Rapid Fluid Removal During Dialysis is Associated With Cardiovascular Morbidity and Mortality

Jennifer E Flythe; Stephen E Kimmel; Steven M Brunelli

Disclosures

Kidney Int. 2011;79(2):250-257. 

In This Article

Results

Baseline Characteristics of Cohort

Demographic, clinical, and biochemical characteristics of the study population are shown in Table 1. Overall, the cohort consisted of 1,846 patients with a mean age of 57.6±14.0 years; 56.2% were women and 62.6% were black. At baseline 39.7% of the patients carried a diagnosis of congestive heart failure, 39.3% had ischemic heart disease and 44.6% were diabetic.

The mean UFR for the cohort was 12.1±4.6 ml/h/kg; 644 (34.9%), 517 (28.0%), and 685 (37.1%) patients had UFR ≤10, 10–13, and >13 ml/h/kg, respectively. Overall, UFR groups were similar in terms of sex, race, dialysis vintage, smoking status, access type, treatment group assignment (flux and Kt/V), diabetes, ischemic heart disease, peripheral vascular disease, serum albumin, and use of most classes of antihypertensive agents (Table 1). At baseline, patients with high UFRs were younger, more likely to have congestive heart failure and oliguria, and less likely to have cerebrovascular disease; they tended to have higher systolic blood pressures, serum creatinine and phosphate concentrations, and lower hematocrits. Not surprisingly, high UFR was associated with increased interdialytic weight gain and shorter HD session length.

Associations between UFR and All-cause and CV Mortality

Overall, 871 deaths occurred during 5,233 patient-years of at-risk time; 343 of these deaths were due to CV causes. The median survival time was 2.5 years. Compared with UFR ≤10 ml/h/kg, UFR >13 ml/h/kg was significantly associated with all-cause mortality: unadjusted hazard ratio (HR) (95% confidence interval (CI)) 1.20 (1.03–1.41) (Figure 1). When multivariable adjustment was used to account for baseline differences between groups, this association was greatly potentiated: HR (95% CI) 1.59 (1.29–1.96). UFR 10–13 ml/h/kg bore an intermediate association with CV mortality that was not statistically significant: adjusted HR (95% CI) 1.06 (0.87–1.28). Results were similar when UFRs following the long interdialytic break were excluded from consideration, when the referent group was restricted to participants with UFR 8–10 ml/h/kg (data not shown), and when flux and Kt/V treatment group assignments were included as covariates in the statistical model (Supplementary Table SA online).

Figure 1.

Unadjusted and adjusted associations between ultrafiltration rate (UFR) and all-cause mortality based on Cox regression models. Multivariable models were adjusted for age, sex, interdialytic weight gain, race (black, non-black), smoking status (never, past, current), vintage (<1, 1–2, 2–4, ≥4 years), access type (graft, fistula, catheter), systolic blood pressure (<120, 120–140, 140–160, 160–180, ≥180 mm Hg), residual urine output (≤ versus >200 ml/day), diabetes, congestive heart failure, peripheral vascular disease, ischemic heart disease, cerebrovascular disease, serum albumin, creatinine, hematocrit (<30, 30–33, 33–36, ≥36%), and phosphorus, and use of α-adrenergic blocker, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker, β-blocker, calcium channel blocker, nitrates, and other antihypertensives. Two-way cross-product terms with time were included for albumin and systolic blood pressure due to non-proportional hazards. Abbreviations: ref., reference; CI, confidence interval; HR, hazard ratio.

Similarly, compared with UFR ≤10 ml/h/kg, UFR >13 ml/h/kg was associated with increased CV mortality: unadjusted HR (95% CI) 1.33 (1.03–1.72) (Figure 2). Upon multivariable adjustment, this association was greatly potentiated: adjusted HR (95% CI) 1.71 (1.23–2.38). UFR 10–13 ml/h/kg bore an intermediate association with CV mortality that was not statistically significant: adjusted HR (95% CI) 1.06 (0.78–1.44). Again, results were similar when UFRs following the long interdialytic break were excluded from consideration, when the referent group was restricted to participants with UFR 8–10 ml/h/kg (data not shown), and when flux and Kt/V treatment group assignments were included as covariates in the statistical model (Supplementary Table SA online).

Figure 2.

Unadjusted and adjusted associations between ultrafiltration rate (UFR) and cardiovascular (CV)-related mortality based on Cox regression models. Multivariable models were adjusted for age, sex, interdialytic weight gain, race (black, non-black), smoking status (never, past, current), vintage (<1, 1–2, 2–4, ≥4 years), access type (graft, fistula, catheter), systolic blood pressure (<120, 120–140, 140–160, 160–180, ≥180 mm Hg), residual urine output (≤ versus >200 ml/day), diabetes, congestive heart failure, peripheral vascular disease, ischemic heart disease, cerebrovascular disease, serum albumin, creatinine, hematocrit (<30, 30–33, 33–36, ≥36%), and phosphorus, and use of α-adrenergic blocker, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker, β-blocker, calcium channel blocker, nitrates, and other antihypertensives. A two-way cross-product term with time was included for albumin due to non-proportional hazards. Abbreviations: ref., reference; CI, confidence interval; HR, hazard ratio.

The data suggested effect modification of the UFR–mortality and UFR–CV mortality associations on the basis of congestive heart failure. Specifically, UFR between 10 and 13 ml/kg/h was associated with greater all-cause mortality and nearly associated with greater CV mortality among patients with congestive heart failure, but was not among patients without congestive heart failure (Table 2). The estimates for UFR >13 ml/kg/h did not appear to be materially affected by the presence or absence of congestive heart failure; point estimates of both groups were similar to those from the primary analyses. No effect modification on the basis of oliguria, arterial disease (coronary, cerebral, or peripheral arterial), or HEMO Study treatment group assignment (flux or dose) was detected (data not shown).

Secondary Analyses

In order to more fully examine the threshold(s) at which UFR may become harmful, we conducted analyses in which we examined the association of UFR, represented as a cubic spline, with CV and all-cause mortality. As demonstrated in Figure 3, the HRs for both CV and all-cause mortality rose sharply at values between 10 and 14 ml/h/kg, and to a less pronounced degree at higher values. Consistent with results of the primary analysis, the HR for CV mortality was greater than that for all-cause mortality at all values of UFR.

Figure 3.

Cubic spline analysis of the associations between ultrafiltration rate (UFR) and cardiovascular (CV) (solid line) and all-cause (dashed line) mortality. Hazard ratios were adjusted for age, sex, interdialytic weight gain, race (black, non-black), smoking status (never, past, current), vintage (<1, 1–2, 2–4, ≥4 years), access type (graft, fistula, catheter), systolic blood pressure (<120, 120–140, 140–160, 160–180, ≥180 mm Hg), residual urine output (≤ versus >200 ml/day), diabetes, congestive heart failure, peripheral vascular disease, ischemic heart disease, cerebrovascular disease, serum albumin, creatinine, hematocrit (<30, 30–33, 33–36, ≥36%), and phosphorus, and use of α-adrenergic blocker, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker, β-blocker, calcium channel blocker, nitrates, and other antihypertensives. Estimates are presented for UFRs between 5.8 ml/h/kg (the 5th percentile of observed UFR in the study sample) and 20.4 ml/h/kg (the 95th percentile).

In order to examine the association between UFR and CV-related morbidity, we conducted time-to-event analyses in which the outcomes of interest were (1) hospitalization for CV disease or all-cause mortality (n=1081); (2) hospitalization for CV disease or CV mortality (n=843); and (3) hospitalization for CV disease (n=742). In total, participants contributed 3762 patient-years of at-risk time with a median survival time of 1.5 years. In each instance, UFR >13 ml/kg/h was potently and significantly associated with a greater hazard for outcome, whereas UFR 10–13 ml/kg/h was not (Figure 4).

Figure 4.

Adjusted association between ultrafiltration rate (UFR) and (1) cardiovascular (CV) hospitalization and all-cause mortality, (2) CV hospitalization and CV-related mortality, and (3) CV hospitalization. Based on Cox regression models adjusted for age, sex, interdialytic weight gain, race (black, non-black), smoking status (never, past, current), vintage (<1, 1–2, 2–4, ≥4 years), access type (graft, fistula, catheter), systolic blood pressure (<120, 120–140, 140–160, 160–180, ≥180 mm Hg), residual urine output (≤ versus >200 ml/day), diabetes, congestive heart failure, peripheral vascular disease, ischemic heart disease, cerebrovascular disease, serum albumin, creatinine, hematocrit (<30, 30–33, 33–36, ≥36%), and phosphorus, and use of α-adrenergic blocker, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker, β-blocker, calcium channel blocker, nitrates, and other antihypertensives. Unadjusted estimates (not shown) for the relationship between UFR (10–13 and >13 ml/h/kg, respectively) and outcomes were: 0.88 (0.76–1.03; P=0.13) and 1.21 (1.05–1.40; P=0.01) for CV hospitalization and all-cause mortality; 0.90 (0.76–1.08; P=0.26) and 1.23 (1.05–1.45; P=0.01) for CV hospitalization and CV-related mortality; and 0.88 (0.73–1.06; P=0.18) and 1.14 (0.96–1.36; P=0.13) for CV hospitalization. Abbreviations: ref., reference; CI, confidence interval.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....