Results
Incidence of WRF
We studied an elderly, severely symptomatic HF population with a significant burden of comorbidities (Table I). The proportions of patients in chronic kidney disease stages 1 (eGFR ≥90 mL/min per 1.73 m), 2 (60–89), 3 (30–59), 4 (15–29), and 5 (<15 or dialysis) at baseline were 4%, 32%, 55%, 9%, and 0%. The incidence of WRF defined as a rise in sCr >0.2 mg/dL, >0.3 mg/dL, and >0.5 mg/dL at any time during the first 6 months was 299 (53%) of 566, 229 (40%) of 566, and 124 (22%) of 566, respectively. The incidence of WRF I, WRF II, and WRF III was 70 (12%) of 566, 105 (19%) of 566, and 124 (22%) of 566, respectively, which occurred at all time points during the first 6 months of the study.
Prognostically Relevant WRF
During the entire 18-month study period, 105 (19%) of 566 patients died. The risk of death in patients with WRF III was twice as high as in patients without WRF (hazard ratio [HR] 1.98 [95% CI 1.27–3.07], P = .002), whereas the risk of death in patients with WRF I and II was not significantly different from those without WRF (Figure 1). Accordingly, further analyses were focused on WRF III.
(Enlarge Image)
Figure 1.
Survival in patients who developed WRF I to III and those who did not. For definitions, see text.
Univariable Predictors of WRF III
Baseline characteristics of patients subsequently experiencing WRF III and those not doing so are depicted in Table I . Patients with subsequent WRF III were more likely to have a history of renal failure, anemia, and orthopnea; were more likely to have edema; and had higher blood urea nitrogen (BUN) and NT-proBNP plasma concentrations and lower hemoglobin level. Groups did not differ in terms of age, gender, HF etiology, left ventricular ejection fraction (LVEF), and sCr or eGFR at baseline. Patients subsequently experiencing WRF III were more likely to be treated with loop diuretics than those who did not, whereas other medications did not differ. These patients also had higher doses of loop diuretics, whereas doses of other medications did not differ (data not shown).
Changes in medication recommended within the initial 3 months were more common in patients who subsequently developed WRF III than in those who did not ( Table II ). In particular, recommendations to increase doses of loop diuretics or to start spironolactone were accompanied with WRF III, whereas recommendations on angiotensin-converting enzyme inhibitors (ACEI)/angiotensin receptor blockers (ARB) and β-blockers were not. Patients who developed WRF III had higher doses of loop diuretics throughout the first 6 months (Figure 2A) and larger increases in loop diuretic doses ( Table II ). In addition, the proportion of patients on spironolactone at any time during the first 6 months was larger in patients experiencing WRF III ( Table II ). In contrast, the groups did not differ with respect to allocation to NT-proBNP–guided management or changes of doses of ACEI/ARB and β-blockers ( Table II ).
(Enlarge Image)
Figure 2.
Panel A, Box plots representing loop diuretic doses expressed as milligram furosemide equivalents in patients who developed WRF III and those who did not. Boxes represent medians with interquartile ranges, whiskers indicate 10th and the 90th percentiles. Panel B, Proportion of patients with any signs of congestion (cf. Table I) from baseline to month 3 among patients who developed WRF III and those who did not. *P < .05, P < .1 versus no WRF III.
Multivariable Predictors of WRF III
In the multivariable analysis ( Table III ), a history of renal failure, any treatment with spironolactone during the first 6 months, higher baseline loop diuretic dose, and higher maximal increase in loop diuretic dose were independently associated with the occurrence of WRF III. If WRF III was treated as a time-dependent variable, that is, Cox regression with WRF III occurring at month 1, 3, or 6, the results did not differ significantly (data not shown).
Time Course of HF Characteristics in Relation to WRF III
Patients who developed WRF III had more signs of congestion throughout the first 3 months (Figure 2B) and higher NT-proBNP throughout the first 6 months (Figure 3A) than those not doing so. Although systolic blood pressure and eGFR at baseline did not differ between groups, systolic blood pressure and eGFR at months 1, 3, and 6 were lower in patients developing WRF III (Figure 3B and C).
(Enlarge Image)
Figure 3.
N-terminal pro–B-type natriuretic peptide (panel A, logarithmic scale), systolic blood pressure (panel B), and eGFR (panel C) from baseline to month 6 in patients who developed WRF III and those who did not. Means and SDs are shown. *P < .05 versus no WRF III.
WRF III and Outcomes
Patients who developed WRF III had higher rates of death, death or any hospitalization, and death or HF hospitalization at 18 months compared with those who did not (Figure 4). After extensive adjustment for baseline characteristics including sCr, eGFR, BUN, and symptoms and signs of congestion, WRF III was still predictive of death (HR 2.00 [1.24–3.23], P = .004), death or any hospitalization (HR 1.56 [1.06–1.78], P = .02), and death or HF hospitalization (HR 1.70 [1.16–2.47], P = .006).
(Enlarge Image)
Figure 4.
Survival, hospitalization-free survival, and HF hospitalization-free survival in patients who developed WRF III and those who did not.
In the multivariable model apart from WRF III (HR 1.87 [1.23–2.86], P = .004), history of anemia (HR 1.63 [1.18–2.07], P = .002), higher baseline NT-proBNP (HR 1.37 [1.08–1.73] per ln unit, P = .008), higher BUN (HR 1.05 [1.02–1.09] per mmol/L, P = .001), the presence of rales at baseline (HR 1.56 [1.24–1.96] per scale point 0–3, P < .001), and baseline spironolactone treatment (HR 1.64 [1.09–2.48], P = .02) were independently associated with mortality, whereas sCr and eGFR were not.
There was no interaction between treatment allocation (NT-proBNP–guided vs symptom-guided) and prognostic value of WRF III in any of these analyses (data not shown).There was, however, an interaction between WRF III and baseline sCr and mortality (P = .03), which is illustrated in Figure 5. Patients with both supramedian baseline sCr and subsequent WRF III had a substantially higher risk of death than patients with inframedian baseline sCr and no subsequent WRF III, whereas in patients with inframedian baseline sCr and with subsequent WRF III and patients with supramedian baseline sCr and without subsequent WRF III, the risk was moderately elevated.
(Enlarge Image)
Figure 5.
Survival according to baseline serum creatinine (sCr, ↓: inframedian, ↑: supramedian) and occurrence of worsening renal function grade III (WRF III). BL, baseline.