Professional Documents
Culture Documents
research-article2017
ACC0010.1177/2048872617700874European Heart Journal: Acute Cardiovascular CareJobs et al.
Abstract
Background: The prognostic impact of pneumonia and signs of systemic inflammation in patients with acute
decompensated heart failure (ADHF) has not been fully elucidated yet. The aim of the present study was thus to
investigate the association of pneumonia and the inflammation surrogate C-reactive protein with all-cause mortality in
patients admitted for ADHF.
Methods: We analysed data of 1939 patients admitted for ADHF. Patients were dichotomised according to the
presence or absence of pneumonia. The primary endpoint of all-cause mortality was determined by death registry
linkage.
Results: In total, 412 (21.2%) patients had concomitant pneumonia. Median C-reactive protein levels were higher in
patients with compared to patients without pneumonia (24.9 versus 9.8 mg/l, respectively; P<0.001). All-cause mortality
was significantly higher in patients with pneumonia (P<0.001). In adjusted Cox regression models, pneumonia as well
as C-reactive protein were independently associated with in-hospital mortality. Only C-reactive protein remained as
independent predictor for long-term mortality.
Conclusion: Pneumonia is relatively common in ADHF and a predictor for in-hospital mortality. However, inflammation
in general seems to be more important than pneumonia itself for long-term prognosis. Compared to community-acquired
pneumonia studies, C-reactive protein levels were rather low and therefore pneumonia might be over-diagnosed in
ADHF patients.
Keywords
Acute decompensated heart failure, pneumonia, inflammation, C-reactive protein, prognosis
Date received: 3 November 2016; accepted: 2 March 2017; Final Disposition Set: 2 March 2017
Introduction
Heart failure is the leading cause of hospital admission
1University Heart Center Lübeck, Department of Cardiology, Angiology
among adults older than 65 years in western countries.1,2
and Intensive Care Medicine, Germany
The risk of death is considerably increased over a pro- 2German Center for Cardiovascular Research (DZHK), Partner Site
or diet).4–6 In large registries, respiratory infection was one present, or definitely present according to reports by board-
of the most prevalent factors precipitating ADHF occur- certified radiologists. Possibly and definitely present were
ring in 15% to 29% of cases4,7,8 and was independently combined to present for regression analysis. Pneumonia
associated with inhospital mortality in some analyses.4,7 was defined based on the documentation of ICD-10 codes
Diagnosing pneumonia is often challenging in patients J15.* and J18.* and the presence of infiltrates on admission
admitted for ADHF as dyspnoea and rales are cardinal chest radiograph. Worsening renal function was defined as
symptoms for both heart failure and pneumonia.9,10 an increase in creatinine concentration of 1.5 or greater
Clinically diagnosing pneumonia becomes even more chal- times from baseline or 26.5 µmol/l or more within 48 hours,
lenging in elderly patients because respiratory and non- as recently suggested.16
respiratory symptoms are less often reported in this subset
of patients.11 Based on current guidelines, the presence of
Mortality data
infiltrates demonstrated by chest radiograph or another
lung imaging modality is mandatory for the diagnosis of The primary endpoint of all-cause mortality was prospec-
Table 2. Predictors of pneumonia in logistic regression analysis before and after stepwise backward selection.
Pneumonia
Univariable Multivariable
Due to missing data in >10% of all patients the following variables were not included in logistic regression analyses: ejection fraction, mitral
regurgitation, tricuspid regurgitation, and high-sensitive troponin T.
CI: confidence interval; COPD: chronic obstructive pulmonary disease; eGFR: estimated glomerular filtration rate; MDRD: modification of diet in
renal disease study equation; OR: odds ratio.
adjustment for demographics, comorbidities as well as Multivariable logistic regression analysis identified
laboratory and radiological findings available on admis- COPD, estimated glomerular filtration rate (eGFR), and
sion presented in Table 1. pulmonary congestion as independent predictors for pneu-
All statistical tests were two-sided and considered sig- monia (Table 2).
nificant if P<0.05. All statistical analyses were done using
R software (www.r-project.org; version 3.1.3).
Clinical outcome
Median follow-up was 16 (IQR 5–35) months. In-hospital
Results mortality was 4.0% (n=78) and 753 patients (38.8%) died
within 3 years. Pneumonia was a marker of all-cause mor-
Patient characteristics tality (hazard ratio (HR) 1.34, 95% confidence interval (CI)
In total, 2180 unique patients diagnosed with ADHF hospi- 1.13–1.58; P<0.001; Figure 1a).
talised for three or more days were identified. Of these, 1939 Infiltrates were also associated with all-cause mortality
patients (88.9%) had a chest radiograph within 24 hours after (HR 1.20, 95% CI 1.04–1.40; P=0.02). This association
admission and were included in the final analysis. was present for infiltrates reported as possibly present (HR
The cohort was dichotomised into 1527 (78.8%) patients 1.31, 95% CI 1.08–1.60; P=0.007) as well as definitely pre-
without and 412 patients (21.2%) with pneumonia. Baseline sent (HR 1.30, 95% CI 1.10–1.51; P=0.002, Figure 1b).
characteristics are presented in Table 1. Patients with pneu- Considering only patients without pneumonia, infiltrates
monia were older and had a higher prevalence of comorbidi- were not associated with all-cause mortality (HR 1.17, 95%
ties such as diabetes mellitus or chronic obstructive CI 0.97–1.42; P=0.10).
pulmonary disease (COPD). Moreover, their chest radio- Elevated CRP levels above the median were also linked to
graphs were less often performed biplane in the upright all-cause mortality (HR 1.59, 95% CI 1.37–1.84; P<0.001).
position and more often showed pulmonary congestion. In a stratified analysis, CRP was only associated with adverse
However, radiologists reported infiltrates in 569 (37.3%) clinical outcome in patients without pneumonia (pneumonia:
patients without ICD-10 coded pneumonia as determined by HR 1.00, 95% CI 0.72–1.39; P=0.99; no pneumonia: HR
treating physicians. Baseline CRP did not differ between 1.71, 95% CI 1.44–2.02; P<0.001, Figure 1c).
these patients and patients without pneumonia and without Predicted hazard ratios for all-cause mortality did not
infiltrates (9.6, IQR 4.1–20.2 versus 9.9, IQR 3.8–24.5 mg/l; increase linearly with increasing baseline CRP in a
P=0.26). The median baseline CRP was 12.1 (IQR 4.6–28.7) restricted cubic spline model (Figure 2a) in contrast to the
mg/l in the total population and was higher in patients with logarithm of baseline CRP (Figure 2b), which was there-
pneumonia compared to those without (24.9 versus 9.8 fore used for regression analysis. In an adjusted logistic
mg/l, respectively; P<0.001, Table 1). Considering patients regression model, pneumonia and the logarithm of base-
with pneumonia, 29% and 11% had a baseline CRP above line CRP but not demonstrable infiltrates on baseline
50 mg/l and 100 mg/l, respectively. Echocardiographic chest radiograph were independently associated with in-
characteristics did not differ between patients with and with- hospital mortality (Table 3). The logarithm of baseline
out concomitant pneumonia. CRP was identified as the only significant predictor for
366 European Heart Journal: Acute Cardiovascular Care 7(4)
Figure 1. Kaplan–Meier analysis of all-cause mortality stratified for (a) diagnosis of pneumonia (no versus yes); (b) confidence of
infiltrate diagnosis (no infiltrate, possible infiltrate, or definite infiltrate); and (c) diagnosis of pneumonia (no versus yes) and baseline
C-reactive protein (below or equal to/above median).
long-term mortality in an adjusted Cox regression model independently associated with in-hospital mortality but not
(Table 3). with long-term mortality.
Pneumonia occurred in 21.2% patients of our ADHF
cohort, which is in the range of reported frequencies in pre-
Discussion viously published analyses of large registries investigating
This retrospective cohort study demonstrates that pneumo- precipitating factors for ADHF (15.3–8.5% in the Organized
nia is common in patients admitted for ADHF, occurs more Program to Initiate Lifesaving Treatment in Hospitalized
often in patients with pronounced comorbidities, and is Patients With Heart Failure (OPTIMIZE-HF), 28.2% in Get
Jobs et al. 367
Table 3. Odds and hazard ratios for in-hospital and long-term mortality, respectively, for pneumonia, log CRP, and infiltrates in
adjusted regression models.
OR/HR (95% CI) P value OR/HR (95% CI) P value OR/HR (95% CI) P value
In-hospital mortality 1.91 (1.16–3.09) 0.009 1.66 (1.37–2.02) <0.001 1.01 (0.62–1.63) 0.980
Long-term mortality 1.15 (0.97–1.36) 0.117 1.18 (1.12–1.25) <0.001 1.11 (0.95–1.29) 0.184
CI: confidence interval; COPD: chronic obstructive pulmonary disease; CRP: C-reactive protein; eGFR: estimated glomerular filtration rate; HR:
hazard ratio; MDRD: modification of diet in renal disease study equation.
With The Guidelines-HF (GWTG-HF), and 29% in the for community-acquired pneumonia,19,20 which underlines
Spanish National Registry on Heart Failure (RICA)).4,7,8 the plausibility of the current results.
However, if pneumonia is truly a precipitating factor for Pneumonia as a precipitating factor was independently
ADHF or rather a superimposed complication cannot be associated with in-hospital mortality but not with mid-
clearly determined. term mortality in two large registries.4,7 Our analysis of
Patients with pneumonia more often had COPD and pul- concomitant pneumonia in ADHF is in line with this
monary congestion on baseline chest radiograph. Together observation, as pneumonia was independently associated
with eGFR, both factors were identified as independent with in-hospital mortality but not with long-term mortal-
predictors for pneumonia in a multivariable logistic regres- ity after adjusting for demographics, comorbidities, labo-
sion model. Interestingly, early experimental data indicated ratory and radiographic characteristics.
that pulmonary congestion may facilitate the growth of Pneumonia is commonly diagnosed based on clinical
bacteria commonly seen in pneumonia (i.e. Streptococcus features suggestive of acute lower respiratory tract infec-
pneumonia and Staphylococcus aureus).17,18 Moreover, tion and demonstrable infiltrates on chest radiograph.12–14
COPD and chronic kidney disease are known risk factors In addition, the diagnosis is often aided by elevated CRP
368 European Heart Journal: Acute Cardiovascular Care 7(4)
levels. One major finding of our analysis is that infiltrates Hospitalized with Acute Decompensated Heart Failure and
were also reported in 37% of patients without pneumonia Volume Overload to Assess Treatment Effect on Congestion
and that baseline CRP did not differ in these patients com- and Renal Function (PROTECT) trial despite the exclusion
pared to patients without infiltrates. Unfortunately, clinical of patients with antibiotic treatment.28 Joffe et al.29 com-
signs and symptoms as well as antibiotic treatment could pared admission CRP levels between patients admitted for
not be assessed retrospectively. In summary, the diagnosis ADHF to patients admitted for community-acquired pneu-
of pneumonia and the decision for or against an antibiotic monia. While the CRP level in their ADHF population was
therapy is challenging in ADHF despite available chest radi- similar to ours (13.5±13.5 mg/l), it was considerably higher
ograph findings. Interpretation of chest radiographs is well in their pneumonia population (127±84 mg/l).29 Several
known to have technical limitations and to be prone to inter- other studies investigating CRP for aiding community-
observer and intraobserver variability.21,22 Assessing pulmo- acquired pneumonia diagnosis found that most patients had
nary infiltration by means of a chest radiograph may be CRP levels above 50 mg/l or 100 mg/l on admission.
hampered in ADHF due to pulmonary congestion. In line Moreover, such values were highly specific for the diagno-
independently associated with in-hospital mortality but 11. Metlay JP, Schulz R, Li YH, et al. Influence of age on symp-
not with long-term mortality. With respect to prognosis, toms at presentation in patients with community-acquired
baseline CRP seems to be more important than pneumo- pneumonia. Arch Intern Med 1997; 157: 1453–1459.
nia. This is probably due to the fact that CRP covers other 12. Ewig S, Hoffken G, Kern WV, et al. [Management of adult
community-acquired pneumonia and prevention – update
types of infections and also the severity of inflammation
2016]. Pneumologie 2016; 70: 151–200.
caused by ADHF itself. CRP levels are rather low com-
13. Mandell LA, Wunderink RG, Anzueto A, et al. Infectious
pared to CRP levels reported from community-acquired Diseases Society of America/American Thoracic Society
pneumonia populations. We therefore speculated that consensus guidelines on the management of community-
pneumonia is over-diagnosed in the setting of ADHF. acquired pneumonia in adults. Clin Infect Dis 2007; 44
(Suppl 2): S27–S72.
Conflict of interest 14. Watkins RR and Lemonovich TL. Diagnosis and manage-
The authors declare that there is no conflict of interest. ment of community-acquired pneumonia in adults. Am Fam
Physician 2011; 83: 1299–1306.
identifies high risk acute heart failure patients. Int J 31. Hansson LO, Hedlund JU and Ortqvist AB. Sequential
Cardiol 2016; 204: 164–171. changes of inflammatory and nutritional markers in patients
29. Joffe E, Justo D, Mashav N, et al. C-reactive protein to dis- with community-acquired pneumonia. Scand J Clin Lab
tinguish pneumonia from acute decompensated heart failure. Invest 1997; 57: 111–118.
Clin Biochem 2009; 42: 1628–1634. 32. Castro-Guardiola A, Armengou-Arxe A, Viejo-Rodriguez A,
3 0. Smith RP, Lipworth BJ, Cree IA, et al. C-reactive protein. et al. Differential diagnosis between community-acquired
A clinical marker in community-acquired pneumonia. Chest pneumonia and non-pneumonia diseases of the chest in the
1995; 108: 1288–1291. emergency ward. Eur J Intern Med 2000; 11: 334–339.