Article
Kidney Function and Prevalent and Incident Frailty Lorien S. Dalrymple, Ronit Katz, Dena E. Rifkin, David Siscovick, Anne B. Newman, Linda F. Fried, Mark J. Sarnak, Michelle C. Odden, and Michael G. Shlipak
Summary Background and objectives Kidney disease is associated with physiologic changes that may predispose to frailty. This study sought to investigate whether lower levels of kidney function were associated with prevalent or incident frailty in Cardiovascular Health Study (CHS) participants. Design, setting, participants, & measurements CHS enrolled community-dwelling adults age $65 years between 1989–1990 and 1992–1993. To examine prevalent frailty, included were 4150 participants without stroke, Parkinson disease, prescribed medications for Alzheimer disease or depression, or severely impaired cognition. To examine incident frailty, included were a subset of 3459 participants without baseline frailty or development of exclusion criteria during follow-up. The primary predictor was estimated GFR (eGFR) calculated using serum cystatin C (eGFRcys). Secondary analyses examined eGFR using serum creatinine (eGFRSCr). Outcomes were prevalent frailty and incident frailty at 4 years of follow-up. Frailty was ascertained on the basis of weight loss, exhaustion, weakness, slowness, and low physical activity. Results The mean age was 75 years and the median eGFRcys was 73 ml/min per 1.73 m2. Among participants with an eGFRcys ,45 ml/min per 1.73 m2, 24% had prevalent frailty. In multivariable analysis and compared with eGFRcys $90 ml/min per 1.73 m2, eGFRcys categories of 45–59 (odds ratio [OR], 1.80; 95% confidence interval [CI], 1.17 to 2.75) and 15–44 (OR, 2.87; 95% CI, 1.72 to 4.77) were associated with higher odds of frailty, whereas 60–75 (OR, 1.14; 95% CI, 0.76 to 1.70) was not. In multivariable analysis, eGFRcys categories of 60–75 (incidence rate ratio [IRR], 1.72; 95% CI, 1.07 to 2.75) and 15–44 (IRR, 2.28; 95% CI, 1.23 to 4.22) were associated with higher incidence of frailty whereas 45–59 (IRR, 1.53; 95% CI, 0.90 to 2.60) was not. Lower levels of eGFRSCr were not associated with higher risk of prevalent or incident frailty.
Due to the number of contributing authors, the affiliations are provided in the Supplemental Material. Correspondence: Dr. Lorien S. Dalrymple, Department of Internal Medicine, Division of Nephrology, University of California Davis, 4150 V Street, #3500 PSSB, Sacramento, CA 95817. Email: lorien. dalrymple@ucdmc. ucdavis.edu
Conclusions In community-dwelling elders, lower eGFRcys was associated with a higher risk of prevalent and incident frailty whereas lower eGFRSCr was not. These findings highlight the importance of considering non-GFR determinants of kidney function. Clin J Am Soc Nephrol 8: 2091–2099, 2013. doi: 10.2215/CJN.02870313
Introduction CKD affects a large segment of the aged population (1), and older adults may be more vulnerable to the metabolic and hormonal disturbances associated with kidney disease. Although CKD is known to affect cardiovascular disease risk, in older adults it has also been associated with noncardiovascular complications, poor physical functioning, and unsuccessful aging (2–5). Of interest is whether kidney function is independently associated with frailty, defined as a lack of physiologic reserve. Identifying and clearly delineating whether kidney function is associated with frailty are important for the clinical care of older adults living with CKD. Frailty is also a strong risk factor for the development of disability and death (6). Few studies have examined whether kidney function is independently associated with frailty, and these studies have been limited by the use of serum creatinine–based estimates of kidney function, the examination of a referred population, and/or the use of modified frailty criteria (7–9). The misclassification of www.cjasn.org Vol 8 December, 2013
kidney function with use of serum creatinine is a particular concern for examining frailty because unintentional weight loss, poor physical function, and weakness are more likely to be associated with low muscle mass and low serum creatinine independent of kidney function, therefore obscuring the relation between kidney function and frailty. In this study, we examined whether lower levels of kidney function, as estimated by serum cystatin C, were associated with the risk of prevalent or incident frailty in Cardiovascular Health Study (CHS) participants. In secondary analyses, we examined the association between kidney function, as estimated by serum creatinine, and prevalent or incident frailty.
Materials and Methods Study Population The CHS is a prospective cohort study of adults 65 years and older living in the community. Participants were initially recruited between 1989 and 1990 from Forsyth County, North Carolina; Sacramento County, Copyright © 2013 by the American Society of Nephrology
2091
2092
Clinical Journal of the American Society of Nephrology
Figure 1. | Cohort selection. The selection of Cardiovascular Health Study participants to examine the association between kidney function and prevalent frailty. eGFRcys, estimated GFR using cystatin C; MMSE, Mini-Mental State Examination; 3MSE, modified Mini-Mental State Examination.
California; Washington County, Maryland; and Pittsburgh, Pennsylvania. An additional 687 African American participants were enrolled between 1992 and 1993. To be eligible participants had to provide informed consent; had to intend to remain within the enrollment area for at least 3 years; and could not be institutionalized, using a wheelchair within the home, receiving chemotherapy or radiation for cancer, or receiving hospice care. The study design and recruitment have been described in detail (10,11). The CHS was approved by institutional review boards at each participating clinical center and the University of Washington coordinating center. Participants provided informed consent. To examine the association between kidney function and the prevalence and incidence of frailty, we restricted the CHS cohort to participants who presented for the year 3 (1992–1993) evaluation. We chose year 3 as the baseline to ensure equal length of follow-up for incident frailty
between the original cohort enrolled at year 0 (1989– 1990) and the African American cohort enrolled at year 3 (1992–1993). We further excluded participants for the following: Parkinson disease, stroke, prescribed medications for Parkinson disease, Alzheimer disease or depression, a Mini-Mental State Examination score ,18, or a modified Mini-Mental State Examination score ,60; these exclusion criteria were applied to be consistent with the exclusion criteria used to validate the frailty phenotype in CHS (6). We further excluded patients with fewer than three frailty components measured, missing serum cystatin C, or baseline estimated GFR (eGFR) (calculated using serum cystatin C [eGFRcys]) ,15 ml/min per 1.73 m2. To examine the incidence of frailty, the cohort was further restricted to those without baseline frailty and those without interval development of the exclusion criteria used to validate the frailty phenotype between years 3 and 7 of the study.
Clin J Am Soc Nephrol 8: 2091–2099, December, 2013
Kidney Function and Frailty, Dalrymple et al.
2093
Table 1. Included Cardiovascular Health Study participant characteristics at 1992–1993 evaluation
Category of eGFRcys Characteristic
Age (yr) Women, n (%) Black, n (%) Smoking, n (%)a Never Former Current Education, n (%)a # Grade 9 High school Professional/ vocational BMI (kg/m2) Diabetes mellitus, n (%) Hypertension, n (%) Coronary heart disease, n (%) Heart failure, n (%) C-reactive protein (mg/L) Hemoglobin (g/dl) Albumin (g/dl) Total cholesterol (mg/dl) HDL cholesterol (mg/dl) LDL cholesterol (mg/dl) Triglycerides (mg/dl)
$90 ml/min per 1.73 m2 (n=715)
76–89 ml/min per 1.73 m2 (n=1155)
60–75 ml/min per 1.73 m2 (n=1366)
45–59 ml/min per 1.73 m2 (n=693)
15–44 ml/min per 1.73 m2 (n=221)
7364 506 (71) 213 (30)
7464 721 (62) 213 (18)
7565 729 (53) 161 (12)
7766 369 (53) 71 (10)
7866 107 (48) 26 (12)
338 (48) 301 (43) 62 (9)
548 (49) 472 (42) 109 (10)
609 (45) 602 (45) 136 (10)
287 (42) 336 (49) 58 (9)
92 (42) 100 (46) 25 (12)
105 (15) 255 (36) 353 (50)
157 (14) 452 (39) 544 (47)
198 (15) 522 (38) 643 (47)
146 (21) 248 (36) 298 (43)
47 (21) 90 (41) 84 (38)
26.064.4 123 (17)
26.664.6 137 (12)
27.264.7 189 (14)
27.765.0 125 (18)
27.565.2 40 (18)
373 (52)
600 (52)
749 (55)
435 (63)
158 (71)
95 (13)
188 (16)
285 (21)
205 (30)
73 (33)
19 (3)
38 (3)
45 (3)
72 (10)
34 (15)
1.99 (0.93, 4.72)
2.37 (1.09, 5.26)
2.49 (1.25, 5.45)
3.47 (1.52, 7.86)
4.88 (1.89, 9.15)
13.661.2
13.761.3
13.961.3
13.661.4
13.061.4
3.9660.26 211635
3.9260.27 210635
3.9260.27 208638
3.8960.27 206643
3.8460.27 202645
60615
55614
51613
48612
47614
127633
129632
129635
126634
120637
102 (77, 144)
118 (87, 162)
126 (93, 175)
138 (100, 199)
146 (113, 225)
Data presented as n (%), mean 6SD, or median (interquartile range). eGFRcys, estimated GFR using serum cystatin C. a Column totals less than total number of participants because of missing data: smoking history (n=75), education (n=8).
Data Collection Extensive baseline data were collected, including information on age, sex, self-reported race, coexisting illnesses, body composition, habits, medications, laboratory measures, and physical and cognitive function. Participants underwent semi-annual evaluation by telephone contact alternating with comprehensive annual examinations. Details regarding measures, data collection, and laboratory specimen collection and processing have previously been described in detail (10,12,13). Exposure and Outcome The primary exposure of interest was kidney function estimated by serum cystatin C. Serum cystatin C concentration was measured by particle-enhanced immunonephelometric assay (N Latex Cystatin C, Siemens; www.usa. siemens.com) and GFR was estimated using the 2008 CKD-Epidemiology Collaboration (CKD-EPI) cystatin C
equation without demographic coefficients: eGFRcys = 76.7 3 CysC21.19 (14). Although more recent equations have been developed, the 2008 CKD-EPI cystatin C equation was used in numerous prior reports from the CHS cohort and thus offers continuity with studies evaluating other outcomes. In sensitivity analyses, we compared results with the 2012 CKD-EPI cystatin C equation, which includes coefficients for age and sex (15). In secondary analyses, we estimated kidney function using the 2009 CKD-EPI creatinine equation (eGFRSCr) (16). The primary outcome of interest was frailty. Frailty was ascertained using previously validated criteria and defined as having at least three of five components: (1) shrinking: unintentional weight loss of $10 pounds or 5% in the past year; (2) poor endurance and energy: assessed using two questions from the Centers for Epidemiologic Studies– Depression scale; (3) weakness: grip strength in the lowest 20% adjusted for sex and body mass index; (4) slowness:
2094
Clinical Journal of the American Society of Nephrology
Figure 2. | Prevalence of frailty by level of kidney function. (A) Prevalence of frailty by categories of estimated GFR using cystatin C (eGFRcys). (B) Prevalence of frailty by categories of estimated GFR using cystatin C stratified by sex.
slowest 20% in the 15-foot walk adjusted for sex and height; and (5) low physical activity: lowest quintile for physical activity for each sex (6). The components of frailty were examined at the beginning of our study (year 3 of CHS) and after 4 years (year 7 of CHS). Statistical Analyses We examined the association between kidney function and prevalent frailty using logistic regression. Kidney function was examined continuously and by categories: 15–44, 45–59, 60–75, 76–89, and $90 ml/min per 1.73 m2. We examined stepwise logistic regression models: model 1, adjusted for age, sex, race; model 2, further adjusted for potential confounders, including diabetes mellitus, hypertension, coronary heart disease, heart failure, smoking, and body mass index; and model 3, further adjusted for
potential mediators, including C-reactive protein, hemoglobin, albumin, LDL cholesterol, and HDL cholesterol. Potential mediators were selected on the basis of prior studies. We examined the association between kidney function and the incidence of frailty using stepwise Poisson regression models, similar to the analyses of prevalent frailty. Poisson regression models were tested for overdispersion, of which there was no evidence. Participants with missing data were excluded from multivariate models requiring those elements. The numbers of participants with incomplete data were as follows: smoking history (n=75), education (n=8), hypertension (n=6), and diabetes history (n=56). The overall prevalence of frailty did not differ between those with and without missing data. To examine whether the association between kidney function and prevalent or incident frailty was modified by age, sex,
1.00 (reference) 0.88 (0.58 to 1.34) 1.15 (0.76 to 1.72) 1.75 (1.13 to 2.71) 2.68 (1.58 to 4.54) OR, odds ratio; CI, confidence interval; eGFRcys, estimated GFR using serum cystatin C. a Age, sex, and race. b Further adjusted for smoking, body mass index, diabetes mellitus, hypertension, coronary heart disease, and heart failure. c Further adjusted for C-reactive protein, hemoglobin, albumin, LDL cholesterol, and HDL cholesterol. d P=0.001.
1.00 (reference) 0.89 (0.59 to 1.36) 1.14 (0.76 to 1.70) 1.80 (1.17 to 2.75) 2.87 (1.72 to 4.77) 1.00 (reference) 0.92 (0.61 to 1.39) 1.22 (0.83 to 1.81) 2.09 (1.38 to 3.15) 3.56 (2.18 to 5.81) 44 67 108 106 54 715 1155 1366 693 221
1.00 (reference) 0.92 (0.61 to 1.38) 1.29 (0.89 to 1.88) 2.60 (1.77 to 3.83) 4.93 (3.14 to 7.77)
1.20 (1.11 to 1.29) 1.21 (1.13 to 1.30)d 1.26 (1.17 to 1.35) 1.33 (1.25 to 1.42) 379 4150
Unadjusted OR (95% CI) Patients with Frailty (n) Total Patients (n) Variable
Table 2. The association between kidney function (estimated GFR using cystatin C) and prevalent frailty
We included 4150 CHS participants in our study of prevalent frailty (Figure 1). The mean age was 75 years, 41% of participants were male, 17% were black, and the median eGFRcys was 73 ml/min per 1.73 m2. Of these patients, 22% had an eGFRcys of 15–59 ml/min per 1.73 m2. Participants with the lowest kidney function tended to be older and more likely to have hypertension, coronary heart disease, and heart failure (Table 1). The prevalence of frailty was higher in those with lower levels of kidney function (Figure 2A). For each level of eGFRcys, women had a higher prevalence of frailty than men (Figure 2B). There was a graded association between categories of eGFRcys and prevalent frailty, with the highest odds of frailty in those with the lowest kidney function. In multivariable analysis adjusting for potential confounders, eGFRcys categories of 45–59 and 15–44 ml/min per 1.73 m2 were associated with 80% and 187% higher odds of frailty, respectively, compared with an eGFRcys $ 90 ml/min per 1.73 m2. Further adjustment for potential mediators did not materially alter these associations (Table 2). The association between kidney function and prevalent frailty was not modified by age, sex, or race (P for interaction = 0.41, 0.94, and 0.56). When the individual components of frailty were examined, lower kidney function was associated with higher odds of weight loss, exhaustion, walking slowness, and low physical activity (Table 3). The cohort was further restricted to examine the incidence of frailty. Participants with prevalent frailty (n=379) or those who developed stroke or cognitive deficits or were prescribed medications for depression, Parkinson disease, or Alzheimer disease during follow-up (n=312) were excluded, resulting in a cohort of 3459 participants. During follow-up, 214 (6%) participants became frail and 318 (9%) died. The rate of death was highest in participants with the lowest level of kidney function. During follow-up, 32% of participants with an eGFRcys of 15–44 ml/min per 1.73 m2 died compared with only 5% of those with an eGFRcys $90 ml/min per 1.73 m2. In the demographic-adjusted analysis, the eGFRcys categories ,76 ml/min per 1.73 m2 were associated with higher risk of incident frailty. After adjustment for potential confounders, eGFRcys categories of 60–75 and 15–44 were associated with 72% and 128% higher incidence of frailty, respectively, compared with an eGFRcys $90 ml/min per 1.73 m2. Further adjusting for potential mediators did not materially alter these associations (Table 4). The association between kidney function and incident frailty did not differ by age, sex, or race (P=0.11, 0.78, and 0.13). When we repeated analyses using the 2012 CKD-EPI cystatin C equation, associations with prevalent frailty were smaller in magnitude for the eGFRCKD-EPIcys 45–59 (odds ratio [OR], 1.47; 95% confidence interval [CI], 0.89 to 2.43) and 15–44 (OR, 2.44; 95% CI, 1.43 to 4.19) categories and no longer statistically significant for those with an eGFRCKD-EPIcys $45 ml/min per 1.73 m2 (Supplemental
OR Adjusted for Demographic Characteristicsa (95% CI)
Results
Continuous eGFRcys per 10 ml/min per 1.73 m2 decrease Category of eGFRcys $90 ml/min per 1.73 m2 76–89 ml/min per 1.73 m2 60–75 ml/min per 1.73 m2 45–59 ml/min per 1.73 m2 15–44 ml/min per 1.73 m2
or race, interaction terms were included and tested for significance. Statistical analyses were performed using S-Plus (version 8.0, Tibco, Seattle, WA) and SPSS statistical software (version 15.0.1.1, SPSS, Inc., Chicago, IL).
OR Further Adjusted for Potential Mediatorsc (95% CI)
Kidney Function and Frailty, Dalrymple et al.
OR Further Adjusted for Confoundersb (95% CI)
Clin J Am Soc Nephrol 8: 2091–2099, December, 2013
2095
2096
Clinical Journal of the American Society of Nephrology
Table 3. Cross-sectional association between kidney function (estimated GFR using cystatin C) and individual components of frailty
Frailty Component
Unadjusted OR (95% CI)
OR Adjusted for Demographic Characteristicsa (95% CI)
OR Further Adjusted for Confoundersb (95% CI)
OR Further Adjusted for Potential Mediatorsc (95% CI)
Exhaustion Low physical activity Walking slowness Weakness Weight loss
1.09 (1.04 to 1.13) 1.12 (1.08 to 1.17)
1.11 (1.05 to 1.16) 1.13 (1.08 to 1.18)
1.08 (1.03 to 1.13) 1.09 (1.04 to 1.14)
1.07 (1.02 to 1.13) 1.09 (1.04 to 1.14)
1.26 (1.20 to 1.31) 1.14 (1.09 to 1.19) 1.14 (1.05 to 1.22)
1.25 (1.19 to 1.31) 1.06 (1.01 to 1.11) 1.09 (1.01 to 1.18)
1.20 (1.14 to 1.26) 1.03 (0.98 to 1.08) 1.15 (1.06 to 1.25)
1.18 (1.12 to 1.24) 1.02 (0.97 to 1.07) 1.13 (1.04 to 1.23)
Odds ratio for each frailty component associated with continuous eGFRcys per 10 ml/min per 1.73 m2 decrease. OR, odds ratio; CI, confidence interval. a Age, sex, and race b Further adjusted for smoking, body mass index, diabetes mellitus, hypertension, coronary heart disease, and heart failure. c Further adjusted for C-reactive protein, hemoglobin, albumin, LDL cholesterol, and HDL cholesterol.
Table 1). An eGFRCKD-EPIcys ,76 ml/min per 1.73 m2 was moderately associated with incident frailty in the demographic-adjusted model but not in the fully adjusted models (Supplemental Table 2). In secondary analyses, we examined the association between serum creatinine–based estimates of kidney function and frailty in participants with measured serum creatinine (n=4085). We found an eGFRSCr of 60–89 ml/min per 1.73 m2 was associated with a significantly lower risk of prevalent frailty compared with an eGFRSCr $90 ml/min per 1.73 m2. Lower levels of eGFRSCr were not significantly associated with prevalent frailty. An eGFRSCr of 76–89 was associated with a significantly lower risk of incident frailty compared with an eGFRSCr $ 90 ml/min per 1.73 m2 whereas other levels of eGFRSCr were not associated with incident frailty (Supplemental Tables 3 and 4). We found substantial reclassification of kidney function when using eGFRcys compared with eGFRSCr, with 55% of participants reclassified into a different category of kidney function based on the following groups: 15–44, 45–59, 60– 75, 76–89, and $90 ml/min per 1.73 m2 (Supplemental Table 5).
Discussion In summary, we found that lower levels of kidney function, as measured by eGFRcys, were independently associated with higher risk of prevalent and incident frailty among older community-dwelling adults. Nearly one fourth of older adults with an eGFRcys ,45 ml/min per 1.73 m2 had prevalent frailty. The components of prevalent frailty associated with low levels of kidney function were exhaustion, low physical activity, walking slowness, and weight loss. Older adults with an eGFRcys ,45 ml/min per 1.73 m2 had a doubling of the risk of incident frailty compared with those with normal kidney function. The findings were attenuated when we used the 2012 CKD-EPI cystatin C equation adjusted for demographic characteristics. However, the 2012 CKD-EPI cystatin C equation was not derived in an elderly cohort, and the age coefficient may not be optimal in elderly cohorts; using the
2012 CKD-EPI cystatin C equation resulted in a higher number of participants being classified into lower levels of kidney function. Of note, the odds ratios and incidence rate ratios associated with each 10 ml/min per 1.73 m2 lower level of kidney function were nearly identical when we compared the 2008 CKD-EPI cystatin C equation without demographic coefficients and the 2012 CKD-EPI cystatin C equation. We did not observe lower levels of eGFRSCr to be associated with a higher risk of prevalent frailty. Instead, we found an eGFRSCr of 60–89 ml/min per 1.73 m2 was associated with a significantly lower odds of prevalent frailty compared with an eGFRSCr $90 ml/min per 1.73 m2. Similarly, an eGFRSCr of 76–89 ml/min per 1.73 m2 was associated with a significantly lower risk of incident frailty compared with an eGFRSCr $90 ml/min per 1.73 m2. One explanation for the difference in findings between eGFRSCr and eGFRcys is the reclassification of kidney function with use of serum creatinine–based versus cystatin C– based estimates of kidney function, as 55% of participants in our cohort were in discrepant categories of kidney function based on the two measures. These findings highlight the importance of considering non-GFR determinants of kidney function measurements, in particular muscle mass, for examining outcomes such as frailty. Of note, eGFRSCr may not be an accurate measurement of kidney function in elderly adults with low muscle mass, and alternative markers of glomerular filtration, such as cystatin C, may be particularly important in this setting (17). Similar to the 2012 CKD-EPI cystatin C equation, the CKD-EPI eGFRSCr equation was not developed in an elderly cohort, and therefore the age coefficient may not be accurate in elderly cohorts such as ours. Because creatinine is known to be influenced by health status and muscle mass, we believe cystatin C is a better marker of GFR. We chose the 2008 CKD-EPI cystatin C equation because it does not impose interactions of age and sex with cystatin C in estimating kidney function. Similar to our findings in this study, we have previously shown that eGFRcys estimated using the 2008 CKD-EPI equation without demographic coefficients is highly linked to adverse outcomes in the
1.00 (reference) 1.37 (0.83 to 2.27) 1.65 (1.02 to 2.67) 1.49 (0.87 to 2.57) 2.08 (1.09 to 3.98) IRR, incidence rate ratio; eGFRcys, estimated glomerular filtration rate using serum cystatin C. a Age, sex, and race b Further adjusted for smoking, body mass index, diabetes mellitus, hypertension, coronary heart disease, and heart failure. c Further adjusted for C-reactive protein, hemoglobin, albumin, LDL cholesterol, and HDL cholesterol. c P=0.04.
1.00 (reference) 1.50 (0.92 to 2.45) 1.72 (1.07 to 2.75) 1.53 (0.90 to 2.60) 2.28 (1.23 to 4.22) 1.00 (reference) 1.58 (0.98 to 2.54) 1.90 (1.20 to 3.00) 2.00 (1.21 to 3.31) 3.08 (1.67 to 5.68) 23 55 78 41 17 622 992 1160 537 148
1.00 (reference) 1.49 (0.92 to 2.39) 1.82 (1.15 to 2.87) 2.19 (1.33 to 3.59) 3.69 (2.04 to 6.68)
1.09 (1.00 to 1.18) 1.09 (1.00 to 1.17)c 1.18 (1.09 to 1.27) 214 3459
Continuous eGFRcys per 10 ml/min per 1.73 m2 decrease Category of eGFRcys $90 ml/min per 1.73 m2 76–89 ml/min per 1.73 m2 60–75 ml/min per 1.73 m2 45–59 ml/min per 1.73 m2 15–44 ml/min per 1.73 m2
Variable
1.15 (1.06 to 1.24)
IRR Further Adjusted for Potential Mediatorsc (95% CI) Unadjusted IRR (95% CI) Patients with Frailty (n) Total Patients (n)
Table 4. Association between kidney function (estimated GFR using cystatin C) and incident frailty
IRR Adjusted for Demographic Characteristicsa (95% CI)
IRR Further Adjusted for Potential Confoundersb (95% CI)
Clin J Am Soc Nephrol 8: 2091–2099, December, 2013
Kidney Function and Frailty, Dalrymple et al.
2097
elderly and thus offers continuity with prior studies evaluating other outcomes (2,18). Frailty is exceptionally common in patients with stage 5 CKD who are undergoing dialysis (19). The extent to which kidney dysfunction independently increases the risk of frailty and the level of kidney function at which the prevalence of frailty substantially rises are important to understand. Similar to our findings, previous studies have shown an association between lower levels of kidney function and prevalent frailty. Wilhelm-Leen et al. (9) estimated kidney function using the Mayo quadratic equation and found that 21% of participants in the Third National Health and Nutrition Examination Survey with an eGFR ,45 ml/min per 1.73 m2 were frail according to modified frailty criteria. In a previous study of CHS participants, we found serum creatinine $1.5 mg/dl in men and 1.3 mg/dl in women to be associated with a nearly 80% higher odds of prevalent frailty (8). Roshanravan et al. examined 336 Seattle Kidney Study participants, a cohort of patients referred for nephrology care, and estimated kidney function using the CKD-EPI equation with cystatin C, age, sex, and race (7). In their study, the relative prevalence of frailty was 2.1- and 2.8-fold higher for participants with an eGFR ,30 and 30–44 ml/min per 1.73 m2, respectively, compared with an eGFR .60 ml/min per 1.73 m2 (7). Compared with these earlier studies using serum creatinine, we used a more precise measure of kidney function, serum cystatin C, and compared with Roshanravan et al. we examined a community-based cohort. In addition, we evaluated incident frailty, which to our knowledge has not been studied previously, and found that an eGFRcys of 60–75 and 15–44 ml/min per 1.73 m2 were independently associated with a higher risk of incident frailty. The lack of a statistically significant association in those with an eGFRcys of 45–59 ml/min per 1.73 m2 may be due to a lack of power given the small number of participants with more advanced CKD or chance variability resulting in a nonsignificant association. Our study makes important contributions to the current field of research. We examined the association between kidney function and frailty in an elderly cohort using a more precise estimate of kidney function, cystatin C, and found a strong and graded association. This finding was completely obscured when evaluated by a more common estimate of kidney function based on serum creatinine. Our study is one of only two studies to examine the association between cystatin C and prevalent frailty, and ours is unique because of its setting within a sample of community-dwelling elders and use of the original validated criteria for frailty. Our study is the first to evaluate incident frailty and to uncover an independent association between lower levels of kidney function and a higher risk of incident frailty—an important finding that warrants further study. Our findings highlight that frailty is not a phenomena that starts with dialysis initiation, but, like many complications of CKD, the processes begin before the need for renal replacement therapy. Our study makes important contributions to the care of older patients with CKD because it highlights the notably high prevalence of frailty and the heightened risk for incident frailty in this special
2098
Clinical Journal of the American Society of Nephrology
population and informs their care and management. Reduced kidney function may be related to accelerated loss of lean muscle mass (20), which is thought to be an underlying process in frailty (21). Further examining whether specific interventions in this population, aimed at mitigating or preventing the development of frailty, warrant specific study as frailty in CKD has been linked to a higher risk of dialysis therapy or death (7). Our study had many strengths. We used a measure of kidney function, serum cystatin C, that is not related to muscle mass, a particularly important consideration for examining conditions such as frailty. We had comprehensive ascertainment of coexisting illnesses, allowing for appropriate exclusions and adjustment for potential confounders. We used validated and well accepted criteria to identify frailty. Last, we examined incident frailty, allowing us to further elaborate on the relation between kidney function and frailty. Our study had limitations that warrant consideration. Our examination of the association between kidney function and prevalent frailty was cross-sectional in nature and as such cannot establish causality. Frailty was reassessed at 4 years after the start of our study, and we cannot exclude that participants developed frailty and died before reassessment for frailty; this may have occurred disproportionately among those with the lowest kidney function, obscuring the association between kidney function and frailty and contributing to an underestimate of the true incidence of frailty. In addition, some diseases we considered as confounders, such as hypertension and heart failure, may actually be mediators, thereby attenuating the observed association between kidney function and frailty. Additionally, the longitudinal analysis examined kidney function only at one point in time. Participants with worse kidney function may be more likely to have subsequent declines in kidney function, and our study does not capture the impact of trajectories of decline in kidney function on incident frailty. Furthermore, we cannot exclude the possibility that unmeasured non-GFR determinants of cystatin C account for some of the observed association between eGFRcys and frailty. Last, we did not have measures of urinary protein and therefore cannot examine whether the association between levels of kidney function and frailty are modified by the presence of proteinuria. In summary, kidney function is independently associated with prevalent and incident frailty. Our findings highlight the importance of considering and assessing for frailty in older adults with CKD.
Acknowledgments This research was supported by contracts HHSN268201200036C, HHSN268200800007C, N01 HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, and grant HL080295 from the National Heart, Lung, and Blood Institute (NHLBI), with additional contribution from the National Institute of Neurological Disorders and Stroke (NINDS). Additional support was provided by AG023629 from the National Institute on Aging (NIA). L.S.D. was supported by UL1 RR024146 from the National Center for Research for Medical Research. A full list of principal CHS investigators and institutions can be found at https://chs-nhlbi. org/pi. The contents are solely the responsibility of the authors and
do not necessarily represent the official view of National Center for Research Resources or National Institutes of Health. The Cardiovascular Health Study was funded through contracts with the NHLBI and included substantial NHLBI involvement in the design and conduct of the study and the collection and management of the data. The funding source was not involved in data analysis, interpretation of the data, preparation, review, or approval of the manuscript. Findings from this study were presented as a Poster Presentation at the American Society of Nephrology Renal Week in San Diego, California, on November 2, 2012. Disclosures None. References 1. Stevens LA, Li S, Wang C, Huang C, Becker BN, Bomback AS, Brown WW, Burrows NR, Jurkovitz CT, McFarlane SI, Norris KC, Shlipak M, Whaley-Connell AT, Chen SC, Bakris GL, McCullough PA: Prevalence of CKD and comorbid illness in elderly patients in the United States: Results from the Kidney Early Evaluation Program (KEEP). Am J Kidney Dis 55(3 Suppl 2):S23–33, 2010 2. Dalrymple LS, Katz R, Kestenbaum B, de Boer IH, Fried L, Sarnak MJ, Shlipak MG: The risk of infection-related hospitalization with decreased kidney function. Am J Kidney Dis 59: 356–363, 2012 3. Fried LF, Katz R, Sarnak MJ, Shlipak MG, Chaves PH, Jenny NS, Stehman-Breen C, Gillen D, Bleyer AJ, Hirsch C, Siscovick D, Newman AB: Kidney function as a predictor of noncardiovascular mortality. J Am Soc Nephrol 16: 3728–3735, 2005 4. Fried LF, Lee JS, Shlipak M, Chertow GM, Green C, Ding J, Harris T, Newman AB: Chronic kidney disease and functional limitation in older people: Health, aging and body composition study. J Am Geriatr Soc 54: 750–756, 2006 5. Sarnak MJ, Katz R, Fried LF, Siscovick D, Kestenbaum B, Seliger S, Rifkin D, Tracy R, Newman AB, Shlipak MG; Cardiovascular Health Study: Cystatin C and aging success. Arch Intern Med 168: 147–153, 2008 6. Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener J, Seeman T, Tracy R, Kop WJ, Burke G, McBurnie MA; Cardiovascular Health Study Collaborative Research Group: Frailty in older adults: Evidence for a phenotype. J Gerontol A Biol Sci Med Sci 56: M146–M156, 2001 7. Roshanravan B, Khatri M, Robinson-Cohen C, Levin G, Patel KV, de Boer IH, Seliger S, Ruzinski J, Himmelfarb J, Kestenbaum B: A prospective study of frailty in nephrology-referred patients with CKD. Am J Kidney Dis 60: 912–921, 2012 8. Shlipak MG, Stehman-Breen C, Fried LF, Song X, Siscovick D, Fried LP, Psaty BM, Newman AB: The presence of frailty in elderly persons with chronic renal insufficiency. Am J Kidney Dis43: 861–867, 2004 9. Wilhelm-Leen ER, Hall YN, K Tamura M, Chertow GM: Frailty and chronic kidney disease: The Third National Health and Nutrition Evaluation Survey. Am J Med 122: 664–671, e662, 2009 10. Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller LH, Manolio TA, Mittelmark MB, Newman A, O’Leary DH, Psaty B, Rautaharju P, Tracy RP, Weiler PG: The Cardiovascular Health Study: Design and rationale. Ann Epidemiol 1: 263– 276, 1991 11. Tell GS, Fried LP, Hermanson B, Manolio TA, Newman AB, Borhani NO: Recruitment of adults 65 years and older as participants in the Cardiovascular Health Study. Ann Epidemiol 3: 358–366, 1993 12. Cushman M, Cornell ES, Howard PR, Bovill EG, Tracy RP: Laboratory methods and quality assurance in the Cardiovascular Health Study. Clin Chem 41: 264–270, 1995 13. Psaty BM, Kuller LH, Bild D, Burke GL, Kittner SJ, Mittelmark M, Price TR, Rautaharju PM, Robbins J: Methods of assessing prevalent cardiovascular disease in the Cardiovascular Health Study. Ann Epidemiol 5: 270–277, 1995
Clin J Am Soc Nephrol 8: 2091–2099, December, 2013
14. Stevens LA, Coresh J, Schmid CH, Feldman HI, Froissart M, Kusek J, Rossert J, Van Lente F, Bruce RD 3rd, Zhang YL, Greene T, Levey AS: Estimating GFR using serum cystatin C alone and in combination with serum creatinine: A pooled analysis of 3,418 individuals with CKD. Am J Kidney Dis 51: 395–406, 2008 15. Inker LA, Schmid CH, Tighiouart H, Eckfeldt JH, Feldman HI, Greene T, Kusek JW, Manzi J, Van Lente F, Zhang YL, Coresh J, Levey AS; CKD-EPI Investigators: Estimating glomerular filtration rate from serum creatinine and cystatin C. N Engl J Med 367: 20– 29, 2012 16. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J; CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration): A new equation to estimate glomerular filtration rate. Ann Intern Med 150: 604–612, 2009 17. Shlipak MG, Matsushita K, Arnlov J, Inker LA, Katz R, Polkinghorne KR, Rothenbacher D, Sarnak MJ, Astor BC, Coresh J, Levey AS, Gansevoort RT: Cystatin C versus creatinine in determining risk based on kidney function. N Engl J Med 2013 (in press) 18. Peralta CA, Katz R, Sarnak MJ, Ix J, Fried LF, De Boer I, Palmas W, Siscovick D, Levey AS, Shlipak MG: Cystatin C identifies chronic kidney disease patients at higher risk for complications. J Am Soc Nephrol 22: 147–155, 2011
Kidney Function and Frailty, Dalrymple et al.
2099
19. Bao Y, Dalrymple L, Chertow GM, Kaysen GA, Johansen KL: Frailty, dialysis initiation, and mortality in end-stage renal disease. Arch Intern Med 172: 1071–1077, 2012 20. Fried LF, Boudreau R, Lee JS, Chertow G, Kurella-Tamura M, Shlipak MG, Ding J, Sellmeyer D, Tylavsky FA, Simsonick E, Kritchevsky SB, Harris TB, Newman AB; Health, Aging and Body Composition Study: Kidney function as a predictor of loss of lean mass in older adults: Health, aging and body composition study. J Am Geriatr Soc 55: 1578–1584, 2007 21. Walston J, Hadley EC, Ferrucci L, Guralnik JM, Newman AB, Studenski SA, Ershler WB, Harris T, Fried LP: Research agenda for frailty in older adults: Toward a better understanding of physiology and etiology: Summary from the American Geriatrics Society/National Institute on Aging Research Conference on Frailty in Older Adults. J Am Geriatr Soc 54: 991–1001, 2006 Received: March 14, 2013 Accepted: August 8, 2013 Published online ahead of print. Publication date available at www. cjasn.org. This article contains supplemental material online at http://cjasn. asnjournals.org/lookup/suppl/doi:10.2215/CJN.02870313/-/ DCSupplemental.
Kidney Function and Prevalent and Incident Frailty Lorien S. Dalrymple, Ronit Katz, Dena E. Rifkin, David Siscovick, Anne B. Newman, Linda F. Fried, Mark J. Sarnak, Michelle C. Odden, and Michael G. Shlipak
Summary Background and objectives Kidney disease is associated with physiologic changes that may predispose to frailty. This study sought to investigate whether lower levels of kidney function were associated with prevalent or incident frailty in Cardiovascular Health Study (CHS) participants. Design, setting, participants, & measurements CHS enrolled community-dwelling adults age $65 years between 1989–1990 and 1992–1993. To examine prevalent frailty, included were 4150 participants without stroke, Parkinson disease, prescribed medications for Alzheimer disease or depression, or severely impaired cognition. To examine incident frailty, included were a subset of 3459 participants without baseline frailty or development of exclusion criteria during follow-up. The primary predictor was estimated GFR (eGFR) calculated using serum cystatin C (eGFRcys). Secondary analyses examined eGFR using serum creatinine (eGFRSCr). Outcomes were prevalent frailty and incident frailty at 4 years of follow-up. Frailty was ascertained on the basis of weight loss, exhaustion, weakness, slowness, and low physical activity. Results The mean age was 75 years and the median eGFRcys was 73 ml/min per 1.73 m2. Among participants with an eGFRcys ,45 ml/min per 1.73 m2, 24% had prevalent frailty. In multivariable analysis and compared with eGFRcys $90 ml/min per 1.73 m2, eGFRcys categories of 45–59 (odds ratio [OR], 1.80; 95% confidence interval [CI], 1.17 to 2.75) and 15–44 (OR, 2.87; 95% CI, 1.72 to 4.77) were associated with higher odds of frailty, whereas 60–75 (OR, 1.14; 95% CI, 0.76 to 1.70) was not. In multivariable analysis, eGFRcys categories of 60–75 (incidence rate ratio [IRR], 1.72; 95% CI, 1.07 to 2.75) and 15–44 (IRR, 2.28; 95% CI, 1.23 to 4.22) were associated with higher incidence of frailty whereas 45–59 (IRR, 1.53; 95% CI, 0.90 to 2.60) was not. Lower levels of eGFRSCr were not associated with higher risk of prevalent or incident frailty.
Due to the number of contributing authors, the affiliations are provided in the Supplemental Material. Correspondence: Dr. Lorien S. Dalrymple, Department of Internal Medicine, Division of Nephrology, University of California Davis, 4150 V Street, #3500 PSSB, Sacramento, CA 95817. Email: lorien. dalrymple@ucdmc. ucdavis.edu
Conclusions In community-dwelling elders, lower eGFRcys was associated with a higher risk of prevalent and incident frailty whereas lower eGFRSCr was not. These findings highlight the importance of considering non-GFR determinants of kidney function. Clin J Am Soc Nephrol 8: 2091–2099, 2013. doi: 10.2215/CJN.02870313
Introduction CKD affects a large segment of the aged population (1), and older adults may be more vulnerable to the metabolic and hormonal disturbances associated with kidney disease. Although CKD is known to affect cardiovascular disease risk, in older adults it has also been associated with noncardiovascular complications, poor physical functioning, and unsuccessful aging (2–5). Of interest is whether kidney function is independently associated with frailty, defined as a lack of physiologic reserve. Identifying and clearly delineating whether kidney function is associated with frailty are important for the clinical care of older adults living with CKD. Frailty is also a strong risk factor for the development of disability and death (6). Few studies have examined whether kidney function is independently associated with frailty, and these studies have been limited by the use of serum creatinine–based estimates of kidney function, the examination of a referred population, and/or the use of modified frailty criteria (7–9). The misclassification of www.cjasn.org Vol 8 December, 2013
kidney function with use of serum creatinine is a particular concern for examining frailty because unintentional weight loss, poor physical function, and weakness are more likely to be associated with low muscle mass and low serum creatinine independent of kidney function, therefore obscuring the relation between kidney function and frailty. In this study, we examined whether lower levels of kidney function, as estimated by serum cystatin C, were associated with the risk of prevalent or incident frailty in Cardiovascular Health Study (CHS) participants. In secondary analyses, we examined the association between kidney function, as estimated by serum creatinine, and prevalent or incident frailty.
Materials and Methods Study Population The CHS is a prospective cohort study of adults 65 years and older living in the community. Participants were initially recruited between 1989 and 1990 from Forsyth County, North Carolina; Sacramento County, Copyright © 2013 by the American Society of Nephrology
2091
2092
Clinical Journal of the American Society of Nephrology
Figure 1. | Cohort selection. The selection of Cardiovascular Health Study participants to examine the association between kidney function and prevalent frailty. eGFRcys, estimated GFR using cystatin C; MMSE, Mini-Mental State Examination; 3MSE, modified Mini-Mental State Examination.
California; Washington County, Maryland; and Pittsburgh, Pennsylvania. An additional 687 African American participants were enrolled between 1992 and 1993. To be eligible participants had to provide informed consent; had to intend to remain within the enrollment area for at least 3 years; and could not be institutionalized, using a wheelchair within the home, receiving chemotherapy or radiation for cancer, or receiving hospice care. The study design and recruitment have been described in detail (10,11). The CHS was approved by institutional review boards at each participating clinical center and the University of Washington coordinating center. Participants provided informed consent. To examine the association between kidney function and the prevalence and incidence of frailty, we restricted the CHS cohort to participants who presented for the year 3 (1992–1993) evaluation. We chose year 3 as the baseline to ensure equal length of follow-up for incident frailty
between the original cohort enrolled at year 0 (1989– 1990) and the African American cohort enrolled at year 3 (1992–1993). We further excluded participants for the following: Parkinson disease, stroke, prescribed medications for Parkinson disease, Alzheimer disease or depression, a Mini-Mental State Examination score ,18, or a modified Mini-Mental State Examination score ,60; these exclusion criteria were applied to be consistent with the exclusion criteria used to validate the frailty phenotype in CHS (6). We further excluded patients with fewer than three frailty components measured, missing serum cystatin C, or baseline estimated GFR (eGFR) (calculated using serum cystatin C [eGFRcys]) ,15 ml/min per 1.73 m2. To examine the incidence of frailty, the cohort was further restricted to those without baseline frailty and those without interval development of the exclusion criteria used to validate the frailty phenotype between years 3 and 7 of the study.
Clin J Am Soc Nephrol 8: 2091–2099, December, 2013
Kidney Function and Frailty, Dalrymple et al.
2093
Table 1. Included Cardiovascular Health Study participant characteristics at 1992–1993 evaluation
Category of eGFRcys Characteristic
Age (yr) Women, n (%) Black, n (%) Smoking, n (%)a Never Former Current Education, n (%)a # Grade 9 High school Professional/ vocational BMI (kg/m2) Diabetes mellitus, n (%) Hypertension, n (%) Coronary heart disease, n (%) Heart failure, n (%) C-reactive protein (mg/L) Hemoglobin (g/dl) Albumin (g/dl) Total cholesterol (mg/dl) HDL cholesterol (mg/dl) LDL cholesterol (mg/dl) Triglycerides (mg/dl)
$90 ml/min per 1.73 m2 (n=715)
76–89 ml/min per 1.73 m2 (n=1155)
60–75 ml/min per 1.73 m2 (n=1366)
45–59 ml/min per 1.73 m2 (n=693)
15–44 ml/min per 1.73 m2 (n=221)
7364 506 (71) 213 (30)
7464 721 (62) 213 (18)
7565 729 (53) 161 (12)
7766 369 (53) 71 (10)
7866 107 (48) 26 (12)
338 (48) 301 (43) 62 (9)
548 (49) 472 (42) 109 (10)
609 (45) 602 (45) 136 (10)
287 (42) 336 (49) 58 (9)
92 (42) 100 (46) 25 (12)
105 (15) 255 (36) 353 (50)
157 (14) 452 (39) 544 (47)
198 (15) 522 (38) 643 (47)
146 (21) 248 (36) 298 (43)
47 (21) 90 (41) 84 (38)
26.064.4 123 (17)
26.664.6 137 (12)
27.264.7 189 (14)
27.765.0 125 (18)
27.565.2 40 (18)
373 (52)
600 (52)
749 (55)
435 (63)
158 (71)
95 (13)
188 (16)
285 (21)
205 (30)
73 (33)
19 (3)
38 (3)
45 (3)
72 (10)
34 (15)
1.99 (0.93, 4.72)
2.37 (1.09, 5.26)
2.49 (1.25, 5.45)
3.47 (1.52, 7.86)
4.88 (1.89, 9.15)
13.661.2
13.761.3
13.961.3
13.661.4
13.061.4
3.9660.26 211635
3.9260.27 210635
3.9260.27 208638
3.8960.27 206643
3.8460.27 202645
60615
55614
51613
48612
47614
127633
129632
129635
126634
120637
102 (77, 144)
118 (87, 162)
126 (93, 175)
138 (100, 199)
146 (113, 225)
Data presented as n (%), mean 6SD, or median (interquartile range). eGFRcys, estimated GFR using serum cystatin C. a Column totals less than total number of participants because of missing data: smoking history (n=75), education (n=8).
Data Collection Extensive baseline data were collected, including information on age, sex, self-reported race, coexisting illnesses, body composition, habits, medications, laboratory measures, and physical and cognitive function. Participants underwent semi-annual evaluation by telephone contact alternating with comprehensive annual examinations. Details regarding measures, data collection, and laboratory specimen collection and processing have previously been described in detail (10,12,13). Exposure and Outcome The primary exposure of interest was kidney function estimated by serum cystatin C. Serum cystatin C concentration was measured by particle-enhanced immunonephelometric assay (N Latex Cystatin C, Siemens; www.usa. siemens.com) and GFR was estimated using the 2008 CKD-Epidemiology Collaboration (CKD-EPI) cystatin C
equation without demographic coefficients: eGFRcys = 76.7 3 CysC21.19 (14). Although more recent equations have been developed, the 2008 CKD-EPI cystatin C equation was used in numerous prior reports from the CHS cohort and thus offers continuity with studies evaluating other outcomes. In sensitivity analyses, we compared results with the 2012 CKD-EPI cystatin C equation, which includes coefficients for age and sex (15). In secondary analyses, we estimated kidney function using the 2009 CKD-EPI creatinine equation (eGFRSCr) (16). The primary outcome of interest was frailty. Frailty was ascertained using previously validated criteria and defined as having at least three of five components: (1) shrinking: unintentional weight loss of $10 pounds or 5% in the past year; (2) poor endurance and energy: assessed using two questions from the Centers for Epidemiologic Studies– Depression scale; (3) weakness: grip strength in the lowest 20% adjusted for sex and body mass index; (4) slowness:
2094
Clinical Journal of the American Society of Nephrology
Figure 2. | Prevalence of frailty by level of kidney function. (A) Prevalence of frailty by categories of estimated GFR using cystatin C (eGFRcys). (B) Prevalence of frailty by categories of estimated GFR using cystatin C stratified by sex.
slowest 20% in the 15-foot walk adjusted for sex and height; and (5) low physical activity: lowest quintile for physical activity for each sex (6). The components of frailty were examined at the beginning of our study (year 3 of CHS) and after 4 years (year 7 of CHS). Statistical Analyses We examined the association between kidney function and prevalent frailty using logistic regression. Kidney function was examined continuously and by categories: 15–44, 45–59, 60–75, 76–89, and $90 ml/min per 1.73 m2. We examined stepwise logistic regression models: model 1, adjusted for age, sex, race; model 2, further adjusted for potential confounders, including diabetes mellitus, hypertension, coronary heart disease, heart failure, smoking, and body mass index; and model 3, further adjusted for
potential mediators, including C-reactive protein, hemoglobin, albumin, LDL cholesterol, and HDL cholesterol. Potential mediators were selected on the basis of prior studies. We examined the association between kidney function and the incidence of frailty using stepwise Poisson regression models, similar to the analyses of prevalent frailty. Poisson regression models were tested for overdispersion, of which there was no evidence. Participants with missing data were excluded from multivariate models requiring those elements. The numbers of participants with incomplete data were as follows: smoking history (n=75), education (n=8), hypertension (n=6), and diabetes history (n=56). The overall prevalence of frailty did not differ between those with and without missing data. To examine whether the association between kidney function and prevalent or incident frailty was modified by age, sex,
1.00 (reference) 0.88 (0.58 to 1.34) 1.15 (0.76 to 1.72) 1.75 (1.13 to 2.71) 2.68 (1.58 to 4.54) OR, odds ratio; CI, confidence interval; eGFRcys, estimated GFR using serum cystatin C. a Age, sex, and race. b Further adjusted for smoking, body mass index, diabetes mellitus, hypertension, coronary heart disease, and heart failure. c Further adjusted for C-reactive protein, hemoglobin, albumin, LDL cholesterol, and HDL cholesterol. d P=0.001.
1.00 (reference) 0.89 (0.59 to 1.36) 1.14 (0.76 to 1.70) 1.80 (1.17 to 2.75) 2.87 (1.72 to 4.77) 1.00 (reference) 0.92 (0.61 to 1.39) 1.22 (0.83 to 1.81) 2.09 (1.38 to 3.15) 3.56 (2.18 to 5.81) 44 67 108 106 54 715 1155 1366 693 221
1.00 (reference) 0.92 (0.61 to 1.38) 1.29 (0.89 to 1.88) 2.60 (1.77 to 3.83) 4.93 (3.14 to 7.77)
1.20 (1.11 to 1.29) 1.21 (1.13 to 1.30)d 1.26 (1.17 to 1.35) 1.33 (1.25 to 1.42) 379 4150
Unadjusted OR (95% CI) Patients with Frailty (n) Total Patients (n) Variable
Table 2. The association between kidney function (estimated GFR using cystatin C) and prevalent frailty
We included 4150 CHS participants in our study of prevalent frailty (Figure 1). The mean age was 75 years, 41% of participants were male, 17% were black, and the median eGFRcys was 73 ml/min per 1.73 m2. Of these patients, 22% had an eGFRcys of 15–59 ml/min per 1.73 m2. Participants with the lowest kidney function tended to be older and more likely to have hypertension, coronary heart disease, and heart failure (Table 1). The prevalence of frailty was higher in those with lower levels of kidney function (Figure 2A). For each level of eGFRcys, women had a higher prevalence of frailty than men (Figure 2B). There was a graded association between categories of eGFRcys and prevalent frailty, with the highest odds of frailty in those with the lowest kidney function. In multivariable analysis adjusting for potential confounders, eGFRcys categories of 45–59 and 15–44 ml/min per 1.73 m2 were associated with 80% and 187% higher odds of frailty, respectively, compared with an eGFRcys $ 90 ml/min per 1.73 m2. Further adjustment for potential mediators did not materially alter these associations (Table 2). The association between kidney function and prevalent frailty was not modified by age, sex, or race (P for interaction = 0.41, 0.94, and 0.56). When the individual components of frailty were examined, lower kidney function was associated with higher odds of weight loss, exhaustion, walking slowness, and low physical activity (Table 3). The cohort was further restricted to examine the incidence of frailty. Participants with prevalent frailty (n=379) or those who developed stroke or cognitive deficits or were prescribed medications for depression, Parkinson disease, or Alzheimer disease during follow-up (n=312) were excluded, resulting in a cohort of 3459 participants. During follow-up, 214 (6%) participants became frail and 318 (9%) died. The rate of death was highest in participants with the lowest level of kidney function. During follow-up, 32% of participants with an eGFRcys of 15–44 ml/min per 1.73 m2 died compared with only 5% of those with an eGFRcys $90 ml/min per 1.73 m2. In the demographic-adjusted analysis, the eGFRcys categories ,76 ml/min per 1.73 m2 were associated with higher risk of incident frailty. After adjustment for potential confounders, eGFRcys categories of 60–75 and 15–44 were associated with 72% and 128% higher incidence of frailty, respectively, compared with an eGFRcys $90 ml/min per 1.73 m2. Further adjusting for potential mediators did not materially alter these associations (Table 4). The association between kidney function and incident frailty did not differ by age, sex, or race (P=0.11, 0.78, and 0.13). When we repeated analyses using the 2012 CKD-EPI cystatin C equation, associations with prevalent frailty were smaller in magnitude for the eGFRCKD-EPIcys 45–59 (odds ratio [OR], 1.47; 95% confidence interval [CI], 0.89 to 2.43) and 15–44 (OR, 2.44; 95% CI, 1.43 to 4.19) categories and no longer statistically significant for those with an eGFRCKD-EPIcys $45 ml/min per 1.73 m2 (Supplemental
OR Adjusted for Demographic Characteristicsa (95% CI)
Results
Continuous eGFRcys per 10 ml/min per 1.73 m2 decrease Category of eGFRcys $90 ml/min per 1.73 m2 76–89 ml/min per 1.73 m2 60–75 ml/min per 1.73 m2 45–59 ml/min per 1.73 m2 15–44 ml/min per 1.73 m2
or race, interaction terms were included and tested for significance. Statistical analyses were performed using S-Plus (version 8.0, Tibco, Seattle, WA) and SPSS statistical software (version 15.0.1.1, SPSS, Inc., Chicago, IL).
OR Further Adjusted for Potential Mediatorsc (95% CI)
Kidney Function and Frailty, Dalrymple et al.
OR Further Adjusted for Confoundersb (95% CI)
Clin J Am Soc Nephrol 8: 2091–2099, December, 2013
2095
2096
Clinical Journal of the American Society of Nephrology
Table 3. Cross-sectional association between kidney function (estimated GFR using cystatin C) and individual components of frailty
Frailty Component
Unadjusted OR (95% CI)
OR Adjusted for Demographic Characteristicsa (95% CI)
OR Further Adjusted for Confoundersb (95% CI)
OR Further Adjusted for Potential Mediatorsc (95% CI)
Exhaustion Low physical activity Walking slowness Weakness Weight loss
1.09 (1.04 to 1.13) 1.12 (1.08 to 1.17)
1.11 (1.05 to 1.16) 1.13 (1.08 to 1.18)
1.08 (1.03 to 1.13) 1.09 (1.04 to 1.14)
1.07 (1.02 to 1.13) 1.09 (1.04 to 1.14)
1.26 (1.20 to 1.31) 1.14 (1.09 to 1.19) 1.14 (1.05 to 1.22)
1.25 (1.19 to 1.31) 1.06 (1.01 to 1.11) 1.09 (1.01 to 1.18)
1.20 (1.14 to 1.26) 1.03 (0.98 to 1.08) 1.15 (1.06 to 1.25)
1.18 (1.12 to 1.24) 1.02 (0.97 to 1.07) 1.13 (1.04 to 1.23)
Odds ratio for each frailty component associated with continuous eGFRcys per 10 ml/min per 1.73 m2 decrease. OR, odds ratio; CI, confidence interval. a Age, sex, and race b Further adjusted for smoking, body mass index, diabetes mellitus, hypertension, coronary heart disease, and heart failure. c Further adjusted for C-reactive protein, hemoglobin, albumin, LDL cholesterol, and HDL cholesterol.
Table 1). An eGFRCKD-EPIcys ,76 ml/min per 1.73 m2 was moderately associated with incident frailty in the demographic-adjusted model but not in the fully adjusted models (Supplemental Table 2). In secondary analyses, we examined the association between serum creatinine–based estimates of kidney function and frailty in participants with measured serum creatinine (n=4085). We found an eGFRSCr of 60–89 ml/min per 1.73 m2 was associated with a significantly lower risk of prevalent frailty compared with an eGFRSCr $90 ml/min per 1.73 m2. Lower levels of eGFRSCr were not significantly associated with prevalent frailty. An eGFRSCr of 76–89 was associated with a significantly lower risk of incident frailty compared with an eGFRSCr $ 90 ml/min per 1.73 m2 whereas other levels of eGFRSCr were not associated with incident frailty (Supplemental Tables 3 and 4). We found substantial reclassification of kidney function when using eGFRcys compared with eGFRSCr, with 55% of participants reclassified into a different category of kidney function based on the following groups: 15–44, 45–59, 60– 75, 76–89, and $90 ml/min per 1.73 m2 (Supplemental Table 5).
Discussion In summary, we found that lower levels of kidney function, as measured by eGFRcys, were independently associated with higher risk of prevalent and incident frailty among older community-dwelling adults. Nearly one fourth of older adults with an eGFRcys ,45 ml/min per 1.73 m2 had prevalent frailty. The components of prevalent frailty associated with low levels of kidney function were exhaustion, low physical activity, walking slowness, and weight loss. Older adults with an eGFRcys ,45 ml/min per 1.73 m2 had a doubling of the risk of incident frailty compared with those with normal kidney function. The findings were attenuated when we used the 2012 CKD-EPI cystatin C equation adjusted for demographic characteristics. However, the 2012 CKD-EPI cystatin C equation was not derived in an elderly cohort, and the age coefficient may not be optimal in elderly cohorts; using the
2012 CKD-EPI cystatin C equation resulted in a higher number of participants being classified into lower levels of kidney function. Of note, the odds ratios and incidence rate ratios associated with each 10 ml/min per 1.73 m2 lower level of kidney function were nearly identical when we compared the 2008 CKD-EPI cystatin C equation without demographic coefficients and the 2012 CKD-EPI cystatin C equation. We did not observe lower levels of eGFRSCr to be associated with a higher risk of prevalent frailty. Instead, we found an eGFRSCr of 60–89 ml/min per 1.73 m2 was associated with a significantly lower odds of prevalent frailty compared with an eGFRSCr $90 ml/min per 1.73 m2. Similarly, an eGFRSCr of 76–89 ml/min per 1.73 m2 was associated with a significantly lower risk of incident frailty compared with an eGFRSCr $90 ml/min per 1.73 m2. One explanation for the difference in findings between eGFRSCr and eGFRcys is the reclassification of kidney function with use of serum creatinine–based versus cystatin C– based estimates of kidney function, as 55% of participants in our cohort were in discrepant categories of kidney function based on the two measures. These findings highlight the importance of considering non-GFR determinants of kidney function measurements, in particular muscle mass, for examining outcomes such as frailty. Of note, eGFRSCr may not be an accurate measurement of kidney function in elderly adults with low muscle mass, and alternative markers of glomerular filtration, such as cystatin C, may be particularly important in this setting (17). Similar to the 2012 CKD-EPI cystatin C equation, the CKD-EPI eGFRSCr equation was not developed in an elderly cohort, and therefore the age coefficient may not be accurate in elderly cohorts such as ours. Because creatinine is known to be influenced by health status and muscle mass, we believe cystatin C is a better marker of GFR. We chose the 2008 CKD-EPI cystatin C equation because it does not impose interactions of age and sex with cystatin C in estimating kidney function. Similar to our findings in this study, we have previously shown that eGFRcys estimated using the 2008 CKD-EPI equation without demographic coefficients is highly linked to adverse outcomes in the
1.00 (reference) 1.37 (0.83 to 2.27) 1.65 (1.02 to 2.67) 1.49 (0.87 to 2.57) 2.08 (1.09 to 3.98) IRR, incidence rate ratio; eGFRcys, estimated glomerular filtration rate using serum cystatin C. a Age, sex, and race b Further adjusted for smoking, body mass index, diabetes mellitus, hypertension, coronary heart disease, and heart failure. c Further adjusted for C-reactive protein, hemoglobin, albumin, LDL cholesterol, and HDL cholesterol. c P=0.04.
1.00 (reference) 1.50 (0.92 to 2.45) 1.72 (1.07 to 2.75) 1.53 (0.90 to 2.60) 2.28 (1.23 to 4.22) 1.00 (reference) 1.58 (0.98 to 2.54) 1.90 (1.20 to 3.00) 2.00 (1.21 to 3.31) 3.08 (1.67 to 5.68) 23 55 78 41 17 622 992 1160 537 148
1.00 (reference) 1.49 (0.92 to 2.39) 1.82 (1.15 to 2.87) 2.19 (1.33 to 3.59) 3.69 (2.04 to 6.68)
1.09 (1.00 to 1.18) 1.09 (1.00 to 1.17)c 1.18 (1.09 to 1.27) 214 3459
Continuous eGFRcys per 10 ml/min per 1.73 m2 decrease Category of eGFRcys $90 ml/min per 1.73 m2 76–89 ml/min per 1.73 m2 60–75 ml/min per 1.73 m2 45–59 ml/min per 1.73 m2 15–44 ml/min per 1.73 m2
Variable
1.15 (1.06 to 1.24)
IRR Further Adjusted for Potential Mediatorsc (95% CI) Unadjusted IRR (95% CI) Patients with Frailty (n) Total Patients (n)
Table 4. Association between kidney function (estimated GFR using cystatin C) and incident frailty
IRR Adjusted for Demographic Characteristicsa (95% CI)
IRR Further Adjusted for Potential Confoundersb (95% CI)
Clin J Am Soc Nephrol 8: 2091–2099, December, 2013
Kidney Function and Frailty, Dalrymple et al.
2097
elderly and thus offers continuity with prior studies evaluating other outcomes (2,18). Frailty is exceptionally common in patients with stage 5 CKD who are undergoing dialysis (19). The extent to which kidney dysfunction independently increases the risk of frailty and the level of kidney function at which the prevalence of frailty substantially rises are important to understand. Similar to our findings, previous studies have shown an association between lower levels of kidney function and prevalent frailty. Wilhelm-Leen et al. (9) estimated kidney function using the Mayo quadratic equation and found that 21% of participants in the Third National Health and Nutrition Examination Survey with an eGFR ,45 ml/min per 1.73 m2 were frail according to modified frailty criteria. In a previous study of CHS participants, we found serum creatinine $1.5 mg/dl in men and 1.3 mg/dl in women to be associated with a nearly 80% higher odds of prevalent frailty (8). Roshanravan et al. examined 336 Seattle Kidney Study participants, a cohort of patients referred for nephrology care, and estimated kidney function using the CKD-EPI equation with cystatin C, age, sex, and race (7). In their study, the relative prevalence of frailty was 2.1- and 2.8-fold higher for participants with an eGFR ,30 and 30–44 ml/min per 1.73 m2, respectively, compared with an eGFR .60 ml/min per 1.73 m2 (7). Compared with these earlier studies using serum creatinine, we used a more precise measure of kidney function, serum cystatin C, and compared with Roshanravan et al. we examined a community-based cohort. In addition, we evaluated incident frailty, which to our knowledge has not been studied previously, and found that an eGFRcys of 60–75 and 15–44 ml/min per 1.73 m2 were independently associated with a higher risk of incident frailty. The lack of a statistically significant association in those with an eGFRcys of 45–59 ml/min per 1.73 m2 may be due to a lack of power given the small number of participants with more advanced CKD or chance variability resulting in a nonsignificant association. Our study makes important contributions to the current field of research. We examined the association between kidney function and frailty in an elderly cohort using a more precise estimate of kidney function, cystatin C, and found a strong and graded association. This finding was completely obscured when evaluated by a more common estimate of kidney function based on serum creatinine. Our study is one of only two studies to examine the association between cystatin C and prevalent frailty, and ours is unique because of its setting within a sample of community-dwelling elders and use of the original validated criteria for frailty. Our study is the first to evaluate incident frailty and to uncover an independent association between lower levels of kidney function and a higher risk of incident frailty—an important finding that warrants further study. Our findings highlight that frailty is not a phenomena that starts with dialysis initiation, but, like many complications of CKD, the processes begin before the need for renal replacement therapy. Our study makes important contributions to the care of older patients with CKD because it highlights the notably high prevalence of frailty and the heightened risk for incident frailty in this special
2098
Clinical Journal of the American Society of Nephrology
population and informs their care and management. Reduced kidney function may be related to accelerated loss of lean muscle mass (20), which is thought to be an underlying process in frailty (21). Further examining whether specific interventions in this population, aimed at mitigating or preventing the development of frailty, warrant specific study as frailty in CKD has been linked to a higher risk of dialysis therapy or death (7). Our study had many strengths. We used a measure of kidney function, serum cystatin C, that is not related to muscle mass, a particularly important consideration for examining conditions such as frailty. We had comprehensive ascertainment of coexisting illnesses, allowing for appropriate exclusions and adjustment for potential confounders. We used validated and well accepted criteria to identify frailty. Last, we examined incident frailty, allowing us to further elaborate on the relation between kidney function and frailty. Our study had limitations that warrant consideration. Our examination of the association between kidney function and prevalent frailty was cross-sectional in nature and as such cannot establish causality. Frailty was reassessed at 4 years after the start of our study, and we cannot exclude that participants developed frailty and died before reassessment for frailty; this may have occurred disproportionately among those with the lowest kidney function, obscuring the association between kidney function and frailty and contributing to an underestimate of the true incidence of frailty. In addition, some diseases we considered as confounders, such as hypertension and heart failure, may actually be mediators, thereby attenuating the observed association between kidney function and frailty. Additionally, the longitudinal analysis examined kidney function only at one point in time. Participants with worse kidney function may be more likely to have subsequent declines in kidney function, and our study does not capture the impact of trajectories of decline in kidney function on incident frailty. Furthermore, we cannot exclude the possibility that unmeasured non-GFR determinants of cystatin C account for some of the observed association between eGFRcys and frailty. Last, we did not have measures of urinary protein and therefore cannot examine whether the association between levels of kidney function and frailty are modified by the presence of proteinuria. In summary, kidney function is independently associated with prevalent and incident frailty. Our findings highlight the importance of considering and assessing for frailty in older adults with CKD.
Acknowledgments This research was supported by contracts HHSN268201200036C, HHSN268200800007C, N01 HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, and grant HL080295 from the National Heart, Lung, and Blood Institute (NHLBI), with additional contribution from the National Institute of Neurological Disorders and Stroke (NINDS). Additional support was provided by AG023629 from the National Institute on Aging (NIA). L.S.D. was supported by UL1 RR024146 from the National Center for Research for Medical Research. A full list of principal CHS investigators and institutions can be found at https://chs-nhlbi. org/pi. The contents are solely the responsibility of the authors and
do not necessarily represent the official view of National Center for Research Resources or National Institutes of Health. The Cardiovascular Health Study was funded through contracts with the NHLBI and included substantial NHLBI involvement in the design and conduct of the study and the collection and management of the data. The funding source was not involved in data analysis, interpretation of the data, preparation, review, or approval of the manuscript. Findings from this study were presented as a Poster Presentation at the American Society of Nephrology Renal Week in San Diego, California, on November 2, 2012. Disclosures None. References 1. Stevens LA, Li S, Wang C, Huang C, Becker BN, Bomback AS, Brown WW, Burrows NR, Jurkovitz CT, McFarlane SI, Norris KC, Shlipak M, Whaley-Connell AT, Chen SC, Bakris GL, McCullough PA: Prevalence of CKD and comorbid illness in elderly patients in the United States: Results from the Kidney Early Evaluation Program (KEEP). Am J Kidney Dis 55(3 Suppl 2):S23–33, 2010 2. Dalrymple LS, Katz R, Kestenbaum B, de Boer IH, Fried L, Sarnak MJ, Shlipak MG: The risk of infection-related hospitalization with decreased kidney function. Am J Kidney Dis 59: 356–363, 2012 3. Fried LF, Katz R, Sarnak MJ, Shlipak MG, Chaves PH, Jenny NS, Stehman-Breen C, Gillen D, Bleyer AJ, Hirsch C, Siscovick D, Newman AB: Kidney function as a predictor of noncardiovascular mortality. J Am Soc Nephrol 16: 3728–3735, 2005 4. Fried LF, Lee JS, Shlipak M, Chertow GM, Green C, Ding J, Harris T, Newman AB: Chronic kidney disease and functional limitation in older people: Health, aging and body composition study. J Am Geriatr Soc 54: 750–756, 2006 5. Sarnak MJ, Katz R, Fried LF, Siscovick D, Kestenbaum B, Seliger S, Rifkin D, Tracy R, Newman AB, Shlipak MG; Cardiovascular Health Study: Cystatin C and aging success. Arch Intern Med 168: 147–153, 2008 6. Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener J, Seeman T, Tracy R, Kop WJ, Burke G, McBurnie MA; Cardiovascular Health Study Collaborative Research Group: Frailty in older adults: Evidence for a phenotype. J Gerontol A Biol Sci Med Sci 56: M146–M156, 2001 7. Roshanravan B, Khatri M, Robinson-Cohen C, Levin G, Patel KV, de Boer IH, Seliger S, Ruzinski J, Himmelfarb J, Kestenbaum B: A prospective study of frailty in nephrology-referred patients with CKD. Am J Kidney Dis 60: 912–921, 2012 8. Shlipak MG, Stehman-Breen C, Fried LF, Song X, Siscovick D, Fried LP, Psaty BM, Newman AB: The presence of frailty in elderly persons with chronic renal insufficiency. Am J Kidney Dis43: 861–867, 2004 9. Wilhelm-Leen ER, Hall YN, K Tamura M, Chertow GM: Frailty and chronic kidney disease: The Third National Health and Nutrition Evaluation Survey. Am J Med 122: 664–671, e662, 2009 10. Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, Kuller LH, Manolio TA, Mittelmark MB, Newman A, O’Leary DH, Psaty B, Rautaharju P, Tracy RP, Weiler PG: The Cardiovascular Health Study: Design and rationale. Ann Epidemiol 1: 263– 276, 1991 11. Tell GS, Fried LP, Hermanson B, Manolio TA, Newman AB, Borhani NO: Recruitment of adults 65 years and older as participants in the Cardiovascular Health Study. Ann Epidemiol 3: 358–366, 1993 12. Cushman M, Cornell ES, Howard PR, Bovill EG, Tracy RP: Laboratory methods and quality assurance in the Cardiovascular Health Study. Clin Chem 41: 264–270, 1995 13. Psaty BM, Kuller LH, Bild D, Burke GL, Kittner SJ, Mittelmark M, Price TR, Rautaharju PM, Robbins J: Methods of assessing prevalent cardiovascular disease in the Cardiovascular Health Study. Ann Epidemiol 5: 270–277, 1995
Clin J Am Soc Nephrol 8: 2091–2099, December, 2013
14. Stevens LA, Coresh J, Schmid CH, Feldman HI, Froissart M, Kusek J, Rossert J, Van Lente F, Bruce RD 3rd, Zhang YL, Greene T, Levey AS: Estimating GFR using serum cystatin C alone and in combination with serum creatinine: A pooled analysis of 3,418 individuals with CKD. Am J Kidney Dis 51: 395–406, 2008 15. Inker LA, Schmid CH, Tighiouart H, Eckfeldt JH, Feldman HI, Greene T, Kusek JW, Manzi J, Van Lente F, Zhang YL, Coresh J, Levey AS; CKD-EPI Investigators: Estimating glomerular filtration rate from serum creatinine and cystatin C. N Engl J Med 367: 20– 29, 2012 16. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J; CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration): A new equation to estimate glomerular filtration rate. Ann Intern Med 150: 604–612, 2009 17. Shlipak MG, Matsushita K, Arnlov J, Inker LA, Katz R, Polkinghorne KR, Rothenbacher D, Sarnak MJ, Astor BC, Coresh J, Levey AS, Gansevoort RT: Cystatin C versus creatinine in determining risk based on kidney function. N Engl J Med 2013 (in press) 18. Peralta CA, Katz R, Sarnak MJ, Ix J, Fried LF, De Boer I, Palmas W, Siscovick D, Levey AS, Shlipak MG: Cystatin C identifies chronic kidney disease patients at higher risk for complications. J Am Soc Nephrol 22: 147–155, 2011
Kidney Function and Frailty, Dalrymple et al.
2099
19. Bao Y, Dalrymple L, Chertow GM, Kaysen GA, Johansen KL: Frailty, dialysis initiation, and mortality in end-stage renal disease. Arch Intern Med 172: 1071–1077, 2012 20. Fried LF, Boudreau R, Lee JS, Chertow G, Kurella-Tamura M, Shlipak MG, Ding J, Sellmeyer D, Tylavsky FA, Simsonick E, Kritchevsky SB, Harris TB, Newman AB; Health, Aging and Body Composition Study: Kidney function as a predictor of loss of lean mass in older adults: Health, aging and body composition study. J Am Geriatr Soc 55: 1578–1584, 2007 21. Walston J, Hadley EC, Ferrucci L, Guralnik JM, Newman AB, Studenski SA, Ershler WB, Harris T, Fried LP: Research agenda for frailty in older adults: Toward a better understanding of physiology and etiology: Summary from the American Geriatrics Society/National Institute on Aging Research Conference on Frailty in Older Adults. J Am Geriatr Soc 54: 991–1001, 2006 Received: March 14, 2013 Accepted: August 8, 2013 Published online ahead of print. Publication date available at www. cjasn.org. This article contains supplemental material online at http://cjasn. asnjournals.org/lookup/suppl/doi:10.2215/CJN.02870313/-/ DCSupplemental.
