Relation Between Kidney Function, Proteinuria, and Adverse Outcomes
Relation Between Kidney Function, Proteinuria, and Adverse Outcomes
Relation Between Kidney Function, Proteinuria, and Adverse Outcomes
teinuria could be used together to identify individuals at high risk. We studied a large cohort of individuals receiving routine clinical care in a single Canadian province, in which all residents are covered by governmentsponsored health insurance. We examined the association between reduced eGFR, proteinuria, and adverse clinical outcomes, including all-cause mortality, myocardial infarction, and progression to kidney failure. We hypothesized that patients with both reduced eGFR and proteinuria would be at higher risk of these outcomes than participants with one or neither characteristic.
METHODS The study population included all adults 18 years and older with at least 1 outpatient serum creatinine measurement in
Author Affiliations: Departments of Medicine (Drs Hemmelgarn, Manns, James, and Quinn) and Community Health Sciences (Drs Hemmelgarn, Manns, and James), University of Calgary, Calgary, Alberta, Canada; Departments of Medicine (Mss Lloyd and Wiebe and Drs Klarenbach and Tonelli) and Public Health Sciences (Drs Klarenbach and Tonelli), University of Alberta, Edmonton, Alberta; and Division of Nephrology, Foothills Medical Centre, Calgary, Alberta (Drs Hemmelgarn, Manns, James, and Quinn). A list of the Alberta Kidney Disease Network members appears at http://www.AKDN.info. Corresponding Author: Brenda R. Hemmelgarn, MD, PhD, Division of Nephrology, Foothills Medical Centre, 1403 29th St NW, Calgary, AB T2N 2T9, Canada ([email protected]). 423
the province of Alberta, Canada, between May 1, 2002, and December 31, 2006, for 7 of the 9 geographically based provincial health regions, and between July 1, 2003, and January 1, 2005, and December 31, 2006, respectively, for the other 2 regions. Patients were excluded if they were treated with dialysis or a kidney transplant at baseline or if the baseline estimate of kidney function was clinically implausible (serum creatinine 0.28 mg/dL [multiply by 88.4 to get mol/L]). To be eligible for inclusion, patients also had to have had at least 1 outpatient measure of proteinuria as described in this section. This study was facilitated by a previously described8 provincial laboratory repository: the Alberta Kidney Disease Network (AKDN).
Measurement of Kidney Function, Proteinuria, and Albuminuria
The eGFR for each patient was estimated using the 4-variable Modification of Diet in Renal Disease (MDRD) Study equation.9 Although data on race were not available, misclassification of eGFR was expected to be minimal because less than 1% of the Alberta population is black.10 Baseline kidney function (index eGFR) was estimated using all outpatient serum creatinine measurements taken within a 6-month period of the first creatinine measurement, with the index eGFR defined as the mean of the measurements in this 6-month period. The date of the last serum creatinine measurement in the 6-month period was used as the index date for individuals with more than a single measurement.8 Index eGFR was categorized as 60 mL/min/1.73 m2 or greater, 45 to 59.9 mL/min/1.73 m2, 30 to 44.9 mL/min/1.73 m2, and 15 to 29.9 mL/min/1.73 m2. Because of inaccuracies in assessment of kidney function using the MDRD Study equation at higher levels of kidney function, and to permit comparisons with related studies,11 we categorized individuals with higher levels of function into 1 category (eGFR 60 mL/min/1.73 m2). Proteinuria was captured by urine dipstick as well as albumin-creatinine ratio (ACR) based on outpatient ran424
dom spot urine measurements. In the primary analysis, we included all patients with at least 1 urine dipstick measurement and defined proteinuria as normal (urine dipstick reading negative), mild (urine dipstick reading trace or 1 ), or heavy (urine dipstick reading 2 ).12 In sensitivity analyses, we considered an alternate definition of proteinuria based on ACR, defined as normal (ACR 30 mg/g), mild (ACR 30-300 mg/g), or heavy (ACR 300 mg/g).12 All outpatient urine dipstick and ACR measurements in the 6-month periods before and after the index eGFR were used to establish baseline proteinuria and albuminuria. Analyses used proteinuria or albuminuria as an ordinal variable according to these 3 categories, with the median of all respective measurements selected for each patient with multiple measurements.
Covariates
Patients were followed up from their index date until study end (March 31, 2007). The primary outcome of interest was all-cause mortality, as identified from the Alberta Health and Wellness Registry file. Secondary outcomes were first hospitalization for acute myocardial infarction17; occurrence of endstage renal disease, defined as the date of registration for chronic dialysis or renal transplantation18; and the occurrence of an outpatient serum creatinine measurement that was twice as high as the first creatinine measurement during the study period (corresponding to a 50% decline in kidney function), assessed at the end of follow-up.
Statistical Analyses
Demographic data were determined from the administrative data files of the provincial health ministry (Alberta Health and Wellness). Aboriginal race/ ethnicity was determined from First Nations status in the registry file; it was not possible to identify other race/ ethnic groups, although more than 85% of the Alberta population is white.10 Socioeconomic status was categorized as high income (annual adjusted taxable family income CaD $39 250 [US $37 695]), low income (annual adjusted taxable family income CaD $39 250), and low income with subsidy (receiving social assistance) based on government records. 13 Diabetes mellitus and hypertension were identified from hospital discharge records and physician claims based on validated algorithms.14,15 Other comorbid conditions were identified using validated International Classification of Diseases, Ninth Revision, Clinical Modification, and International Statistical Classification of Diseases, Tenth Revision (ICD-10), coding algorithms applied to physician claims and hospitalization data.16 The presence of 1 or more diagnostic code in any position up to
Poisson regression was used to evaluate the association between the renal risk factors and each of the outcomes of interest, with output expressed as the rate per 1000 person-years. If the Poisson assumption that variance equals the mean was not met, a negative binomial model was used. We first calculated ageadjusted rates for each of the outcomes (all-cause mortality, hospitalization for myocardial infarction, end-stage renal disease, and doubling of serum creatinine) by level of eGFR and proteinuria, considering urine dipstick reading and ACR separately to classify proteinuria. We then calculated fully adjusted event rates for each of the outcomes, adjusting for the sociodemographic variables and comorbidities listed in TABLE 1. Two-way interactions between eGFR and proteinuria were assessed for all 4 clinical outcomes. The primary analysis was based on the cohort of participants who had data for proteinuria available from dipstick urinalysis. This analysis had greater than 99% statistical power (for =.05) to detect a 10% increase in the likelihood of death among (1) individuals with eGFR of 60 mL/min/1.73 m2 or greater compared with those with eGFR of 15 to 29.9 mL/min/1.73 m2 and (2) individuals with
heavy proteinuria compared with those with no proteinuria. In sensitivity analyses, we repeated statistical models for the subset of participants who had data for proteinuria based on urinary ACRs. We repeated analyses examining the relation between proteinuria and adverse outcomes in 2 subgroups of clinical interest: those with eGFR 45 to 59.9 mL/ min/1.73 m2 (who account for the large majority of people with CKD) and those with mildly reduced eGFR as defined by current guidelines (eGFR 60-89.9 mL/ min/1.73 m2).
We performed sensitivity analyses in strata defined by participant age ( 65 and 65 years). In all analyses, we performed tests for linear trend across categories of proteinuria and eGFR. The variables used to calculate the tests for trend in eGFR and ACR were defined by the median values of these parameters in each category. The variable used to calculate the test for trend in dipstickmeasured proteinuria was defined by values of 1, 2, and 3 for normal, mild, and heavy proteinuria, respectively.20 Finally, we repeated the analysis using
eGFR and ACR as continuous variables, with ACR log-transformed because of its skewed distribution. Statistical analyses were performed using SAS version 9.2 (SAS Institute Inc, Cary, North Carolina) and Stata version 10.1 (StataCorp, College Station, Texas). A P value of .05 was used to indicate statistical significance without adjustment for multiple comparisons. The institutional review boards of the University of Calgary and University of Alberta approved the study and granted waiver of patient consent.
Table 1. Demographic and Clinical Characteristics of Participants by Level of Kidney Function or Proteinuria a
Primary Analysis, % (N = 920 985) eGFR, mL/min/1.73 m2 b Proteinuria Measured by Dipstick Sensitivity Analysis, % (n = 102 701) Proteinuria Measured by ACR
Age, mean (SD), y Female sex Aboriginal Diabetes Hypertension Cerebrovascular disease Peripheral vascular disease CHF Cancer COPD Dementia Diabetes-C Diabetes-NC AIDS/HIV Metastatic solid tumor Myocardial infarction Mild liver disease Moderate/severe liver disease Paralysis Peptic ulcer disease Renal disease Rheumatic disease Socioeconomic status c Low Low with subsidy
60 45-59.9 30-44.9 15-29.9 None Mild Heavy None Mild Heavy (n = 820 571) (n = 79 845) (n = 16 713) (n = 3856) (n = 836 550) (n = 71 557) (n = 12 878) (n = 77 280) (n = 20 217) (n = 5204) 46.4 (15.4) 65.8 (14.0) 75.1 (12.2) 74.7 (13.9) 48.4 (16.3) 50.8 (19.7) 55.4 (20.3) 55.8 (14.7) 60.5 (15.5) 60.1 (15.8) 55 64 65 61 56 52 44 46 45 40 2 1 1 2 2 4 4 3 4 6 6 13 25 36 6 14 31 49 67 74 18 49 76 82 21 32 50 46 60 69 2 6 12 15 2 4 8 3 6 8 1 1 4 13 1 0 3 0 0 1 1 0 0 2 0 1 4 7 8 18 3 1 7 0 1 5 1 0 1 3 3 2 9 20 13 25 8 5 15 0 2 12 2 0 1 5 18 4 14 33 16 30 11 11 26 0 3 18 2 1 2 7 52 5 1 2 4 13 1 0 3 0 0 2 1 0 0 2 1 1 3 5 7 18 3 1 7 0 1 4 2 0 1 3 3 2 6 11 10 21 4 6 18 0 2 8 2 0 1 4 14 3 2 4 5 15 1 2 21 0 0 4 1 0 0 3 2 1 5 8 7 19 2 6 32 0 1 7 2 0 1 3 5 2 8 14 7 22 2 14 43 0 1 10 2 0 1 4 16 2
16 2
38 2
60 2
60 3
18 2
25 4
33 4
24 3
34 3
36 5
Abbreviations: ACR, albumin-creatinine ratio; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; diabetes-C, diabetes with end organ damage; diabetes-NC, diabetes without end organ damage; eGFR, estimated glomerular filtration rate; HIV, human immunodeficiency virus. a Totals do not always add to 100 because of rounding. b Among patients with proteinuria measured by dipstick. c Socioeconomic status was categorized as high (annual adjusted taxable family income CaD $39 250 [US $37 695]), low (annual adjusted taxable family income CaD $39 250), and low with subsidy (receiving social assistance) based on Government of Alberta health care insurance records.19
425
RESULTS A total of 1 530 447 participants had at least 1 outpatient serum creatinine measurement during the study period. We excluded 2345 people with end-stage renal disease prior to cohort entry and 1383 with index eGFR of less than 15 mL/min/1.73 m2. An additional 282 individuals were excluded because they either died or reached end of follow-up on their index date. Of the 1 526 437 participants, 920 985 (60.3%) had at least 1 urine dipstick measurement and 102 701 (6.7%) had at least 1 ACR measurement. Characteristics of the participants by level of eGFR and proteinuria are shown in Table 1. The majority of individuals (89.1%) in the primary analysis with proteinuria measured by urine dipstick had an eGFR of 60 mL/min/1.73 m2 or greater. A total of 102 701 participants had at least 1 urinary ACR measurement performed; individuals in this subset were older (mean [SD] age, 57.0 [15.0]
years vs 48.0 [16.6] years) and more likely to be male (54.5% vs 43.4%) or diabetic (54.1% vs 3.4%) and had a higher mean (SD) Charlson score (0.94 [1.6] vs 0.43 [1.1]) than those without such measurements (all P .001; 2 test and t test for categorical and continuous variables, respectively). A higher proportion of participants in this subset had mild (19.7% vs 7.3%) or heavy proteinuria (5.1% vs 1.1%) than in those without measurements of urinary ACR (both P .001, 2 test).
Age-Adjusted Likelihood of Clinical Outcomes by Level of eGFR and Proteinuria
outcomes were all increased at lower levels of eGFR and at heavier proteinuria (eTable 2 and eTable 3).
Adjusted Likelihood of Clinical Outcomes by Level of eGFR and Proteinuria
During median follow-up of 35 months (range, 0-59 months), 27 959 participants (3.0%) died, 5772 (0.6%) were hospitalized for myocardial infarction, 771 (0.08%) initiated renal replacement therapy, and 2514 (0.4%) experienced a doubling of serum creatinine. The age-adjusted rates of these
Within each stratum of eGFR, there was substantial variability in risk with participants who had heavier proteinuria having markedly increased adjusted rates of all 4 adverse outcomes (TABLE 2; eFigure 1, available at http: //www.jama.com). The adjusted mortality risk was more than 2-fold higher among individuals with heavy proteinuria and eGFR of 60 mL/min/1.73 m2 or greater as compared with those with eGFR of 45 to 59.9 mL/min/1.73 m2 and normal protein excretion (rate ratio, 2.5; 95% confidence interval [CI], 2.32.7). Significant interactions between eGFR and proteinuria were observed for death, initiation of renal replacement, and doubling of serum creatinine
Table 2. Adjusted Rates Per 1000 Person-Years of Clinical Outcomes by Level of eGFR and Proteinuria Measured by Dipstick a
Proteinuria All-Cause Normal eGFR 60 d Events, No. 12 157 Patients, No. 754 158 Rate (95% CI) 2.7 (2.6-2.8) eGFR 45-59.9 d Events, No. 4513 Patients, No. 68 768 Rate (95% CI) 2.9 (2.7-3.0) eGFR 30-44.9 d Events, No. Patients, No. Rate (95% CI) eGFR 15-29.9 d Events, No. Patients, No. Rate (95% CI) 2162 11 823 4.0 (3.7-4.2) Mortality b Heavy 722 8013 7.2 (6.6-7.8) 514 2294 7.2 (6.5-7.8) 511 1594 7.5 (6.8-8.2) Myocardial Normal Infarction b Heavy End-stage Renal Disease b Normal Mild Heavy 35 8013 1.0 (0.7-1.4) 39 2294 4.3 (3.1-6.1) 103 1594 16.1 (12.5-20.7) Doubling of Serum Creatinine c Normal 739 487 335 0.6 (0.5-0.6) 269 58 562 0.9 (0.8-1.0) 178 10 926 2.0 (1.7-2.4) 73 1685 4.5 (3.5-5.9) Mild 223 39 835 1.6 (1.4-1.9) 106 7635 2.1 (1.7-2.5) 133 3004 4.7 (3.9-5.7) Heavy 146 5867 5.9 (5.0-7.0) 144 2011 10.0 (8.3-11.9) 177 1462 12.8 (10.7-15.3)
Mild 3191 58 400 5.8 (5.5-6.0) 1598 8783 5.2 (4.9-5.5) 1059 3296 5.8 (5.4-6.2)
Mild
3171 474 103 62 11 754 158 58 400 8013 754 158 58 400 0.9 1.3 1.6 0.03 0.05 (0.9-1.0) (1.2-1.5) (1.3-2.0) (0.02-0.03) (0.03-0.09) 1011 200 73 68 768 8783 2294 1.2 1.3 1.8 (1.1-1.2) (1.1-1.5) (1.4-2.3) 359 116 73 11 823 3296 1594 1.4 1.5 2.1 (1.3-1.6) (1.2-1.8) (1.6-2.7) 27 68 768 0.2 (0.1-0.2) 36 11 823 1.3 (0.9-1.8) 61 1801 12.7 (9.3-17.3) 19 8783 0.7 (0.5-1.2) 40 3296 4.2 (3.0-6.0)
644 481 407 84 49 59 1801 1078 977 1801 1078 977 6.7 9.1 10.4 2.1 2.2 3.3 (6.2-7.3) (8.2-10.0) (9.3-11.6) (1.6-2.6) (1.6-2.9) (2.5-4.3)
Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate; HIV, human immunodeficiency virus. a Adjusted for age; sex; diabetes; hypertension; socioeconomic status; and history of cancer, cerebrovascular disease, congestive heart failure, chronic obstructive pulmonary disease, dementia, diabetes with end organ damage, diabetes without chronic complication, AIDS/HIV, metastatic solid tumor, myocardial infarction, mild liver disease, moderate or severe liver disease, paralysis, peptic ulcer disease, peripheral vascular disease, renal disease, and rheumatic disease. In this analysis, dipstick urinalysis was used to classify participants with respect to proteinuria: normal (urine dipstick negative), mild (urine dipstick trace or 1 ), or heavy (urine dipstick 2 ). b n=920 985 for all-cause mortality, myocardial infarction, and end-stage renal disease. c n=620 231 for doubling of serum creatinine at end of follow-up. d Unit of measure for eGFR is mL/min/1.73 m2. The tests for linear trend across eGFR categories and across proteinuria categories were all significant at the P .001 level.
426
such that the additional risk of heavier proteinuria was reduced at lower eGFR (all P for interaction statistically significant at .001)but not for myocardial infarction (P for interaction, .08). However, the difference in risk associated with moderate or heavy proteinuria (as compared with those without proteinuria) appeared clinically relevant within every eGFR stratum and for all 4 clinical outcomes.
Sensitivity Analyses
Results were consistent when analyses were restricted to the subset of 102 701 participants who had urinary ACR measurements performed (TABLE 3; eFigure 2). Specifically, risk increased progressively at levels of eGFR below 60 mL/min/1.73 m2 and with mild or heavy proteinuria within all eGFR stratafor all 4 clinical outcomes (adjusted rate ratio for mortality, 2.3 [95% CI, 2.0-2.6] for individuals with heavy proteinuria and eGFR of 60 mL/min/1.73 m2 or greater as compared with those with
eGFR of 45 to 59.9 mL/min/1.73 m2 and normal protein excretion). Next we repeated analyses using a more conservative definition of heavy proteinuria (ACR 2000 mg/g). Compared with those without significant proteinuria, participants with ACRs greater than 2000 mg/g had markedly elevated rates of adverse outcomes. For example, among participants with eGFRs of 45 to 59.9 mL/min/ 1.73 m2, those with heavy proteinuria by this definition had adjusted rates of 21.5 (95% CI,15.5-29.9), 11.2 (95% CI, 6.419.8), and 27.9 (95% CI, 17.6-44.2) per 1000 person-years, respectively, for mortality, myocardial infarction, and initiation of renal placement therapy, as compared with rates of 7.0 (95% CI, 6.37.6), 3.7 (95% CI, 3.2-4.3), and 0.3 (95% CI, 0.2-0.6), respectively, for those without proteinuria. Because current guidelines for the classification of CKD describe eGFR between 60 and 90 mL/min/1.73 m2 as mildly reduced, we examined the prognostic value of proteinuria within
this category specifically. Among the 597 870 participants, a graded increase in the adjusted rate of all-cause mortality was seen with rates of 2.2 (95% CI, 2.1-2.3), 4.3 (95% CI, 4.1-4.6), and 5.1 (95% CI, 4.7-5.6) per 1000 personyears among participants with no, mild, or heavy proteinuria, respectively (P for trend .001). Similar findings were seen for the outcomes of myocardial infarction (rates, 1.0 [95% CI, 0.9-1.0], 1.4 [95% CI, 1.2-1.5], and 1.6 [95% CI, 1.21.9]; P for trend .001), initiation of renal replacement therapy (rates, 0.02 [95% CI, 0.02-0.03], 0.04 [95% CI, 0.02-0.09], and 0.8 [95% CI, 0.5-1.3]; P for trend .001), or doubling of serum creatinine (rates, 0.3 [95% CI, 0.30.4], 0.9 [95% CI, 0.7-1.1], and 2.8 [95% CI, 2.2-3.6]; P for trend .001). Because there has been controversy about whether the prognostic implications of CKD vary in younger and older populations, we repeated analyses stratifying on age. All findings were similar among participants who were 65 years
Table 3. Adjusted Rates Per 1000 Person-Years of Clinical Outcomes by Level of eGFR and Proteinuria Measured by Albumin-Creatinine Ratio a
Proteinuria All-Cause Normal eGFR 60 d Events, No. Patients, No. Rate (95% CI) eGFR 45-59.9 d Events, No. Patients, No. Rate (95% CI) eGFR 30-44.9 d Events, No. Patients, No. Rate (95% CI) eGFR 15-29.9 d Events, No. Patients, No. Rate (95% CI) 1611 64 146 6.3 (6.0-6.7) 643 10 316 7.0 (6.4-7.6) Mortality b Heavy Myocardial Normal Infarction b Heavy End-stage Renal Disease b Normal Mild Heavy Doubling of Serum Creatinine c Normal Mild Heavy
Mild
Mild
809 268 619 249 77 13 5 30 137 104 111 14 597 2805 64 146 14 597 2805 64 146 14 597 2805 51 249 12 672 2539 9.9 15.9 3.0 4.2 6.4 0.06 0.09 2.45 1.0 2.8 13.4 (9.2-10.8) (14.0-18.1) (2.8-3.3) (3.7-4.8) (5.1-8.1) (0.03-0.10) (0.04-0.23) (1.70-3.59) (0.9-1.2) (2.3-3.4) (11.0-16.4) 490 206 211 138 52 9 9 4 3520 1126 10 316 3520 1126 10 316 3520 1126 11.9 18.0 3.7 5.9 7.3 0.3 0.9 8.3 (10.7-13.2) (15.6-20.9) (3.2-4.3) (4.9-7.2) (5.5-9.7) (0.17-0.64) (0.49-1.82) (5.9-11.9) 10 2474 1.7 (0.8-3.2) 21 1624 4.8 (3.1-7.5) 49 58 110 9547 3298 1067 1.6 4.8 25.0 (1.2-2.1) (3.6-6.2) (20.2-30.5)
336 339 213 91 80 49 2474 1624 837 2474 1624 837 10.0 14.1 18.9 5.3 6.7 8.4 (8.9-11.3) (12.4-15.9) (16.2-21.9) (4.3-6.6) (5.2-8.6) (6.2-11.3)
7 37 42 120 837 2360 1549 800 27.3 4.1 6.6 33.4 (20.9-35.8) (2.9-5.7) (4.8-9.1) (27.1-41.2) 128 436 97.3 (75-127) 9 27 108 333 452 422 6.2 13.8 51.8 (3.2-12.0) (9.2-20.7) (40.8-66.5)
91 166 154 13 28 27 8 35 344 476 436 344 476 436 344 476 16.3 22.0 24.6 5.1 8.6 9.7 9.0 27.6 (13.0-20.5) (18.5-26.0) (20.5-29.6) (2.9-9.0) (5.7-12.8) (6.4-14.6) (4.4-18.5) (18.7-40.4)
Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate; HIV, human immunodeficiency virus. a Adjusted for age; sex; diabetes; hypertension; socioeconomic status; and history of cancer, cerebrovascular disease, congestive heart failure, chronic obstructive pulmonary disease, dementia, diabetes with end organ damage, diabetes without chronic complication, AIDS/HIV, metastatic solid tumor, myocardial infarction, mild liver disease, moderate or severe liver disease, paralysis, peptic ulcer disease, peripheral vascular disease, renal disease, and rheumatic disease. In this analysis, only urinary albumin-creatinine ratio (ACR) was used to classify participants with respect to proteinuria: normal (ACR 30 mg/g), mild (ACR 30-300 mg/g), or heavy (ACR 300 mg/g). b n=102 701 for all-cause mortality, myocardial infarction, and end-stage renal disease. c n=86 288 for doubling of serum creatinine at end of follow-up. d Unit of measure for eGFR is mL/min/1.73 m2. The tests for linear trend across eGFR categories and across proteinuria categories were all significant at the P .001 level.
427
and older as compared with those who were younger. Specifically, the risk of all 4 clinical outcomes increased significantly in both age strata with declining eGFR (all P for trend .001), as well as with heavier proteinuria (all P for trend .001). Finally, results with eGFR and ACR as continuous variables were consistent with categorical analyses. The increase in adjusted rate per 10mL/min/1.73 m2 decrease in eGFR was most pronounced for the outcome end-stage renal disease, followed by doubling of serum creatinine, myocardial infarction, and all-cause mortality (increase in rates, 2.17 [95% CI, 2.02-2.34], 1.15 [95% CI, 1.10-1.19], 1.09 [95% CI, 1.05-1.12], and 1.04 [95% CI, 1.03-1.06], respectively). Similar findings were seen per 10fold increase in ACR with an increase in adjusted rates of 1.92 (95% CI, 1.812.04), 1.76 (95% CI, 1.70-1.82), 1.18 (95% CI, 1.14-1.21), and 1.22 (95% CI, 1.21-1.24), respectively, for initiation of renal replacement therapy, doubling of serum creatinine, myocardial infarction, and all-cause mortality. COMMENT In this large, community-based cohort of all adults undergoing laboratory testing in a single Canadian province, we demonstrated that prognosis associated with a given level of eGFR varies substantially based on the presence and severity of proteinuria. In fact, patients with heavy proteinuria but without overtly abnormal eGFR appeared to have worse clinical outcomes than those with moderately reduced eGFR but without proteinuria. Results were consistent for 2 different measures of proteinuria; consistent for several clinically relevant outcomes, including all-cause mortality, myocardial infarction, and the need for renal replacement; and robust to multivariable adjustment and a variety of sensitivity analyses. These findings are important because current guidelines for the classification and staging of CKD are based on eGFR without explicit consideration of the severity of concomitant proteinuria.1 In addition, computerized re428
porting of eGFR (generally without consideration of proteinuria) is increasingly used to assist physicians in identifying patients at high risk of adverse outcomesor those who might benefit from specialist care.21 Although our findings do not directly address which patients would benefit from referral to a nephrologist, they do suggest that risk stratification performed in terms of eGFR alone is relatively insensitive to clinically relevant gradients in risk. Staging systems for the classification of disease are often used to group affected persons into categories that are associated with similar prognoses, generally in a fashion that assigns people with worse prognoses to more advanced stages.22 Although the introduction of the NKF-K/DOQI (National Kidney Foundation/Dialysis Outcomes Quality Initiative) scheme for classification of CKD represented a major advance for researchers and clinicians, our findings suggest that this scheme does not meet these 2 criteria. For example, the age-adjusted rates of all-cause mortality and kidney failure appear to vary up to 4- and 50-fold (depending on the severity of proteinuria) within a given stage as defined by the current scheme. Similarly, a patient with an eGFR of 80 mL/min/1.73 m2 and 3 proteinuria on dipstick reading (or ACR of 400 mg/g) would be assigned to stage 1 CKD under the current systemeven though his or her age-adjusted risks of death and the need for renal replacement therapy would be approximately 2 and 10 times higher, respectively, than an otherwise similar patient with an eGFR of 50 mL/min/ 1.73 m2 but no evidence of proteinuria (stage 3 disease). This latter finding is particularly striking given the high prevalence of stage 3 CKD (defined by eGFRs of 3059.9 mL/min/1.73 m2 with or without proteinuria) in our study, which accounts for the large majority of North American individuals with CKD.6 An additional finding of our analysis is that the risk is heterogeneous within this stage, even when it is defined by eGFR alone. As previously reported, the risk
of all-cause mortality in our study was markedly higher among participants with eGFRs of 30 to 44.9 mL/min/ 1.73 m2 than among those with eGFRs of 45 to 59.9 mL/min/1.73 m2.11 Our data extend this finding to other adverse outcomes, including myocardial infarction and progression to kidney failure. The heterogeneity of risk among the large number of people currently classified as having stage 3 CKD (even when stratified by proteinuria) suggest that consideration should be given to subdividing this stage as done in our analysis, as well as by proteinuria. Focusing clinical attention on people at highest risk (as defined by the intersection of eGFR and proteinuria) may prove to be a more cost-effective approach to preventing the complications of CKD, although further work is required to confirm this hypothesis. Although other equations23 and serum markers24 are available for estimating GFR, we used the MDRD Study equation because it is the most widely used at present. Current practice in Western countries emphasizes the use of ACR rather than dipstick urinalysis in the assessment of CKD.1 Although dipstick urinalysis has less favorable diagnostic properties than ACR for the assessment of proteinuria,25 it is also considerably less expensive. Our results suggest that dipstick urinalysis adds considerable prognostic information to that associated with eGFR alone and the magnitude of excess risk observed with heavy proteinuria appeared similar whether assessed by dipstick or by ACR. Because the majority of people with CKD worldwide live in low- or middle-income nations,26 our data support the further study of dipstick urinalysis as a valid alternative to ACR for risk stratification in resource-limited settings. Our study has limitations due to its observational nature. First, the cohort was limited to individuals who had an outpatient serum creatinine measurement and a measure of urinary protein performed as part of routine care and therefore does not include individuals who did not use medical
services. However, since we studied nearly 1 million individuals, and considering the universal nature of health care coverage in Alberta, this limitation is unlikely to invalidate our finding that proteinuria adds substantial prognostic value to that associated with eGFR alone. Second, proteinuria and albuminuria may have been misclassified because of the known variability of these measurements based on a single measurement.12 However, we attempted to reduce this misclassification by using all urine measurements in a 6-month period before and after the index eGFR. In addition, results were robust to use of 2 different measures of proteinuria and were consistent for multiple clinically relevant outcomes. Third, we assessed the incidence of doubling of serum creatinine during follow-up, which may have included some participants with acute renal failure as well as those with progression of CKD. However, since we excluded inpatient measurements of serum creatinine, and given that the prevalence of acute renal failure in the community is less than 1%,27 this likely accounted for the minority of such eventsalthough this degree of kidney function loss is clinically relevant whether due to acute or chronic disease. 28 Fourth, the follow-up in our study was relatively short to assess progression to kidney failure, especially for people with higher levels of baseline eGFR, although this is unlikely to have threatened the validity of our conclusions. Finally, we did not have information on characteristics such as use of alcohol, tobacco, and antihypertensive medications, which may have resulted in residual confounding. However, given the magnitude of the effect sizes observed in our study, it is unlikely that further adjustment for these covariates would negate the observed associations. In conclusion, we found that the risks of death, myocardial infarction, and progression to kidney failure at a given level of eGFR were independently increased in individuals with higher levels of proteinuria. These findings suggest that future revisions of the classification
Roth D; Modification of Diet in Renal Disease Study Group. A more accurate method to estimate glomerular filtration rate from serum creatinine. Ann Intern Med. 1999;130(6):461-470. 10. Ethnocultural portrait of Canada highlight tables, 2006 census. Statistics Canada. http://www12.statcan .ca/english/census06/data/highlights/ethnic/index .cfm?Lang=E. Accessed June 15, 2009. 11. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney disease and the risks of death, cardiovascular events, and hospitalization. N Engl J Med. 2004;351(13):1296-1305. 12. Lamb EJ, MacKenzie F, Stevens PE. How should proteinuria be detected and measured? Ann Clin Biochem. 2009;46(pt 3):205-217. 13. Sin DD, Svenson LW, Cowie RL, Man SF. Can universal access to health care eliminate health inequities between children of poor and nonpoor families? Chest. 2003;124(1):51-56. 14. Hux JE, Ivis F, Flintoft V, Bica A. Diabetes in Ontario. Diabetes Care. 2002;25(3):512-516. 15. Quan H, Khan N, Hemmelgarn BR, et al; Hypertension Outcome and Surveillance Team of the Canadian Hypertension Education Programs. Validation of a case definition to define hypertension using administrative data. Hypertension. 2009;54(6): 1423-1428. 16. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43 (11):1130-1139. 17. Austin PC, Daly PA, Tu JV. A multicenter study of the coding accuracy of hospital discharge administrative data for patients admitted to cardiac care units in Ontario. Am Heart J. 2002;144(2):290296. 18. Manns BJ, Mortis GP, Taub KJ, McLaughlin K, Donaldson C, Ghali WA. The Southern Alberta Renal Program database. Clin Invest Med. 2001;24(4): 164-170. 19. Premium assistance program: premium subsidy. Alberta Health and Wellness. http://www.health .alberta.ca/AHCIP/premium-subsidy.html. Accessed May 2, 2008. 20. Altman D. Practical Statistics for Medical Research (Statistics Texts). New York, NY: Chapman & Hall/CRC; 1990. 21. Akbari A, Swedko PJ, Clark HD, et al. Detection of chronic kidney disease with laboratory reporting of estimated glomerular filtration rate and an educational program. Arch Intern Med. 2004;164(16): 1788-1792. 22. Chen ML, Hsu CY. Should the K/DOQI definition of chronic kidney disease be changed? Am J Kidney Dis. 2003;42(4):623-625. 23. Levey AS, Stevens LA, Schmid CH, et al; CKDEPI (Chronic Kidney Disease Epidemiology Collaboration). A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9): 604-612. 24. Shlipak MG, Sarnak MJ, Katz R, et al. Cystatin C and the risk of death and cardiovascular events among elderly persons. N Engl J Med. 2005;352(20):20492060. 25. Ciavarella A, Silletti A, Forlani G, et al. A screening test for microalbuminuria in type 1 (insulindependent) diabetes. Diabetes Res Clin Pract. 1989; 7(4):307-312. 26. Hossain MP, Goyder EC, Rigby JE, El Nahas M. CKD and poverty. Am J Kidney Dis. 2009;53(1): 166-174. 27. Feest TG, Round A, Hamad S. Incidence of severe acute renal failure in adults. BMJ. 1993;306 (6876):481-483. 28. Coca SG, Yusuf B, Shlipak MG, Garg AX, Parikh CR. Long-term risk of mortality and other adverse outcomes after acute kidney injury. Am J Kidney Dis. 2009; 53(6):961-973. 429