RCTs indicates randomized controlled trials; AMI, acute myocardial infarction; ACS, acute coronary syndrome; CHF, congestive heart failure; ACC/AHA, American College of Cardiology/American Heart Association.
Coca SG, Krumholz HM, Garg AX, Parikh CR. Underrepresentation of Renal Disease in Randomized Controlled Trials of Cardiovascular Disease. JAMA. 2006;296(11):1377-1384. doi:10.1001/jama.296.11.1377
Author Affiliations: Clinical Epidemiology Research Center, Veterans Affairs Medical Center, West Haven, Conn (Drs Coca and Parikh); Department of Medicine, Yale University School of Medicine, New Haven, Conn (Drs Coca, Krumholz, and Parikh); and Departments of Medicine and Epidemiology and Biostatistics, University of Western Ontario, London (Dr Garg).
Context Patients with renal disease are at high risk for cardiovascular mortality. Determining which interventions best offset this risk remains a health priority.
Objective To quantify the representation of patients with renal disease in randomized controlled trials for interventions proven efficacious for cardiovascular disease.
Data Sources We searched MEDLINE for trials published from 1985 through 2005 in 11 major medical and subspecialty journals.
Study Selection Randomized controlled trials for chronic congestive heart failure and acute myocardial infarction of treatments that are currently listed as class I or II recommendations in the current American College of Cardiology/American Heart Association guidelines were included.
Data Extraction Two reviewers independently abstracted data on study and patient characteristics, renal measurements, outcomes, and prognostic features.
Data Synthesis A total of 153 trials were reviewed. Patients with renal disease were reported as excluded in 86 (56%) trials. Patients with renal disease were more likely to be excluded from trials that were multicenter; of moderate enrollment size; North American; that tested renin-angiotensin-aldosterone system antagonists and anticoagulants; and that tested chronic congestive heart failure. Only 8 (5%) original articles reported the proportion of enrolled patients with renal disease, and only 15 (10%) reported mean baseline renal function. While 81 (53%) trials performed subgroup analyses of some baseline characteristic in the original article, only 4 (3%) subgroup analyses of treatment stratified by renal disease were performed.
Conclusion Major cardiovascular disease trials frequently exclude patients with renal disease and do not provide adequate information on the renal function of enrollees or the effect of interventions on patients with renal disease.
Despite the therapeutic advances in the latter half of the 20th century, cardiovascular disease (CVD) remains the leading cause of death among people with chronic kidney disease. A large proportion of the population, currently exceeding 9 million people in the United States, has chronic kidney disease.1 In patients with CVD, specifically coronary artery disease and chronic congestive heart failure (CHF), the prevalence of renal disease ranges between 30% and 60%.2- 8 Approximately 30% to 50% of patients who present with acute myocardial infarction (AMI) have an estimated glomerular filtration rate (eGFR) of less than 60 mL/min per 1.73 m2, and 10% to 15% have an eGFR of less than 30 mL/min per 1.73 m2.3,5- 7
In patients with CHF, the prevalence of renal disease is slightly greater. Observational data from cohorts with CHF reveal that 40% to 60% have an eGFR of less than 60 mL/min per 1.73 m2 and about 15% have an eGFR of less than 30 mL/min per 1.73 m2.2,3,8 While this relationship may simply represent an epiphenomenon, recent studies suggest that renal disease, in fact, independently portends increased morbidity and mortality with CVD.6,9- 15
In this regard, impaired renal function is as great or a greater predictor of mortality than both left ventricular ejection fraction and New York Heart Association (NYHA) class in chronic CHF.12,15 Several factors to explain the relationship between renal disease and CVD have been proposed, including increased levels of inflammatory factors,16 proteinuria,17 anemia,18- 20 elevated homocysteine levels,21 higher lipoprotein (a) levels,21 arterial stiffness,22 and oxidative stress and altered nitric oxide/endothelin balance leading to endothelial dysfunction.23- 27
Because renal disease is so prevalent in CVD and the pathophysiologic processes of CVD may differ from those with normal renal function, it is important to have information on the subgroup of renal disease. It is possible that patients with renal disease have similar risk-benefit profiles from cardiovascular therapies compared with the general population. While there may be higher absolute benefit in patients with renal disease, the absolute risks may also be higher leading to questionable risk benefit. For example, bleeding complications are often higher in renal disease because of platelet dysfunction.28 Therefore, the benefits of fibrinolytic therapy, antiplatelet agents, and anticoagulants for AMI may be negated or outweighed by these complications.11,29- 32 Similarly, inhibitors of the renin-angiotensin-aldosterone system (RAAS) may have a higher risk in the setting of renal disease because of complications relating to hyperkalemia and worsening renal function.33 The apparent benefits of percutaneous coronary intervention (PCI) may be abolished in patients with renal disease because of different atherosclerotic plaque morphology compared with those who have normal renal function.34 In this regard, PCI in patients with renal disease is accompanied by a higher rate of early and late complications of bleeding,4,35- 37 restenosis, and death.4,36,37 Therefore, in the setting of CVD, the presence of renal disease may identify a segment of the population with a worse prognosis who may be less responsive to the benefits of certain interventions.
While there are strong recommendations in the guidelines for the management of CVD, it is often difficult to make evidence-based medical decisions for management of patients with renal disease and CVD because of the uncertain distribution and representation of renal disease in major CVD trials. We aimed to establish the frequency of exclusion of patients with renal disease from major CVD trials. We also sought to determine the frequency at which baseline renal function was reported and the proportion of patients with renal disease in these trials.
We performed a systematic search of the MEDLINE database to identify all recent major clinical trials of treatment for chronic CHF and AMI. The search strategy included articles from 1985 through the end of 2005 for articles indexed under the subject headings chronic heart failure, myocardial infarction, or acute coronary syndrome and randomized controlled trials. We chose to start with the year 1985 because this marked the beginning of the era of CVD randomized controlled trials (RCTs).
The search was restricted to 11 major medical journals. These included 4 general medicine journals (New England Journal of Medicine, Lancet, Annals of Internal Medicine, and JAMA), 4 major cardiology journals (Circulation, Journal of the American College of Cardiology, American Heart Journal, and European Heart Journal), and 3 major nephrology journals (Journal of the American Society of Nephrology, Kidney International, and American Journal of Kidney Disease). A total of 1848 citations were obtained.
We selected RCTs of drugs, procedures, or devices for treatment of chronic CHF or AMI if they included 100 or more participants and the treatments implemented in the trial are currently listed as either class I or class II recommendations in the current American College of Cardiology/American Heart Association (ACC/AHA) guidelines on the management of chronic heart failure in the adult, management of patients with ST-elevation MI, or management of unstable angina and non–ST-segment elevation MI.38- 40 Trials were excluded if they did not have mortality as either a primary or secondary end point or were a subgroup (eg, women only) or post-hoc analyses of the original study. After application of the inclusion and exclusion criteria, 153 articles remained appropriate for further analysis (Figure).
We entered data from the trials into an electronic database (Microsoft Access, version 2003, Microsoft Corp, Redmond, Wash) that had validity checks and internal logic that assisted with data entry. The data abstraction and data entry were confirmed by a second reviewer who cross-checked 100% of selected articles. The κ statistic was greater than 95% for all of the variables abstracted between the 2 reviewers.
Cardiovascular disease trials were assigned to 4 periods (1985-1990, 1991-1995, 1996-2000, and 2001-2005) based on their year of publication. Types of interventions were grouped into 9 categories: fibrinolytic therapy, PCIs, devices, RAAS antagonists, β-blockers, antiplatelet agents (oral and intravenous), anticoagulants (oral and intravenous), statins, and other.
Variables including exclusion of renal disease, baseline serum creatinine level, proportion of enrollees with kidney disease, and subgroup analyses of renal disease were abstracted. For each trial, the first complete published report was the index article. However, for missing variables, other published articles on the study (not limited to the 11 journals listed above), including methods articles and articles published on post-hoc analyses pertaining to one of the selected index articles, were also reviewed to supplement and complete the data fields. We searched review articles, bibliographies, and the Web of Science database to find all relevant follow-up articles.
Exclusion of renal disease referred to an exclusion criterion of any form of renal disease or level of renal dysfunction for the study. If an absolute serum creatinine level or creatinine clearance was referred to in the exclusion criteria in the original report, no further investigations were necessary. If a qualitative term for renal disease or no mention of renal disease occurred in the methods section, then the primary author of the article was contacted to clarify if patients with renal disease were excluded and what the criteria were that they used to exclude them.
We studied the general characteristics of CVD trials and evaluated the variation of exclusion and reporting of renal disease over the prespecifed time periods by size of the trial, number of centers involved, location (by continent of primary author), therapeutic class of drug or device, funding source (government, industry, or both), journal, and diagnostic category. The Fisher exact test was used to evaluate differences in exclusion for renal disease among categorical variables. All variables listed in Table 1 with significant P values on univariate analysis were included in the multivariate logistic regression analysis. Referent groups for each variable were chosen by clinical judgment or by the group with the lowest frequency of exclusion. All calculations were performed using the software system SAS version 9.1 (SAS Institute Inc, Cary, NC). P values of <.05 were considered statistically significant.
A total of 153 trials were reviewed; their characteristics are summarized in Table 1. Overall, 86 (56%) trials excluded patients with renal disease. As seen in Table 1, patients with renal disease were more likely to be excluded from trials that were multicenter; of moderate enrollment size (500-999 and 1000-4999 participants); North American; that tested RAAS antagonists and anticoagulants; and that tested chronic CHF. Frequency of exclusion did not differ by time period of publication, funding source, or journal. On multivariable analysis, trials of 1000 to 4999 patients (reference group, 100-499 patients; odds ratio [OR], 10.4; 95% confidence interval [CI], 1.8-59.5), trials that were North American (reference group, Europe; OR, 5.1; 95% CI, 1.2-21.3), and trials of RAAS antagonists (reference group, β-blockers; OR, 55.4; 95% CI, 2.5-999) remained as independent predictors of exclusion of renal disease.
The thresholds for exclusion of patients with renal disease are listed in Table 2. Several trials used nonspecific exclusion criteria for renal disease such as “renal dysfunction,” “renal failure,” “renal insufficiency,” “severe renal disorders,” “severe renal impairment,” “known kidney disease,” and “clinically important renal disease.”41- 52 When we contacted several authors, we discovered that only 5 of these 29 trials ultimately used absolute definitions (eg, threshold serum creatinine level) for renal disease in their protocols. In most of the remaining cases, the criterion for exclusion of “renal disease” or equivalent term was left up to interpretation by the individual site coordinator. Patients with renal disease were also excluded from several trials (13/80 [16%]) that did not mention renal disease as an exclusion criterion in the primary article or methods article. A few descriptive examples of these broader exclusion criteria include “any disorder that the investigator judged placed the person at increased risk,” “other disease likely to limit survival,” “high-risk for bleeding,” and “any condition other than cardiac disease that was associated with a high likelihood of death.”
Few trials reported on renal disease or renal function in the baseline characteristics (Table 2). Only 8 (5%) original articles reported the proportion of patients enrolled in the trial with renal disease as a categorical variable, specified by either generic terms (eg, renal failure) or by reporting on the percentage of enrollees with serum creatinine levels, creatinine clearances, or eGFRs above or below a threshold level. From subsequent articles, the proportion of renal disease enrollees was determined for an additional 14 studies. Although the reported proportion of enrollees with renal disease varied from 6% to 44%, the definitions for renal disease used by the authors in these 22 trials were extremely heterogeneous. Only 15 (10%) original articles reported a mean or median serum creatinine level, mean creatinine clearance, or eGFR of the enrollees. The reporting of renal disease (P = .91) or baseline renal function (P = .53) did not change over the prespecified time periods.
More than 50% of all trials performed subgroup analyses of 1 or more baseline characteristics (eg, age, hypertension, diabetes), but only 4 (3%) subgroup analyses of treatment stratified by renal disease were performed in the original article.53- 56 Six additional subgroup analyses of treatment stratified by renal disease were reported in subsequent articles resulting from the primary trial.57- 62 In general, in these 10 studies, the absolute risk reduction in mortality was often greater than those with renal disease, while the relative risk reduction was usually similar. Approximately 60% of CVD trials involving the use of RAAS inhibitors reported on adverse events related to the medication, particularly worsening renal dysfunction and hyperkalemia. Three of 31 RAAS antagonist trials54,63,64 reported adverse events in patients with and without underlying renal disease.
Underrepresentation of certain subsets of the population in cardiac trials, such as women, blacks, and the elderly, has been well described.65- 67 Lack of evidence in specific populations can contribute to undertreatment of these cohorts in clinical practice.68,69 Our findings in this systematic review reinforce current concerns about the representation and reporting of renal disease in CVD trials.
We demonstrate that patients with renal disease in clinical trials for CVD were frequently excluded from participation. The exclusion of renal disease from cardiac trials remains high over the past 20 years despite increasing recognition of the prevalence and importance of renal disease in patients with CVD.4- 13,15 Patients with renal disease are more likely to die of CVD than to progress to end-stage renal disease.70 The in-hospital mortality for patients with renal disease presenting with AMI is increased by nearly 5-fold compared with those who have normal GFR,4,7 which translates in absolute terms to an in-hospital mortality of approximately 20% to 30%.5,7 The 1-year absolute mortality for patients with renal disease and CHF is approximately 20%.2 Overall, at 1 year, the odds of dying are approximately 1.5-fold in patients with moderate chronic renal disease (eGFR, 30-60 mL/min per 1.73 m2) and CHF or AMI and greater than 2-fold in patients with severe chronic renal disease (GFR, 15-30 mL/min per 1.73 m2) and CHF or AMI.2,3,6- 8,71
Because patients with renal disease have the highest burden of CVD and the highest risk for recurrent CVD events and death, they may stand to benefit the most from therapies to reduce CVD morbidity and mortality. Unfortunately, in clinical practice, it is more common for these lifesaving therapies to be withheld from patients with renal disease, probably because of concern for adverse events.72- 74 Contrary to this generally applied theory of practice, the benefits of therapy in patients with renal disease may be significant, and those who do not receive therapy can actually suffer more adverse outcomes.72- 74 Although these observational studies suggest the benefits of interventions outweigh the risks, in an era of evidence-based medicine, it is unacceptable that we lack solid evidence from RCTs of therapy in this cohort of patients. This lack of data is highlighted by the relative insignificance of renal disease in current guidelines. For example, in the ACC/AHA Practice Guidelines for chronic CHF, under the section entitled “Patients with heart failure with concomitant disorders,” there are only 2 paragraphs on patients with renal insufficiency.38 In the ACC/AHA Practice Guidelines for ST-elevation MI, it is stated that angiotensin-converting enzyme inhibitors should not be used if “clinically relevant renal failure is present.”39 However, it remains unclear as to what defines “clinically relevant renal failure.” Finally, in the ACC/AHA Practice Guidelines for the management of patients with unstable angina and non–ST-elevation MI, it is stated that “difficult decisions” exist for revascularization of “patients who have moderate or severe renal failure” (without references to studies pertaining to this issue).40 Furthermore, in the Kidney Disease Outcomes Quality Initiative (KDOQI) Clinical Practice Guidelines for Cardiovascular Disease, the level of evidence is C (consensus opinion only, no evidence from RCTs or observational studies) for nearly all pharmacological treatments of acute coronary syndrome and cardiomyopathy in renal disease patients (www.kidney.org/professionals/kdoqi/guidelines.cfm).
In addition to problems with exclusion of patients with renal disease from CVD trials, the paucity of reporting on renal disease as a baseline characteristic is troubling. In 1996, the Consolidated Standards of Reporting Trials (CONSORT) statement was first published.75 CONSORT is published on the Internet (www.consort-statement.org) in several different languages, including Dutch, English, French, German, Japanese, and Spanish. In addition, revised recommendations were published in 2001 in 3 major journals.76- 78 The goal of CONSORT is to improve the quality of reporting of RCTs. CONSORT consists of a 22-item checklist and flow diagram for reporting an RCT. Item number 15 of CONSORT relates to the reporting of “baseline demographic and clinical characteristics of each group,” which “allows readers . . . to judge how relevant the results of a trial might be to a particular patient.” Reporting of renal disease in CVD trials should be commonplace for 3 reasons. The first reason relates to the ease of obtaining information on the kidney function of enrollees. Clearly all patients have baseline serum chemistries, which universally include a serum creatinine concentration. From this, the proportion of patients with a serum creatinine concentration above a prespecified threshold, or the proportion with an eGFR below 60 mL/min per 1.73 m2 (current KDOQI definition for chronic kidney disease), can be easily incorporated into the published report. Calculating the eGFR for a given patient only requires knowledge of serum creatinine concentration, age, race, and sex, all variables that are collected for every patient. The second reason that reporting on renal disease should be routine is because of the high prevalence of renal disease in CVD, as discussed above. Finally, baseline kidney function should be reported because of the impact of renal disease on the recurrence and outcomes of CVD. There has been an explosion of literature on this interrelationship over the past several years.4- 15 Yet according to the results of our study, renal disease is not included in the baseline characteristics more than 90% of the time, unlike other baseline characteristics such as hypertension, hyperlipidemia, angina, previous MI, history of CHF, and smoking status, which are routinely reported.
We strongly urge trialists to universally adopt new standards for reporting of renal disease in CVD trials. First, CVD trials should have specific inclusion/exclusion criteria based on serum creatinine concentration and eGFR. Second, we strongly recommend that both median creatinine and eGFR distributions be presented in the table summarizing baseline characteristics of the participants. Third, if subgroup analyses of known cardiac risk factors or known effect modifiers are performed on the study population, treatment effects stratified by kidney function strata (National Kidney Foundation kidney disease classification) should be included. Finally, studies should present adverse effects by the same kidney function strata. Implementation of these guidelines would aid expert panels who draft consensus guidelines to base treatment recommendations on proven efficacy for patients within certain strata of renal function, rather than solely recommending therapies for all patients who meet inclusion criteria of the representative trials.
There are some limitations to this study. First, this review only included trials from 11 major medical journals. Some important studies of CVD may have been missed because of this prespecified inclusion criteria. However, the majority of large and important CVD trials are published in these major journals and thus we most likely captured the essence of our questions. In this regard, because the proportion of CVD trials that excluded patients with renal disease was similar among each of the major journals, we doubt that the addition of extra journals would have meaningfully changed our results. Second, the true rate of exclusion of renal disease may in fact be higher than what we report. While we attempted to contact all of the primary authors from studies that did not clearly mention a specific cutoff for renal function (102 authors) as a criterion for exclusion in the article, 27 authors did not reply to our request for information. We counted these 27 ambiguous cases as studies that have not excluded renal disease patients. However, because nearly 20% of authors from trials that did not mention renal disease indicated that renal disease was in fact excluded, the lack of responses would only bias our results toward decreased overall exclusion of renal disease than actually occurred. Finally, there were an insufficient number of subgroup analyses of similar types of interventions to allow us to perform meta-analysis and pooled relative risks of these therapies in patients with and without renal disease.
In conclusion, patients with renal disease are often excluded from CVD trials. Furthermore, inadequate data on the distribution and representation of patients with renal disease and on enrollees' baseline renal function are available from primary or secondary reports of these trials. The consequence is that we lack evidence on interventions for this growing, high-risk population. Patients who are at the highest risk for cardiovascular events should not be denied potentially lifesaving therapy a priori without sufficient evidence for harm or benefit. Both the inclusion and reporting of renal disease in large CVD trials must improve or, alternatively, CVD trials need to be designed exclusively for patients with chronic renal disease to accurately assess the risks and benefits in this population and determine optimal treatment strategies. The funding agencies should make these studies a priority to gather evidence in this increasingly important subgroup of patients.
Corresponding Author: Chirag R. Parikh, MD, PhD, Section of Nephrology, VAMC and Yale University, Clinical Epidemiology Research Center, 950 Campbell Ave, Mail Code 151B, Bldg 35 A, Room 219, West Haven, CT 06516 (firstname.lastname@example.org).
Author Contributions: Dr Parikh and Dr Coca had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Coca, Krumholz, Parikh.
Acquisition of data: Coca, Parikh.
Analysis and interpretation of data: Coca, Krumholz, Garg, Parikh.
Drafting of the manuscript: Coca, Parikh.
Critical revision of the manuscript for important intellectual content: Coca, Krumholz, Garg, Parikh.
Statistical analysis: Coca, Parikh.
Obtained funding: Parikh.
Administrative, technical, or material support: Parikh.
Study supervision: Garg, Parikh.
Financial Disclosures: Dr Krumholz is on the advisory board for Amgen, Alere, and United Health Care and has contracts from the American College of Cardiology and the Colorado Foundation for Medical Care.
Funding/Support: Dr Parikh is supported by career development award K23-DK064689 from the National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases. Dr Garg is supported by a Clinician Scientist Award from the Canadian Institutes of Health Research. Dr Krumholz has received grants from the National Institutes of Health, Robert Wood Johnson Foundation, and Fanny Ripple Foundation.
Role of the Sponsor: The funding organizations had no role in the design and conduct of the study; in the collection, management, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.
Acknowledgment: We thank all of the authors of included studies who generously confirmed and provided information for this review.