially for people with low HDL cholesterol and high TG.
How the intervention might work
Fibrates have been recognised to influence plasma lipids and lipoproteins by regulating the expression of several genes involved in lipoprotein metabolism (Staels 1992). They act via a member of a family of nuclear hormone receptors called the peroxisome proliferator‐activated receptors (PPARs). Three PPAR genes have been described (α, δ, and γ) (Pineda Torra 1999). PPARα is predominantly expressed in tissues that metabolise fatty acids, such as liver, kidney, and muscle (Staels 1998). It transmits signals from lipid‐soluble factors, such as fatty acids, hormones, and vitamins, to genes in the nucleus by binding to DNA within specific response elements. The molecules of fibrates structurally resemble short‐chain fatty acids, and so they are ligands for PPARα. Once fibrates bind to PPARα, they will form a complex binding to specific DNA sites and then lead to gene expression (Fazio 2004; Staels 1998). In this way, fibrates reduce serum TG and raise HDL cholesterol levels. The usage of fibrates were usually oral in clinical trials and clinical practice, and their dosage varied (Acheson 1962; Derosa 2004). Regarding potential side effects, previous trial evidence indicates that fibrates may be associated with myopathy, decreased creatinine clearance and estimated glomerular filtration rate, increased serum creatinine and homocysteine (EASCP 2011).
Why it is important to do this review
Although fibrates have been used for decades, evidence demonstrating their efficacy for prevention of serious vascular events is unclear (AHA/ASA 2010; Nissen 2011). Recently, a meta‐analysis found that fibrates can reduce the risk of major CVD events predominantly by prevention of coronary events (Jun 2010). Despite this, there is still much uncertainty about their separate role in the primary and secondary prevention of CVD. A systematic review on fibrates for primary prevention of CVD is currently being conducted, and the protocol has been published in Cochrane Library (Nordmann 2012). As for secondary prevention, several trials undertaken in such countries as the United States, Sweden, Israel, Australia, New Zealand, and Finland show inconsistent results. Results from the Bezafibrate Coronary Atherosclerosis Intervention Trial (BECAIT) study showed that bezafibrate could reduce coronary events in young survivors of myocardial infarction (MI) and slow the progression of focal coronary atherosclerosis (Ericsson 1996). However, the Bezafibrate Infarction Prevention (BIP) study did not find a statistically significant reduction in coronary heart disease (defined as fatal or non‐fatal coronary events) (Anonymous 2000). In the Veterans Affairs High‐Density Lipoprotein Intervention Trial (VA‐HIT) study gemfibrozil showed a 22% relative risk reduction of the combined endpoint of non‐fatal MI or death from coronary causes and no increase in overall or non‐CVD mortality (VA‐HIT 2001). Furthermore, there was a 25% relative risk reduction in the incidence of new strokes and a 59% reduction in transient ischaemic attacks (TIA) (VA‐HIT 2001). In addition, the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) study found that fenofibrate did not reduce coronary heart disease events in the subgroup of participants with previous CVD (FIELD 2005; Scott 2009). Results from individual monotherapy outcomes trials have been variable (EASCP 2011), and to date no attempt to evaluate the overall efficacy of fibrates for secondary prevention using a systematic approach has been made despite many existing reviews (AHA/ASA 2010; Khoury 2011; Nissen 2011). Despite fibrates appearing to be inexpensive, there is little data on economic effectiveness. We therefore sought to perform a systematic review of randomised controlled trials (RCTs) comparing fibrates with control in people with a history of CVD (including coronary heart disease and stroke) to determine whether there is any clear evidence of their benefit in terms of major CVD outcomes.
Objectives
To assess the efficacy and safety of fibrates for the prevention of serious vascular events in people with previous cardiovascular disease (CVD), including coronary heart disease and stroke.
Methods
Criteria for considering studies for this review
Types of studies
We included RCTs in which a fibrate (for example gemfibrozil, fenofibrate) was compared with placebo or no‐treatment arm, regardless of the duration of treatment and follow‐up. We considered any RCTs with at least one clinical outcome for this review. We excluded RCTs with only laboratory outcomes. We also excluded trials comparing two different fibrates without a placebo or no‐treatment control.
Types of participants
Eligible RCTs included participants at high recurrent risk of CVD events and stroke due to previous clinical manifestations of CVD (that is coronary heart disease or stroke), irrespective of prior lipid levels and prior lipid‐modifying treatment or diet. We did not include trials with participants with other conditions such as hypertension or diabetes but without established CVD. We considered both genders and all ages.
Types of interventions
An eligible intervention was a fibrate (any type and dosage) versus placebo or no treatment. We accepted drug treatments and other interventions provided they were given to both fibrate and comparator arms. Adjuvant treatment with one additional drug if a participant developed excessively abnormal lipids during the trial was acceptable. We included trials if statins were given to both arms. We excluded trials comparing fibrates with statins.
Types of outcome measures
We aimed to extract from each trial the number of participants randomly allocated to each intervention group to allow an intention‐to‐treat analysis. In both groups, we considered the number of participants that had one of the outcomes. We included outcomes if they occurred within the follow‐up treatment period, regardless of whether participants were taking their allocated treatment.
Primary outcomes
The primary outcome was the composite outcome of non‐fatal stroke (ischaemic or hemorrhagic stroke), non‐fatal MI, and vascular death. We included each participant only once in this composite outcome (for example if a participant had a non‐fatal event followed by a fatal event, it was counted as a fatal event only).
Secondary outcomes
-
The separate outcomes of stroke (ischaemic or haemorrhagic, non‐fatal or fatal), MI (non‐fatal or fatal), vascular death, and death from all causes:
-
Stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period. We took the definition of non‐fatal stroke from each particular trial (rather than impose a set duration of survival following the stroke).
-
MI (non‐fatal or fatal) during the treatment and scheduled follow‐up period. We took the definition of non‐fatal MI from each particular trial (rather than impose a set duration of survival following the MI).
-
Death from vascular causes during the treatment and scheduled follow‐up period. We defined vascular causes as stroke or any complication of stroke (e.g. pneumonia, pulmonary embolism, etc), coronary heart disease (e.g. MI, congestive cardiac failure, sudden death), peripheral vascular disease, haemorrhage (intracranial or extracranial), and other vascular cause.
-
Death from any cause during the treatment and scheduled follow‐up period.
-
-
Change in plasma levels of total cholesterol concentration, TG, HDL‐C, and LDL‐C.
-
Quality of life.
-
Adverse events considered included: myopathy, worsening of diabetes control, increased serum creatinine, altered liver function tests, and altered estimated glomerular filtration rate. We evaluated the number of participants developing at least one adverse event listed above.
-
Costs.
Search methods for identification of studies
Electronic searches
We searched the following sources in order to identify trials:
-
Cochrane Central Register of Controlled Trials (CENTRAL; Issue 9, 2014) on the Cochrane Library on 13 October 2014;
-
Database of Abstracts of Reviews of Effects (DARE; Issue 3, 2014) on the Cochrane Library on 13 October 2014;
-
MEDLINE (OVID, 1946 to October week 1 2014) on 13 October 2014;
-
EMBASE (OVID, 1980 to 2014 week 41) on 13 October 2014;
-
Conference Proceedings Citation Index ‐ Science (CPCI‐S) on Web of Science Core Collection (Thomson Reuters, 1990 to 10 October 2014) on 13 October 2014;
-
The China Biological Medicine Database (CBM) (1978 to 13 October 2014);
-
The Chinese National Knowledge Infrastructure (CNKI) (1979 to 13 October 2014);
-
Chinese Science and Technique Journals Database (VIP) (1989 to 13 October 2014).
Appendix 1 lists the search strategies used. We applied the Cochrane sensitivity‐maximising RCT filter to MEDLINE and adaptations of it to EMBASE and Web of Science (Lefebvre 2011).
Searching other resources
In an effort to identify further published, unpublished, and ongoing studies, we:
-
searched the following ongoing trials registers: WHO International Clinical Trial Registry Platform (ICTRP) (http://apps.who.int/trialsearch/, last searched on 13 October 2014), ClinicalTrials.gov (http://www.clinicaltrial.gov/, last searched on 13 October 2014), Current Controlled Trials (http://www.controlled‐trials.com, last searched on 13 October 2014);
-
searched all reference lists from relevant articles and reviews;
-
contacted experts in the field for information on ongoing and unpublished trials and study authors to obtain further data when necessary.
We searched for relevant trials in all languages and arranged translation of trial reports published in languages other than English.
Data collection and analysis
Selection of studies
We merged the search results using Endnote X1, and removed duplicate records of the same report. Two review authors (DW and BL) independently examined each title and abstract to exclude obviously irrelevant reports. A third review author (ML) arbitrated if there were differences in opinion in the screening of titles and abstracts. We retrieved full text of the potentially relevant reports, and two review authors (DW and BL) independently examined the full texts for eligibility. We included duplicate publication just once. If duplicate publication was uncertain, we attempted to contact the authors of the reports. Disagreements about whether a study should be included were resolved by discussion. When the disagreement was due to a difference in interpretation, a third review author (ML) was invited to arbitrate. If we needed additional information to resolve disagreements, we attempted to obtain this information from the study investigators.
Data extraction and management
Two review authors (DW and BL) independently extracted data from each report on electronic data collection forms. We entered data into RevMan 5. We planned to collect data including details of the study, participants, interventions, and outcome results. We resolved any disagreements among the authors by discussion. A third review author (ML) arbitrated if there were differences in opinion. When necessary, we tried to contact study authors for additional information.
Assessment of risk of bias in included studies
We assessed the risk of bias in all included studies using the recommended tool by The Cochrane Collaboration as described in Chapter 8 of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We made judgements of 'low risk' of bias, 'high risk' of bias, or 'unclear risk' of bias for the following seven domains:
-
random sequence generation (selection bias);
-
allocation concealment (selection bias);
-
blinding of participants and personnel (performance bias);
-
blinding of outcome assessment (detection bias);
-
incomplete outcome data (attrition bias);
-
selective outcome reporting (reporting bias);
We reported these judgements in the 'Risk of bias' table and then generated two figures: 'Risk of bias' graph (Figure 1) and 'Risk of bias' summary (Figure 2). Two review authors (DW and BL) independently performed the assessment of risk of bias in included studies. Any disagreements arising at any stage between authors were resolved through discussion or by involving a third review author (ML) when necessary.
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.
Measures of treatment effect
We expressed results for dichotomous outcomes as risk ratios with 95% confidence intervals, and expressed results for continuous outcomes as mean difference (if the same scale for each trial was available) or standardised mean difference (if different scales were used). We used RevMan 5.3 to analyse the data.
Unit of analysis issues
In cases of studies with non‐standard designs (for example cross‐over trials, cluster‐randomised trials), we managed the data according to the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011).
Dealing with missing data
If data were missing, we contacted the investigators for additional information. If data remained unavailable, we considered both best‐case and worst‐case scenarios.
Assessment of heterogeneity
We assessed statistical heterogeneity by using the I2 statistic (Higgins 2011). If the I2 was less than 30%, we considered heterogeneity as not important; 30% to 50% moderate; and more than 50% substantial. In cases where there was substantial statistical heterogeneity, we looked for the potential source(s) of the heterogeneity (that is clinical or methodological heterogeneity).
Assessment of reporting biases
We assessed small‐study bias and potential publication bias for the primary composite outcome using a funnel plot (Egger 1997).
Data synthesis
We used RevMan 5.3 for all data entry and analysis. If I2 statistic was less than 50%, we used fixed‐effect meta‐analysis. If heterogeneity was substantial, that is equal to or greater than 50%, we undertook both fixed‐effect and random‐effects meta‐analysis, and we reported the most conservative result.
Subgroup analysis and investigation of heterogeneity
As clofibrate is no longer commercially available, we compared meta‐analysis results including all fibrate trials versus excluding clofibrate trials for all outcomes.
If appropriate data were available, we intended to perform the following subgroup analysis:
-
older (>= 65 years) versus younger (< 65 years)
-
male versus female
-
with diabetes mellitus versus without diabetes mellitus
-
coronary heart disease versus stroke
-
placebo versus no treatment
-
each fibrate versus placebo or no treatment
-
different fibrate dosages (low and high dose)
Sensitivity analysis
We performed sensitivity analyses based on:
-
trials with low risk of bias (well‐concealed randomisation and blinded outcome assessment) versus trials with high‐risk of bias;
-
trials with follow‐up of more than one year versus those with shorter follow‐up;
-
older trials (before 1990) versus newer trials (since 1990);
-
trials in which fibrates were given in the presence of statin therapy (i.e. statin therapy was given to both treated and control groups) versus trials where no statin therapy was given;
-
re‐analysing the data by including and excluding studies with assumed values to replace missing data.
Results
Description of studies
Results of the search
We searched 8287 records after initial electronic searches and included 13 RCTs involving a total of 16,112 participants in the review (Acheson 1962; BECAIT 1985; BIP 1990; CDP 1968; Derosa 2004; LOCAT 1997; LEADER 1992; Newcastle Trial 1964; Scottish Trials 1964; VACSA 1966; HHS Ancillary Study 1987; VA‐HIT 1991; ACCORD 1999). All included RCTs were of parallel‐group design with participant‐level allocation to study arms. See Characteristics of included studies. We tried to identify further published, unpublished, and ongoing studies by searching other resources outlined in the Methods section. However, we identified no additional studies. For details, see Figure 3.
Study flow diagram.
However, we did identify one ongoing study that is investigating the efficacy and safety of the addition of fenofibrate to atorvastatin compared with atorvastatin alone in mixed hypercholesteraemic and coronary heart disease patients (CKD 2013). The study started in December 2013 and was expected to be completed in February 2015. We will seek to include this study when updating the review (see Characteristics of ongoing studies).
Included studies
One trial, BECAIT 1985, recruited participants who were all younger than 45 years, and one trial reported insufficient information on age of participants (Acheson 1962). In the remaining 11 trials, the age of the participants ranged from 18 to 90 years. Six trials included only male participants (BECAIT 1985; CDP 1968; LOCAT 1997; LEADER 1992; VACSA 1966; VA‐HIT 1991), and the others included participants of both sexes.
Two trials recruited participants with history of cerebrovascular diseases (Acheson 1962; VACSA 1966); one trial recruited participants with cardiovascular disease (CVD) (ACCORD 1999); nine trials recruited participants with history of coronary heart disease (BIP 1990; BECAIT 1985; CDP 1968; Derosa 2004; Newcastle Trial 1964; Scottish Trials 1964; HHS Ancillary Study 1987; VA‐HIT 1991; LOCAT 1997); and one trial recruited participants with lower extremity arterial disease and controlled angina (LEADER 1992).
Clofibrate was used as intervention in five trials (Acheson 1962; CDP 1968; Newcastle Trial 1964; Scottish Trials 1964; VACSA 1966); the dosage of clofibrate was 1800 mg per day in one trial (CDP 1968), 1000 to 2000 mg per day in three trials (Acheson 1962; Newcastle Trial 1964; Scottish Trials 1964), and 2000 mg per day in one trial (VACSA 1966). Bezafibrate was used as intervention in three trials (BECAIT 1985; BIP 1990; LEADER 1992); the dosage of bezafibrate was 400 mg per day in two trials, BIP 1990 and LEADER 1992, and 600 mg per day in one trial (BECAIT 1985). Fenofibrate was used as intervention in two trials (ACCORD 1999; Derosa 2004); the dosage of fenofibrate was 160 mg per day in ACCORD 1999 and 200 mg per day in Derosa 2004. Gemfibrozil was used as intervention in three trials at a dosage of 1200 mg day (HHS Ancillary Study 1987; LOCAT 1997; VA‐HIT 1991). Two trials used statins in both arms (ACCORD 1999; Derosa 2004), one trial used no treatment as control (Derosa 2004), and the other 10 trials used placebo as control.
The included studies differed in duration of treatment. The shortest reported duration was 12 months for the Derosa 2004 study, followed by the LOCAT 1997 study with average duration of 2.5 years, 3 to 8 years for the LEADER 1992 study, 4.5 years for the VACSA 1966 study, 5 years for BECAIT 1985, Newcastle Trial 1964, HHS Ancillary Study 1987, VA‐HIT 1991, and ACCORD 1999 studies, 6 years for BIP 1990, CDP 1968, and Scottish Trials 1964 studies, and 8 years for the Acheson 1962 study.
Excluded studies
see Characteristics of excluded studies.
We excluded nine studies on the following grounds: four trials were focused on primary prevention of CVD (Hanefeld 1991; SENDCAP 1998; HHS 1987; Rottiers 1975); three trials reported both the primary and secondary prevention of CVD, but the data of secondary prevention was not available even after we sought to obtain detailed data from the authors (DAIS 2001; FIELD 1998; FIRST 2008); one trial reported only laboratory outcomes without clinical endpoints (Tonelli 2004); and one trial did not report clinical endpoints (Li 2013).
Risk of bias in included studies
Allocation
All of the included trials reported they were randomised, but only two trials reported the details of random sequence generation (ACCORD 1999; CDP 1968), and seven trials reported the details of allocation concealment (ACCORD 1999; CDP 1968; Derosa 2004; LEADER 1992; Newcastle Trial 1964; Scottish Trials 1964; VA‐HIT 1991). The remaining trials did not report the method of random sequence generation and allocation concealment, hence we graded random sequence generation and allocation concealment for these trials as 'unclear risk'.
Blinding
Twelve trials reported they were double blind, and one trial, Acheson 1962, reported no information on blinding. Of the 12 trials reported as double blind, 10 gave information about who was blinded. In all these cases, we judged both participants and outcome assessors to be blinded (ACCORD 1999; BECAIT 1985; BIP 1990; CDP 1968; Derosa 2004; HHS Ancillary Study 1987; LEADER 1992; Scottish Trials 1964; VACSA 1966; VA‐HIT 1991). For one trial, we judged that outcome assessors were blinded, but we were uncertain whether participants were blinded (Newcastle Trial 1964). One trial reported as double‐blind did not give details of who was blinded (LOCAT 1997).
Incomplete outcome data
All the included trials reported losses to follow‐up or withdrawal. Five trials reported an intention‐to‐treat analysis (ACCORD 1999; BIP 1990; HHS Ancillary Study 1987; LEADER 1992; VA‐HIT 1991). We categorised one trial as 'unclear risk' for incomplete data due to insufficient information of exclusion and withdrawal (LOCAT 1997). We categorised nine trials as 'low risk' for incomplete data due to withdrawal or loss to follow‐up of less than 20% (ACCORD 1999; Acheson 1962; BECAIT 1985; BIP 1990; CDP 1968; Derosa 2004; Newcastle Trial 1964; VA‐HIT 1991; Scottish Trials 1964). We categorised the remaining three trials as 'high risk' for incomplete data.
Selective reporting
The protocols of seven trials were available and all the prespecified outcomes were reported (ACCORD 1999; BECAIT 1985; BIP 1990; CDP 1968; LEADER 1992; VA‐HIT 1991;LOCAT 1997); the protocols of five trials were not available (Acheson 1962; HHS Ancillary Study 1987; Newcastle Trial 1964; Scottish Trials 1964; VACSA 1966), but all the published reports included all expected outcomes, so we categorised them as 'low risk' for selective reporting. One or more outcomes of interest in this review was reported incompletely by one trial (Derosa 2004), so we categorized this trial as 'high risk' for selective reporting.
Other potential sources of bias
The funnel plot for the primary outcome showed some evidence of asymmetry (see Figure 4), indicating that some level of publication bias may exist.
Funnel plot of comparison: 1 Fibrates vs control in cardiovascular disease and stroke, outcome: 1.1 Composite outcome of non‐fatal stroke, non‐fatal myocardial infarction, and vascular death without clofibrate.
Effects of interventions
See: Summary of findings for the main comparison Fibrates compared with control for secondary prevention of cardiovascular disease and stroke; Summary of findings 2 Fibrates compared with control for secondary prevention of cardiovascular disease and stroke ‐ without clofibrate data
Primary outcome
Composite outcome of non‐fatal stroke, non‐fatal myocardial infarction, and vascular death
All 13 included trials recorded the CVD events during treatment and follow‐up period. No major cardiovascular events occurred during follow‐up in one trial (Derosa 2004). We could include 12 trials with 16,064 participants into meta‐analysis of the primary composite outcome of non‐fatal stroke, non‐fatal myocardial infarction (MI), and vascular death (ACCORD 1999; Acheson 1962; BECAIT 1985; BIP 1990; CDP 1968; HHS Ancillary Study 1987; LEADER 1992; LOCAT 1997; Newcastle Trial 1964; Scottish Trials 1964; VACSA 1966; VA‐HIT 1991). Of these 12 trials, 5 used clofibrate as intervention, 3 bezafibrate, 3 gemfibrozil, and 1 fenofibrate. Meta‐analysis performed using fixed‐effect modelling showed evidence for a protective effect of fibrates for composite outcome of non‐fatal stroke, non‐fatal MI, and vascular death (risk ratio (RR) 0.88, 95% confidence interval (CI) 0.83 to 0.94; participants = 16,064; studies = 12; I2 = 45%) (Analysis 1.1). Random‐effects meta‐analysis excluding clofibrate trials showed that fibrates were no longer effective in preventing composite outcome of non‐fatal stroke, non‐fatal MI, and vascular death (RR 0.90, 95% CI 0.79 to 1.03; participants = 10,320; studies = 7; I2 = 50%).
Secondary outcome
1. Death from any cause during the treatment and scheduled follow‐up period
Data on death from any cause from 10 trials with 13,653 participants was available (Acheson 1962; BECAIT 1985; BIP 1990; CDP 1968; HHS Ancillary Study 1987; LEADER 1992; Newcastle Trial 1964; Scottish Trials 1964; VACSA 1966; VA‐HIT 1991). Of these 10 trials, 5 used clofibrate as intervention, 3 bezafibrate, and 2 gemfibrozil. Meta‐analysis performed using fixed‐effect modelling showed no evidence of effect of fibrates for the prevention of death from any cause during the treatment and scheduled follow‐up period (RR 0.98, 95% CI 0.91 to 1.06; participants = 13,653; studies = 10; I2 = 23%) (Analysis 1.7). Meta‐analysis excluding clofibrate data showed fibrates remained ineffective against all‐cause mortality (RR 1.01, 95% CI 0.91 to 1.12; participants = 7909; studies = 5; I2 = 1%, fixed effect) (Analysis 1.2).
2. Death from vascular causes during the treatment and scheduled follow‐up period
Data on vascular death from 10 trials with 13,653 participants was available (Acheson 1962; BECAIT 1985; BIP 1990; CDP 1968; HHS Ancillary Study 1987; LEADER 1992; Newcastle Trial 1964; Scottish Trials 1964; VACSA 1966; VA‐HIT 1991). Of these 10 trials, 5 used clofibrate as intervention, 3 bezafibrate, and 2 gemfibrozil. Meta‐analysis performed using fixed‐effect modelling showed no evidence of effect of fibrates for the prevention of death from vascular causes during the treatment and scheduled follow‐up period (RR 0.95, 95% CI 0.86 to 1.05; participants = 13,653; studies = 10; I2 = 11%) (Analysis 1.8). Meta‐analysis excluding clofibrate data showed fibrates remained ineffective against vascular death (RR 0.98, 95% CI 0.84 to 1.15; participants = 7909; studies = 5; I2 = 44%, fixed effect) (Analysis 1.3).
3. MI (non‐fatal or fatal) during the treatment and scheduled follow‐up period
Data on MI risk from 10 trials with 13,942 participants was available (BECAIT 1985; BIP 1990; CDP 1968; HHS Ancillary Study 1987; LEADER 1992; LOCAT 1997; Newcastle Trial 1964; Scottish Trials 1964; VACSA 1966; VA‐HIT 1991). Of these 10 trials, 4 used clofibrate as intervention, 3 bezafibrate, and 3 gemfibrozil. Meta‐analysis performed using fixed‐effect modelling showed strong evidence of effect of fibrates for the prevention of MI (non‐fatal or fatal) during the treatment and scheduled follow‐up period (RR 0.86, 95% CI 0.80 to 0.93; participants = 13,942; studies = 10; I2 = 24%) (Analysis 1.9). Meta‐analysis excluding clofibrate data showed fibrates continued to be effective for preventing MI (RR 0.85, 95% CI 0.76 to 0.94; participants = 8304; studies = 6; I2 = 47%, fixed effect) (Analysis 1.4).
4. Stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period
A range of stroke events were recorded in 6 trials with 11,719 participants (Acheson 1962; BIP 1990; CDP 1968; LEADER 1992; VACSA 1966; VA‐HIT 1991). Of these 6 trials, 3 used clofibrate as intervention, 2 bezafibrate, and 1 gemfibrozil. Meta‐analysis performed using fixed‐effect modelling showed no evidence of effect of fibrates for the prevention of stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period (RR 1.03, 95% CI 0.91 to 1.16; participants = 11,719; studies = 6; I2 = 11%) (Analysis 1.10). Meta‐analysis excluding clofibrate data showed fibrates remained ineffective for preventing stroke (RR 0.94, 95% CI 0.78 to 1.14; participants = 7189; studies = 3; I2 = 44%, fixed effect) (Analysis 1.5).
5. Change in plasma levels of total cholesterol concentration, triglyceride, high‐density lipoprotein cholesterol, and low‐density lipoprotein cholesterol
All 13 included trials reported change in plasma lipids. However, we could only include data of three trials into meta‐analysis of the change in total cholesterol concentration, triglyceride, and high‐density lipoprotein cholesterol (HDL‐C) (Derosa 2004; LOCAT 1997; VA‐HIT 1991), and two trials into meta‐analysis of the change in low‐density lipoprotein cholesterol (LDL‐C) (Derosa 2004; LOCAT 1997). Meta‐analysis performed using random‐effects modelling showed evidence of an effect of fibrates in reducing total cholesterol concentration (mean difference (MD) ‐15.00, 95% CI ‐29.56 to ‐0.44; participants = 2974; studies = 3; I2 = 93%) (Analysis 1.11) and triglyceride concentration (MD ‐43.79, 95% CI ‐73.75 to ‐13.83; participants = 2974; studies = 3; I2 = 83%) (Analysis 1.15) and increasing HDL‐C concentration (MD 4.00, 95% CI 1.72 to 6.28; participants = 2974; studies = 3; I2 = 84%) (Analysis 1.13). Meta‐analysis performed using fixed‐effect modelling showed evidence that fibrates increased the plasmal level of LDL‐C (MD ‐17.69, 95% CI ‐21.81 to ‐13.58; participants = 443; studies = 2; I2 = 0%) (Analysis 1.12).
6. Quality of life
None of the included trials collected data on quality of life.
7. Adverse events
In the two trials with 3138 participants that recorded the risk of myopathy (Derosa 2004; BIP 1990), there was no significant difference between the fibrates group and the placebo group (RR 0.86, 95% CI 0.31 to 2.35; I2 = 0%, fixed effect). One trial reported that five participants receiving placebo and three participants receiving bezafibrate developed diabetes mellitus (BECAIT 1985), but there was no significant difference between arms (RR 0.57, 95% CI 0.15 to 2.26). One trial reported significant liver enzyme elevations in nine gemfibrozil participants and none of the placebo participants (LOCAT 1997). One trial reported raised creatinine concentration in 25 bezafibrate participants and five placebo participants, with a significant difference (RR 5.01, 95% CI 1.93 to 13.03) (LEADER 1992). None of the included trials reported altered estimated glomerular filtration rate (Analysis 1.14).
The most commonly recorded adverse event was gastrointestinal events, which six trials reported (BECAIT 1985; CDP 1968; LOCAT 1997; Newcastle Trial 1964; Scottish Trials 1964; VA‐HIT 1991). A total of 1048 participants developed gastrointestinal events in the placebo group and 2566 participants developed gastrointestinal events in the fibrate group. There was no statistically significant difference (RR 1.02, 95% CI 1.00 to 1.04; I2 = 42%, fixed effect). In addition, one trial recorded that one participant treating with bezafibrate presented with malignant melanoma, and one presented with hypernephroma at nine months after the end of placebo treatment (BECAIT 1985). One trial diagnosed cancer in seven participants with placebo and three participants with gemfibrozil during the study period (LOCAT 1997).
8. Costs
Only one included trial reported costs (Derosa 2004). The cost of fenofibrate plus fluvastatin therapy amounted to GBP 569.4 per year (USD 624.35 per year), and the cost of only fluvastatin therapy amounted to GBP 460.26 per year (USD 460.26 per year). Considering the efficacy of the two therapeutic strategies in reducing LDL‐C level and increasing HDL‐C level, the incremental cost‐effectiveness ratio showed that each 1% decrease in LDL‐C level with the combination fenofibrate plus fluvastatin added a cost of GBP 14.97 per year (USD 16.41 per year), and each 1% increase in HDL‐C level added a cost of GBP 7.48 per year (USD 8.20 per year).
Subgroup analysis
1. Older versus younger than 65 years
Subgroup analyses of different ages showed evidence for a protective effect of fibrates for the primary outcome in both the older (65 years or older) subgroup (RR 0.77, 95% CI 0.63 to 0.93; participants = 1266; studies = 1; I2 = 0%) and younger (less than 65 years) subgroup (RR 0.76, 95% CI 0.59 to 0.99; participants = 5746; studies = 4; I2 = 77%) ( Analysis 2.1).
2. Gender
Subgroup analyses of different genders showed evidence for a protective effect of fibrates for the primary outcome in both the male subgroup (RR 0.83, 95% CI 0.73 to 0.94; participants = 5092; studies = 4; I2 = 27%) and the female subgroup (RR 0.30, 95% CI 0.16 to 0.56; participants = 616; studies = 3; I2 = 0%) (Analysis 3.1).
3. With diabetes mellitus versus without
Subgroup analyses of participants with or without diabetes mellitus showed evidence for a protective effect of fibrates for the primary outcome in both the subgroup with diabetes mellitus (RR 0.85, 95% CI 0.73 to 0.99; participants = 2643; studies = 2; I2 = 0%) and the subgroup without diabetes mellitus (RR 0.75, 95% CI 0.65 to 0.86; participants = 3118; studies = 3; I2 = 0%) (Analysis 4.1).
4. Coronary heart disease versus stroke
Subgroup analyses showed no evidence for a protective effect of fibrates for the primary outcome either in the subgroup with coronary heart disease (RR 0.83, 95% CI 0.68 to 1.01; participants = 8357; studies = 6; I2 = 75%) or the subgroup with previous stroke (RR 0.99, 95% CI 0.82 to 1.20; participants = 638; studies = 2; I2 = 0%) (Analysis 5.1).
5. Each single fibrate versus placebo or no treatment
Four kinds of fibrates were used in the included trials: fenofibrate (Derosa 2004; ACCORD 1999), bezafibrate (BECAIT 1985; BIP 1990; LEADER 1992), clofibrate (Acheson 1962; CDP 1968; Newcastle Trial 1964; Scottish Trials 1964; VACSA 1966), and gemfibrozil (HHS Ancillary Study 1987; LOCAT 1997; VA‐HIT 1991).
We could include data for three kinds of fibrates into meta‐analysis: clofibrate, bezafibrate, and gemfibrozil. Subgroup analyses showed no evidence of difference in the preventive effect on the primary outcome aross fibrates, that is clofibrate trials (RR 0.86, 95% CI 0.74 to 1.00; participants = 5744; studies = 5; I2 = 51%), bezafibrate trials (RR 0.90, 95% CI 0.73 to 1.12; participants = 4750; studies = 3; I2 = 53%), and gemfibrozil trials (RR 1.01, 95% CI 0.62 to 1.64; participants = 3554; studies = 3; I2 = 67%) (Analysis 6.1).
We did not undertake the subgroup analysis of fibrate dose wbecause of insufficient data.
Sensitivity analysis
1. Trials with low risk of bias versus trials with high risk of bias
We classified trials with 'low risk' for sequence generation and blinded outcome assessment as trials with low risk of bias and trials with 'unclear risk' or 'high risk' for sequence generation and blinded outcome as trials with high risk of bias. Sensitivity analysis showed evidence of an effect of fibrates in preventing the primary outcome in trials with low risk of bias (RR 0.85, 95% CI 0.79 to 0.91; participants = 11,221; studies = 6; I2 = 41%) (Analysis 7.1).
2. New trials versus old (before 1990) trials
We classified trials conducted before 1990 as 'old trials' (Acheson 1962; BECAIT 1985; CDP 1968; HHS Ancillary Study 1987; Newcastle Trial 1964; Scottish Trials 1964; VACSA 1966). Sensitivity analysis showed evidence of an effect of fibrates in preventing the primary outcome in new (1990 and after) trials (RR 0.88, 95% CI 0.81 to 0.95; participants = 9600; studies = 6; I2 = 0%) (Analysis 7.2).
3. Trials with the addition of fibrate to statin therapy
Two trials compared the addition of a fibrate to statin therapy to statin therapy alone. Given its small sample size (N = 48) and follow‐up of only 12 months, one of these trials reported no cardiovascular events (Derosa 2004). The second trial compared the addition of fenofibrate to simvastatin to simvastatin alone (ACCORD 1999). There was no difference in primary composite outcome between arms (RR 0.90, 95% CI 0.74 to 1.09; participants 5518).
We could not undertake other sensitivity analysis (handling of missing data and effect of length of follow‐up) due to insufficient data.
Discussion
Summary of main results
We found that, when compared primarily to placebo, fibrates as a class of drugs can have a protective effect for composite outcome of non‐fatal stroke, non‐fatal MI, and vascular death in people with history of CVD. This protective effect appeared to primarily be the result of a reduction in non‐fatal or fatal MI events. Fibrates were not effective in the prevention of stroke or death from any cause or vascular causes. Apart from clofibrate, we found no single fibrate to have a protective effect for secondary prevention of cardiovascular disease and stroke. There was no significant difference between fibrates group and placebo group in reported adverse events. We found no data on the cost‐effectiveness of fibrates.
Overall completeness and applicability of evidence
This systematic review included all trials of fibrate therapy. However, as clofibrate was withdrawn in 2002 due to safety concerns, we also undertook an analysis excluding clofibrate trials. Meta‐analysis without the clofibrate data showed that fibrates were no longer effective for preventing composite outcome of non‐fatal stroke, non‐fatal MI, and vascular death in people with history of cardiovascular disease and stroke (RR 0.90, 95% CI 0.79 to 1.03; participants = 10,320; studies = 7; I2 = 50%, random effects). However, without clofibrate data, fibrates were still effective in preventing MI (RR 0.85, 95% CI 0.76 to 0.94; participants = 8304; studies = 6; I2 = 47%, fixed effect).
Data in this review were primarily from trials without background treatment with statins (current standard of care) (Keene 2014). The effect of combined statins and fibrates compared with statins or fibrates alone for CVD therefore remains uncertain.
In addition, the majority of participants in the included trials were diagnosed with coronary heart disease. Only two trials recruited 627 participants with cerebrovascular disease, so caution is required in extrapolating the results of this review to people with stroke.
Quality of the evidence
According to GRADE approach, the quality of evidence with the clofibrate data was moderate for the composite (non‐fatal stroke, non‐fatal MI, and vascular death) and MI (non‐fatal or fatal) outcomes and low for stroke (ischaemic or haemorrhagic, non‐fatal or fatal) and death from vascular or any cause during the treatment and scheduled follow‐up period (summary of findings Table for the main comparison). The quality of evidence without the clofibrate data was moderate for the MI (non‐fatal or fatal) outcome and low for the composite (non‐fatal stroke, non‐fatal MI, and vascular death), stroke (ischaemic or haemorrhagic, non‐fatal or fatal), and death from vascular or any cause outcomes during the treatment and scheduled follow‐up period (summary of findings Table 2). The moderate‐quality evidence was due to studies (more than 75%) not reporting the details of randomisation and four trials withdrawn or loss to follow‐up of less than 20%. Only two trials reported a random component in the sequence generation process, and only seven trials provided details of allocation concealment method. Eleven of the included trials reported they were double blind. Withdrawals and attrition was less than 20% in 11 trials (Derosa 2004). The low‐quality evidence was a result of the above description in addition to the imprecision of the effect estimate.
Agreements and disagreements with other studies or reviews
Our results are consistent with two previous systematic reviews. Jun et al assessed effects of fibrates on cardiovascular outcomes in the context of both primary and secondary prevention (Jun 2010). They found fibrate therapy produced a 10% RR reduction (95% CI 0 to 18%; P = 0.05) for major CVD events and a 13% RR reduction (95% CI 7 to 19; P < 0.0001) for coronary events, but had no benefit on the risk of stroke. They also found no effect of fibrate therapy on the risk of all‐cause mortality, cardiovascular mortality, sudden death, or non‐vascular mortality. More recently, the systematic review conducted by Keene et al also included participants with or without CVD. This later review found no significant difference in all‐cause mortality with fibrate treatment (odds ratio (OR) 0.98, 95% CI 0.89 to 1.08, P = 0.66). In addition, neither coronary heart disease mortality (OR 0.92, 95% CI 0.81 to 1.04, P = 0.19) nor stroke (OR 1.01, 95% CI 0.90 to 1.13, P = 0.84) were found to be significantly affected by fibrates across all trials. However, non‐fatal MI was reduced (OR 0.80, 0.74 to 0.87, P < 0.001) (Keene 2014). Based on the results of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial (ACCORD 1999), the Keene review highlighted that the use fibrates did not improve CVD outcomes in individuals with CVD already treated with a statin.
In contrast to the findings of this Cochrane review, the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) trial showed no significant reduction in CVD events (that is CVD death, MI, stroke, coronary, or carotid revascularisation) in the subgroup of participants with previous CVD (FIELD 1998). We excluded this study from our review as it focussed on primary prevention of CVD, and the data for secondary prevention of CVD was not available despite our contacting the authors of this study. The use of statin therapy could partly explain the negative result of this study. Decisions about changes in lipid‐lowering therapy were at the discretion of the participant's primary‐care or specialist physician, and there was a greater use of statin therapy in placebo participants than in those on fenofibrate (FIELD 1998).
The role of fibrates for the prevention of CVD events remains controversial. The 2014 National Institute for Health and Care Excellence (NICE) in United Kingdom guidance did not recommend fibrates for either primary or secondary prevention of CVD (NICE 2014). However, the International Atherosclerosis Society recommends that if non‐HDL‐C and triglycerides remain elevated when the LDL‐C goal is achieved, consideration could be given to adding a fibrate (IAS 2014), albeit RCTs have not definitively shown efficacy of using fibrates combined with another lipid‐lowering drug in the secondary prevention of CVD.
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.
Funnel plot of comparison: 1 Fibrates vs control in cardiovascular disease and stroke, outcome: 1.1 Composite outcome of non‐fatal stroke, non‐fatal myocardial infarction, and vascular death without clofibrate.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 1 Composite outcome of non‐fatal stroke, non‐fatal MI, and vascular death without clofibrate.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 2 Death from any cause during the treatment and scheduled follow‐up period without clofibrate.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 3 Death from vascular causes during the treatment and scheduled follow‐up period without clofibrate.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 4 MI (non‐fatal or fatal) during the treatment and scheduled follow‐up period without clofibrate.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 5 Stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period without clofibrate.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 6 Composite outcome of non‐fatal stroke, non‐fatal MI, and vascular death.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 7 Death from any cause during the treatment and scheduled follow‐up period.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 8 Death from vascular causes during the treatment and scheduled follow‐up period.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 9 MI (non‐fatal or fatal) during the treatment and scheduled follow‐up period.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 10 Stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 11 change in plasma levels of total cholesterol concentration.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 12 Change in plasma levels of LDL‐C.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 13 Change in plasma levels of HDL‐C.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 14 adverse events.
Comparison 1 Fibrates vs control in cardiovascular disease and stroke, Outcome 15 Change in plasma levels of TG.
Comparison 2 Subgroup analysis 1: different ages, Outcome 1 primary outcome.
Comparison 3 Subgroup analysis 2: different genders, Outcome 1 primary outcome.
Comparison 4 Subgroup analysis 3: with versus without diabetes mellitus, Outcome 1 primary outcome.
Comparison 5 Subgroup analysis 4: coronary heart disease versus stroke, Outcome 1 primary outcome.
Comparison 6 Subgroup analysis 5: each fibrate versus placebo or no treatment, Outcome 1 primary outcome.
Comparison 7 Sensitivity analysis, Outcome 1 Trials with low‐risk of bias versus with high‐risk of bias on primary outcome.
Comparison 7 Sensitivity analysis, Outcome 2 New trials versus old trials on primary outcome.
| Fibrates compared with control for secondary prevention of cardiovascular disease and stroke | ||||||
| Patient or population: People with previous clinical manifestations of cardiovascular disease, transient ischaemic attack, or stroke | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Control | Fibrates | |||||
| Composite outcome of non‐fatal stroke, non‐fatal myocardial infarction, and vascular death | Study population | RR 0.88 | 16,064 | ⊕⊕⊕⊝ | ||
| 233 per 1000 | 205 per 1000 | |||||
| Moderate‐risk population | ||||||
| 261 per 1000 | 230 per 1000 | |||||
| Death from any cause during the treatment and scheduled follow‐up period | Study population | RR 0.98 | 13,653 | ⊕⊕⊝⊝ | ||
| 185 per 1000 | 182 per 1000 | |||||
| Moderate‐risk population | ||||||
| 219 per 1000 | 215 per 1000 | |||||
| Death from vascular causes during the treatment and scheduled follow‐up period | Study population | RR 0.95 | 13,653 | ⊕⊕⊝⊝ | ||
| 132 per 1000 | 126 per 1000 | |||||
| Moderate‐risk population | ||||||
| 142 per 1000 | 135 per 1000 | |||||
| Myocardial infarction (non‐fatal or fatal) during the treatment and scheduled follow‐up period | Study population | RR 0.86 | 13,942 | ⊕⊕⊕⊝ | ||
| 190 per 1000 | 163 per 1000 | |||||
| Moderate‐risk population | ||||||
| 227 per 1000 | 195 per 1000 | |||||
| Stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period | Study population | RR 1.03 | 11,719 | ⊕⊕⊝⊝ | ||
| 83 per 1000 | 86 per 1000 | |||||
| Moderate‐risk population | ||||||
| 61 per 1000 | 63 per 1000 | |||||
| *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). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1More than 75% of included studies did not report the details of randomisation, and four trials withdrawn or loss to follow‐up less than 20%. 2Confidence interval overlaps no effect. | ||||||
| Fibrates compared with control for secondary prevention of cardiovascular disease and stroke ‐ without clofibrate data | ||||||
| Patient or population: People with previous clinical manifestations of cardiovascular disease, transient ischaemic attack, or stroke | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Control in cardiovascular disease and stroke | Fibrates | |||||
| Composite outcome of non‐fatal stroke, non‐fatal myocardial infarction, and vascular death without clofibrate | Study population | RR 0.9 | 10,320 | ⊕⊕⊝⊝ | ||
| 204 per 1000 | 183 per 1000 | |||||
| Moderate‐risk population | ||||||
| 204 per 1000 | 184 per 1000 | |||||
| Death from any cause during the treatment and scheduled follow‐up period without clofibrate | Study population | RR 1.01 | 7909 | ⊕⊕⊝⊝ | ||
| 146 per 1000 | 148 per 1000 | |||||
| Moderate‐risk population | ||||||
| 99 per 1000 | 100 per 1000 | |||||
| Death from vascular causes during the treatment and scheduled follow‐up period without clofibrate | Study population | RR 0.98 | 7909 | ⊕⊕⊝⊝ | ||
| 73 per 1000 | 71 per 1000 | |||||
| Moderate‐risk population | ||||||
| 74 per 1000 | 73 per 1000 | |||||
| Myocardial infarction (non‐fatal or fatal) during the treatment and scheduled follow‐up period without clofibrate | Study population | RR 0.85 | 8304 | ⊕⊕⊕⊝ | ||
| 149 per 1000 | 126 per 1000 | |||||
| Moderate‐risk population | ||||||
| 132 per 1000 | 112 per 1000 | |||||
| Stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period without clofibrate | Study population | RR 0.94 | 7189 | ⊕⊕⊝⊝ | ||
| 56 per 1000 | 53 per 1000 | |||||
| Moderate‐risk population | ||||||
| 60 per 1000 | 56 per 1000 | |||||
| *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). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1Studies did not report the details of randomisation. | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Composite outcome of non‐fatal stroke, non‐fatal MI, and vascular death without clofibrate Show forest plot | 7 | 10320 | Risk Ratio (M‐H, Random, 95% CI) | 0.90 [0.79, 1.03] |
| 2 Death from any cause during the treatment and scheduled follow‐up period without clofibrate Show forest plot | 5 | 7909 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.01 [0.91, 1.12] |
| 3 Death from vascular causes during the treatment and scheduled follow‐up period without clofibrate Show forest plot | 5 | 7909 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.98 [0.84, 1.15] |
| 4 MI (non‐fatal or fatal) during the treatment and scheduled follow‐up period without clofibrate Show forest plot | 6 | 8304 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.85 [0.76, 0.94] |
| 5 Stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period without clofibrate Show forest plot | 3 | 7189 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.94 [0.78, 1.14] |
| 6 Composite outcome of non‐fatal stroke, non‐fatal MI, and vascular death Show forest plot | 12 | 16064 | Risk Ratio (M‐H, Random, 95% CI) | 0.88 [0.81, 0.97] |
| 7 Death from any cause during the treatment and scheduled follow‐up period Show forest plot | 10 | 13653 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.98 [0.91, 1.06] |
| 8 Death from vascular causes during the treatment and scheduled follow‐up period Show forest plot | 10 | 13653 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.95 [0.86, 1.05] |
| 9 MI (non‐fatal or fatal) during the treatment and scheduled follow‐up period Show forest plot | 10 | 13942 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.86 [0.80, 0.93] |
| 10 Stroke (ischaemic or haemorrhagic, non‐fatal or fatal) during the treatment and scheduled follow‐up period Show forest plot | 6 | 11719 | Risk Ratio (M‐H, Fixed, 95% CI) | 1.03 [0.91, 1.16] |
| 11 change in plasma levels of total cholesterol concentration Show forest plot | 3 | 2974 | Mean Difference (IV, Random, 95% CI) | ‐15.00 [‐29.56, ‐0.44] |
| 12 Change in plasma levels of LDL‐C Show forest plot | 2 | 443 | Mean Difference (IV, Fixed, 95% CI) | ‐17.69 [‐21.81, ‐13.58] |
| 13 Change in plasma levels of HDL‐C Show forest plot | 3 | 2974 | Mean Difference (IV, Random, 95% CI) | 4.00 [1.72, 6.28] |
| 14 adverse events Show forest plot | 8 | Risk Ratio (M‐H, Random, 95% CI) | Subtotals only | |
| 14.1 myopathy | 2 | 3138 | Risk Ratio (M‐H, Random, 95% CI) | 0.86 [0.31, 2.39] |
| 14.2 worsening of diabetes control | 1 | 92 | Risk Ratio (M‐H, Random, 95% CI) | 0.57 [0.15, 2.26] |
| 14.3 altered liver function tests | 1 | 395 | Risk Ratio (M‐H, Random, 95% CI) | 19.10 [1.12, 325.86] |
| 14.4 Gastrointestinal events | 5 | 5593 | Risk Ratio (M‐H, Random, 95% CI) | 1.29 [0.78, 2.14] |
| 14.5 raised creatinine concentration | 1 | 1568 | Risk Ratio (M‐H, Random, 95% CI) | 5.01 [1.93, 13.03] |
| 15 Change in plasma levels of TG Show forest plot | 3 | 2974 | Mean Difference (IV, Random, 95% CI) | ‐43.79 [‐73.75, ‐13.83] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 primary outcome Show forest plot | 4 | 7012 | Risk Ratio (M‐H, Random, 95% CI) | 0.77 [0.64, 0.94] |
| 1.1 older than 65 years | 1 | 1266 | Risk Ratio (M‐H, Random, 95% CI) | 0.77 [0.63, 0.93] |
| 1.2 younger than 65 years | 4 | 5746 | Risk Ratio (M‐H, Random, 95% CI) | 0.76 [0.59, 0.99] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 primary outcome Show forest plot | 5 | 5708 | Risk Ratio (M‐H, Random, 95% CI) | 0.75 [0.61, 0.92] |
| 1.1 males | 4 | 5092 | Risk Ratio (M‐H, Random, 95% CI) | 0.83 [0.73, 0.94] |
| 1.2 female | 3 | 616 | Risk Ratio (M‐H, Random, 95% CI) | 0.30 [0.16, 0.56] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 primary outcome Show forest plot | 4 | 5761 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.79 [0.72, 0.88] |
| 1.1 with diabetes mellitus | 2 | 2643 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.85 [0.73, 0.99] |
| 1.2 without diabetes mellitus | 3 | 3118 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.75 [0.65, 0.86] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 primary outcome Show forest plot | 8 | 8995 | Risk Ratio (M‐H, Random, 95% CI) | 0.87 [0.75, 1.01] |
| 1.1 Ischaemic heart disease and coronary heart disease | 6 | 8357 | Risk Ratio (M‐H, Random, 95% CI) | 0.83 [0.68, 1.01] |
| 1.2 stroke | 2 | 638 | Risk Ratio (M‐H, Random, 95% CI) | 0.99 [0.82, 1.20] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 primary outcome Show forest plot | 11 | 14048 | Risk Ratio (M‐H, Random, 95% CI) | 0.88 [0.79, 0.98] |
| 1.1 clofibrate | 5 | 5744 | Risk Ratio (M‐H, Random, 95% CI) | 0.86 [0.74, 1.00] |
| 1.2 bezafibrate | 3 | 4750 | Risk Ratio (M‐H, Random, 95% CI) | 0.90 [0.73, 1.12] |
| 1.3 gemfibrozil | 3 | 3554 | Risk Ratio (M‐H, Random, 95% CI) | 1.01 [0.62, 1.64] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Trials with low‐risk of bias versus with high‐risk of bias on primary outcome Show forest plot | 12 | 16064 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.88 [0.83, 0.94] |
| 1.1 low‐risk of bias | 6 | 11221 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.85 [0.79, 0.91] |
| 1.2 high‐risk of bias | 6 | 4843 | Risk Ratio (M‐H, Fixed, 95% CI) | 0.96 [0.86, 1.07] |
| 2 New trials versus old trials on primary outcome Show forest plot | 12 | 16064 | Risk Ratio (M‐H, Random, 95% CI) | 0.88 [0.81, 0.97] |
| 2.1 new trials | 5 | 9600 | Risk Ratio (M‐H, Random, 95% CI) | 0.88 [0.81, 0.95] |
| 2.2 old trials | 7 | 6464 | Risk Ratio (M‐H, Random, 95% CI) | 0.88 [0.73, 1.05] |
