Contrast-Induced Acute Kidney Injury and Risk of Adverse Clinical Outcomes After Coronary Angiography
A Systematic Review and Meta-Analysis
Background—Contrast-induced acute kidney injury (CI-AKI) has been associated with mortality, although it has been suggested this association may be attributable to confounding. We performed a systematic review and meta-analysis to characterize the associations between CI-AKI and subsequent clinical outcomes.
Methods and Results—We identified studies using MEDLINE (1950 to June 2011) and Embase (1980 to June 2011), manual bibliographic searches, and contact with experts. We included observational studies that characterized outcomes among patients with and without AKI (based on changes in serum creatinine) after coronary angiography. Eligible studies reported at least 1 of mortality, cardiovascular events, end-stage renal disease, or length of hospital stay. Thirty-nine observational studies met inclusion criteria. Of 34 studies reporting mortality (including 139 603 participants), 33 reported an increased risk of death in those with CI-AKI, although the effect size varied between studies (I2=93.5%). Between-study heterogeneity was partially explained by whether adjustment for confounding features was performed (11 studies without adjustment; pooled crude risk ratio, 8.19; 95% confidence interval, 4.30–15.60; I2=77.3% versus 23 studies with adjustment; pooled adjusted risk ratio, 2.39; 95% confidence interval, 1.98–2.90; I2=88.3%). CI-AKI was consistently associated with an increased risk of cardiovascular events in 14 studies, end-stage renal disease in 3 studies, and prolonged hospitalization in 11 studies.
Conclusions—CI-AKI is associated with an increased risk of mortality, cardiovascular events, renal failure, and prolonged hospitalization. However, the association between CI-AKI and mortality is strongly confounded by baseline clinical characteristics that simultaneously predispose to both kidney injury and mortality, and the risk attributable to CI-AKI is much lower than that reported from unadjusted studies.
Acute kidney injury (AKI) after coronary angiography is often attributed to radiocontrast-associated kidney injury,1,2 the third leading cause of AKI in hospitalized patients.3 Patients with preexisting comorbidities, including those with diabetes mellitus, chronic kidney disease (CKD), and heart failure, are at particularly high risk of contrast-induced AKI (CI-AKI).4–6 The primary manifestation is a small decline in kidney function, occurring 1 to 3 days after the procedure.1 Kidney function usually returns to preexisting levels within 7 days,7 and AKI after radiocontrast administration rarely requires acute dialysis treatment.1,8
Several observational studies suggest that these small declines in kidney function after contrast media exposure are associated with adverse clinical outcomes, including longer hospital admission, subsequent cardiovascular events, and increased mortality.2,4–6,9 Although some of these findings have been summarized in narrative reviews,2,6,10 the interpretation of these findings has remained controversial given the correlation between preexisting clinical variables that are associated with both CI-AKI and adverse clinical outcomes, variability in adjustment for potential confounders across observational studies, and uncertainty whether reported risks are indeed attributable to CI-AKI.
To address these knowledge gaps, we conducted a systematic review and meta-analysis of observational studies that examined the association between CI-AKI after coronary angiography and adverse clinical outcomes, including mortality, cardiovascular events, end-stage renal disease (ESRD), and prolongation of hospitalization. Specifically, we sought to clarify the quality of existing studies, and study features that may contribute to the heterogeneity of results across published studies.
WHAT IS KNOWN
Contrast-induced acute kidney injury (CI-AKI) has been associated with high morbidity and mortality, although it remains controversial to what degree this association is confounded by preexisting clinical features associated with both CI-AKI and adverse clinical outcomes.
WHAT THE STUDY ADDS
In this systematic review and meta-analysis, significant associations between CI-AKI and adverse clinical outcomes were observed.
In 11 studies without adjustment for confounding features, CI-AKI was associated with an 8-fold increase in risk of mortality (pooled RR, 8.19; 95% CI, 4.30–15.60); however, this association was attenuated among 23 studies that performed adjustment for confounders (pooled RR, 2.39; 95% CI, 1.98–2.90) and publication bias (RR, 1.79; 95% CI, 1.47–2.18).
These findings suggest that the association between CI-AKI and death is related to baseline clinical characteristics and that the risk attributable to CI-AKI is lower than that implied from unadjusted analyses.
We adhered to the Meta-analysis of Observational Studies in Epidemiology (MOOSE) guidelines11 and followed a prespecified study protocol.
Data Sources and Searches
We systematically searched MEDLINE (1950 to June 2011) and Embase (1980 to June 2011) for studies describing the association between CI-AKI (identified based on changes in serum creatinine concentration) and death, cardiovascular events (including cardiovascular mortality, myocardial infarction, target vessel reocclusion or need for revascularization, cerebrovascular accident or heart failure), chronic dialysis, or ESRD, and length of hospital stay among patients undergoing coronary angiography. We also searched the reference lists of all identified relevant publications and contacted experts in coronary angiography and AKI. We limited inclusion to studies published in English.
Three search themes were combined using the Boolean operator and. The first theme, coronary angiography, combined exploded versions of Medical Subject Headings angiography, contrast media, angiocardiography, heart catheterization, angioplasty, transluminal, percutaneous coronary angioplasty, or myocardial revascularization, or text words coronary angiography, cardiac catheterization, percutaneous coronary intervention, PCI, angiography, coronary revascularization, or cardiac angiography. The second theme, combined exploded version of the Medical Subject Headings terms acute kidney failure or creatinine or text words acute kidney injury, acute kidney failure, acute renal failure, acute renal insufficiency nephropathy, contrast nephropathy, or contrast-induced nephropathy. We used the approach of Egger et al12 to identify studies with an observational design.
Two reviewers independently identified potentially eligible articles by performing an initial screen of titles and abstracts. Articles were considered for inclusion if they reported data from an original study (review articles were excluded) and reported on clinical outcomes according to CI-AKI status after diagnostic or therapeutic coronary angiography. We used broad inclusion criteria for studies, including varying definitions for AKI data and information on any clinical outcomes as they were defined by the primary studies. Articles were retained when either of the reviewers believed that it should be retained or when there was uncertainty as to eligibility based on title and abstract alone.
Selected articles were subsequently screened based on a full-text review. To be included, studies had to be observational studies of participants following diagnostic or interventional coronary angiography, with a comparison between those with CI-AKI (based on a relative or absolute change in serum creatinine) and those without CI-AKI. Studies that reported outcomes only on the basis of contrast volume or as a ratio of contrast dose to serum creatinine were not included. We included any study reporting on 1 or more of mortality, cardiovascular events (including including cardiovascular mortality, myocardial infarction, target vessel reocclusion or need for revascularization, stroke, or heart failure), ESRD (including chronic dialysis), or length of hospital stay. When >1 publication was identified from the same cohort examining the same study outcome, we included data from the article with the larger sample size. Information from randomized trials was included where publications reported associations between CI-AKI and outcomes of interest.
Data Extraction and Quality Assessment
Two reviewers independently extracted data on baseline patient characteristics, procedural characteristics, criteria to define CI-AKI, and duration of follow-up. We also collected data on methodological features indicative of study quality, following the MOOSE guidelines.11 These included specification of the inclusion/exclusion criteria, the inclusion of consecutive participants in the cohort, losses to follow-up <10% or appropriate handling of losses to follow-up, blinding of exposure status for outcome assessment, and statistical adjustment for confounders. The exposure variable of interest was CI-AKI, and the reference group was those without CI-AKI in each study. Most studies (n=36) identified dichotomous groups (with CI-AKI versus without CI-AKI); however 3 studies categorized severity of CI-AKI based on magnitude of the change in serum creatinine, and 1 study categorized CI-AKI based on whether it persisted for >7 days. To enable pooling, we combined results for these additional categories within a single exposure group (with CI-AKI), and performed meta-regression and subgroup analyses, according to the serum creatinine criteria used to define groups with CI-AKI (ie, increase in serum creatinine concentration >25% or 0.5 mg/dL versus >50% or 1.0 m/dL) in each study.
The primary outcome was all-cause mortality. Secondary outcomes included cardiovascular events, ESRD, and length of hospital stay. We considered major adverse cardiovascular events to include cardiovascular mortality, myocardial infarction, target vessel reocclusion or need for revascularization, stroke, heart failure, or a composite of these events. We defined ESRD as the requirement for chronic dialysis after hospital discharge. Length of stay was defined based on the number of days from either hospital admission or angiography to discharge, depending on the study design. The duration of follow-up for clinical outcomes varied across studies. We grouped studies on the basis of short-term (in-hospital or at 30 days) or long-term follow-up (postdischarge or ≥6 months) and performed meta-regression and subgroup analysis on the basis of this distinction.
We recorded risk ratios (RR), hazard ratios, or odds ratios (OR) for quantifying the association between CI-AKI and the dichotomous clinical outcomes of interest (mortality, major adverse cardiovascular events, and ESRD), and the means and SDs for measured continuous outcomes (days in hospital) for patients with CI-AKI compared with those without CI-AKI. Adjusted values were obtained wherever reported. We collected crude values if adjusted results were not presented.
Data Synthesis and Analysis
We pooled RRs for dichotomous outcomes, and means for continuous outcomes across studies. To transform ORs to RRs, we used the formula RR=OR/[(1-P0)+(P0×OR)], where P0 is the incidence of the outcome of interest in the unexposed group.13 Estimates of variance of ORs were converted to RRs using the Miettinen test-based approach; variance lnRR=variance lnOR×(lnRR/lnOR).14 We pooled the natural logarithm of the RRs of binary outcomes and determined the weighted mean difference of continuous outcomes using the random effects model of DerSimonian and Laird.15 We used the Cochrane Q statistic (at a significance level of P<0.10), and the I2 statistic to assess for heterogeneity across studies.16,17 Subgroup analyses, stratified by study population characteristics and study methodology criteria, were also performed. We evaluated funnel plots and used Begg test to detect small study effects suggestive of publication bias.18,19 We used the Duval and Tweedie20 nonparametric trim-and-fill procedure to determine the possible effect of publication bias on pooled estimates by imputing the estimate of effect of hypothetical missing studies, and imputing a pooled estimate that included these studies. All analyses were performed in Stata version 11 (StataCorp, College Station, Texas) using the metan, metareg, metabias, and metatrim commands.
Our search strategy yielded 4142 unique citations. We excluded 4058 citations based on screening of title and abstract, leaving 84 articles for full-text review. We subsequently excluded 45 studies that did not meet inclusion criteria; of which 14 articles comprised overlapping cohorts of patients and were excluded to avoid duplicate inclusion of data (Figure 1). There was good agreement between reviewers on the final articles eligible for inclusion (κ=0.824).
Characteristics of the 39 studies included in the systematic review are provided in Tables I and II in the online-only Data Supplement. Studies were published between 1990 and 2011, and 19 were from North America, 13 from Europe, 5 from Asia, and 2 from Israel. The number of participants ranged from 78 to 27 608 (152 459 participants in total), and the mean age ranged from 56.4 to 75.4 years across studies. Most studies included patients with and without CKD at baseline (range 3.2%–69.6% according to individual study definitions of CKD), although 4 studies included only patients with impaired baseline kidney function, and 2 studies excluded patients with elevated serum creatinine at baseline. Twenty-six studies included only patients receiving percutaneous coronary interventions (including 8 studies of patients receiving primary percutaneous intervention for ST segment elevation myocardial infarction), whereas 9 studies included patients receiving diagnostic coronary angiography.
The definition of CI-AKI was based on a relatively small increase in serum creatinine in all studies; 32 studies identified CI-AKI on the basis of an increase in serum creatinine concentration from baseline exceeding either 25% or 0.5 mg/dL, whereas 5 studies identified CI-AKI based on a >50% or 1.0 mg/dL increase in serum creatinine concentration, and 2 studies categorized the severity of CI-AKI based on the magnitude of this increase (Table I in the online-only Data Supplement). The duration of follow-up varied among studies, with 12 studies using follow-up to hospital discharge, 2 studies with follow-up 30 days postprocedure, and 25 studies with long-term follow-up ranging between 6 months and 5 years.
Inclusion and exclusion criteria were clearly specified in all but 2 studies, and all but 2 studies enrolled consecutive patients in the cohort. There were no losses to follow-up in 32 studies, 2 studies did not report on losses to follow-up, and 6 studies reported losses to follow-up ranging from <1% to 36% (Table II in the online-only Data Supplement). Study personnel who evaluated outcomes were blinded to exposure status in only 1 study.
RRs of mortality were adjusted for baseline severity of illness variables that may be confounders in 23 studies, whereas only RRs of mortality that were unadjusted were reported in 11 studies (Table II in the online-only Data Supplement). One study did not contribute to the pooled analysis for mortality because no deaths were recorded. RRs of major adverse cardiovascular events were adjusted for potential confounders in 9 studies, but only unadjusted RRs were reported in 5 studies (Table II in the online-only Data Supplement). Cardiovascular events included cardiovascular mortality from 4 studies, myocardial infarction from 10 studies, coronary artery reocclusion or revascularization from 7 studies, heart failure from 2 studies, and stroke from 2 studies. Only 1 study reported the adjusted RR of ESRD, and 2 studies provided only data to determine the unadjusted RR. Ten studies reported the mean length of hospital stay without adjustment for other baseline severity of illness factors; however, only 1 study reported the adjusted increase in length of hospital stay associated with CI-AKI after adjustment for potential confounders. Adjustment was performed for age in 24 studies, diabetes mellitus in 20 studies, severity of coronary artery disease in 14 studies, heart failure (based on left ventricular function, pulmonary edema, or cardiogenic shock) in 24 studies, and baseline kidney function (serum creatinine or estimated glomerular filtration rate) in 15 studies (Table II in the online-only Data Supplement).
Association Between CI-AKI and Mortality
Results from 34 studies examining mortality, including a total of 139 603 participants, showed evidence of statistical heterogeneity (Q statistic P<0.001; I2=93.5%). However, 33 of the 34 studies reported an increased risk of death in those with CI-AKI after coronary angiography. RR from studies that did not adjust for baseline severity of illness variables were significantly greater than those obtained from studies that adjusted estimates for these potential confounders (meta-regression, P<0.001). Based on 11 studies (with 27 190 participants) that reported unadjusted results, the pooled crude RR of death was 8.19 (95% CI, 4.30–15.60; Q statistic P=0.008; I2=77.3%), whereas the pooled adjusted RR from 23 studies (with 112 413 participants) with adjusted results was 2.39 (95% CI, 1.98–2.90; Q statistic P<0.001; I2=88.3%; Figure 2).
We performed further meta-regression to explore reasons for remaining heterogeneity across the 23 studies that reported adjusted RRs. Studies with short follow-up (to hospital discharge or 30 days postprocedure) reported higher adjusted RRs of death associated with CI-AKI than studies with longer (≥6 month) follow-up (Table III in the online-only Data Supplement). Different definitions of CI-AKI (increase in serum creatinine concentration >50% or 1.0 mg/dL versus >25% or 0.5 mg/dL) did not influence the risk of mortality associated with AKI, although both studies that categorized the severity of AKI reported higher risks of mortality with larger increases in serum creatinine concentration.9,21 Results from studies without substantial (<10%) losses to follow-up (18 studies; 104 825 participants; RR, 2.28; 95% CI, 1.86–2.80; Q statistic P<0.001; I2=90.4%) did not significantly differ from those with >10% of participants lost to follow-up or no reporting of losses to follow-up. Results were also similar regardless of differences in the study distributions of age, diabetes mellitus, presence or absence of CKD before angiography, ST segment elevation myocardial infarction, and type of procedure (percutaneous coronary intervention only versus diagnostic or interventional procedure).
The funnel plot for studies reporting adjusted RRs for mortality was asymmetrical (Figure IA in the online-only Data Supplement). The possibility of publication bias was further suggested by Egger test (z=1.74; P=0.034). When the trim-and-fill procedure was used to impute results for hypothesized unpublished studies20 (Figure IB in the online-only Data Supplement), the pooled adjusted RR was attenuated but continued to show a significant association between AKI and mortality (RR, 1.79; 95% CI, 1.47–2.18).
Association Between CI-AKI and Cardiovascular Events, ESRD, and Length of Hospitalization
Of 14 studies (70 031 participants) reporting on cardiovascular events, all reported an increased risk associated with CI-AKI after coronary angiography (Figure 3). The pooled RR from these studies for cardiovascular events was 2.42 (95% CI, 1.62–3.64), although there was again evidence of statistical heterogeneity (Q statistic P<0.001; I2=96.0%). The reporting of crude versus adjusted RRs did not significantly alter estimates of risk of cardiovascular events associated with CI-AKI (meta-regression P=0.745). Based on 6 studies (with 43 959 participants) that reported unadjusted results, the pooled crude RR of cardiovascular events was 2.59 (95% CI, 1.05–6.27; Q statistic P<0.001; I2=98.3%), whereas the pooled adjusted RR from 8 studies (with 26 072 participants) with adjusted results was 1.98 (95% CI, 1.52–2.59; Q statistic P=0.001; I2=71.1%). Neither study design features nor study population characteristics further explained heterogeneity (P<0.10 for all analyses). There was no evidence of funnel plot asymmetry among studies evaluating cardiovascular events (Egger test, z=1.81; P=0.225), suggesting no publication bias was present.
Three studies (18 457 participants) reported on the risk of progression to ESRD, which ranged from 0% to 0.2% in those without CI-AKI, and from 0.2% to 4.5% in those with CI-AKI. CI-AKI was associated with ESRD without evidence of heterogeneity (Figure 3; Q statistic P=0.804; I2=0%). Two studies reported only unadjusted risks of ESRD (pooled crude RR, 15.26; 95% CI, 1.86–125.01), whereas 1 study reported adjusted results (adjusted RR, 6.95; 95% CI, 2.51–19.26). All studies examining length of hospital stay reported longer admissions in patients with CI-AKI compared with those without AKI. Ten studies (19 674 participants) reported an unadjusted mean length of hospital stay that ranged from 0.5 to 8.3 days longer for participants with AKI, although there was heterogeneity in the size of this difference (Q statistic P<0.001; I2=99.2%), with substantial variability in length of hospitalization between studies (Figure II in the online-only Data Supplement). Only 1 study reported an increase in length of hospital stay that was adjusted for baseline severity of illness variables, and this difference was equivalent to 1.6 additional days of hospitalization attributable to CI-AKI (P=0.005).
We identified several observational studies examining the risks of mortality, cardiovascular events, and kidney failure associated with CI-AKI after coronary angiography. CI-AKI was associated with a pooled 8-fold increase in the risk of death after coronary angiography among studies without adjustment. Among studies that adjusted for baseline clinical characteristics that simultaneously predispose to both CI-AKI and mortality, CI-AKI remained independently predictive of long-term mortality, but with a reduced pooled RR of about 2. Our results show that although CI-AKI has been consistently associated with mortality after coronary angiography, the strength of this association has varied between studies, and that the risk associated with CI-AKI in adjusted studies is a fraction of the overall risk reported from unadjusted observational studies. These results suggest that the independent association between CI-AKI and mortality is more modest, and further attenuated after accounting for publication bias, resulting in a remaining independent 79% increase in the adjusted risk of death associated with CI-AKI in our meta-analysis.
Even with adjustment, observational studies remain inherently limited by the potential for residual confounding and cannot prove a causal relationship between CI-AKI and death. We did observe features in-keeping with a causal relationship including a temporal relationship between CI-AKI and adverse clinical outcomes after coronary angiography. Furthermore, the associations between CI-AKI and mortality were similar, regardless of the definition used for CI-AKI and, in 2 studies that categorized CI-AKI severity according to the magnitude of serum creatinine increase after coronary angiography, larger changes in creatinine, representative of more severe AKI, were associated with a higher risk of death.9,21 Such a dose–response relationship with risk of death has also been seen in other studies of AKI outside the setting of coronary angigraphy.22–24 However, the biological mechanism by which CI-AKI may lead to death remains unclear. Severe forms of AKI (that often require dialysis) could predispose to early mortality after coronary angiography because of volume overload, electrolyte disturbances, or uremia; however, it is less clear how relatively small changes in kidney function (that subsequently resolve) might increase this risk. It has been hypothesized that patients who develop CI-AKI after radiocontrast exposure are treated more conservatively so as to preserve remaining kidney function.25,26 Further research is needed to delineate how patients who develop CI-AKI are subsequently managed, including studies to examine whether there are any disparities in subsequent use of cardiovascular therapies.
Because these observational studies are limited by the potential for residual confounding, it also remains possible that CI-AKI is a marker of severity of illness that accompanies hemodynamic instability and ischemia, rather than a causal agent itself. Thus, there is a need for experimental studies to determine whether CI-AKI prevention strategies also lead to improved survival. Randomized trials of prevention strategies for CI-AKI will require large sample sizes to achieve adequate power to detect differences in clinical outcomes and are best suited to determining the efficacy of specific interventions in this setting. Reductions in the rates of clinical outcomes, such as death, cardiovascular events, and ESRD in the intervention arms of such trials, would be important to demonstrate, although interventions could reduce the risk of clinical outcomes via mechanisms separate from prevention of CI-AKI.
We identified several studies that examined cardiovascular outcomes after coronary angiography and found that CI-AKI was also associated with an increased risk of these events. The risks of coronary vascular disease and cardiovascular events associated with CKD are well established27,28 and are thought to involve mechanisms, including accelerated atherosclerosis, vascular calcification, left-ventricular hypertrophy, and sudden death.29,30 It remains unclear whether CI-AKI is associated with these processes or whether the relationship between CI-AKI and these cardiovascular events is mediated through preexisting cardiovascular risk factors or CKD.28,31
The effect of CI-AKI on long-term renal outcomes after angiography has been controversial. Most studies have reported resolution of renal impairment within days in most patients,7 although some observational data have suggested an increased risk of ESRD associated with radiocontrast media exposure.32 We identified 3 studies that reported an increased risk of progression to ESRD in participants with CI-AKI, and even small changes in kidney function that defined CI-AKI in this systematic review were a prognostic factor for an increased long-term risk of renal failure. Further research is needed to determine whether progressive CKD mediates the increased long-term risks of cardiovascular events and death associated with CI-AKI.
The results of this systematic review should be interpreted with some caveats. First, we observed substantial statistical heterogeneity in the magnitude of effect sizes between studies for several outcomes. For differences in the length of hospital stay, this heterogeneity may be explained by differences in clinical practices between institutions, whereas, for all outcomes, variations in study design (including the nature of adjustment for confounders and the length of follow-up) appeared to explain some of this variation. Nonetheless, despite quantitative differences in the magnitude of risk, the qualitative findings from our systematic review were consistent for all outcomes examined. A further limitation was evidence of publication bias detected among studies that reported on mortality. However, we found that the relation between CI-AKI and risk of death remained after applying imputation techniques to adjust for potential unpublished studies, suggesting that such bias is unlikely to completely explain this association. Finally, our systematic review included observational studies that remain vulnerable to residual confounding, even when adjustment has been performed.
In conclusion, this systematic review demonstrates that CI-AKI is consistently associated with mortality, cardiovascular events, kidney failure, and prolongation of hospital stay. However, there is substantial attenuation of the relative risk of death associated with CI-AKI among studies that adjusted for baseline clinical characteristics that simultaneously predispose to both CI-AKI and mortality. This meta-analysis suggests that the relationships between CI-AKI and subsequent clinical outcomes are substantially influenced by confounding, and that the risk of these outcomes attributable to CI-AKI appears to be lower than has been widely emphasized. These findings highlight important caveats in interpreting the results of prognostic studies of CI-AKI.
Sources of Funding
Dr James was supported by a Kidney Research Scientist Core Education and National Training (KRESCENT) fellowship and an Alberta Innovates Health Solutions (AIHS) Fellowship Award. M. A. Manning. was supported by an AIHS Summer Studentship Award. Dr Samuel was supported by a joint Canadian Child Health Clinician Scientist Program Career Development Award and KRESCENT New Investigator Award. Dr Tonelli and Dr Ghali were supported by a Health Scholar Award from AIHS. Dr Knudtson received part support from the Libin Trust Fund. Dr Hemmelgarn was supported by a Population Health Investigator Award from AIHS. The authors have no financial interests in the information contained in this article. Sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the article.
Dr Knudtson has received honoraria for presentations from the Canadian Cardiovascular Society and Medtronic.
The online-only Data Supplement is available at http://circinterventions.ahajournals.org/lookup/suppl/doi:10.1161/CIRCINTERVENTIONS.112.974493/-/DC1.
- Received May 4, 2012.
- Accepted November 19, 2012.
- © 2013 American Heart Association, Inc.
- Finn WF
- Mehran R,
- Aymong ED,
- Nikolsky E,
- Lasic Z,
- Iakovou I,
- Fahy M,
- Mintz GS,
- Lansky AJ,
- Moses JW,
- Stone GW,
- Leon MB,
- Dangas G
- Weisbord SD,
- Chen H,
- Stone RA,
- Kip KE,
- Fine MJ,
- Saul MI,
- Palevsky PM
- Stroup DF,
- Berlin JA,
- Morton SC,
- Olkin I,
- Williamson GD,
- Rennie D,
- Moher D,
- Becker BJ,
- Sipe TA,
- Thacker SB
- Egger M,
- Davey Smith G,
- O'Rourke K
- Miettinen O
- Higgins JP,
- Thompson SG,
- Deeks JJ,
- Altman DG
- Egger M,
- Davey Smith G,
- Schneider M,
- Minder C
- James MT,
- Ghali WA,
- Knudtson ML,
- Ravani P,
- Tonelli M,
- Faris P,
- Pannu N,
- Manns BJ,
- Klarenbach SW,
- Hemmelgarn BR
- Newsome BB,
- Warnock DG,
- McClellan WM,
- Herzog CA,
- Kiefe CI,
- Eggers PW,
- Allison JJ
- Chertow GM,
- Normand SL,
- McNeil BJ
- Tonelli M,
- Wiebe N,
- Culleton B,
- House A,
- Rabbat C,
- Fok M,
- McAlister F,
- Garg AX
- Anavekar NS,
- McMurray JJ,
- Velazquez EJ,
- Solomon SD,
- Kober L,
- Rouleau JL,
- White HD,
- Nordlander R,
- Maggioni A,
- Dickstein K,
- Zelenkofske S,
- Leimberger JD,
- Califf RM,
- Pfeffer MA