Patients with ESRD experience markedly advanced atherosclerotic disease of the cerebral vasculature (1–4). Although we recently reported a 5- to 10-fold risk of hospitalized ischemic and hemorrhagic stroke among ESRD patients compared with non-ESRD individuals (5), little is known regarding potential stroke risk factors among the ESRD population. Risk factors for stroke in this population are likely to differ from those in the non-ESRD population. For example, elevated BP and body mass index are risk factors for stroke in the general population, whereas in the dialysis population, they are associated with a lower risk of adverse outcomes such as all-cause and cardiac death (6–9). Only one small study, conducted in a Japanese dialysis population, has examined risk factors for stroke (10), identifying hypertension as the only significant predictor. However, the study had limited power, and results from the younger and healthier Japanese dialysis population (11) may not be generalizable to the US dialysis population. In addition, although blacks have much higher rates of stroke in the general population, no previous studies have examined black race as a risk factor for incident stroke in the dialysis population. We used data collected by the United States Renal Data System (USRDS) to identify patient characteristics associated with the risk of hospitalized or fatal stroke in dialysis patients, with specific focus on black race, hypertension, and malnutrition as risk factors.
Materials and Methods
We used data from the USRDS Dialysis Morbidity and Mortality Studies, Waves 2 to 4 (DMMS-2 to -4). Details of the studies performed by the USRDS are described elsewhere (12). Briefly, the USRDS collects demographic and clinical data on all patients who have survived >90 d of renal replacement therapy for ESRD. DMMS-2 was a prospective observational study of a sample of adult patients who initiated dialysis in 1996 and early 1997, with deliberate oversampling of patients on peritoneal dialysis. DMMS-3 and -4 were retrospective studies of random samples of hemodialysis patients who were alive on December 31, 1993. Data collection techniques and content were kept consistent across the three studies, allowing them to be combined for use in epidemiologic research. The study population for this analysis included all patients who were in DMMS-2 to -4 and treated with dialysis and were Medicare-insured with no previous history of stroke or transient ischemic attack. Following the recommendations of the USRDS for studies of hospitalization rates (13), patients for whom fee-for-service Medicare was not the primary insurer were excluded from our analysis, because hospitalization data for these patients are incomplete in the USRDS database (see the Outcome section).
Ascertainment of Baseline Patient Characteristics
Baseline patient data were abstracted by dialysis facility personnel from each patient’s medical record and through patient interview. Patient characteristics ascertained included demographic (age, gender, and race), laboratory (albumin, cholesterol, hemoglobin, calcium, parathyroid hormone, and phosphorous), clinical (cause of renal disease, history of cardiovascular disease (CVD), history of stroke or transient ischemic attack (TIA), dialysis vintage, smoking status, and a subjective assessment of undernourishment), and other measurements (height, weight, and BP). Previous CVD was defined as any diagnosis of coronary artery disease, myocardial infarction, coronary artery bypass, angioplasty, cardiac arrest, or congestive heart failure. The average of up to three measurements of BP during three consecutive dialysis sessions, recorded either before dialysis (in hemodialysis patients) or randomly (in peritoneal dialysis patients), were used in this analysis. For DMMS-2 patients, only single measurements of laboratory variables were available. For patients in DMMS-3 and -4, multiple values for laboratory measurements were available for up to 3 mo preceding the start of the study; for these patients, all available values were averaged before inclusion into a statistical model.
The primary outcome was defined as first hospitalized stroke or fatal nonhospitalized stroke. Hospitalization for stroke was determined by linking Medicare hospital billing records to each patient through unique identifier codes supplied by the USRDS. Diagnosis of stroke was based on the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) diagnosis codes contained in these billing records. On the basis of previous studies that examined the relative accuracy of different ICD-9-CM diagnosis codes in identifying patients with actual hospitalized stroke (14–18), we a priori considered the following five codes to identify acute stroke: 430 (subarachnoid hemorrhage), 431 (intracerebral hemorrhage), 433.X1 (occlusion of precerebral arteries with infarction), 434.X1 (occlusion of cerebral arteries with infarction), and 436 (acute cerebrovascular attack), where “X” can be any integer from 0 to 9 and refers to specific arterial syndromes. A hospitalization in which one of these five diagnosis codes was listed in either the primary or the nine secondary positions reported by the USRDS was considered stroke related.
For information on fatal stroke, survival status and cause of death were linked to the DMMS data from the USRDS Patients Standard Analysis File via unique patient identifiers. The date and cause of death listed in a patient’s Standard Analysis File was obtained from information submitted to the USRDS by the patient’s nephrologist (form HCFA 2746). Fatal nonhospitalized stroke was defined as a primary cause of death from either “cerebrovascular disease” or “cerebrovascular accident including intracranial hemorrhage” without a preceding hospitalization for stroke.
Secondary outcomes included hospitalized hemorrhagic stroke (ICD-9 codes 430, 431) and ischemic stroke (codes 433.X1, 434.X1, or 436). Data on hospitalization and mortality were available through December 31, 1999.
The Cox proportional hazards model for censored survival data was used to assess the association between the primary risk factors of interest and incident stroke after adjustment for potential confounders. Primary predictor variables of interest included race (white, black, Asian, other), mean BP (MBP; calculated as [systolic BP + 2*diastolic BP]/3), and markers of malnutrition (serum albumin, body weight, and a subjective assessment of undernourishment). Adjustment variables to control confounding were a priori chosen on the basis of their potential relationship with the outcome of interest. Cigarette smoking was not included as an adjustment covariate in the primary model because of the high degree of missing information (15%) for this variable in the USRDS; rather, the potential confounding effect of smoking was assessed in an exploratory analysis. Renal replacement modality (hemodialysis, peritoneal dialysis, transplant) was modeled as time-dependent covariates, allowing patients to switch risk groups over the course of follow-up. All models were further adjusted for DMMS study via stratification.
Following recommendations from the USRDS (13), incident patients were not considered at risk for hospitalizations until day 90 of ESRD. Patients were followed from day 90 of ESRD or (for prevalent patients) from the DMMS study start date and were censored at loss to follow-up, nonstroke death, or the end of the study period (December 31, 1999).
Formal and graphic techniques were used to confirm the presence of proportional hazards and to identify potential outliers. We hypothesized that all continuous covariates would be linearly related to the outcome of interest; however, exploratory residual analyses were performed to investigate functional form further. In particular, we explored linear and nonlinear forms of MBP to determine whether there was a “U”- or “J”-shaped relationship between BP and stroke. Effect modification by age, gender, CVD, and DMMS wave was explored and tested via stratification and the use of multiplicative interactions.
A total of 13,716 patients were included in DMMS-2 to -4. A total of 712 patients were excluded because their DMMS study start date was missing or implausible (n = 120) or because they did not have information on mortality or treatment history (n = 365), were younger than 18 yr (n = 22), did not survive until day 90 of ESRD care (n = 125), had a functioning transplant at the study start date (n = 30), or had other errors in their baseline data collection (n = 50). Finally, because we were interested in risk factors for incident stroke only, we excluded 1958 patients with a reported history of stroke or TIA before study start date. After these exclusions, 11,046 patients remained. Of these, 2126 (19%) patients did not have Medicare as a primary insurer at the start of follow-up and were excluded. Baseline characteristics of the final study cohort of 8920 patients are presented in Table 1. There were 6862 patients with complete data available for inclusion in the fully adjusted multivariate models; baseline characteristics were similar in this group compared with the total cohort (Table 1).
In the total cohort, 915 hospitalized or fatal nonhospitalized strokes occurred over a median follow-up of 3.1 yr, with an incidence density of 33/1000 person-years (95% confidence interval [CI]: 31 to 35). Model-based unadjusted and adjusted estimates of the association of patient characteristics with incident stroke are presented in Table 2. The association between race and incident stroke differed significantly among individuals with and without prevalent CVD (P = 0.001 for test of interaction). Among patients without prevalent CVD, blacks were estimated to be at higher risk of stroke when compared with whites (hazard ratio [HR] = 1.24; 95% CI = 0.96 to 1.6), whereas among patients with prevalent CVD, blacks were at significantly lower risk (HR = 0.74; 95% CI = 0.60 to 0.92). Patients of other races were not found to have experienced a significantly different risk of stroke, regardless of CVD status (Table 2).
Among our other predictors of interest, markers of malnutrition all were strongly associated with a higher risk of stroke. Patients who were described by dialysis staff as being undernourished were estimated to have a 27% higher risk of stroke (HR = 1.27; 95% CI = 1.01 to 1.61). A 1-g/dl decrement in serum albumin was associated with a 43% higher risk of stroke (HR = 1.43; 95% CI = 1.17 to 1.74), and there was a trend toward higher risk of stroke associated with low body weight, after adjustment for height (per 25% relative decrease, HR = 1.08; 95% CI = 1.00 to 1.18; P = 0.057).
MBP was also predictive of incident stroke with an increase of 10 mmHg in MBP associated with an 11% increased risk of stroke (HR = 1.11; 95% CI = 1.05 to 1.18). There was no evidence of excess risk at very low BP (e.g., MBP <75 mmHg), although there were relatively few patients in this category (n = 183, eight strokes). The association of MBP with incident stroke was similar among hemodialysis (when measured before dialysis) and peritoneal dialysis patients (measured randomly) and among patients of different races, types of renal disease, and duration of ESRD (P > 0.05 for all tests of interaction; data not shown). Among hemodialysis patients, postdialysis MBP was not significantly associated with incident stroke (per 10 mmHg increment, HR = 1.03; 95% CI = 0.95 to 1.12), and adjustment for this BP measurement did not meaningfully change the HR estimate for predialysis MBP. Further adjustment for smoking status did not materially change the associations between any of the primary predictors and stroke, and smoking was not a significant independent predictor of stroke (data not shown).
In exploratory analyses, we assessed the association with stroke of other laboratory parameters that could possibly contribute to the risk of stroke in the dialysis population. There was no relationship between baseline cholesterol (per 10 mg/dl increment, HR = 1.00), serum calcium, phosphorous, or parathyroid hormone and incident stroke (data not shown). Among hemodialysis patients, the change in MBP from predialysis to postdialysis was not associated with incident stroke (P = 0.7). Hemoglobin levels were modestly associated with stroke, after adjustment for other covariates. In comparison with patients with hemoglobin levels of 10 to 12 g/dl, those with very low hemoglobin (<9 g/dl) were at a 22% higher risk for stroke (HR = 1.22, 95% CI = 1.00 to 1.49).
A total of 131 incident hospitalized hemorrhagic strokes occurred during the follow-up period, with an estimated incidence density of 4.6 per 1000 person-years (95% CI = 3.9 to 5.5). Patient characteristics associated with risk of hemorrhagic stroke are shown in Table 3. The association between race and hemorrhagic stroke differed significantly among individuals with and without prevalent CVD (P = 0.002 for test of interaction). Compared with whites without CVD, blacks without CVD had twice the rate of hemorrhagic stroke (HR = 2.19; 95% CI = 1.21 to 3.98). Among individuals without CVD, however, blacks had much lower rates of hemorrhagic stroke (HR = 0.52; 95% CI = 0.24 to 1.10).
Higher MBP was associated with a higher stroke risk (per 10 mmHg increase, HR = 1.32; 95% CI = 1.15 to 1.53). Among the markers of malnutrition, undernourishment showed a trend toward an association with a higher stroke rate (HR = 1.76; 95% CI = 0.98 to 3.18); serum albumin and height-adjusted weight were not predictive of stroke. Patients with polycystic kidney disease were at a 2.5-fold risk for hemorrhagic stroke compared with patients with primary glomerulonephritis (HR = 2.55; 95% CI = 0.94 to 6.86), although this association was short of statistical significance (P = 0.07).
There were 700 incident hospitalized ischemic strokes observed over the course of follow-up, corresponding to an incidence density of 25.2/1000 person-years (95% CI = 23.4 to 27.1). For ischemic stroke, the effects of the major predictors of interest differed among patients with and without prevalent CVD (P < 0.05 for interaction terms), so separate multivariate models were developed for these two subgroups (Table 4). Among patients with prevalent CVD, blacks were at 23% lower risk of ischemic stroke when compared with whites (HR = 0.77; 95% CI = 0.60 to 0.98), and there were no significant associations between BP and stroke. Among the markers of malnutrition, only a subjective assessment of undernourishment was associated with higher stroke risk (HR = 1.40; 95% CI = 1.02 to 1.92). Among patients without prevalent CVD, blacks were at equal risk to whites, and higher MBP was associated with an increased risk for ischemic stroke (HR per 10 mmHg increment, HR = 1.16; 95% CI = 1.04 to 1.30). Low serum albumin levels in this subgroup were also associated with a higher ischemic stroke risk (per 1-g/dl decrease, HR = 1.84; 95% CI = 1.28 to 2.64), although height-adjusted body weight was not.
In a cohort of incident and prevalent US dialysis patients, the incidence of hospitalized and fatal stroke was 33/1000 person-years. Markers of malnutrition (low height-adjusted body weight, hypoalbuminemia, and undernourishment) and elevated MBP were predictive of incident stroke, as was low hemoglobin. Stroke risk among blacks relative to whites differed among those with and without clinical cardiac disease: their risk was somewhat higher among individuals without cardiac disease but was significantly lower among individuals with cardiac disease.
Most studies of stroke in the general population have found that blacks are at a 50 to 200% higher risk than whites (14,15,19–24). This increased risk seems to be independent of traditional stroke risk factors such as hypertension and diabetes (14,20,24). To our knowledge, no study has examined whether these racial differences in stroke rate differ by cardiac disease status. In the dialysis population, although racial differences in stroke incidence have not been assessed, previous studies reported a lower rate of all-cause mortality, cardiac-specific mortality (25), cerebrovascular mortality (26), and coronary artery disease prevalence (27). Our finding of a significant and large interaction between race and prevalent cardiac disease on the risk of stroke raises the question of whether a similar interaction exists with regard to other important outcomes.
The reason for this interaction is unclear. One possible explanation is a nondifferential misclassification of cardiac disease between blacks and whites. Information on comorbid illness in DMMS was obtained by dialysis facility personnel from the patients themselves and from notation in the medical record available in the dialysis centers; there was no independent verification of comorbid status. If cardiac disease was less accurately characterized among black patients (as a result of less self-knowledge of their medical conditions or because of lower rates of screening for cardiac disease) than among whites, then resulting selective misclassification would create the appearance of an interaction between black race and cardiac disease.
If the observed interaction between race and CVD is not due to bias, then one interpretation of this interaction is that blacks are less “susceptible” to the adverse effects of CVD on risk of stroke. Some authors have suggested that a selection process occurs in which blacks who have early renal disease and survive long enough to reach ESRD represent a relatively healthier subgroup, as a result of a competing risk between a premature death and ESRD among blacks (28). In this scenario, blacks with cardiac disease and early renal insufficiency would be more likely to have a premature death before reaching ESRD than whites with similar cardiac disease, such that only blacks who are in some way more “resistant” to the effects of this cardiac disease would survive to ESRD. Although this is a plausible hypothesis, there is currently neither supporting nor refuting evidence.
Consistent with data from the general population (29–32), elevated BP was associated with an increased risk for stroke. We found no evidence of a U- or J-shaped effect of MBP. Our finding is also in agreement with those of Iseki et al. (10), who reported a 120% increased stroke risk associated with hypertension in a Japanese dialysis cohort. The effect of BP on stroke risk seems to differ from its effect on all-cause mortality, as suggested by several studies in which predialysis BP (especially systolic BP) was inversely related to mortality risk, with an excess risk at low BP (6–8). Among hemodialysis patients, we did not find an association between postdialysis BP and stroke, in contrast to previous reports suggesting a U-shaped association between postdialysis BP and mortality (6,7).
The finding of a positive relationship between BP and stroke should be interpreted with some caution. Our analysis did not distinguish those patients using antihypertensive medications, and it is possible that the effect of BP is confounded by the use of these medications or differs among patients according to their use of antihypertensives. In addition, the observed effect of BP could be biased by a competing risk between mortality and stroke, in which patients with low BP are less likely to survive long enough to be at risk for stroke.
In an exploratory analysis, profound anemia (hemoglobin <9.0 g/dl) was associated with a significant 22% increase risk of stroke, compared with a hemoglobin level of 10 to 12 g/dl. This is in strong contrast to findings from the general population, in which high, not low, hemoglobin was associated with an increased risk (33–36). However, these studies may have had insufficient power to detect an association between very low hemoglobin levels and stroke given the low prevalence of severe anemia in the non-ESRD population. Our findings may represent a type I error, given that several secondary risk factors were tested for association with incident stroke. However, an increased risk of stroke from profound anemia in ESRD is biologically plausible and could be mediated either through the direct effects of low oxygen-carrying capacity in regions of the brain already poorly perfused from vascular disease or dialysis-related hypotension or through the detrimental effects of chronic anemia on arterial and cardiac hypertrophy (37), which are associated with increased stroke risk in the non-ESRD population (38,39).
In the current study, several markers of malnutrition, including low serum albumin, low height-adjusted body weight, and a subjective assessment of undernourishment, were associated with a higher risk of incident stroke. This is in contrast to the general population, in which obesity, rather than malnutrition, confers a higher stroke risk (40). However, malnutrition has been well recognized as a strong risk factor for total (41–44) and cardiovascular-specific (9) mortality in the dialysis population. Several authors have suggested that malnutrition in ESRD patients reflects not merely poor nutrient intake but also the effects of a chronic microinflammatory state (45,46). Elevated inflammatory markers have been associated with higher rates of stroke in the general population (47–49). It is possible that the mechanism that leads to the association of inflammation and stroke in the general population exerts a similar effect in the dialysis population and might explain the observed association between malnutrition and stroke in the present study.
This study has a number of limitations. First, acute strokes were detected through hospital discharge diagnosis codes and ESRD cause-of-death forms; it was not possible to validate these events in the data sources used. It therefore is likely that some patients in this study were misclassified with regard to their stroke status. However, the combination of ICD-9-CM codes used to identify stroke hospitalizations in our study has been shown to have high sensitivity and specificity in identifying true hospitalized stroke in the general population (18). In addition, one would expect such misclassification to be nondifferential with regard to baseline patient characteristics, which would result in an underestimation of the true relative risk of stroke associated with specific risk factors. Our study excluded patients for whom fee-for-service Medicare was not the primary insurer, and therefore the results cannot be generalized to this generally healthier and younger subgroup of dialysis patients. However, the overwhelming majority (81%) of patients in the DMMS studies had fee-for-service Medicare as the primary insurer.
Despite these limitations, this is the first study to examine risk factors for incident stroke among the US dialysis population. Because the study cohort included a national sample of dialysis patients, the results of this study are not limited to a single center or geographic region. These results confirm the extraordinarily high rates of stroke in this population and identify several modifiable risk factors, including malnutrition, anemia, and hypertension, as potential targets for preventive therapies.
This study was supported by a Veterans Affairs Career Development Award and PHS grants (DK07721-6 and DK63079-01) from the National Institutes of Health, Bethesda, MD. The data reported here have been supplied by the USRDS. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy of or interpretation by the US government.
1. Kennedy R, Case C, Fathi R, Johnson D, Isbel N, Marwick TH: Does renal failure cause an atherosclerotic milieu in patients with end-stage renal disease? Am J Med 110: 198–204, 2001
2. Kawagishi T, Nishizawa Y, Konishi T, Kawasaki K, Emoto M, Shoji T, Tabata T, Inoue T, Morii H: High-resolution B-mode ultrasonography in evaluation of atherosclerosis in uremia. Kidney Int 48: 820–826, 1995
3. Zoungas S, Ristevski S, Lightfoot P, Liang YL, Branley P, Shiel LM, Kerr P, Atkins R, McNeil JJ, McGrath BP: Carotid artery intima-medial thickness is increased in chronic renal failure. Clin Exp Pharmacol Physiol 27: 639–641, 2000
4. Savage T, Clarke AL, Giles M, Tomson CR, Raine AE: Calcified plaque is common in the carotid and femoral arteries of dialysis patients without clinical vascular disease. Nephrol Dial Transplant 13: 2004–2012, 1998
5. Seliger S, Gillen D, Kestenbaum B, Ball A, Wasse H, Stehman-Breen C: A comparison of stroke rates among the dialysis and general populations [Abstract]. J Am Soc Nephrol 13: 438A, 2002
6. Zager PG, Nikolic J, Brown RH, Campbell MA, Hunt WC, Peterson D, Van Stone J, Levey A, Meyer KB, Klag MJ, Johnson HK, Clark E, Sadler JH, Teredesai P: “U” curve association of blood pressure and mortality in hemodialysis patients. Medical Directors of Dialysis Clinic, Inc. Kidney Int 54: 561–569, 1998
7. Port FK, Hulbert-Shearon TE, Wolfe RA, Bloembergen WE, Golper TA, Agodoa LY, Young EW: Predialysis blood pressure and mortality risk in a national sample of maintenance hemodialysis patients. Am J Kidney Dis 33: 507–517, 1999
8. Klassen PS, Lowrie EG, Reddan DN, DeLong ER, Coladonato JA, Szczech LA, Lazarus JM, Owen WF Jr: Association between pulse pressure and mortality in patients undergoing maintenance hemodialysis. JAMA 287: 1548–1555, 2002
9. Fung F, Sherrard DJ, Gillen DL, Wong C, Kestenbaum B, Seliger S, Ball A, Stehman-Breen C: Increased risk for cardiovascular mortality among malnourished end-stage renal disease patients. Am J Kidney Dis 40: 307–314, 2002
10. Iseki K, Fukiyama K: Predictors of stroke in patients receiving chronic hemodialysis. Kidney Int 50: 1672–1675, 1996
11. Goodkin DA, Mapes DL, Held PJ: The dialysis outcomes and practice patterns study (DOPPS): How can we improve the care of hemodialysis patients? Semin Dial 14: 157–159, 2001
12. United States Renal Data System: Researcher’s Guide to the USRDS Database, Bethesda, MD, The National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, 1999
13. United States Renal Data System: Researcher’s Guide to the USRDS Database, Bethesda, MD, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, 2002
14. Rosamond WD, Folsom AR, Chambless LE, Wang CH, McGovern PG, Howard G, Copper LS, Shahar E: Stroke incidence and survival among middle-aged adults: 9-year follow- up of the Atherosclerosis Risk in Communities (ARIC) cohort. Stroke 30: 736–743, 1999
15. Broderick J, Brott T, Kothari R, Miller R, Khoury J, Pancioli A, Gebel J, Mills D, Minneci L, Shukla R: The Greater Cincinnati/Northern Kentucky Stroke Study: Preliminary first-ever and total incidence rates of stroke among blacks. Stroke 29: 415–421, 1998
16. Ives DG, Fitzpatrick AL, Bild DE, Psaty BM, Kuller LH, Crowley PM, Cruise RG, Theroux S: Surveillance and ascertainment of cardiovascular events. The Cardiovascular Health Study. Ann Epidemiol 5: 278–285, 1995
17. Goldstein LB: Accuracy of ICD-9-CM coding for the identification of patients with acute ischemic stroke: Effect of modifier codes. Stroke 29: 1602–1604, 1998
18. Tirschwell DL, Longstreth WT Jr: Validating administrative data in stroke research. Stroke 33: 2465–2470, 2002
19. Gaines K, Burke G: Ethnic differences in stroke: Black-white differences in the United States population. SECORDS Investigators. Southeastern Consortium on Racial Differences in Stroke. Neuroepidemiology 14: 209–239, 1995
20. Giles WH, Kittner SJ, Hebel JR, Losonczy KG, Sherwin RW: Determinants of black-white differences in the risk of cerebral infarction. The National Health and Nutrition Examination Survey Epidemiologic Follow-up Study. Arch Intern Med 155: 1319–1324, 1995
21. Sacco RL, Boden-Albala B, Gan R, Chen X, Kargman DE, Shea S, Paik MC, Hauser WA: Stroke incidence among white, black, and Hispanic residents of an urban community: The Northern Manhattan Stroke Study. Am J Epidemiol 147: 259–268, 1998
22. Broderick J, Brott T, Tomsick T, Huster G, Miller R: The risk of subarachnoid and intracerebral hemorrhages in blacks as compared with whites. N Engl J Med 326: 733–736, 1992
23. Friday G, Lai S, Alter M, Sobel E, LaRue L, Gil-Peralta A, McCoy R, Levitt L, Isack T: Stroke in the Lehigh Valley: Racial/ethnic differences. Neurology 39: 1165–1168, 1989
24. Kittner SJ, White LR, Losonczy KG, Wolf PA, Hebel JR: Black-white differences in stroke incidence in a national sample. The contribution of hypertension and diabetes mellitus. JAMA 264: 1267–1270, 1990
25. Bloembergen WE, Port FK, Mauger EA, Wolfe RA: Causes of death in dialysis patients: Racial and gender differences. J Am Soc Nephrol 5: 1231–1242, 1994
26. United States Renal Data System: USRDS 2001 Annual Data Report: Atlas of End-Stage Renal Disease in the United States, Bethesda, Maryland, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, 2001
27. Trespalacios FC, Taylor AJ, Agodoa LY, Abbott KC: Incident acute coronary syndromes in chronic dialysis patients in the United States. Kidney Int 62: 1799–1805, 2002
28. Price DA, Owen WF Jr: African-Americans on maintenance dialysis: A review of racial differences in incidence, treatment, and survival. Adv Ren Replace Ther 4: 3–12, 1997
29. Collins R, Peto R, MacMahon S, Hebert P, Fiebach NH, Eberlein KA, Godwin J, Qizilbash N, Taylor JO, Hennekens CH: Blood pressure, stroke, and coronary heart disease. Part 2: Short-term reductions in blood pressure: Overview of randomised drug trials in their epidemiological context. Lancet 335: 827–838, 1990
30. MacMahon S, Peto R, Cutler J, Collins R, Sorlie P, Neaton J, Abbott R, Godwin J, Dyer A, Stamler J: Blood pressure, stroke, and coronary heart disease. Part 1: Prolonged differences in blood pressure: Prospective observational studies corrected for the regression dilution bias. Lancet 335: 765–774, 1990
31. Wolf PA, D’Agostino RB, Belanger AJ, Kannel WB: Probability of stroke: A risk profile from the Framingham Study. Stroke 22: 312–318, 1991
32. Prevention of stroke by antihypertensive drug treatment in older persons with isolated systolic hypertension. Final results of the Systolic Hypertension in the Elderly Program (SHEP). SHEP Cooperative Research Group. JAMA 265: 3255–3264, 1991
33. Di Mascio R, Marchioli R, Vitullo F, Tognoni G: A positive relation between high hemoglobin values and the risk of ischemic stroke. Progetto 3A Investigators. Eur Neurol 36: 85–88, 1996
34. Kiyohara Y, Ueda K, Hasuo Y, Fujii I, Yanai T, Wada J, Kawano H, Shikata T, Omae T, Fujishima M: Hematocrit as a risk factor of cerebral infarction: Long-term prospective population survey in a Japanese rural community. Stroke 17: 687–692, 1986
35. Tohgi H, Yamanouchi H, Murakami M, Kameyama M: Importance of the hematocrit as a risk factor in cerebral infarction. Stroke 9: 369–374, 1978
36. Kannel WB, Gordon T, Wolf PA, McNamara P: Hemoglobin and the risk of cerebral infarction: The Framingham Study. Stroke 3: 409–420, 1972
37. Metivier F, Marchais SJ, Guerin AP, Pannier B, London GM: Pathophysiology of anaemia: Focus on the heart and blood vessels. Nephrol Dial Transplant 15: 14–18, 2000
38. Verdecchia P, Porcellati C, Reboldi G, Gattobigio R, Borgioni C, Pearson TA, Ambrosio G: Left ventricular hypertrophy as an independent predictor of acute cerebrovascular events in essential hypertension. Circulation 104: 2039–2044, 2001
39. Chambless LE, Folsom AR, Clegg LX, Sharrett AR, Shahar E, Nieto FJ, Rosamond WD, Evans G: Carotid wall thickness is predictive of incident clinical stroke: The Atherosclerosis Risk in Communities (ARIC) study. Am J Epidemiol 151: 478–487, 2000
40. Leys D, Deplanque D, Mounier-Vehier C, Mackowiak-Cordoliani MA, Lucas C, Bordet R: Stroke prevention: Management of modifiable vascular risk factors. J Neurol 249: 507–517, 2002
41. Avram M, Sreedhara R, Fein P, Oo K, Chattopadhyay J, Mittman N: Survival on hemodialysis and peritoneal dialysis over 12 years with emphasis on nutritional parameters. Am J Kidney Dis 37: s77–s80, 2001
42. Combe C, Chauveau P, Laville M, Fouque D, Azar R, Cano N, Canaud B, Roth H, Leverve X, Aparicio M: Influence of nutritional factors and hemodialysis adequacy on the survival of 1,610 French patients. Am J Kidney Dis 37: s81–s88, 2001
43. Lowrie EG, Lew NL: Death risk in hemodialysis patients: The predictive value of commonly measured variables and an evaluation of death rate differences between facilities. Am J Kidney Dis 15: 458–482, 1990
44. Pifer TB, McCullough KP, Port FK, Goodkin DA, Maroni BJ, Held PJ, Young EW: Mortality risk in hemodialysis patients and changes in nutritional indicators: DOPPS. Kidney Int 62: 2238–2245, 2002
45. Stenvinkel P: Malnutrition and chronic inflammation as risk factors for cardiovascular disease in chronic renal failure. Blood Purif 19: 143–151, 2001
46. Kalantar-Zadeh K, Kopple J: Relative contributions of nutrition and inflammation to clinical outcome in dialysis patients. Am J Kidney Dis 38: 1343–1350, 2001
47. Ridker PM, Buring JE, Shih J, Matias M, Hennekens CH: Prospective study of C-reactive protein and the risk of future cardiovascular events among apparently healthy women. Circulation 98: 731–733, 1998
48. Rost NS, Wolf PA, Kase CS, Kelly-Hayes M, Silbershatz H, Massaro JM, D’Agostino RB, Franzblau C, Wilson PWF: Plasma concentration of C-reactive protein and risk of ischemic stroke and transient ischemic attack: The Framingham Study. Stroke 32: 2575–2579, 2001
49. Ford ES, Giles WH: Serum C-reactive protein and self-reported stroke: Findings from the Third National Health and Nutrition Examination Survey. Arterioscler Thromb Vasc Biol 20: 1052–1056, 2000