Chronic Kidney Disease and the Risks of Death, Cardiovascular Events, and Hospitalization
http://www.100md.com
《新英格兰医药杂志》
ABSTRACT
Background End-stage renal disease substantially increases the risks of death, cardiovascular disease, and use of specialized health care, but the effects of less severe kidney dysfunction on these outcomes are less well defined.
Methods We estimated the longitudinal glomerular filtration rate (GFR) among 1,120,295 adults within a large, integrated system of health care delivery in whom serum creatinine had been measured between 1996 and 2000 and who had not undergone dialysis or kidney transplantation. We examined the multivariable association between the estimated GFR and the risks of death, cardiovascular events, and hospitalization.
Results The median follow-up was 2.84 years, the mean age was 52 years, and 55 percent of the group were women. After adjustment, the risk of death increased as the GFR decreased below 60 ml per minute per 1.73 m2 of body-surface area: the adjusted hazard ratio for death was 1.2 with an estimated GFR of 45 to 59 ml per minute per 1.73 m2 (95 percent confidence interval, 1.1 to 1.2), 1.8 with an estimated GFR of 30 to 44 ml per minute per 1.73 m2 (95 percent confidence interval, 1.7 to 1.9), 3.2 with an estimated GFR of 15 to 29 ml per minute per 1.73 m2 (95 percent confidence interval, 3.1 to 3.4), and 5.9 with an estimated GFR of less than 15 ml per minute per 1.73 m2 (95 percent confidence interval, 5.4 to 6.5). The adjusted hazard ratio for cardiovascular events also increased inversely with the estimated GFR: 1.4 (95 percent confidence interval, 1.4 to 1.5), 2.0 (95 percent confidence interval, 1.9 to 2.1), 2.8 (95 percent confidence interval, 2.6 to 2.9), and 3.4 (95 percent confidence interval, 3.1 to 3.8), respectively. The adjusted risk of hospitalization with a reduced estimated GFR followed a similar pattern.
Conclusions An independent, graded association was observed between a reduced estimated GFR and the risk of death, cardiovascular events, and hospitalization in a large, community-based population. These findings highlight the clinical and public health importance of chronic renal insufficiency.
More than 400,000 Americans have end-stage renal disease, and over 300,000 of these patients require maintenance dialysis.1 Mortality rates remain above 20 percent per year with the use of dialysis, with more than half of the deaths related to cardiovascular disease. The annual direct medical costs for end-stage renal disease are nearly $23 billion.1 Although an estimated 8 million adults in the United States have chronic kidney disease of at least stage 3 (as defined by an estimated glomerular filtration rate of less than 60 ml per minute per 1.73 m2 of body-surface area),2 less is known about the rates of death, cardiovascular disease, and resource use among persons with a reduced estimated GFR who are not yet receiving maintenance dialysis.
Several, but not all, previous studies suggested that mild-to-moderate elevations in serum creatinine levels are associated with increased rates of death from any cause3,4,5,6,7,8,9 and from cardiovascular causes,5,7,10,11,12,13 but whether chronic kidney disease independently increases the risk of any type of cardiovascular disease has not been established.4,5,6,7,14 Furthermore, previous studies have been limited by the inclusion of relatively small numbers of persons with kidney disease,3,4,5,6,7,8,11,12,13,14,15,16,17 the use of dichotomous groups of estimated kidney function,3,4,5,6,7,8,15 the use of the serum creatinine level alone as a proxy for GFR and nonuniform cutoff values to define kidney disease,3,4,5,6,7,8,9,10,11,15 lack of information on longitudinal changes in GFR and coexisting conditions,3,4,5,6,7,8,9,10,11,14,15,16,17 selected populations,6,8,11,14,16 and populations with limited racial or ethnic diversity. In addition, few studies have investigated the association between chronic kidney disease and the risk of hospitalization,18 which has important economic implications.
Using longitudinal measures of estimated GFR, a more accurate method of assessing kidney function than the measurement of serum creatinine alone, we examined the effect of the severity of kidney dysfunction on the risks of death, cardiovascular events, and hospitalization among a large, diverse group of adults. We hypothesized that there would be a graded, independent association between the estimated GFR and the risks of these outcomes.
Methods
Study Sample and Measures of Kidney Function
The Kaiser Permanente Renal Registry included all adult members (20 years of age or older) of Kaiser Permanente of Northern California, a large integrated health care system insuring more than 35 percent of the adult population of the San Francisco Bay Area, whose kidney function was known. To be eligible for the registry, the subjects had to have had one or more outpatient determinations of serum creatinine levels recorded in a health-plan laboratory database between January 1, 1996, and December 31, 2000. We excluded subjects who had already received a kidney transplant or who were receiving maintenance dialysis at entry. Given the nature of the study, the institutional review board of the Kaiser Foundation Research Institute determined that informed consent was not required.
We used the abbreviated Modification of Diet in Renal Disease (MDRD) equation to estimate the GFR.19,20 We calibrated the measurement of serum creatinine by the Kaiser regional laboratory against that of the MDRD core laboratory.21 The date of the first measurement of GFR during the study period was considered the subject's index date (baseline). Changes in GFR during follow-up were estimated from serum creatinine determinations not associated with hospitalizations to reflect more accurately stable estimates of kidney function. We used a modified National Kidney Foundation classification of chronic kidney disease,22 which classifies estimated GFR in the following ranges: at least 60 ml per minute per 1.73 m2, 45 to 59 ml per minute per 1.73 m2 (stage 3a), 30 to 44 ml per minute per 1.73 m2 (stage 3b), 15 to 29 ml per minute per 1.73 m2 (stage 4), and less than 15 ml per minute per 1.73 m2 (stage 5).
Characteristics of the Subjects
Data on age, sex, and racial or ethnic group were obtained from health-plan databases. All names and identifiers were removed before any data were analyzed, according to procedures approved by the institutional review board at Kaiser.
We identified coexisting illnesses using validated methods23,24,25,26 based on health-plan databases for hospitalization-discharge diagnoses, ambulatory diagnoses, laboratory results, and medication prescriptions, as well as regional cancer-registry data24 (diagnostic criteria are listed in the Appendix). These diagnoses included coronary disease, stroke or transient ischemic attack, heart failure, peripheral arterial disease, diabetes mellitus, hypertension, dyslipidemia, lung or liver disease, cancer, and dementia. We also evaluated laboratory-database entries for serum albumin to identify subjects with a value of 3.5 g per deciliter or less. The presence of proteinuria was based on laboratory-database entries of urine dipstick results of 1+ or greater (approximately 30 mg per deciliter or greater) in the absence of a possible urinary tract infection (i.e., concomitant positive test for urinary nitrite or esterase).
Socioeconomic status was assigned with the use of 2000 U.S. Census-block data, which generally correspond to city blocks or neighborhoods.28 Subjects with a low income were defined as those living in a Census-block group with a median annual household income under $35,000, and those with a low level of education were categorized as those living in a Census-block group in which more than 25 percent of residents older than 25 years had less than a 12th-grade education.
Outcomes
Data on subjects were censored if they underwent kidney transplantation or disenrolled from the health plan, which was defined as a continuous gap in membership of more than 90 days without the interim use of services. Incident end-stage renal disease, defined by the receipt of maintenance dialysis or a kidney transplant, was identified from a comprehensive health-plan registry.29 The primary outcomes of interest included death from any cause, cardiovascular events, and hospitalizations through December 31, 2000. Death was identified from a search of health-plan databases and the California death registry.30 A cardiovascular event was defined as hospitalization for coronary disease, heart failure, stroke, or peripheral arterial disease (see the Appendix).
Statistical Analysis
Event rates were directly adjusted for age with the use of the age distribution of the adult source population and are presented as the number of events per 100 person-years, with 95 percent confidence intervals. To evaluate the independent effect of the estimated GFR on outcomes, we used Cox proportional-hazards models with time-dependent covariates for changing GFR and coexisting illnesses. All variables known to be associated with either the estimated GFR or the outcomes were included in the final models, along with any variables associated with a reduced estimated GFR (i.e., less than 60 ml per minute per 1.73 m2) in univariate analyses with a P value of less than 0.01. Age was entered as a categorical variable (20 to 49, 50 to 59, 60 to 69, 70 to 79, and 80 years or older), with other covariates included as dichotomous variables. For recurrent outcomes of cardiovascular events and hospitalizations, we used a "sandwich" estimate of the variance–covariance matrix to obtain standard errors accommodating the clustering of observations on subjects.31 In the model for death, we did not include cardiovascular events or hospitalizations that occurred after the index date, since they were hypothesized to be part of the pathway by which reduced GFR may increase the risk of death.
Each subject and time was assigned an estimated GFR with the use of the last-value-carried-forward method. Given the varying numbers and spacing of measurements of GFR for each subject, this approach may preferentially attribute outcome events to higher levels of GFR. To address this possibility we analyzed a subgroup of subjects who had had regular serum creatinine determinations spaced between 1 and 14 months apart throughout follow-up. Also, although age and sex are important predictors of adverse events and are incorporated into the MDRD equation, we found no interactions among age, sex, and GFR and thus present only the main-model results.
All analyses were conducted with the use of SAS software (version 8.2). The institutional review board of each collaborating institution approved the study.
Results
Baseline Characteristics
We identified 1,120,295 adults who had had one or more outpatient measurements of serum creatinine, had not previously received dialysis or a kidney transplant, and were alive on the index date. The median number of outpatient measurements of serum creatinine per subject during follow-up was three (interquartile range, one to four).
Subjects with a low estimated GFR at baseline were older than those with an estimated GFR of at least 60 ml per minute per 1.73 m2, and there was greater minority-group representation among subjects with a low estimated GFR (Table 1). As compared with the group with an estimated GFR of at least 60 ml per minute per 1.73 m2, the groups with a reduced estimated GFR also had a higher prevalence of prior cardiovascular disease, proteinuria, diabetes, hypertension, a serum albumin level of 3.5 g per deciliter or less, prior hospitalizations, and other coexisting illnesses (Table 1). Among the 60.7 percent of subjects who had one or more urine dipstick determinations at any time during follow-up, 17.6 percent were found to have proteinuria. Among the 21.2 percent of subjects who had one or more serum albumin determinations at any time during follow-up, 27.8 percent were found to have reduced levels.
Table 1. Baseline Characteristics of 1,120,295 Ambulatory Adults, According to the Estimated GFR at Baseline.
Outcomes According to the Estimated GFR
The median follow-up among the 1,120,295 subjects was 2.84 years (interquartile range, 1.65 to 4.01), which amounts to 3,132,192 person-years. Overall, 21.8 percent of subjects disenrolled during follow-up, but those who disenrolled were substantially younger, were relatively unlikely to have a reduced estimated GFR, and had fewer coexisting illnesses than those who remained enrolled (data not shown). The vital status was complete for all subjects.
During follow-up, 3171 (0.28 percent) subjects began maintenance dialysis and 329 (0.03 percent) underwent kidney transplantation. There were 51,424 deaths, 138,291 cardiovascular events, and 554,651 hospitalizations. Age-standardized rates of death, cardiovascular events, and hospitalization increased substantially with progressively lower estimated GFRs (Figure 1).
Figure 1. Age-Standardized Rates of Death from Any Cause (Panel A), Cardiovascular Events (Panel B), and Hospitalization (Panel C), According to the Estimated GFR among 1,120,295 Ambulatory Adults.
A cardiovascular event was defined as hospitalization for coronary heart disease, heart failure, ischemic stroke, and peripheral arterial disease. Error bars represent 95 percent confidence intervals. The rate of events is listed above each bar.
The group of subjects with an estimated GFR of at least 60 ml per minute per 1.73 m2 was used as the reference group in the analysis of the association between the level of the estimated GFR and each outcome. After adjustment for differences in sociodemographic characteristics and the presence or absence of prior cardiovascular disease, prior hospitalizations, diabetes, hypertension, dyslipidemia, lung or liver disease, cancer, a serum albumin level of 3.5 g per deciliter or less, dementia, proteinuria, and the initiation of dialysis during follow-up, the risk of death from any cause increased sharply as the estimated GFR declined, ranging from a 17 percent increase in risk with an estimated GFR of 45 to 59 ml per minute per 1.73 m2 to a nearly 600 percent increase with an estimated GFR of less than 15 ml per minute per 1.73 m2 (Table 2). The adjusted risk of any cardiovascular event also increased as the estimated GFR decreased, ranging from a 43 percent increase in risk with an estimated GFR of 45 to 59 ml per minute per 1.73 m2 to a 343 percent increase with an estimated GFR of less than 15 ml per minute per 1.73 m2 (Table 2). Finally, the adjusted risk of hospitalization increased as the estimated GFR decreased, ranging from an increase of 14 percent with an estimated GFR of 45 to 59 ml per minute per 1.73 m2 to an increase of 315 percent with an estimated GFR of less than 15 ml per minute per 1.73 m2 (Table 2). The presence of documented proteinuria was also an independent predictor of death (adjusted hazard ratio, 1.3; 95 percent confidence interval, 1.3 to 1.4), cardiovascular events (adjusted hazard ratio, 1.3; 95 percent confidence interval, 1.2 to 1.3), and hospitalization (adjusted hazard ratio, 1.4; 95 percent confidence interval, 1.4 to 1.4).
Table 2. Adjusted Hazard Ratio for Death from Any Cause, Cardiovascular Events, and Hospitalization among 1,120,295 Ambulatory Adults, According to the Estimated GFR.
In a subgroup of 172,144 subjects who had regular measurements of serum creatinine during follow-up (mean number of measurements, 3.3±2.1; mean interval between measurements, 7.5±3.4 months), the adjusted risks of adverse outcomes differed materially from those for the entire cohort only among the subjects with an estimated GFR of 45 to 59 ml per minute per 1.73 m2. In the analysis of the subgroup itself, as compared with the subjects with an estimated GFR of at least 60 ml per minute per 1.73 m2, the subjects with an estimated GFR of 45 to 59 ml per minute per 1.73 m2 had a similar adjusted hazard ratio for death (1.0; 95 percent confidence interval, 1.0 to 1.1) and hospitalization (1.0; 95 percent confidence interval, 1.0 to 1.0). The adjusted risk of cardiovascular events was also attenuated (1.2; 95 percent confidence interval, 1.1 to 1.3).
Discussion
Among a large, diverse population of adults, a reduced estimated GFR was associated with increased risks of death, cardiovascular events, and hospitalization that were independent of known risk factors, a history of cardiovascular disease, and the presence of documented proteinuria. Our study demonstrates that these graded risks of adverse events rose sharply for subjects with an estimated GFR of less than 45 ml per minute per 1.73 m2 for each outcome examined both in the overall cohort and in subgroup analyses. Furthermore, in the cohort as a whole, the absolute rates of these outcomes were considerably higher than the risk of end-stage renal disease.
There has been rapidly growing interest in the relation between kidney disease and the risk of death and cardiovascular disease. With recognition that the presence of chronic kidney disease that does not necessitate dialysis is of considerable importance, several studies have examined the association of different cutoff values of serum creatinine with the risks of death from any cause, death from cardiovascular causes, and cardiovascular events, and most,3,4,6,7,8,9,10,11,12,15 but not all,5,13 of these studies have found increased risks with higher serum creatinine levels. Others have noted high rates of hospitalization among persons with elevated serum creatinine levels.18 However, since serum creatinine levels are not linearly associated with GFR,32 the use of predictive equations (the Cockcroft–Gault equation for creatinine clearance33 and the MDRD equation for estimated GFR19) has been proposed as a more accurate means of estimating the GFR, with the MDRD equation having better predictive ability in certain populations.20
Relatively few studies have evaluated the estimated GFR and the risk of outcomes in the general population. In the Second National Health and Nutrition Examination Survey (NHANES II),12 an estimated GFR of less than 70 ml per minute per 1.73 m2 was associated with a 68 percent increase in the risk of death from any cause and a 51 percent increase in the risk of death from cardiovascular causes, as compared with an estimated GFR of at least 90 ml per minute per 1.73 m2. In the Atherosclerosis Risk in Communities Study,17 an estimated GFR of 15 to 59 ml per minute per 1.73 m2 at baseline was associated with a 38 percent increase in the risk of cardiovascular disease, as compared with an estimated GFR of 90 to 150 ml per minute per 1.73 m2. Similar results were obtained in a cohort of older adults (at least 65 years of age)16 for an estimated GFR of 15 to 59 ml per minute per 1.73 m2, as compared with an estimated GFR of 90 to 130 ml per minute per 1.73 m2. However, the NHANES I Epidemiologic Follow-up Study did not find a significant association between an estimated GFR of approximately 30 to 60 ml per minute per 1.73 m2 and the risk of death from any cause or death from cardiovascular causes.13 In addition, in the Framingham Heart Study, an elevated serum creatinine level — 1.5 to 3.0 mg per deciliter (133 to 265 μmol per liter) in men and 1.4 to 3.0 mg per deciliter (124 to 265 μmol per liter) in women — was associated with a significant risk of death among men but not women and was not associated with the risk of cardiovascular events in either sex.5
These studies were limited by the use of a single measurement of kidney function, the use of a broad range of definitions of diminished kidney function, and the inclusion of relatively small numbers of persons with kidney disease, thus limiting their statistical power to examine different levels of reduced GFR. Furthermore, although several studies used the MDRD equation to estimate GFR, serum creatinine measurements were not directly calibrated to the values of the MDRD laboratory, so the absolute threshold level of GFR associated with adverse outcomes cannot be known with confidence.
Our study further delineates the relation between the GFR and the risk of adverse events. We found nonlinear relations between the GFR and the risks of death, cardiovascular events, and hospitalization, with an increased risk associated with an estimated GFR of less than 60 ml per minute per 1.73 m2, which further rose sharply when values dropped below 45 ml per minute per 1.73 m2. We found that an estimated GFR of 15 to 29 ml per minute per 1.73 m2 and an estimated GFR of less than 15 ml per minute per 1.73 m2 in the absence of dialysis were associated with strikingly high age-adjusted mortality rates (11.4 and 14.1 per 100 person-years, respectively). These rates approach the rates among patients with treated end-stage renal disease.1 Although the prevention of end-stage renal disease remains a very important goal in patients with kidney disease, more effective interventions are clearly needed to reduce the disproportionate cardiovascular and economic burden in this population.
Multiple possible explanations exist for the association between chronic kidney disease and increased risks of death and cardiovascular disease. We observed an increased prevalence of prior cardiovascular disease, known risk factors for cardiovascular events and death, and coexisting conditions with lower levels of the estimated GFR. However, a reduced estimated GFR was an independent and strong risk factor for adverse outcomes. Reduced kidney function is also associated with increased levels of inflammatory factors,34,35 abnormal apolipoprotein levels,34 elevated plasma homocysteine,34 enhanced coagulability,35 anemia,36 left ventricular hypertrophy,37 increased arterial calcification,38 endothelial dysfunction,39 and arterial stiffness.40 Whether and how these and other factors interact to increase the risk of adverse outcomes remains unclear but are the focus of ongoing investigations.41 Our large, ethnically diverse population is typical of patients receiving usual clinical care rather than referral populations, recruited cohorts, or clinical-trial participants, and thus, our results are probably more generalizable. The inclusion of large numbers of subjects with a spectrum of kidney disease also enabled us to make a more detailed evaluation of the effect of the level of the GFR on outcomes. Use of outpatient serum creatinine values that were directly calibrated to those of the MDRD core laboratory increased our confidence in the validity of the observed associations of absolute, not just relative, levels of the estimated GFR. Most subjects had serial estimates of the GFR in order to characterize kidney function over time as accurately as possible. We had complete records of deaths and hospitalizations among these subjects that occurred at health-plan and other facilities. Our study also shows that the use of a baseline estimate of the GFR alone may lead to the misattribution of events to a certain level of GFR measured in the distant past and to the subsequent attenuation of the true strength of the association between a reduced GFR and outcomes.
Our study had several limitations. As a study of subjects who received usual clinical care, estimates of the GFR were not available for the entire source population. For example, patients with a reduced GFR who did not use medical services would not have been included. However, over half of adult members of the health plan had serum creatinine measurements, which allowed for a robust assessment of the estimated GFR and outcomes among more than 1.1 million subjects. We did not have information on tobacco or alcohol use, diet, physical activity, other possible unmeasured confounders (e.g., body-mass index), or the severity of certain conditions (e.g., level of blood pressure or severity of diabetes). Nevertheless, residual confounding is unlikely to explain the large effect estimates observed for most categories of a reduced estimated GFR. Finally, since our study was conducted among insured adults in northern California, our results may not be completely generalizable to uninsured persons or persons in other geographic regions.
In conclusion, we found an independent, graded association between lower levels of the estimated GFR and the risks of death, cardiovascular events, and hospitalization. These risks were evident at an estimated GFR of less than 60 ml per minute per 1.73 m2 and substantially increased with an estimated GFR of less than 45 ml per minute per 1.73 m2. Our findings support the validity of the National Kidney Foundation staging system for chronic kidney disease22 but suggest that the system could be further refined, since all persons with stage 3 chronic kidney disease (GFR, 30 to 59 ml per minute per 1.73 m2) may not be at equal risk for each outcome. Our findings highlight the clinical and public health importance of chronic kidney disease that does not necessitate dialysis.
Supported by a grant (R01 DK58411) from the National Institute for Diabetes, Digestive, and Kidney Diseases.
Presented in part at the 36th Annual Meeting of the American Society of Nephrology, San Diego, Calif., November 14 to 17, 2003.
Dr. Go reports having received research support from Amgen. Dr. Chertow reports having received research support from Genzyme, Amgen, and Kureha Chemical and having served on advisory boards for Genzyme and Amgen.
Source Information
From the Division of Research, Kaiser Permanente of Northern California, Oakland (A.S.G., D.F.); and the Departments of Epidemiology and Biostatistics (A.S.G., G.M.C., C.E.M.) and Medicine (A.S.G., G.M.C., C.-y.H.), University of California, San Francisco, San Francisco.
Address reprint requests to Dr. Go at the Division of Research, Kaiser Permanente of Northern California, 2000 Broadway, 3rd Fl., Oakland, CA 94612-2304, or at alan.s.go@kp.org.
References
Renal Data System. USRDS 2003 annual data report: atlas of end-stage renal disease in the United States. Bethesda, Md.: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, 2003.
Coresh J, Astor BC, Greene T, Eknoyan G, Levey AS. Prevalence of chronic kidney disease and decreased kidney function in the adult US population: Third National Health and Nutrition Examination Survey. Am J Kidney Dis 2003;41:1-12.
Damsgaard EM, Froland A, Jorgensen OD, Mogensen CE. Microalbuminuria as predictor of increased mortality in elderly people. BMJ 1990;300:297-300.
Wannamethee SG, Shaper AG, Perry IJ. Serum creatinine concentration and risk of cardiovascular disease: a possible marker for increased risk of stroke. Stroke 1997;28:557-563.
Culleton BF, Larson MG, Wilson PW, Evans JC, Parfrey PS, Levy D. Cardiovascular disease and mortality in a community-based cohort with mild renal insufficiency. Kidney Int 1999;56:2214-2219.
Ruilope LM, Salvetti A, Jamerson K, et al. Renal function and intensive lowering of blood pressure in hypertensive participants of the Hypertension Optimal Treatment (HOT) Study. J Am Soc Nephrol 2001;12:218-225.
Mann JF, Gerstein HC, Pogue J, Bosch J, Yusuf S. Renal insufficiency as a predictor of cardiovascular outcomes and the impact of ramipril: the HOPE randomized trial. Ann Intern Med 2001;134:629-636.
Langford HG, Stamler J, Wassertheil-Smoller S, Prineas RJ. All-cause mortality in the Hypertension Detection and Follow-up Program: findings for the whole cohort and for persons with less severe hypertension, with and without other traits related to risk of mortality. Prog Cardiovasc Dis 1986;29:Suppl 1:29-54.
Drey N, Roderick P, Mullee M, Rogerson M. A population-based study of the incidence and outcomes of diagnosed chronic kidney disease. Am J Kidney Dis 2003;42:677-684.
Shlipak MG, Simon JA, Grady D, Lin F, Wenger NK, Furberg CD. Renal insufficiency and cardiovascular events in postmenopausal women with coronary heart disease. J Am Coll Cardiol 2001;38:705-711.
Shlipak MG, Heidenreich PA, Noguchi H, Chertow GM, Browner WS, McClellan MB. Association of renal insufficiency with treatment and outcomes after myocardial infarction in elderly patients. Ann Intern Med 2002;137:555-562.
Muntner P, He J, Hamm L, Loria C, Whelton PK. Renal insufficiency and subsequent death resulting from cardiovascular disease in the United States. J Am Soc Nephrol 2002;13:745-753.
Garg AX, Clark WF, Haynes RB, House AA. Moderate renal insufficiency and the risk of cardiovascular mortality: results from the NHANES I. Kidney Int 2002;61:1486-1494.
Wang JG, Staessen JA, Fagard RH, Birkenhager WH, Gong L, Liu L. Prognostic significance of serum creatinine and uric acid in older Chinese patients with isolated systolic hypertension. Hypertension 2001;37:1069-1074.
Shulman NB, Ford CE, Hall WD, et al. Prognostic value of serum creatinine and effect of treatment of hypertension on renal function: results from the Hypertension Detection and Follow-up Program. Hypertension 1989;13:Suppl:I-80.
Manjunath G, Tighiouart H, Coresh J, et al. Level of kidney function as a risk factor for cardiovascular outcomes in the elderly. Kidney Int 2003;63:1121-1129.
Manjunath G, Tighiouart H, Ibrahim H, et al. Level of kidney function as a risk factor for atherosclerotic cardiovascular outcomes in the community. J Am Coll Cardiol 2003;41:47-55.
Khan SS, Kazmi WH, Abichandani R, Tighiouart H, Pereira BJ, Kausz AT. Health care utilization among patients with chronic kidney disease. Kidney Int 2002;62:229-236.
Levey AS, Greene T, Kusek JW, Beck GJ. A simplified equation to predict glomerular filtration rate from serum creatinine. J Am Soc Nephrol 2000;11:155A-155A. abstract.
Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Ann Intern Med 1999;130:461-470.
Landis JR, Gaughan C, Joffe M. Inter-laboratory serum creatinine (sCr) calibration study. J Am Soc Nephrol 2003;14:294A-294A. abstract.
K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39:Suppl 1:S1-S266.
Go AS, Hylek EM, Borowsky LH, Phillips KA, Selby JV, Singer DE. Warfarin use among ambulatory patients with nonvalvular atrial fibrillation: the anticoagulation and risk factors in atrial fibrillation (ATRIA) study. Ann Intern Med 1999;131:927-934.
Fireman BH, Fehrenbacher L, Gruskin EP, Ray GT. Cost of care for patients in cancer clinical trials. J Natl Cancer Inst 2000;92:136-142.
Alexander M, Tekawa I, Hunkeler E, et al. Evaluating hypertension control in a managed care setting. Arch Intern Med 1999;159:2673-2677.
Selby JV, Ray GT, Zhang D, Colby CJ. Excess costs of medical care for patients with diabetes in a managed care population. Diabetes Care 1997;20:1396-1402.
Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Executive summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 2001;285:2486-2497.
Bureau of the Census. 2000 Census of population and housing: summary files 1 to 3. Washington, D.C.: Department of Commerce, 2001-2003.
Karter AJ, Ferrara A, Liu JY, Moffet HH, Ackerson LM, Selby JV. Ethnic disparities in diabetic complications in an insured population. JAMA 2002;287:2519-2527.
Arellano MG, Petersen GR, Petitti DB, Smith RE. The California Automated Mortality Linkage System (CAMLIS). Am J Public Health 1984;74:1324-1330.
Therneau TM, Grambsch PM. Modeling survival data: extending the Cox model. New York: Springer, 2000.
Shemesh O, Golbetz H, Kriss JP, Myers BD. Limitations of creatinine as a filtration marker in glomerulopathic patients. Kidney Int 1985;28:830-838.
Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31-41.
Muntner P, Hamm LL, Kusek JW, Chen J, Whelton PK, He J. The prevalence of nontraditional risk factors for coronary heart disease in patients with chronic kidney disease. Ann Intern Med 2004;140:9-17.
Shlipak MG, Fried LF, Crump C, et al. Elevations of inflammatory and procoagulant biomarkers in elderly persons with renal insufficiency. Circulation 2003;107:87-92.
Hsu CY, McCulloch CE, Curhan GC. Epidemiology of anemia associated with chronic renal insufficiency among adults in the United States: results from the Third National Health and Nutrition Examination Survey. J Am Soc Nephrol 2002;13:504-510.
Levin A, Thompson CR, Ethier J, et al. Left ventricular mass index increase in early renal disease: impact of decline in hemoglobin. Am J Kidney Dis 1999;34:125-134.
Raggi P, Boulay A, Chasan-Taber S, et al. Cardiac calcification in adult hemodialysis patients: a link between end-stage renal disease and cardiovascular disease? J Am Coll Cardiol 2002;39:695-701.
Blacher J, Safar ME, Guerin AP, Pannier B, Marchais SJ, London GM. Aortic pulse wave velocity index and mortality in end-stage renal disease. Kidney Int 2003;63:1852-1860.
London GM, Guerin AP, Marchais SJ, Metivier F, Pannier B, Adda H. Arterial media calcification in end-stage renal disease: impact on all-cause and cardiovascular mortality. Nephrol Dial Transplant 2003;18:1731-1740.
Feldman HI, Appel LJ, Chertow GM, et al. The Chronic Renal Insufficiency Cohort (CRIC) Study: design and methods. J Am Soc Nephrol 2003;14:Suppl 2:S148-S153.
Appendix
The following criteria and International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM), and Current Procedural Terminology (CPT) codes, if relevant, were used to define coexisting illnesses (Table 3).
Related Letters:
Chronic Renal Disease and Cardiovascular Risk
Risch L., Sagmeister M., Huber A., Cheng H., Go A. S., Chertow G. M., Hsu C.-y., Anavekar N. S., McMurray J. J.V., Pfeffer M. A.(Alan S. Go, M.D., Glenn M)
Background End-stage renal disease substantially increases the risks of death, cardiovascular disease, and use of specialized health care, but the effects of less severe kidney dysfunction on these outcomes are less well defined.
Methods We estimated the longitudinal glomerular filtration rate (GFR) among 1,120,295 adults within a large, integrated system of health care delivery in whom serum creatinine had been measured between 1996 and 2000 and who had not undergone dialysis or kidney transplantation. We examined the multivariable association between the estimated GFR and the risks of death, cardiovascular events, and hospitalization.
Results The median follow-up was 2.84 years, the mean age was 52 years, and 55 percent of the group were women. After adjustment, the risk of death increased as the GFR decreased below 60 ml per minute per 1.73 m2 of body-surface area: the adjusted hazard ratio for death was 1.2 with an estimated GFR of 45 to 59 ml per minute per 1.73 m2 (95 percent confidence interval, 1.1 to 1.2), 1.8 with an estimated GFR of 30 to 44 ml per minute per 1.73 m2 (95 percent confidence interval, 1.7 to 1.9), 3.2 with an estimated GFR of 15 to 29 ml per minute per 1.73 m2 (95 percent confidence interval, 3.1 to 3.4), and 5.9 with an estimated GFR of less than 15 ml per minute per 1.73 m2 (95 percent confidence interval, 5.4 to 6.5). The adjusted hazard ratio for cardiovascular events also increased inversely with the estimated GFR: 1.4 (95 percent confidence interval, 1.4 to 1.5), 2.0 (95 percent confidence interval, 1.9 to 2.1), 2.8 (95 percent confidence interval, 2.6 to 2.9), and 3.4 (95 percent confidence interval, 3.1 to 3.8), respectively. The adjusted risk of hospitalization with a reduced estimated GFR followed a similar pattern.
Conclusions An independent, graded association was observed between a reduced estimated GFR and the risk of death, cardiovascular events, and hospitalization in a large, community-based population. These findings highlight the clinical and public health importance of chronic renal insufficiency.
More than 400,000 Americans have end-stage renal disease, and over 300,000 of these patients require maintenance dialysis.1 Mortality rates remain above 20 percent per year with the use of dialysis, with more than half of the deaths related to cardiovascular disease. The annual direct medical costs for end-stage renal disease are nearly $23 billion.1 Although an estimated 8 million adults in the United States have chronic kidney disease of at least stage 3 (as defined by an estimated glomerular filtration rate of less than 60 ml per minute per 1.73 m2 of body-surface area),2 less is known about the rates of death, cardiovascular disease, and resource use among persons with a reduced estimated GFR who are not yet receiving maintenance dialysis.
Several, but not all, previous studies suggested that mild-to-moderate elevations in serum creatinine levels are associated with increased rates of death from any cause3,4,5,6,7,8,9 and from cardiovascular causes,5,7,10,11,12,13 but whether chronic kidney disease independently increases the risk of any type of cardiovascular disease has not been established.4,5,6,7,14 Furthermore, previous studies have been limited by the inclusion of relatively small numbers of persons with kidney disease,3,4,5,6,7,8,11,12,13,14,15,16,17 the use of dichotomous groups of estimated kidney function,3,4,5,6,7,8,15 the use of the serum creatinine level alone as a proxy for GFR and nonuniform cutoff values to define kidney disease,3,4,5,6,7,8,9,10,11,15 lack of information on longitudinal changes in GFR and coexisting conditions,3,4,5,6,7,8,9,10,11,14,15,16,17 selected populations,6,8,11,14,16 and populations with limited racial or ethnic diversity. In addition, few studies have investigated the association between chronic kidney disease and the risk of hospitalization,18 which has important economic implications.
Using longitudinal measures of estimated GFR, a more accurate method of assessing kidney function than the measurement of serum creatinine alone, we examined the effect of the severity of kidney dysfunction on the risks of death, cardiovascular events, and hospitalization among a large, diverse group of adults. We hypothesized that there would be a graded, independent association between the estimated GFR and the risks of these outcomes.
Methods
Study Sample and Measures of Kidney Function
The Kaiser Permanente Renal Registry included all adult members (20 years of age or older) of Kaiser Permanente of Northern California, a large integrated health care system insuring more than 35 percent of the adult population of the San Francisco Bay Area, whose kidney function was known. To be eligible for the registry, the subjects had to have had one or more outpatient determinations of serum creatinine levels recorded in a health-plan laboratory database between January 1, 1996, and December 31, 2000. We excluded subjects who had already received a kidney transplant or who were receiving maintenance dialysis at entry. Given the nature of the study, the institutional review board of the Kaiser Foundation Research Institute determined that informed consent was not required.
We used the abbreviated Modification of Diet in Renal Disease (MDRD) equation to estimate the GFR.19,20 We calibrated the measurement of serum creatinine by the Kaiser regional laboratory against that of the MDRD core laboratory.21 The date of the first measurement of GFR during the study period was considered the subject's index date (baseline). Changes in GFR during follow-up were estimated from serum creatinine determinations not associated with hospitalizations to reflect more accurately stable estimates of kidney function. We used a modified National Kidney Foundation classification of chronic kidney disease,22 which classifies estimated GFR in the following ranges: at least 60 ml per minute per 1.73 m2, 45 to 59 ml per minute per 1.73 m2 (stage 3a), 30 to 44 ml per minute per 1.73 m2 (stage 3b), 15 to 29 ml per minute per 1.73 m2 (stage 4), and less than 15 ml per minute per 1.73 m2 (stage 5).
Characteristics of the Subjects
Data on age, sex, and racial or ethnic group were obtained from health-plan databases. All names and identifiers were removed before any data were analyzed, according to procedures approved by the institutional review board at Kaiser.
We identified coexisting illnesses using validated methods23,24,25,26 based on health-plan databases for hospitalization-discharge diagnoses, ambulatory diagnoses, laboratory results, and medication prescriptions, as well as regional cancer-registry data24 (diagnostic criteria are listed in the Appendix). These diagnoses included coronary disease, stroke or transient ischemic attack, heart failure, peripheral arterial disease, diabetes mellitus, hypertension, dyslipidemia, lung or liver disease, cancer, and dementia. We also evaluated laboratory-database entries for serum albumin to identify subjects with a value of 3.5 g per deciliter or less. The presence of proteinuria was based on laboratory-database entries of urine dipstick results of 1+ or greater (approximately 30 mg per deciliter or greater) in the absence of a possible urinary tract infection (i.e., concomitant positive test for urinary nitrite or esterase).
Socioeconomic status was assigned with the use of 2000 U.S. Census-block data, which generally correspond to city blocks or neighborhoods.28 Subjects with a low income were defined as those living in a Census-block group with a median annual household income under $35,000, and those with a low level of education were categorized as those living in a Census-block group in which more than 25 percent of residents older than 25 years had less than a 12th-grade education.
Outcomes
Data on subjects were censored if they underwent kidney transplantation or disenrolled from the health plan, which was defined as a continuous gap in membership of more than 90 days without the interim use of services. Incident end-stage renal disease, defined by the receipt of maintenance dialysis or a kidney transplant, was identified from a comprehensive health-plan registry.29 The primary outcomes of interest included death from any cause, cardiovascular events, and hospitalizations through December 31, 2000. Death was identified from a search of health-plan databases and the California death registry.30 A cardiovascular event was defined as hospitalization for coronary disease, heart failure, stroke, or peripheral arterial disease (see the Appendix).
Statistical Analysis
Event rates were directly adjusted for age with the use of the age distribution of the adult source population and are presented as the number of events per 100 person-years, with 95 percent confidence intervals. To evaluate the independent effect of the estimated GFR on outcomes, we used Cox proportional-hazards models with time-dependent covariates for changing GFR and coexisting illnesses. All variables known to be associated with either the estimated GFR or the outcomes were included in the final models, along with any variables associated with a reduced estimated GFR (i.e., less than 60 ml per minute per 1.73 m2) in univariate analyses with a P value of less than 0.01. Age was entered as a categorical variable (20 to 49, 50 to 59, 60 to 69, 70 to 79, and 80 years or older), with other covariates included as dichotomous variables. For recurrent outcomes of cardiovascular events and hospitalizations, we used a "sandwich" estimate of the variance–covariance matrix to obtain standard errors accommodating the clustering of observations on subjects.31 In the model for death, we did not include cardiovascular events or hospitalizations that occurred after the index date, since they were hypothesized to be part of the pathway by which reduced GFR may increase the risk of death.
Each subject and time was assigned an estimated GFR with the use of the last-value-carried-forward method. Given the varying numbers and spacing of measurements of GFR for each subject, this approach may preferentially attribute outcome events to higher levels of GFR. To address this possibility we analyzed a subgroup of subjects who had had regular serum creatinine determinations spaced between 1 and 14 months apart throughout follow-up. Also, although age and sex are important predictors of adverse events and are incorporated into the MDRD equation, we found no interactions among age, sex, and GFR and thus present only the main-model results.
All analyses were conducted with the use of SAS software (version 8.2). The institutional review board of each collaborating institution approved the study.
Results
Baseline Characteristics
We identified 1,120,295 adults who had had one or more outpatient measurements of serum creatinine, had not previously received dialysis or a kidney transplant, and were alive on the index date. The median number of outpatient measurements of serum creatinine per subject during follow-up was three (interquartile range, one to four).
Subjects with a low estimated GFR at baseline were older than those with an estimated GFR of at least 60 ml per minute per 1.73 m2, and there was greater minority-group representation among subjects with a low estimated GFR (Table 1). As compared with the group with an estimated GFR of at least 60 ml per minute per 1.73 m2, the groups with a reduced estimated GFR also had a higher prevalence of prior cardiovascular disease, proteinuria, diabetes, hypertension, a serum albumin level of 3.5 g per deciliter or less, prior hospitalizations, and other coexisting illnesses (Table 1). Among the 60.7 percent of subjects who had one or more urine dipstick determinations at any time during follow-up, 17.6 percent were found to have proteinuria. Among the 21.2 percent of subjects who had one or more serum albumin determinations at any time during follow-up, 27.8 percent were found to have reduced levels.
Table 1. Baseline Characteristics of 1,120,295 Ambulatory Adults, According to the Estimated GFR at Baseline.
Outcomes According to the Estimated GFR
The median follow-up among the 1,120,295 subjects was 2.84 years (interquartile range, 1.65 to 4.01), which amounts to 3,132,192 person-years. Overall, 21.8 percent of subjects disenrolled during follow-up, but those who disenrolled were substantially younger, were relatively unlikely to have a reduced estimated GFR, and had fewer coexisting illnesses than those who remained enrolled (data not shown). The vital status was complete for all subjects.
During follow-up, 3171 (0.28 percent) subjects began maintenance dialysis and 329 (0.03 percent) underwent kidney transplantation. There were 51,424 deaths, 138,291 cardiovascular events, and 554,651 hospitalizations. Age-standardized rates of death, cardiovascular events, and hospitalization increased substantially with progressively lower estimated GFRs (Figure 1).
Figure 1. Age-Standardized Rates of Death from Any Cause (Panel A), Cardiovascular Events (Panel B), and Hospitalization (Panel C), According to the Estimated GFR among 1,120,295 Ambulatory Adults.
A cardiovascular event was defined as hospitalization for coronary heart disease, heart failure, ischemic stroke, and peripheral arterial disease. Error bars represent 95 percent confidence intervals. The rate of events is listed above each bar.
The group of subjects with an estimated GFR of at least 60 ml per minute per 1.73 m2 was used as the reference group in the analysis of the association between the level of the estimated GFR and each outcome. After adjustment for differences in sociodemographic characteristics and the presence or absence of prior cardiovascular disease, prior hospitalizations, diabetes, hypertension, dyslipidemia, lung or liver disease, cancer, a serum albumin level of 3.5 g per deciliter or less, dementia, proteinuria, and the initiation of dialysis during follow-up, the risk of death from any cause increased sharply as the estimated GFR declined, ranging from a 17 percent increase in risk with an estimated GFR of 45 to 59 ml per minute per 1.73 m2 to a nearly 600 percent increase with an estimated GFR of less than 15 ml per minute per 1.73 m2 (Table 2). The adjusted risk of any cardiovascular event also increased as the estimated GFR decreased, ranging from a 43 percent increase in risk with an estimated GFR of 45 to 59 ml per minute per 1.73 m2 to a 343 percent increase with an estimated GFR of less than 15 ml per minute per 1.73 m2 (Table 2). Finally, the adjusted risk of hospitalization increased as the estimated GFR decreased, ranging from an increase of 14 percent with an estimated GFR of 45 to 59 ml per minute per 1.73 m2 to an increase of 315 percent with an estimated GFR of less than 15 ml per minute per 1.73 m2 (Table 2). The presence of documented proteinuria was also an independent predictor of death (adjusted hazard ratio, 1.3; 95 percent confidence interval, 1.3 to 1.4), cardiovascular events (adjusted hazard ratio, 1.3; 95 percent confidence interval, 1.2 to 1.3), and hospitalization (adjusted hazard ratio, 1.4; 95 percent confidence interval, 1.4 to 1.4).
Table 2. Adjusted Hazard Ratio for Death from Any Cause, Cardiovascular Events, and Hospitalization among 1,120,295 Ambulatory Adults, According to the Estimated GFR.
In a subgroup of 172,144 subjects who had regular measurements of serum creatinine during follow-up (mean number of measurements, 3.3±2.1; mean interval between measurements, 7.5±3.4 months), the adjusted risks of adverse outcomes differed materially from those for the entire cohort only among the subjects with an estimated GFR of 45 to 59 ml per minute per 1.73 m2. In the analysis of the subgroup itself, as compared with the subjects with an estimated GFR of at least 60 ml per minute per 1.73 m2, the subjects with an estimated GFR of 45 to 59 ml per minute per 1.73 m2 had a similar adjusted hazard ratio for death (1.0; 95 percent confidence interval, 1.0 to 1.1) and hospitalization (1.0; 95 percent confidence interval, 1.0 to 1.0). The adjusted risk of cardiovascular events was also attenuated (1.2; 95 percent confidence interval, 1.1 to 1.3).
Discussion
Among a large, diverse population of adults, a reduced estimated GFR was associated with increased risks of death, cardiovascular events, and hospitalization that were independent of known risk factors, a history of cardiovascular disease, and the presence of documented proteinuria. Our study demonstrates that these graded risks of adverse events rose sharply for subjects with an estimated GFR of less than 45 ml per minute per 1.73 m2 for each outcome examined both in the overall cohort and in subgroup analyses. Furthermore, in the cohort as a whole, the absolute rates of these outcomes were considerably higher than the risk of end-stage renal disease.
There has been rapidly growing interest in the relation between kidney disease and the risk of death and cardiovascular disease. With recognition that the presence of chronic kidney disease that does not necessitate dialysis is of considerable importance, several studies have examined the association of different cutoff values of serum creatinine with the risks of death from any cause, death from cardiovascular causes, and cardiovascular events, and most,3,4,6,7,8,9,10,11,12,15 but not all,5,13 of these studies have found increased risks with higher serum creatinine levels. Others have noted high rates of hospitalization among persons with elevated serum creatinine levels.18 However, since serum creatinine levels are not linearly associated with GFR,32 the use of predictive equations (the Cockcroft–Gault equation for creatinine clearance33 and the MDRD equation for estimated GFR19) has been proposed as a more accurate means of estimating the GFR, with the MDRD equation having better predictive ability in certain populations.20
Relatively few studies have evaluated the estimated GFR and the risk of outcomes in the general population. In the Second National Health and Nutrition Examination Survey (NHANES II),12 an estimated GFR of less than 70 ml per minute per 1.73 m2 was associated with a 68 percent increase in the risk of death from any cause and a 51 percent increase in the risk of death from cardiovascular causes, as compared with an estimated GFR of at least 90 ml per minute per 1.73 m2. In the Atherosclerosis Risk in Communities Study,17 an estimated GFR of 15 to 59 ml per minute per 1.73 m2 at baseline was associated with a 38 percent increase in the risk of cardiovascular disease, as compared with an estimated GFR of 90 to 150 ml per minute per 1.73 m2. Similar results were obtained in a cohort of older adults (at least 65 years of age)16 for an estimated GFR of 15 to 59 ml per minute per 1.73 m2, as compared with an estimated GFR of 90 to 130 ml per minute per 1.73 m2. However, the NHANES I Epidemiologic Follow-up Study did not find a significant association between an estimated GFR of approximately 30 to 60 ml per minute per 1.73 m2 and the risk of death from any cause or death from cardiovascular causes.13 In addition, in the Framingham Heart Study, an elevated serum creatinine level — 1.5 to 3.0 mg per deciliter (133 to 265 μmol per liter) in men and 1.4 to 3.0 mg per deciliter (124 to 265 μmol per liter) in women — was associated with a significant risk of death among men but not women and was not associated with the risk of cardiovascular events in either sex.5
These studies were limited by the use of a single measurement of kidney function, the use of a broad range of definitions of diminished kidney function, and the inclusion of relatively small numbers of persons with kidney disease, thus limiting their statistical power to examine different levels of reduced GFR. Furthermore, although several studies used the MDRD equation to estimate GFR, serum creatinine measurements were not directly calibrated to the values of the MDRD laboratory, so the absolute threshold level of GFR associated with adverse outcomes cannot be known with confidence.
Our study further delineates the relation between the GFR and the risk of adverse events. We found nonlinear relations between the GFR and the risks of death, cardiovascular events, and hospitalization, with an increased risk associated with an estimated GFR of less than 60 ml per minute per 1.73 m2, which further rose sharply when values dropped below 45 ml per minute per 1.73 m2. We found that an estimated GFR of 15 to 29 ml per minute per 1.73 m2 and an estimated GFR of less than 15 ml per minute per 1.73 m2 in the absence of dialysis were associated with strikingly high age-adjusted mortality rates (11.4 and 14.1 per 100 person-years, respectively). These rates approach the rates among patients with treated end-stage renal disease.1 Although the prevention of end-stage renal disease remains a very important goal in patients with kidney disease, more effective interventions are clearly needed to reduce the disproportionate cardiovascular and economic burden in this population.
Multiple possible explanations exist for the association between chronic kidney disease and increased risks of death and cardiovascular disease. We observed an increased prevalence of prior cardiovascular disease, known risk factors for cardiovascular events and death, and coexisting conditions with lower levels of the estimated GFR. However, a reduced estimated GFR was an independent and strong risk factor for adverse outcomes. Reduced kidney function is also associated with increased levels of inflammatory factors,34,35 abnormal apolipoprotein levels,34 elevated plasma homocysteine,34 enhanced coagulability,35 anemia,36 left ventricular hypertrophy,37 increased arterial calcification,38 endothelial dysfunction,39 and arterial stiffness.40 Whether and how these and other factors interact to increase the risk of adverse outcomes remains unclear but are the focus of ongoing investigations.41 Our large, ethnically diverse population is typical of patients receiving usual clinical care rather than referral populations, recruited cohorts, or clinical-trial participants, and thus, our results are probably more generalizable. The inclusion of large numbers of subjects with a spectrum of kidney disease also enabled us to make a more detailed evaluation of the effect of the level of the GFR on outcomes. Use of outpatient serum creatinine values that were directly calibrated to those of the MDRD core laboratory increased our confidence in the validity of the observed associations of absolute, not just relative, levels of the estimated GFR. Most subjects had serial estimates of the GFR in order to characterize kidney function over time as accurately as possible. We had complete records of deaths and hospitalizations among these subjects that occurred at health-plan and other facilities. Our study also shows that the use of a baseline estimate of the GFR alone may lead to the misattribution of events to a certain level of GFR measured in the distant past and to the subsequent attenuation of the true strength of the association between a reduced GFR and outcomes.
Our study had several limitations. As a study of subjects who received usual clinical care, estimates of the GFR were not available for the entire source population. For example, patients with a reduced GFR who did not use medical services would not have been included. However, over half of adult members of the health plan had serum creatinine measurements, which allowed for a robust assessment of the estimated GFR and outcomes among more than 1.1 million subjects. We did not have information on tobacco or alcohol use, diet, physical activity, other possible unmeasured confounders (e.g., body-mass index), or the severity of certain conditions (e.g., level of blood pressure or severity of diabetes). Nevertheless, residual confounding is unlikely to explain the large effect estimates observed for most categories of a reduced estimated GFR. Finally, since our study was conducted among insured adults in northern California, our results may not be completely generalizable to uninsured persons or persons in other geographic regions.
In conclusion, we found an independent, graded association between lower levels of the estimated GFR and the risks of death, cardiovascular events, and hospitalization. These risks were evident at an estimated GFR of less than 60 ml per minute per 1.73 m2 and substantially increased with an estimated GFR of less than 45 ml per minute per 1.73 m2. Our findings support the validity of the National Kidney Foundation staging system for chronic kidney disease22 but suggest that the system could be further refined, since all persons with stage 3 chronic kidney disease (GFR, 30 to 59 ml per minute per 1.73 m2) may not be at equal risk for each outcome. Our findings highlight the clinical and public health importance of chronic kidney disease that does not necessitate dialysis.
Supported by a grant (R01 DK58411) from the National Institute for Diabetes, Digestive, and Kidney Diseases.
Presented in part at the 36th Annual Meeting of the American Society of Nephrology, San Diego, Calif., November 14 to 17, 2003.
Dr. Go reports having received research support from Amgen. Dr. Chertow reports having received research support from Genzyme, Amgen, and Kureha Chemical and having served on advisory boards for Genzyme and Amgen.
Source Information
From the Division of Research, Kaiser Permanente of Northern California, Oakland (A.S.G., D.F.); and the Departments of Epidemiology and Biostatistics (A.S.G., G.M.C., C.E.M.) and Medicine (A.S.G., G.M.C., C.-y.H.), University of California, San Francisco, San Francisco.
Address reprint requests to Dr. Go at the Division of Research, Kaiser Permanente of Northern California, 2000 Broadway, 3rd Fl., Oakland, CA 94612-2304, or at alan.s.go@kp.org.
References
Renal Data System. USRDS 2003 annual data report: atlas of end-stage renal disease in the United States. Bethesda, Md.: National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, 2003.
Coresh J, Astor BC, Greene T, Eknoyan G, Levey AS. Prevalence of chronic kidney disease and decreased kidney function in the adult US population: Third National Health and Nutrition Examination Survey. Am J Kidney Dis 2003;41:1-12.
Damsgaard EM, Froland A, Jorgensen OD, Mogensen CE. Microalbuminuria as predictor of increased mortality in elderly people. BMJ 1990;300:297-300.
Wannamethee SG, Shaper AG, Perry IJ. Serum creatinine concentration and risk of cardiovascular disease: a possible marker for increased risk of stroke. Stroke 1997;28:557-563.
Culleton BF, Larson MG, Wilson PW, Evans JC, Parfrey PS, Levy D. Cardiovascular disease and mortality in a community-based cohort with mild renal insufficiency. Kidney Int 1999;56:2214-2219.
Ruilope LM, Salvetti A, Jamerson K, et al. Renal function and intensive lowering of blood pressure in hypertensive participants of the Hypertension Optimal Treatment (HOT) Study. J Am Soc Nephrol 2001;12:218-225.
Mann JF, Gerstein HC, Pogue J, Bosch J, Yusuf S. Renal insufficiency as a predictor of cardiovascular outcomes and the impact of ramipril: the HOPE randomized trial. Ann Intern Med 2001;134:629-636.
Langford HG, Stamler J, Wassertheil-Smoller S, Prineas RJ. All-cause mortality in the Hypertension Detection and Follow-up Program: findings for the whole cohort and for persons with less severe hypertension, with and without other traits related to risk of mortality. Prog Cardiovasc Dis 1986;29:Suppl 1:29-54.
Drey N, Roderick P, Mullee M, Rogerson M. A population-based study of the incidence and outcomes of diagnosed chronic kidney disease. Am J Kidney Dis 2003;42:677-684.
Shlipak MG, Simon JA, Grady D, Lin F, Wenger NK, Furberg CD. Renal insufficiency and cardiovascular events in postmenopausal women with coronary heart disease. J Am Coll Cardiol 2001;38:705-711.
Shlipak MG, Heidenreich PA, Noguchi H, Chertow GM, Browner WS, McClellan MB. Association of renal insufficiency with treatment and outcomes after myocardial infarction in elderly patients. Ann Intern Med 2002;137:555-562.
Muntner P, He J, Hamm L, Loria C, Whelton PK. Renal insufficiency and subsequent death resulting from cardiovascular disease in the United States. J Am Soc Nephrol 2002;13:745-753.
Garg AX, Clark WF, Haynes RB, House AA. Moderate renal insufficiency and the risk of cardiovascular mortality: results from the NHANES I. Kidney Int 2002;61:1486-1494.
Wang JG, Staessen JA, Fagard RH, Birkenhager WH, Gong L, Liu L. Prognostic significance of serum creatinine and uric acid in older Chinese patients with isolated systolic hypertension. Hypertension 2001;37:1069-1074.
Shulman NB, Ford CE, Hall WD, et al. Prognostic value of serum creatinine and effect of treatment of hypertension on renal function: results from the Hypertension Detection and Follow-up Program. Hypertension 1989;13:Suppl:I-80.
Manjunath G, Tighiouart H, Coresh J, et al. Level of kidney function as a risk factor for cardiovascular outcomes in the elderly. Kidney Int 2003;63:1121-1129.
Manjunath G, Tighiouart H, Ibrahim H, et al. Level of kidney function as a risk factor for atherosclerotic cardiovascular outcomes in the community. J Am Coll Cardiol 2003;41:47-55.
Khan SS, Kazmi WH, Abichandani R, Tighiouart H, Pereira BJ, Kausz AT. Health care utilization among patients with chronic kidney disease. Kidney Int 2002;62:229-236.
Levey AS, Greene T, Kusek JW, Beck GJ. A simplified equation to predict glomerular filtration rate from serum creatinine. J Am Soc Nephrol 2000;11:155A-155A. abstract.
Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Ann Intern Med 1999;130:461-470.
Landis JR, Gaughan C, Joffe M. Inter-laboratory serum creatinine (sCr) calibration study. J Am Soc Nephrol 2003;14:294A-294A. abstract.
K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis 2002;39:Suppl 1:S1-S266.
Go AS, Hylek EM, Borowsky LH, Phillips KA, Selby JV, Singer DE. Warfarin use among ambulatory patients with nonvalvular atrial fibrillation: the anticoagulation and risk factors in atrial fibrillation (ATRIA) study. Ann Intern Med 1999;131:927-934.
Fireman BH, Fehrenbacher L, Gruskin EP, Ray GT. Cost of care for patients in cancer clinical trials. J Natl Cancer Inst 2000;92:136-142.
Alexander M, Tekawa I, Hunkeler E, et al. Evaluating hypertension control in a managed care setting. Arch Intern Med 1999;159:2673-2677.
Selby JV, Ray GT, Zhang D, Colby CJ. Excess costs of medical care for patients with diabetes in a managed care population. Diabetes Care 1997;20:1396-1402.
Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Executive summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 2001;285:2486-2497.
Bureau of the Census. 2000 Census of population and housing: summary files 1 to 3. Washington, D.C.: Department of Commerce, 2001-2003.
Karter AJ, Ferrara A, Liu JY, Moffet HH, Ackerson LM, Selby JV. Ethnic disparities in diabetic complications in an insured population. JAMA 2002;287:2519-2527.
Arellano MG, Petersen GR, Petitti DB, Smith RE. The California Automated Mortality Linkage System (CAMLIS). Am J Public Health 1984;74:1324-1330.
Therneau TM, Grambsch PM. Modeling survival data: extending the Cox model. New York: Springer, 2000.
Shemesh O, Golbetz H, Kriss JP, Myers BD. Limitations of creatinine as a filtration marker in glomerulopathic patients. Kidney Int 1985;28:830-838.
Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:31-41.
Muntner P, Hamm LL, Kusek JW, Chen J, Whelton PK, He J. The prevalence of nontraditional risk factors for coronary heart disease in patients with chronic kidney disease. Ann Intern Med 2004;140:9-17.
Shlipak MG, Fried LF, Crump C, et al. Elevations of inflammatory and procoagulant biomarkers in elderly persons with renal insufficiency. Circulation 2003;107:87-92.
Hsu CY, McCulloch CE, Curhan GC. Epidemiology of anemia associated with chronic renal insufficiency among adults in the United States: results from the Third National Health and Nutrition Examination Survey. J Am Soc Nephrol 2002;13:504-510.
Levin A, Thompson CR, Ethier J, et al. Left ventricular mass index increase in early renal disease: impact of decline in hemoglobin. Am J Kidney Dis 1999;34:125-134.
Raggi P, Boulay A, Chasan-Taber S, et al. Cardiac calcification in adult hemodialysis patients: a link between end-stage renal disease and cardiovascular disease? J Am Coll Cardiol 2002;39:695-701.
Blacher J, Safar ME, Guerin AP, Pannier B, Marchais SJ, London GM. Aortic pulse wave velocity index and mortality in end-stage renal disease. Kidney Int 2003;63:1852-1860.
London GM, Guerin AP, Marchais SJ, Metivier F, Pannier B, Adda H. Arterial media calcification in end-stage renal disease: impact on all-cause and cardiovascular mortality. Nephrol Dial Transplant 2003;18:1731-1740.
Feldman HI, Appel LJ, Chertow GM, et al. The Chronic Renal Insufficiency Cohort (CRIC) Study: design and methods. J Am Soc Nephrol 2003;14:Suppl 2:S148-S153.
Appendix
The following criteria and International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM), and Current Procedural Terminology (CPT) codes, if relevant, were used to define coexisting illnesses (Table 3).
Related Letters:
Chronic Renal Disease and Cardiovascular Risk
Risch L., Sagmeister M., Huber A., Cheng H., Go A. S., Chertow G. M., Hsu C.-y., Anavekar N. S., McMurray J. J.V., Pfeffer M. A.(Alan S. Go, M.D., Glenn M)