Renal dysfunction (RD) is associated with increased mortality in heart failure (HF). The aim of this study was to identify whether worsened or improved renal function during mid-term follow-up is associated with worsened outcomes in patients with chronic HF. A total of 892 participants from a multicenter cohort study of chronic HF were followed over 3.1 ± 1.9 years of enrollment. Worsened and improved renal functions were tested with multivariate models as independent predictors of HF hospitalization and mortality. Although 12% of subjects experienced a ≥25% decrease in estimated glomerular filtration rate (eGFR), 17% experienced a ≥25% increase in eGFR, and there was stability of kidney function observed in the cohort as a whole. The quartile with the worst RD at any point in time had increased risk of HF hospitalization and mortality. Worsened eGFR was associated with HF outcomes in the unadjusted (hazard ratio = 1.71, 95% confidence interval 1.04 to 2.81, p = 0.035), but not the adjusted analysis. Improvement in eGFR was not associated with outcome (p = 0.453). In chronic HF, the severity of RD predicts risk of poor outcome better than changes in renal function during mid-term follow-up. This suggests that in patients with appropriately treated chronic HF, worsening renal function in itself does not yield useful prognostic information and may not reflect poor outcome.
Heart failure (HF) affects approximately 6 million people in the United States. Co-morbidities clearly impact HF prognosis. Over the last 2 decades, the number of co-morbidities and medications in the average patient with HF has increased substantially, renal failure being among those. Given the high cost of HF hospitalization, identifying risk factors that increase its likelihood is useful. Renal function is considered to be a sensitive marker of decreased organ perfusion and is commonly believed to deteriorate in HF because of chronic hypoperfusion. Recently, several studies have reported an association between worsening renal function (WRF) during inpatient treatment for acute decompensated HF and poor clinical outcomes. In chronic HF, reduced renal perfusion may occur over a long period, and patients may experience few symptoms related to the declining renal function. Several studies have found an association of WRF with mortality in the ambulatory setting. Most studies have included only patients with HF with reduced ejection fraction (HFrEF), and follow-up has typically been short, investigating changes in renal function over no more than a 6-month interval from baseline. Our aim was to assess how kidney function changed during mid-term follow-up in patients with HF, and whether WRF predicts all-cause mortality and HF hospitalization in patients medically treated for chronic HF. We also examined risk factors for WRF and whether improvement in renal function was associated with improved outcomes.
Methods
Subjects were enrolled in the multicenter Penn Heart Failure study. The Penn Heart Failure study began in 2003 at the University of Pennsylvania and subsequently expanded into a multicenter study. This is a prospective observational cohort study of more than 2,000 subjects with HF followed in HF specialty clinics. The study was approved by institutional review committees, and the subjects gave informed consent. Detailed patient information was collected at baseline and patients were followed every 6 months to measure predefined end points (hospitalization, change in therapy, and death). Patients were either seen in clinic or called at intervals of 6 months. Inclusion criteria in this analysis were an available baseline measurement of creatinine (at time of enrollment) and at least 1 follow-up value. At the beginning of the study follow-up kidney function was not routinely collected, and therefore only the subset of patients in whom this information was available was included in this analysis. The primary outcome measures were death or HF hospitalization (primary composite outcome) and death alone. Ten subjects underwent heart transplantation and were counted in the death end point. This was done because the assumed outcome without transplantation is death. HF hospitalization was based on primary discharge diagnosis. Patients with a clinical diagnosis of HF were considered to have HFrEF based on an EF ≤40% as defined in current guidelines. The remaining patients were classified as HF with preserved EF (HFpEF).
Estimated glomerular filtration rate (eGFR) was calculated using the modification of diet in renal disease equation. Change in eGFR was calculated by subtracting the most recent follow-up eGFR from baseline eGFR. For patients with a primary outcome, the most recent eGFR before reaching the primary outcome was used. We used previously defined criteria for WRF: a ≥25% decrease in eGFR or an increase in serum creatinine (SCr) ≥0.3 mg/dl. Improvement in renal function was defined as a ≥25% increase in eGFR or a decrease in SCr ≥0.3 mg/dl.
Participants were divided into quartiles of baseline eGFR. Comparisons between baseline eGFR groups were made with 1-way analysis of variance, Kruskal-Wallis tests, or chi-square tests based on distribution and normality assumptions. Univariate Cox proportional hazards models were used to assess the relation between time to a primary outcome and baseline or follow-up eGFR and/or SCr. WRF status and time to primary composite outcome were also assessed with univariate Cox proportional hazards model. Similarly, univariate Cox models were used for the mortality outcome. To assess for linearity in the coefficients of the Cox model over the entire range of follow-up SCr and eGFR, each group was divided into quartiles and hazard ratios (HRs) calculated using the lowest SCr quartile and highest eGFR quartile as the reference.
Multivariate Cox proportional hazards models were developed by compiling a list of 39 baseline variables of clinical importance and that did not have large numbers of missing values. Univariate Cox models of each baseline variable were created for time to primary composite outcome. Candidate variables were considered to be baseline variables that had chi-square p values >0.05. Backward and forward stepwise models of the candidate variables were run to determine final variables for inclusion in multivariate models. To test for robustness, models were rerun excluding variables that might overcorrect the model, such as New York Heart Association (NYHA) class. This led to no significant changes in the predictive value of the variables in the model, so the best iteration is presented. The previously mentioned analysis was repeated for HFrEF and HFpEF separately.
Results
The analysis cohort included 892 patients. Fifty-two (5.7%) of the 892 subjects were missing data for at least 1 candidate variable in the multivariate analysis. Table 1 illustrates baseline characteristics across eGFR quartiles. The average age was 56 years and 2/3 of the subjects were men. HFrEF was present in 61% of patients. Most patients had NYHA class II or III symptoms. More than 1/3 (36%) of the study population had experienced a hospitalization in the 12 months before enrollment into Penn Heart Failure study. Older patients, those with an ischemic origin, and those with co-morbidities such as diabetes, hypertension, and stroke were more likely to have lower baseline eGFR. NYHA class and the Minnesota Living with Heart Failure Questionnaire score were greater in patients with lower baseline eGFR, indicating higher symptom burden. Mean EF (37%) was not different between groups, nor were blood pressure, heart rate, body mass index, or serum sodium. Loop diuretics, aldosterone antagonists, aspirin, hydralazine, long acting nitrates, and statins were more commonly prescribed in patients with worse baseline renal function. Angiotensin-converting enzyme (ACE) inhibitor use was more common in those with better baseline renal function.
| Variable | Quartile | P | ||||
|---|---|---|---|---|---|---|
| 1 (n=223) | 2 (n=223) | 3 (n=223) | 4 (n=223) | Total Cohort (n=892) | ||
| eGFR (mL*min -1 *1.73m -2 ) median (minimum, maximum) | 95 (85, 628) | 76 (69, 84) | 62 (53, 69) | 40 (6, 53) | 69 (6, 628) | … |
| Age (years) | 48 (14) | 54 (13) | 59 (14) | 64 (12) | 56 (15) | < 0.01 |
| Male | 154 (69%) | 138 (62%) | 135 (61%) | 128 (57%) | 555 (62%) | 0.07 |
| White Race | 155 (71%) | 165 (76%) | 182 (83%) | 151 (70%) | 653 (73%) | < 0.01 |
| Black Race | 58 (27%) | 50 (23%) | 32 (15%) | 58 (27%) | 198 (22%) | |
| Ischemic origin | 33 (15%) | 47 (21%) | 70 (32%) | 78 (36%) | 228 (26%) | < 0.01 |
| Systolic heart failure | 114 (52%) | 130 (58%) | 121 (55%) | 121 (55%) | 486 (55%) | 0.57 |
| Hospitalization in prior 12 months | 68 (30%) | 83 (37%) | 84 (38%) | 91 (41%) | 326 (37%) | 0.14 |
| Diabetes mellitus | 44 (20%) | 49 (22%) | 49 (22%) | 86 (39%) | 228 (26%) | < 0.01 |
| Hypertension | 11 (50%) | 121 (54%) | 131 (59%) | 166 (74%) | 529 (59%) | < 0.01 |
| Stroke | 3 (1%) | 16 (7%) | 10 (4%) | 21 (9%) | 50 (6%) | < 0.01 |
| Follow-up time, years, median (IQR) | 3.0 (1.6, 5.0) | 2.9 (1.8, 4.8) | 3.0 (1.7, 4.4) | 2.5 (1.3, 3.7) | 2.9 (1.5, 4.5) | < 0.01 |
| New York Heart Association Class | ||||||
| II | 122 (55%) | 123 (55%) | 116 (52%) | 113 (52%) | 472 (53%) | |
| III | 31 (14%) | 43 (19%) | 45 (20%) | 71 (33%) | 190 (22%) | < 0.01 |
| IV | 1 (0%) | 0 (0%) | 6 (3%) | 6 (3%) | 13 (1%) | |
| Ejection Fraction (%) | 37 (16) | 36 (17) | 38 (17) | 39 (18) | 37 (17) | 0.53 |
| Body Mass Index (kg/m 2 ) | 30 (7) | 30 (8) | 31 (7) | 32 (9) | 31 (8) | 0.11 |
| Heart rate (beats per minute) | 73 (13) | 72 (13) | 73 (14) | 72 (13) | 72 (13) | 0.78 |
| Systolic blood pressure (mm Hg) | 117 (21) | 116 (21) | 117 (22) | 119 (25) | 118 (22) | 0.59 |
| MLHFQ ∗ score, median (IQR) | 19 (2, 51) | 18 (4, 45) | 24 (6, 50) | 34 (9, 59) | 24 (4, 52) | 0.01 |
| Serum creatinine (mg/dL) | 0.9 (0.1) | 1.0 (0.1) | 1.2 (0.2) | 2.1 (1.4) | 1.3 (0.9) | < 0.01 |
| Serum sodium (mEq/L) | 139 (3) | 140 (2) | 139 (3) | 139 (4) | 139 (3) | 0.19 |
| Potassium-sparing diuretics | 2 (1%) | 3 (1%) | 7 (3%) | 2 (1%) | 14 (2%) | 0.25 |
| Loop diuretics | 118 (53%) | 135 (61%) | 138 (62%) | 168 (75%) | 559 (63%) | < 0.01 |
| ACE inhibitors | 161 (72%) | 163 (73%) | 149 (67%) | 138 (62%) | 611 (68%) | 0.04 |
| Aldosterone antagonist | 59 (26%) | 64 (29%) | 61 (27%) | 87 (39%) | 271 (30%) | 0.01 |
| Angiotensin receptor blockers | 48 (22%) | 50 (22%) | 59 (26%) | 56 (25%) | 213 (24%) | 0.58 |
| Aspirin | 114 (51%) | 122 (55%) | 123 (55%) | 145 (65%) | 504 (57%) | 0.02 |
| β-Blockers | 195 (87%) | 198 (89%) | 194 (87%) | 198 (89%) | 785 (88%) | 0.91 |
| Digoxin | 63 (28%) | 58 (26%) | 57 (26%) | 70 (31%) | 248 (28%) | 0.5 |
| Hydralazine | 14 (6%) | 9 (4%) | 9 (4%) | 36 (16%) | 68 (8%) | < 0.01 |
| Long acting nitrate | 19 (9%) | 21 (9%) | 21 (9%) | 54 (24%) | 115 (13%) | < 0.01 |
| Statin | 89 (40%) | 127 (57%) | 128 (57%) | 143 (64%) | 487 (55%) | < 0.01 |
∗ Minnesota Living with Heart Failure Questionnaire (MLHFQ).
There are 2,767 patient-years of follow-up in the cohort, with a median follow-up time of 2.9 years, and mean follow-up of 3.1 ± 1.9 years. A regression analysis of creatinine values versus time was created for each subject. The mean change in creatinine over time was 0.0074 ± 0.43 mg/dl increase in creatinine per year; the slope of this regression line was not statistically different from zero. Similarly, eGFR did not deteriorate over time in the cohort as a whole. Stage 3 or greater chronic kidney disease (CKD) was present at baseline in 309 (35%) of the 892 subjects. A total of 322 (36%) subjects had Stage 3 or greater CKD at the most recent follow-up visit or just before reaching a primary outcome. A total of 109 (12%) subjects experienced WRF during follow-up using eGFR, 110 (12%) using SCr. A total of 152 (17%) subjects experienced improved eGFR; 108 (12%) had improved SCr. A total of 674 (76%) subjects had stable eGFR. Mean baseline SCr in the worsening eGFR group was 1.56 ± 1.18 mg/dl and was 1.27 ± 0.81 mg/dl in the stable group (p = 0.015). There was a trend toward WRF in patients with lower baseline eGFR, but this did not reach statistical significance (p = 0.076). A total of 110 subjects (12%) had a primary outcome; 26 (2.9%) died. Of 840 subjects with complete data in the multivariate analysis, 107 (12%) had a primary outcome, of whom 26 (3.1%) died and 10 (1.2%) underwent heart transplant.
Table 2 includes the univariate analysis results for the 14 candidate variables and 4 preselected variables (age, gender, diabetes, and ischemic status). Of the preselected variables, only gender did not have a statistically significant association with the primary composite outcome of death or HF hospitalization (p = 0.906). As illustrated in Table 3 , baseline and follow-up renal function demonstrated significant associations with the primary composite outcome in both the unadjusted and adjusted analysis. Separate analyses of HFrEF (HR = 1.05, 95% confidence interval [CI] 1.00 to 1.09, p = 0.05 and HR = 1.21, 95% CI 1.10 to 1.32, p = <0.001 for baseline SCr and eGFR, respectively) and HFpEF (HR = 1.09, 95% CI 1.02 to 1.16, p = 0.009 and HR = 1.20, 95% CI 1.01 to 1.42, p = 0.038 for baseline SCr and eGFR, respectively) showed the same association with the primary outcome in unadjusted but not in the adjusted analysis. Figure 1 demonstrates that risk of the primary outcome rises markedly in the group with most impaired renal function at baseline. The association between follow-up renal function and the primary composite outcome was even stronger.
| Baseline characteristic | Hazard Ratio | 95% CI | P-value | Number (%) | Events (%) |
|---|---|---|---|---|---|
| New York Heart Association class | 882 (99) | 109 (99.1) | |||
| I | reference | — | — | ||
| II | 3.21 | 1.46 – 7.09 | 0.004 | ||
| III | 9.89 | 4.47 – 21.91 | < 0.001 | ||
| IV | 13.47 | 3.93 – 46.2 | < 0.001 | ||
| Race | 867 (97.3) | 108 (98.2) | |||
| Caucasian | reference | — | — | ||
| Black | 1.63 | 1.07 – 2.48 | 0.022 | ||
| Other | 0.54 | 0.07 – 3.86 | 0.536 | ||
| Gender | 892 (100) | 110 (100) | |||
| Female | reference | — | — | ||
| Male | 0.94 | 0.64 – 1.40 | 0.747 | ||
| Age (years) | 1.019 | 1.01 – 1.03 | 0.007 | 892 (100) | 110 (100) |
| Ischemic origin | 1.47 | 0.98 – 2.21 | 0.060 | 876 (98.3) | 109 (99.1) |
| Hospitalization in prior 12 months | 1.22 | 1.14 – 1.32 | < 0.001 | 892 (100) | 110 (100) |
| Diabetes | 1.52 | 1.01 – 2.28 | 0.043 | 892 (100) | 110 (100) |
| Hypertension | 2.07 | 1.37 – 3.12 | 0.001 | 892 (100) | 110 (100) |
| Hyperlipidemia | 1.65 | 1.12 – 2.42 | 0.010 | 892 (100) | 110 (100) |
| Ejection fraction | 0.973 | 0.96 – 0.99 | < 0.001 | 886 (99.4) | 110 (100) |
| Heart rate, beats per minute | 1.017 | 1.00 – 1.03 | 0.021 | 877 (98.4) | 108 (98.2) |
| Systolic blood pressure (mm Hg) | 0.989 | 0.98 – 1.00 | 0.026 | 884 (99.2) | 110 (100) |
| MLHFQ score | 1.017 | 1.01 – 1.02 | < 0.001 | 892 (100) | 110 (100) |
| Loop diuretic | 2.82 | 1.78 – 4.47 | < 0.001 | 892 (100) | 110 (100) |
| Aldosterone antagonist | 2.08 | 1.42 – 3.03 | < 0.001 | 892 (100) | 110 (100) |
| Hydralazine | 3.13 | 1.76 – 5.55 | < 0.001 | 892 (100) | 110 (100) |
| Long acting nitrate | 3.04 | 1.96 – 4.71 | < 0.001 | 892 (100) | 110 (100) |
| Digoxin | 1.57 | 1.07 – 2.31 | 0.020 | 892 (100) | 110 (100) |
| Primary Composite | Unadjusted Analysis | Adjusted Analysis ∗ | ||||
|---|---|---|---|---|---|---|
| HR † | 95% CI | P | HR † | 95% CI | P | |
| Baseline SCr | 1.06 | 1.02 – 1.10 | 0.002 | 1.04 | 0.99 – 1.09 | 0.091 |
| Baseline eGFR | 1.20 | 1.11 – 1.30 | < 0.001 | 1.11 | 1.02 – 1.21 | 0.017 |
| Follow-Up SCr | 1.05 | 1.02 – 1.08 | 0.002 | 1.05 | 1.00 – 1.10 | 0.040 |
| Follow-Up eGFR | 1.24 | 1.15 – 1.33 | < 0.001 | 1.16 | 1.07 – 1.26 | < 0.001 |
| Stable SCr | ref. | |||||
| ≥ 0.3 mg/dL ↓SCr ‡ | 2.02 | 1.20 – 3.40 | 0.008 | 1.11 | 0.64 – 1.93 | 0.713 |
| ≥ 0.3 mg/dL ↑SCr ‡ | 2.21 | 1.36 – 3.57 | 0.001 | 1.27 | 0.76 – 2.13 | 0.368 |
| Stable eGFR | ref. | |||||
| ≥ 25% ↑eGFR ‡ | 1.32 | 0.81 – 2.15 | 0.269 | 0.88 | 0.53 – 1.47 | 0.624 |
| ≥ 25% ↓eGFR ‡ | 1.71 | 1.04 – 2.81 | 0.035 | 0.92 | 0.53 – 1.58 | 0.759 |
| Death | HR | 95% CI | P | HR | 95% CI | P |
|---|---|---|---|---|---|---|
| Baseline SCr | 1.03 | 0.95 – 1.13 | 0.424 | 0.96 | 0.85 – 1.07 | 0.433 |
| Baseline eGFR | 1.21 | 1.05 – 1.39 | 0.009 | 1.05 | 0.90 – 1.23 | 0.514 |
| Follow-Up SCr | 1.05 | 1.00 – 1.11 | 0.052 | 1.03 | 0.94 – 1.14 | 0.477 |
| Follow-Up eGFR | 1.33 | 1.17 – 1.52 | < 0.001 | 1.23 | 1.04 – 1.44 | 0.014 |
| Stable SCr | ref. | |||||
| ≥ 0.3 mg/dL ↓SCr ‡ | 1.92 | 0.72 – 5.14 | 0.192 | 0.81 | 0.28 – 2.31 | 0.689 |
| ≥ 0.3 mg/dL ↑SCr ‡ | 3.1 | 1.41 – 6.82 | 0.005 | 1.48 | 0.61 – 3.57 | 0.387 |
| Stable eGFR | ref. | |||||
| ≥ 25% ↑eGFR ‡ | 1.03 | 0.39 – 2.73 | 0.956 | 0.53 | 0.19 – 1.52 | 0.239 |
| ≥ 25% ↓eGFR ‡ | 2.26 | 1.00 – 5.10 | 0.050 | 1.03 | 0.42 – 2.55 | 0.941 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
