Single baseline serum creatinine measurements predict mortality in critically ill patients hospitalized for acute heart failure

Abstract

Background

Acute heart failure (AHF) is a leading cause of death in critically ill patients and is often accompanied by significant renal dysfunction. Few data exist on the predictive value of measures of renal dysfunction in large cohorts of patients hospitalized for AHF.

Methods

Six hundred and eighteen patients hospitalized for AHF (300 male, aged 73.3 ± 10.3 years, 73% New York Heart Association Class 4, mean hospital length of stay 12.9 ± 7.7 days, 97% non‐ischaemic AHF) were included in a retrospective single‐centre data analysis. Echocardiographic data, serum creatinine/urea levels, estimated glomerular filtration rate (eGFR), and clinical/laboratory markers were recorded. Mean follow‐up time was 2.9 ± 2.1 years. All‐cause mortality was recorded, and univariate/multivariate analyses were performed.

Results

Normal renal function defined as eGFR > 90 mL/min/1.73 m² was noted in only 3% of AHF patients at baseline. A significant correlation of left ventricular ejection fraction with serum creatinine levels and eGFR (all P < 0.002) was noted. All‐cause mortality rates were 12% (90 days) and 40% (at 2 years), respectively. In a multivariate model, increased age, higher New York Heart Association class at admission, higher total cholesterol levels, and lower eGFR independently predicted death. Patients with baseline eGFR < 30 mL/min/1.73 m² had an exceptionally high risk of death (odds ratio 2.80, 95% confidence interval 1.52–5.15, P = 0.001).

Conclusions

In a large cohort of patients with mostly non‐ischaemic AHF, enhanced serum creatinine levels and reduced eGFR independently predict death. It appears that patients with eGFR < 30 mL/min/1.73 m² have poorest survival rates. Our data add to mounting data indicating that impaired renal function is an important risk factor for non‐survival in patients hospitalized for AHF.

Introduction

Acute heart failure (AHF) is a leading cause of death and implies an important burden on healthcare systems worldwide. AHF is of epidemiologic importance as it accounts for over 1 million hospitalizations and about 300 000 heart‐failure‐related deaths annually in the USA alone.1, 2 Among the key findings in critically ill patients with AHF on today's intensive care units is the fact that it may often be accompanied by comorbidities1, 3, 4, 5 such as impaired renal function.4, 6, 7, 8, 9, 10, 11, 12, 13, 14

Although the pathophysiology of AHF‐induced acute kidney injury remains incompletely understood, a number of pathomechanisms including both haemodynamic and neurohormonal mechanisms were proposed.1, 4, 7, 8, 9, 10, 11, 12, 15, 16 Importantly, chronic kidney disease (CKD) may often be present as underlying comorbidity in patients with stable chronic heart failure (CHF).4, 7, 10, 11 Data from large CHF cohorts demonstrate that up to 60% of CHF patients develop some degree of CKD.17 Importantly, in patients with stable CHF, presence of CKD significantly affects survival rates, and previous investigations demonstrate a strong correlation between glomerular filtration rate (GFR) and risk of death.4, 10, 11, 14, 18 In CHF, data indicate that the risk of death may even be stronger correlated to a decline in GFR than to declining left ventricular ejection fraction (LVEF).19

In addition, acute renal failure (ARF) may develop in AHF that may require renal replacement therapy (RRT).12, 20, 21 ARF may induce progressive uraemia that worsens the prognosis of affected patients.22 Importantly, development of ARF accounts for excess mortality not only in heart failure patients but also in a general population of critically ill patients.7, 12, 18, 21, 23, 24 However, in cases of dialysis‐dependent ARF, optimal timing of RRT25 and optimal mode (i.e. continuous vs. intermittent RRT) remain currently unclear.26, 27, 28 This may especially be the case in haemodynamically instable patients with AHF.

Little is known about the predictive value of measures of routine renal function in the subgroup of patients hospitalized for non‐ischaemic AHF. In a large cohort of patients hospitalized for non‐ischaemic AHF, we therefore aimed to investigate the prognostic impact of impaired renal function assessed by serum creatinine/urea levels and estimated GFR (eGFR) on all‐cause mortality.

Methods

Study patients and study design

Between 2001 and 2003, n = 638 Caucasian patients were hospitalized for AHF. Of those, 618 patients [mean age 73.3 ± 10.3 years, 300 male, 449 in New York Heart Association (NYHA) Class 4] had available data for evaluation of renal function (i.e. baseline creatinine). Data from admission charts in a single centre, that is, the General Hospital Murska Sobota (Slovenia), were analysed in a retrospective fashion. The retrospective data analysis was approved by the National Ethics Committee and was performed in accordance with the Declaration of Helsinki. Patients with clinical and radiological evidence of decompensated AHF were included on the basis of clinical symptoms as established in respective international guidelines.29, 30 Patients were hospitalized in medical intensive care/intermediate care units and were treated according to current international recommendations. Patients on ambulatory chronic hemodialysis were excluded from the analysis.

Assessment of renal function, outcome parameters, and laboratory indices

At admission, serum creatinine levels, serum urea levels, eGFR, LVEF, high‐sensitivity C‐reactive protein, white blood cell count, and other prognostically important laboratory indices were assessed. Serum creatinine and serum urea levels were assessed by photometry from heparin plasma samples. For serum creatinine levels, the Jaffé method of detection was used (Roche Diagnostics, Mannheim, Germany). For assessment of outcome, the 28 days mortality, length of hospital stay, total follow‐up time, NYHA status at discharge, and other clinical indices were recorded. eGFR was assessed using the modified diet in renal disease formula31: GFR (mL/min/1.73 m²) = 186 × (S_cr/88.4)^−1.154 × (age)^−0.203 × (0.742 if female). Although the degree of CKD remains unclear in the present cohort (please also refer to limitations section of discussion), study patients were grouped according to disease severity Stages 1–5 equivalent to severity stages proposed for CKD. The following stages applied: Stage 1 (increased risk; GFR > 90 mL/min/1.73 m²), Stage 2 (mild severity; GFR 60–89 mL/min/1.73 m²), Stage 3 (moderate severity; GFR 30–59 mL/min/1.73 m²), Stage 4 (severe disease; GFR 15–29 mL/min/1.73 m²), and Stage 5 (very severe disease; GFR < 15 mL/min/1.73 m²).

Statistical analysis

Statistical analyses were performed using MedCalc 9.0.1.1 software (MedCalc Software, Mariakerke, Belgium) and StatView 5 (SAS Institute, Cary, North Carolina, USA). All data are expressed as means ± SD and were tested for normal distribution using the Kolmogorov–Smirnov test, if not indicated otherwise. The relationship of baseline variables with survival was assessed by Cox proportional‐hazard analysis (univariate and multivariate analyses). Odds ratios (ORs) and 95% confidence interval (CI) for risk factors are given. Kaplan–Meier cumulative survival curves are displayed for illustrative purposes. Comparison between curves was performed by Mantel–Haensel log‐rank test. Significance was assigned when P < 0.05.

Results

Study population

A total of 618 Caucasian patients [mean age 73.3 ± 10.3 years, 300 male (48.5%), mean length of in‐hospital stay 12.9 ± 7.7 days, 100% NYHA Classes 3 and 4, CHADS score 2.4 ± 1.0] hospitalized on a medical intensive care unit for AHF were included in this analysis. Data on patients' demographics in respective equivalent stages of renal dysfunction and laboratory data are presented in Table 1. Of the overall sample, n = 152 (25%) patients presented with clinical signs of severe fluid overload (pleural effusion, pulmonary oedema, and peripheral oedema). Eighteen patients of the overall sample (3%) were considered to have AHF because of reasons related to acute ischaemia, that is, acute myocardial infarction (AMI). n = 160 patients (26%) of the overall samples initially presented in a hypertensive (initial systolic blood pressure ≥ 140 mmHg) state (mean systolic blood pressure in hypertensive subjects 151.4 ± 13.4 mmHg). At initial presentation, atrial fibrillation was present in n = 320 patients (52%), and n = 281 (i.e. 88%) of these patients were found to have chronic atrial fibrillation. At hospital admission, 22% of all patients had previous evidence for ischaemic heart disease or history of AMI (11%). Major concomitant diseases at hospital admission were as follows: arterial hypertension (44%), diabetes mellitus (33%), anaemia 29% (according to World Health Organization definitions), chronic obstructive pulmonary disease (17%), hyper‐lipoproteinaemia (12%), thyroid disease (7%), and gout (5%). In the study population, patients with advanced age and male gender were more likely to be found in higher renal disease severity categories (Table 1). Co‐medication at admission consisted in the following drugs at baseline: angiotensin‐converting‐enzyme inhibitors 78% plus angiotensin II receptor antagonists 8% (mean enalapril equivalent dose 11.0 ± 10.1 mg), 48% anti‐arrhythmic drugs/beta‐blockers, statins (20%), diuretics (83%), spironolactone (44%), and digoxin (37%). Mean amount of total medications 6.2 ± 1.9 per patient (with 4.4 ± 1.6 cardiovascular medications).

Table 1.

Patients' demographics and prognosis relevant indices

	All patients n = 618 (100%)	>90 mL/min/1.73 m2 (equivalent stage 1) n = 20 (3.2%)	60–89 mL/min/1.73 m2 (equivalent stage 2) n = 157 (25.4%)	30–59 mL/min/1.73 m2 (equivalent stage 3) n = 370 (59.9%)	15–29 mL/min/1.73 m2 (equivalent stage 4) n = 62 (10.0%)	<15 mL/min/1.73 m2 (equivalent stage 5) n = 9 (1.5%)
Age	73.3 ± 10.3	69.7 ± 11.9	72.3 ± 11.0	73.2 ± 10.1	78.7 ± 7.3	70.1 ± 9.0
Gender	300 male (48.5%)	6 male (30%)	48 male (31%)	201 male (54%)	39 male (63%)	6 male (67%)
LVEF (%)	43.4 ± 12.1	52.0 ± 14.9	44.2 ± 11.3	43.3 ± 12.4	39.5 ± 9.8	20.0 ± 2.5
NYHA at admission	3.7 ± 0.5	3.6 ± 0.50	3.7 ± 0.48	3.7 ± 0.45	3.9 ± 0.36	4.0 ± 0.0
Atrial fibrillation at admission	n = 320 (51.8%)	n = 9 (45%)	n = 71 (45%)	n = 199 (51%)	n = 33 (53%)	n = 1 (11%)
CHADS score	2.4 ± 0.96	2.6 ± 0.95	2.5 ± 1.1	2.36 ± 0.93	2.46 ± 0.92	2.1 ± 0.6
Length of in‐hospital stay (days)	12.9 ± 7.7	19.0 ± 16.7	12.8 ± 5.5	12.9 ± 7.9	12.4 ± 6.9	13.1 ± 7.3
Serum creatinine (µmol/L)	113.5 ± 65.1	53.5 ± 7.2	75.55 ± 7.4	110.9 ± 18.4	189.0 ± 32.3	504.6 ± 217.5
Serum urea (mg/dL)	9.2 ± 4.8	4.5 ± 1.9	6.6 ± 3.2	9.4 ± 3.8	15.5 ± 5.8	17.2 ± 6.6
eGFR (mL/min/1.73 m2)	51.6 ± 19.6	107.6 ± 20.4	70.75 ± 7.9	45.9 ± 8.3	24.4 ± 4.1	9.45 ± 4.2
Serum potassium (mmol/L)	4.4 ± 0.5	4.1 ± 0.4	4.3 ± 0.5	4.5 ± 0.5	4.6 ± 0.5	4.9 ± 0.6
C‐reactive protein (mg/L)	31.6 ± 43.3	46.0 ± 37.45	33.34 ± 38.75	29.54 ± 46.91	35.46 ± 38.96	23.78 ± 14.42
White blood cell count (x109/L)	8.9 ± 5.3	8.2 ± 3.6	8.8 ± 6.1	8.8 ± 3.5	10.2 ± 10.3	8.5 ± 3.75
Platelet count (x109/L)	236.9 ± 93.8	246.3 ± 94.5	248.6 ± 94.2	230.8 ± 88.6	238.1 ± 91.7	262.1 ± 228.9
Total cholesterol (mg/dL)	4.8 ± 1.6	4.5 ± 1.21	5.07 ± 1.76	4.78 ± 1.44	4.64 ± 2.01	4.01 ± 1.13
Total bilirubin (µmol/L)	20.1 ± 19.4	32.4 ± 49.5	17.9 ± 16.4	20.56 ± 18.94	19.42 ± 11.73	15.4 ± 11.75
Uric acid (mg/dL)	420.2 ± 139.9	293.9 ± 99.2	337.3 ± 110.0	440.5 ± 127.1	548.5 ± 129.9	411.5 ± 233.1
Creatinine phosphokinase (U/L)	1.4 ± 2.4	1.28 ± 0.78	1.39 ± 3.12	1.32 ± 2.08	1.25 ± 0.83	4.22 ± 6.67
Aspartate aminotransferase (U/L)	0.53 ± 2.0	0.51 ± 0.42	0.78 ± 3.92	0.42 ± 0.42	0.63 ± 1.35	0.38 ± 0.41

eGFR, estimated glomerular filtration rate; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association.

Assessment of cardiac function and cardiac injury

Assessment of cardiac function by means of echocardiography was performed in 63% of all cases at admission. A significant decline of LVEF was found with advanced stages of renal dysfunction (Stages 1 vs. 4, P = 0.01, Table 1). A direct correlation of LVEF with serum creatinine (r = −0.2, P = 0.001) and eGFR (r = 0.2, P = 0.002) but not with urea levels (r = −0.07, P = 0.27) was identified. Mean left ventricular end‐diastolic diameter (LVEDD) was 25.9 ± 26.8 mm (median 6.6, 25th–75th percentile: 5.2–55.0), and mean left ventricular end‐systolic diameter (LVESD) was 20.6 ± 22.1 mm (median 5.4, 25th–75th percentile: 3.9–44.0) at admission. LVEDD and LVESD were both found to correlate significantly with LVEF (both P < 0.0003) but not with serum creatinine, serum urea, or eGFR (all P > 0.4). Overall, mean systolic and diastolic blood pressures were 129.3 ± 18.8 and 77.7 ± 10.0 mmHg, respectively. The mean heart rate was 113.3 ± 29.9 beats per minute at admission. For laboratory assessment of cardiac injury, serum creatinine phosphokinase (CPK) levels were assessed in 80.1% of all admissions (Table 1). CPK was not found to correlate with LVEF, eGFR, serum creatinine, or serum urea levels (all P > 0.25).

An explorative subgroup analysis was performed in AHF patients grouped for diastolic (defined as LVEF > 40%) vs. systolic AHF. A significant difference in eGFR was noted between these two subgroups. In detail, eGFR was 53.4 ± 18.9 (diastolic AHF) vs. 47.6 ± 16.0 (systolic AHF) mL/min/1.73 m² (P = 0.0069). Correspondingly, serum creatinine levels were higher in patients with systolic AHF [i.e. 117.7 ± 46.9 vs. 104.3 ± 33.5 (diastolic AHF) mL/min/1.73 m², P = 0.01]. Nevertheless, mean age in patients with systolic AHF was higher (72.6 ± 9.8 vs. 69.7 ± 11.4 years, P = 0.02), and NYHA class at admission was higher also (3.8 ± 0.4 vs. 3.7 ± 0.5, P = 0.004). No between group differences were noted in regard to serum levels of uric acid, total cholesterol, alanine aminotransferase, and C‐reactive protein (all n.s.).

Assessment of renal function, renal injury, and associated comorbidities

Of 618 hospitalized patients for AHF, n = 20 patients (i.e. 3.2%) had an eGFR of above 90 mL/min/1.73 m², which was considered the lower limit of normal (Table 1). The major proportion (59.5%) of patients presented with Stage 3 equivalent renal dysfunction (eGFR 30–59 mL/min/1.73 m²). This demonstrates severe renal impairment in most study patients (Table 1). A highly significant correlation of eGFR with serum creatinine (r = −0.677), serum urea (r = −0.581), and serum potassium (r = −0.2284) was found (all P < 0.0001). Throughout CKD, equivalence stages of renal dysfunction, white blood cell counts, platelet counts, total cholesterol, CPK, aspartate aminotransferase, and alanine aminotransferase (data not shown) were rather unchanged (Table 1).

Impact of measures of renal and cardiac function on outcome prognostication

Details on univariate and multivariate analyses are given in Table 2. In the respective single predictor model for survival, pronounced effects on survival were observed for the following variables: age, NYHA class at admission, eGFR, serum levels of urea, serum levels of uric acid, and total cholesterol (Table 2). Interestingly, higher cholesterol levels were associated with improved survival rates, a fact previously known as the cholesterol paradox.32 Moreover, LVEF, serum creatinine levels, potassium, white blood cell count, haemoglobin, and diastolic blood pressure were noted to significantly impact on survival. In the multivariate model for survival, the following variables were included: age, NYHA class at admission, eGFR, haemoglobin, and total cholesterol (please refer to Table 2). For illustrative purposes, Kaplan–Meier survival estimate curves were constructed for 2 years (Figure 1 A) and 1 year follow‐up (Figure 1 B and C). All‐cause mortality 30 days after hospital admission was 6% and was 12% (Day 90), 18% (Day 180), 27% (after 1 year), 40% (after 2 years), and 49% (after 3 years), respectively. The mean survival time of patients surviving the initial episode of AHF was 1115.5 ± 753.7 days (median 1115.5 days, 25th–75th percentile 322.0–1689.0). For further assessment of impact of renal function on outcome measures, ORs were calculated after grouping of patients to the following eGFR categories: <30 mL/min/1.73 m² and <60 mL/min/1.73 m². In patients with an initial eGFR < 30 mL/min/1.73 m², the OR for death was determined as OR 2.80, 95% CI 1.52–5.15, P = 0.001. In the subgroup of patients with baseline eGFR < 60 mL/min/1.73 m², the odds for death were 1.94 (95% CI 1.36–2.76, P = 0.0003).

Table 2.

Univariate and multivariate survival models in patients hospitalized for acute heart failure

	Single predictor model for non‐survival			Multivariable model for non‐survival
Variable	Hazard ratio (95% CI)	P‐value	χ 2	Hazard ratio (95% CI)	P‐value	χ 2
Age (1 year increase)	1.034 (1.022–1.045)	<0.0001	38.2	1.028 (1.015–1.041)	<0.0001	19.0
Gender (male)	1.09 (0.891–1.332)	0.40	0.7	—	—	—
Aetiology of heart failure (non‐ischaemic)	1.204 (0.936–1.548)	0.15	2.1	—	—	—
LVEF (>40%/≤40%)*, increase by 1%	0.986 (0.973–0.999)	0.03	4.6	*	*	*
NYHA class at admission (per 1 class up)	1.981 (1.547–2.536)	<0.0001	33.0	1.68 (1.261–2.239)	0.0004	12.6
Creatinine (10 µmol/L increase)	1.014 (1.004–1.024)	0.006	6.0	—	—	—
eGFR (per 1 mL/min/1.73 m2 increase)	0.987 (0.981–0.992)	<0.0001	21.2	0.988 (0.982–0.995)	0.0006	11.9
Urea (10 mg/dL increase)	1.062 (1.043–1.082)	<0.0001	34.3	—	—	—
Uric acid (10 µmol increase)	1.017 (1.009–1.025)	<0.0001	17.1	—	—	—
Potassium (1 mmol/L increase)	1.379 (1.131–1.682)	0.0016	9.9	—	—	—
White blood cell count (1/nL increase)	1.027 (1.009–1.046)	0.0033	6.2	—	—	—
Haemoglobin (1 g/dL increase)	0.992 (0.987–0.997)	0.0012	10.0	0.996 (0.991–1.002)	0.20	1.6
Diastolic BP (10 mmHg increase)	0.979 (0.967–0.991)	0.0004	12.4	—	—	—
Total cholesterol (10 mg/dL increase)	0.835 (0.763–0.912)	<0.0001	17.6	0.885 (0.808–0.97)	0.009	6.8

BP, blood pressure; CI, confidence interval; eGFR, estimated glomerular filtration rate; LVEF, left ventricular ejection fraction; NYHA, New York Heart Association.

^— Not included to multivariable model.

^* Not included to multivariable model due to missing data.

Figure 1.

The 2 years (A) and 1 year (B) and (C) Kaplan–Meier survival estimates in patients hospitalized for acute heart failure according to estimated glomerular filtration rate categories are given (overall sample n = 618). (A) The 2 years survival estimates for acute heart failure patients with mild (full line), moderate (dashed line), and severe (dotted line) renal dysfunction. (B) The 1 year survival estimates for patients with normal to mildly reduced (full line), moderately reduced (dashed line), and severely to very severely (dotted line) reduced renal function. (C) The 1 year survival estimates for patients with normal to moderately (full line), severely (dashed line), and very severely (dotted line) reduced renal function.

Discussion

In the current analysis, we investigate the impact of renal dysfunction as assessed by a baseline single‐serum creatinine measurement on the outcomes of critically ill patients hospitalized for decompensated AHF. In a large cohort of AHF patients, we observed that mortality is significantly increased with advancing stages of renal dysfunction when patients were grouped according to GFR stages equivalent to the stages suggested for patients with CKD. Our findings add to few previous reports demonstrating a relationship between outcomes and renal dysfunction in patients hospitalized for AHF. As the vast majority of our study population suffered from non‐ischaemic AHF, our results suggest that these effects apply also in this subpopulation of AHF.

Interestingly, 97% of AHF patients presented with impaired renal function as defined by eGFR of >90 mL/min/1.73 m² at baseline. In detail, the study cohort investigated here consists of AHF patients with mostly moderate kidney disease at admission (60%, Table 1). When compared to previously published data, the overall severity of renal impairment seems in line with most published CHF cohorts, and distribution of age and gender as well as all‐cause mortality rates may be considered comparable.4, 7, 10, 21 Thus, we believe our study cohort reflects a rather typical cohort of patients hospitalized for AHF. Nevertheless, as mentioned before, our cohort may be distinguished from previous investigations as in our investigation, mostly patients with non‐ischaemic AHF were included.

Epidemiological data including data from larger cohorts of patients with CHF demonstrate that CKD may be present in up to 50% of affected patients.7, 17, 22, 33, 34, 35 In addition, it seems pivotal to note that estimation of acute‐on‐chronic renal dysfunction in this specific cohort may be regarded especially difficult. Nevertheless, although we aimed to investigate the effect of initial eGFR on long‐term outcomes, we cannot fully elucidate the degree of CKD in the cohort under investigation. On the other hand, we believe that is important to investigate the impact of renal dysfunction on the cohort under investigation. In our univariate and multivariate outcome models, we identified eGFR as one of the most important risk factors for death in this cohort of acute care patients. Additional independent predictors of death in our multivariate outcome model were increased age and NYHA class at admission as well as decreased haemoglobin levels, and total cholesterol levels. The latter association was previously referred to as ‘cholesterol paradox’.32

The authors are well aware of respective limitations that are primarily driven by study design. First, as discussed before, we cannot comment on the degree of underlying CKD in our cohort because of reasons of missing pre‐clinical serum creatinine data. In addition, assessment of course of both serum creatinine levels and urinary output was beyond the scope of our analysis. We are therefore unable to investigate our cohort in regard to the recently established AKIN/RIFLE criteria and thus focus on initial eGFR as read out. Second, our analysis is of retrospective single‐center nature, and respective limitations apply. Nevertheless, we focused on long‐term follow‐up. With a mean follow‐up period of 2.9 ± 2.1 years, we believe that a rather long observational period may be regarded a strength of our analysis. Third, in the clinical setting of acute decompensated heart failure, serum creatinine levels and eGFR may only partially reflect renal dysfunction. This seems the fact as respective indices should not be considered in a steady state. However, this well‐known effect is a major challenge in regard to all respective investigations both of retrospective and prospective nature. Fourth, we used eGFR rather than measured GFR for assessment of renal dysfunction. Nevertheless, due to the specific kinetics of the underlying condition, GFR estimating equations may be considered even more accurate than 24 h urine creatinine clearance studies. Nevertheless, eGFR equations theoretically assume stable kidney function, and respective data must therefore be interpreted with caution. In addition, a potential best eGFR equation is still under debate. Importantly, we would like to highlight the fact that here, GFR categories are used in equivalence to well‐known severity stages as proposed for CKD and should not be misinterpreted as CKD stages. Fifth, we focused on all‐cause mortality as read‐out because cardiovascular events, re‐hospitalization, or renal‐related outcome measures were not recorded. Sixth, a minority of patients had some previous evidence for ischaemic heart disease or history of AMI. Thus, a minor effect of ischaemia‐induced AHF may influence our data. Seventh, due to the specifics of the underlying condition (i.e. CHF), serum creatinine levels and thus eGFR may be influenced by additional factors such as malnutrition or muscle wasting,36 limb amputation, liver cirrhosis, or others. Although we are convinced that our study population consists in a typical cohort of AHF patients, we cannot exclude an effect of some degree on respective data.

Conclusions

In the present retrospective investigation in a large cohort of patients with mostly non‐ischaemic AHF, we observed that eGFR independently predicts survival. It appears that patients with eGFR < 30 mL/min/1.73 m² have poorest survival rates. Our results add to mounting data indicating that the degree of renal dysfunction is of pivotal importance in a population of patients with (mostly non‐ischaemic) AHF.

Conflict of interest

All authors declare that they have no conflict of interest.

Schefold, J. C. , Lainscak, M. , Hodoscek, L. M. , Blöchlinger, S. , Doehner, W. , and von Haehling, S. (2015) Single baseline serum creatinine measurements predict mortality in critically ill patients hospitalized for acute heart failure. ESC Heart Failure, 2: 122–128. doi: 10.1002/ehf2.12058.