Summary
Background
Acute heart failure (HF) carries high hospital mortality rates in older patients; a multimarker strategy may help identify patients at high risk.
Aims
To investigate prospectively the prognostic relevance of serum albumin and serum total cholesterol (TC) in older patients with severe, acute HF.
Methods
Usual prognostic variables were collected on admission in 207 consecutive patients aged > 70 years with severe, acute HF. Serum albumin and serum TC were obtained soon after clinical improvement.
Results
Hospital mortality rate was 19%. Patients who died were similar to patients who survived in terms of age, sex, heart rate, serum haemoglobin and left ventricular ejection fraction. Patients who died had higher concentrations of B-type natriuretic peptide (BNP), blood urea nitrogen, serum creatinine, C-reactive protein and serum troponin I, lower systolic blood pressure, and lower concentrations of serum albumin and serum TC than patients who survived ( P < 0.01 for all). Serum albumin was the best independent predictor of hospital death (odds ratio 0.82 [0.74–0.90], P < 0.001), with blood urea nitrogen ( P = 0.02) and log (BNP) ( P = 0.02). A simple risk score based on serum albumin (< 3 g/dL; 2 points), BNP (> 840 pg/mL; 1 point) and blood urea nitrogen (> 15.3 mmol/L; 1 point) discriminated patients without (score 0 to 1, hospital death 4%) from patients with (score 2 to 4, hospital death 35%, P < 0.001) a high risk of death.
Conclusion
Hypoalbuminaemia offers powerful additional prognostic information to usual prognostic variables in older patients with severe, acute HF, and deserves further attention in multimarker strategies.
Résumé
Contexte
Le taux de mortalité hospitalière est particulièrement élevé dans l’insuffisance cardiaque aiguë (ICA) du sujet âgé, et une stratégie basée sur plusieurs marqueurs biologiques pourrait contribuer à l’identification des sujets les plus à risque.
Objectif
Évaluer la valeur pronostique de l’albumine sérique et du cholestérol total chez le sujet âgé en ICA sévère.
Méthodes
Les paramètres biologiques usuels ont été dosés à l’admission chez 207 patients consécutifs de plus de 70 ans. L’albumine sérique et le cholestérol total ont été dosés lors de la stabilisation clinique.
Résultats
La mortalité hospitalière était de 19 %. Les patients décédés et survivants étaient similaires en terme d’âge, de sexe, de fréquence cardiaque, d’hémoglobine sérique et de fraction d’éjection. Les patients décédés avaient des concentration en peptide natriurétique de type B (BNP), urée plasmatique, créatinine sanguine, protéine C réactive et troponine I plus élevées, d’albumine sérique et de cholestérol total plus basses ( p < 0,01). L’albumine sérique était le facteur prédictif indépendant de mortalité le plus puissant (OR 0,82 [0,74–0,90], p < 0,001), avec l’urée ( p = 0,02) et le BNP ( p = 0,02). Un score de risque basé sur l’albumine sérique (< 3 g/dL ; 2 points), le BNP (> 840 pg/mL ; 1 point) et l’urée (> 15,3 mmol/L ; 1 point) séparait les patients à bas risque (score 0 à 1, mortalité de 4 %) des patients à haut risque (score 2 à 4, mortalité de 35 %).
Conclusion
L’albumine sérique offre une information pronostique pertinente chez le sujet âgé hospitalisé en ICA sévère, et mérite d’être intégré plus largement dans les algorithmes pronostiques.
Background
Acute HF syndromes are one of the most frequent causes of admission to community hospitals in developed countries. This epidemic medical condition primarily affects older patients and carries high hospital mortality rates . Many elderly patients hospitalized with HF do not benefit from cardiologist care , and may therefore experience poorer outcomes . Severe symptoms and high concentrations of BNP are well-established landmarks of severe, acute HF at hospital admission . A targeted multimarker strategy may, however, be helpful in improving the identification of older patients at high risk of hospital death, who are candidates for cardiologist care and tailored unloading therapy.
Serum albumin and serum TC are two simple and inexpensive markers of malnutrition-inflammation syndrome, which have been recently proposed for the identification of patients with acute HF at risk of adverse outcome ; however, it is unknown whether these two biomarkers offer relevant prognostic information incremental to usual prognosticators in older patients with severe, acute HF. The present study addressed the prognostic relevance of serum albumin and serum TC in the prediction of hospital death in older patients identified as having severe, acute HF at admission to a French community hospital.
Methods
Study population
This prospective, observational study included consecutive elderly patients aged > 70 years, who were admitted to the Department of Cardiology of our institution from January 2009 to February 2011 with the primary diagnosis of acute HF. All the patients had acute dyspnoea at rest, clinical and radiographic signs of pulmonary oedema, which responded favourably to intravenous furosemide therapy, and abnormal concentrations of BNP. Comprehensive Doppler echocardiography was performed in all patients within 24 h of admission. Because low and intermediate BNP concentrations indicate good prognosis at the time of admission, all patients with BNP concentrations < 300 pg/mL associated with a LV ejection fraction > 50%, as well as all patients with BNP concentrations < 600 pg/mL associated with a LV ejection fraction < 50%, were not included in the study . Other exclusion criteria were acute coronary syndromes, acute myocardial infarction with ST-segment elevation, severe left-sided valve disease, neoplasia and liver cirrhosis. Finally, 207 consecutive patients with the final diagnosis of severe, acute HF were included in the study. All patients were managed by two HF specialists during their hospital stay. Patients were discharged from the hospital provided that they were clinically stable under oral therapy for at least 2 days, without residual pulmonary and peripheral fluid overload. All patients were included after informed consent was obtained.
Baseline patient data
Baseline clinical data, serum sodium concentration (Dimension RXL system, Siemens, Munich, Germany; normal range 137 to 145 mmol/L), BNP concentration (Architect I1000 system, Abbott Diagnostics, Abbott Park, IL, USA; 10 to 5000 pg/mL), serum creatinine concentration (Dimension RXL system; normal range 71–115 μmol/L for men and 53–88 μmol/L for women), blood urea nitrogen concentration (Dimension RXL system; normal range 2.5 to 6.4 mmol/L), serum haemoglobin concentration (XE2100 system, Sysmex, Kobe, Japan; normal range 13 to 17 g/dL in men and 12 to 16 g/dL in women) and serum troponin I concentration (Dimension RXL system; normal value < 0.14 pg/mL) were collected on admission. Serum albumin concentration (Dimension RXL system; normal range 3.5 to 5 g/dL), serum TC concentration (Dimension RXL system; normal range 135 to 250 mg/dL) and C-reactive protein concentration (Dimension RXL system; normal value < 3 mg/L) were measured in the same blood sample within 3 days of admission, after clinical stabilization. Clinically relevant hypoalbuminaemia was defined by a value of < 3 g/dL . Creatinine clearance (mL/minute) was calculated according to the MDRD formula, which integrates serum creatinine concentration, age and sex but not weight.
The Boston score for congestive HF was calculated by a cardiologist at admission. This score is based on symptoms (0–4 points), physical examination (0–4) and chest X-ray (0–4). As all of the patients presented with dyspnoea at rest and radiographic pulmonary oedema, their scores ranged from 7 to 12. All patients underwent comprehensive Doppler echocardiography at the bedside within 24 h of admission. LV ejection fraction was measured by Simpson’s method; the combination of visual estimate and endocardial fractional shortening was used in patients with poor echogenicity. An LV ejection fraction ≥ 50% was used to define normal LV systolic function. Diastolic function was assessed in patients in sinus rhythm by the analysis of mitral filling and tissue Doppler imaging at the mitral annulus. Diastolic dysfunction was classified as abnormal relaxation of mitral filling, pseudonormal mitral filling and restrictive mitral filling.
Statistical analysis
Descriptive data with normal distribution are given as mean ± standard deviation. Descriptive data without normal distribution are given as median [interquartile range]. Intergroup comparison used the analysis of variance test, the Kruskal-Wallis test, the Chi 2 test and Fisher’s exact test with bilateral formulation as appropriate. The prespecified endpoint was hospital death. The analysis of variables associated with length of hospital stay in patients discharged alive was designed retrospectively, at the end of the study. Multiple regression analysis was used to identify variables that were associated with serum albumin concentration and serum TC concentration, and the correlation coefficient ( r ) was provided for variables that achieved statistically significant results. Logistic regression analysis was used to determine predictors of hospital death, using hospital death as the dependent variable and age, sex, heart rate, systolic blood pressure, LV ejection fraction, log (BNP), serum troponin I, serum sodium, serum creatinine, creatinine clearance, blood urea nitrogen, serum albumin, serum TC and C-reactive protein as independent variables. Multivariable stepwise logistic regression analysis was used to determine variables that independently predicted hospital death; a variable was entered into the model if its associated significance level was < 0.05 and was removed if its associated significance level was > 0.1. The area under the ROC curve in the prediction of hospital death was given with its 95% confidence interval. Variables that were independently associated with hospital death were used to create a risk score by assigning a specific number of points proportional to regression coefficients. Multiple regression analysis was used to identify predictors of hospital length of stay in patients discharged alive. A P value < 0.05 was considered statistically significant. Medcalc ® software, version 11.1.0.0 (Medcalc ® Software, Mariakerke, Belgium) was used for the purpose of statistical analysis.
Results
Baseline characteristics of the study population
The mean age of the study population was 86 ± 7 (range 71–102) years. Median LV ejection fraction was 55% [40–65] and 144 patients (70%) had normal LV systolic function. Mean serum albumin concentration was 3.05 ± 0.5 (range 1.47–4.52) g/dL and median serum TC was 170 [140–200] (range 70–330) mg/dL. Ninety-seven patients (47%) had clinically relevant hypoalbuminaemia and 45 patients (22%) had hypocholesterolaemia. Among patients in sinus rhythm, 42% had a restrictive mitral filling pattern, 33% had pseudonormal mitral filling and 25% had abnormal relaxation mitral filling. Forty patients died during their hospital stay, soon after a short period of clinical improvement: 32 with refractory congestive HF and eight with sudden cardiac death. Twenty-seven patients (67%) who died had normal LV systolic function. None of the patients died from causes of extracardiac origin during their hospital stay. The length of hospital stay was significantly shorter in patients who died than in those discharged alive ( P < 0.001). Baseline clinical characteristics of the patients are displayed in Table 1 according to outcome.
| Variable | Survivors ( n = 167) | Non-survivors ( n = 40) | p |
|---|---|---|---|
| Age (years) | 86 ± 7 | 87 ± 7 | 0.4 |
| Women | 112 (67) | 28 (70) | 0.8 |
| Presenting characteristics | |||
| Boston score | 10.1 ± 1.5 | 10 ± 1.7 | 0.8 |
| Systolic blood pressure (mmHg) | 152 ± 34 | 132 ± 36 | < 0.01 |
| Sinus rhythm | 93 (56) | 17 (43) | 0.2 |
| Heart rate (beats per minute) | 83 [69–100] | 90 [72–110] | 0.09 |
| LV ejection fraction (%) | 55 [40–65] | 55 [37–65] | 0.8 |
| BNP concentration (pg/mL) | 919 [660–166] | 1194 [757–2790] | < 0.01 |
| Serum sodium (mmol/L) | 138 [136–141] | 140 [134–144] | 0.2 |
| Serum creatinine (μmol/L) | 102 [84–134] | 134 [89–203] | 0.01 |
| Creatinine clearance (mL/minute) | 50 [37–63] | 35 [25–56] | < 0.01 |
| Blood urea nitrogen (mmol/L) | 9.2 [7–12.6] | 14.1 [9.6–23.5] | < 0.001 |
| Serum haemoglobin (g/dL) | 12.1 ± 1.8 | 12 ± 1.9 | 0.6 |
| Positive serum troponin I | 26 (15) | 15 (38) | < 0.01 |
| C-reactive protein (mg/L) | 38 [14–85] | 143 [41–208] | < 0.001 |
| Serum albumin concentration (g/dL) | 3.1 ± 0.5 | 2.6 ± 0.5 | < 0.001 |
| Serum total cholesterol concentration (mg/dL) | 174 [143–204] | 146 [119–172] | < 0.001 |
| Medical history | |||
| Congestive heart failure | 70 (42) | 14 (35) | 0.5 |
| Hypertension | 111 (66) | 22 (55) | 0.2 |
| Coronary artery disease | 53 (32) | 16 (40) | 0.4 |
| Stroke | 31 (18) | 5 (12) | 0.5 |
| Diabetes mellitus | 35 (21) | 13 (32) | 0.1 |
| Pulmonary disease | 28 (17) | 6 (15) | 0.9 |
| Dementia | 39 (23) | 14 (35) | 0.2 |
| Institutionalized status | 80 (48) | 21 (52) | 0.7 |
| Chronic medications | |||
| Diuretics | 93 (56) | 24 (60) | 0.7 |
| ACE-I or ARB | 75 (45) | 17 (42) | 0.9 |
| Beta-blocker | 65 (39) | 17 (42) | 0.8 |
| Statin | 23 (14) | 6 (15) | 0.9 |
| Length of hospital stay (days) | 10 [7–13] | 6 [4–9] | < 0.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
