Journal of Cardiology 64 (2014) 441–449

Contents lists available at ScienceDirect

Journal of Cardiology
journal homepage: www.elsevier.com/locate/jjcc

Original article

New scoring system (APACHE-HF) for predicting adverse outcomes in

patients with acute heart failure: Evaluation of the APACHE II and
Modified APACHE II scoring systems
Hirotake Okazaki (MD) a , Akihiro Shirakabe (MD, PhD) a,∗ , Noritake Hata (MD, PhD) a ,
Masanori Yamamoto (MD, PhD) a , Nobuaki Kobayashi (MD, PhD) a ,
Takuro Shinada (MD, PhD) a , Kazunori Tomita (MD) a , Masafumi Tsurumi (MD) a ,
Masato Matsushita (MD) a , Yoshiya Yamamoto (MD) a , Shinya Yokoyama (MD, PhD) a ,
Kuniya Asai (MD, PhD) b , Wataru Shimizu (MD, PhD, FJCC) b
a
Division of Intensive Care Unit, Chiba Hokusoh Hospital, Nippon Medical School, Chiba, Japan
b
Department of Cardiovascular Medicine, Nippon Medical School, Tokyo, Japan

a r t i c l e i n f o a b s t r a c t

Article history: Background: No scoring system for assessing acute heart failure (AHF) has been reported.
Received 2 December 2013 Methods and results: Data for 824 AHF patients were analyzed. The subjects were divided into an alive
Received in revised form 4 February 2014 (n = 750) and a dead group (n = 74). We constructed a predictive scoring system based on eight significant
Accepted 19 February 2014
APACHE II factors in the alive group [mean arterial pressure (MAP), pulse, sodium, potassium, hematocrit,
Available online 29 April 2014
creatinine, age, and Glasgow Coma Scale (GCS); giving each one point], defined as the APACHE-HF score.
The patients were assigned to five groups by the APACHE-HF score [Group 1: point 0 (n = 70), Group 2:
Keywords:
points 1 and 2 (n = 343), Group 3: points 3 and 4 (n = 294), Group 4: points 5 and 6 (n = 106), and Group
Acute heart failure syndrome
Mortality
5: points 7 and 8 (n = 11)]. A higher optimal balance was observed in the APACHE-HF between sensitivity
Prognosis and specificity [87.8%, 63.9%; area under the curve (AUC) = 0.779] at 2.5 points than in the APACHE II
Scoring (47.3%, 67.3%; AUC = 0.558) at 17.5 points. The multivariate Cox regression model identified belonging
to Group 5 [hazard ratio (HR): 7.764, 95% confidence interval (CI) 1.586–38.009], Group 4 (HR: 6.903,
95%CI 1.940–24.568) or Group 3 (HR: 5.335, 95%CI 1.582–17.994) to be an independent predictor of
3-year mortality. The Kaplan–Meier curves revealed a poorer prognosis, including all-cause death and
HF events (death, readmission-HF), in Group 5 and Group 4 than in the other groups, in Group 3 than in
Group 2 or Group 1, and in Group 2 than in Group 1.
Conclusions: The new scoring system including MAP, pulse, sodium, potassium, hematocrit, creatinine,
age, and GCS (APACHE-HF) can be used to predict adverse outcomes of AHF.
© 2014 Japanese College of Cardiology. Published by Elsevier Ltd. All rights reserved.

Introduction The APACHE II system consists of three parts, including the

The Acute Physiology and Chronic Health Evaluation (APACHE)
scoring system was first established in 1981 to predict the progno-
sis in patients receiving intensive care (Fig. 1A) [1]. Subsequently,
the APACHE II, III, and IV systems were published over the past
20 years [2–4].
∗ Corresponding author at: Division of Intensive Care Unit, Chiba Hokusoh Hospi-
tal, Nippon Medical School, 1715 Kamagari, Inzai, Chiba 270-1694, Japan.
Tel.: +81 476 99 1111; fax: +81 476 99 1911.
E-mail address: s6042@nms.ac.jp (A. Shirakabe).
acute physiology score, chronic health points, and age points.
The total number of points for the three parts is calculated as the
APACHE II score. This score has been reported to be predictive
of adverse outcomes in patients requiring intensive care, such
as those with respiratory disease, severe pancreatitis, or severe
sepsis [5–9]. However, this scoring system involves many factors,
as described above; therefore, it cannot be applied easily, and
clinicians hesitate to use it in every patient.
In previous observational studies, various predictive factors for
detecting adverse outcomes in acute heart failure (AHF) patients
have been identified, including age [10], anemia [11], renal insuf-
ficiency [12,13], poor liver function [14], high uric acid [15], high
lactate [16], low cholesterol [12,16], elevated blood glucose [17],

http://dx.doi.org/10.1016/j.jjcc.2014.03.002
0914-5087/© 2014 Japanese College of Cardiology. Published by Elsevier Ltd. All rights reserved.
442 H. Okazaki et al. / Journal of Cardiology 64 (2014) 441–449

Fig. 1. Definition of each scoring system. (A) The APACHE II scoring system was defined in this study. (B) The Modified APACHE II scoring system was constructed based on the
significant APACHE II factors in the alive group [mean blood pressure (BP), sodium, potassium, creatinine, age, and Glasgow Coma Scale (GCS)] and was given points based on
the APACHE II system. (C) The APACHE-HF scoring system was constructed based on the significant APACHE II factors in the alive group (mean BP, pulse, sodium, potassium,
creatinine, hematocrit, age, and GCS) and was given one point for each cut-off value. The cut-off value for each factor was defined by the receiver-operating characteristic
(ROC) curve as follows: mean BP [91.5 mmHg, area under the ROC curve (AUC) = 0.678, p < 0.001), pulse (110.5 beats/min, AUC = 0.594, p = 0.008), sodium (137.5 mmol/L,
AUC = 0.613, p = 0.001), potassium (4.85 mmol/L, AUC = 0.601, p = 0.004), hematocrit (36.95 mg/dL, AUC = 0.617, p = 0.001), creatinine (1.475 mg/dL, AUC = 0.676, p < 0.001), age
(71.5 years, AUC = 0.572, p = 0.042) and GCS (13.5, AUC = 0.567, p = 0.058)].
 H. Okazaki et al. / Journal of Cardiology 64 (2014) 441–449
hyperkalemia [12], hyponatremia [18], brain-type natriuretic pep-
tide (BNP), and left ventricular ejection fraction. With respect to
patients with AHF, no predictive scoring system has been estab-
lished, and evaluations of the APACHE II system have rarely been
reported. We therefore evaluated the efficacy of the APACHE II and
our newly established scoring system for AHF patients.
Methods
Subjects
Clinical data were collected from 824 patients with AHF who
were admitted to the intensive care unit at Chiba Hokusoh Hospital,
Nippon Medical School between January 2000 and July 2012. AHF
was defined as either new-onset HF or decompensation of chronic
HF with symptoms sufficient to warrant hospitalization [19]. HF
was diagnosed according to the Framingham criteria for a clinical
diagnosis of HF based on the satisfaction of two major criteria or
one major and two minor criteria [20]. All patients had a New York
Heart Association (NYHA) functional class of either Class III or IV.
AHF patients with one of the following criteria were admitted to
the intensive care unit (ICU) by physician’s decision in the present
study: (1) patients who need high projectile oxygen inhalation
(including mechanical support) to treat orthopnea, (2) patients who
need intrope or mechanical support with low blood pressure, (3)
patients who need many types of diuretics to improve the general
or lung edema. Patients with HF caused by acute coronary syn-
drome were excluded from the study. All data were retrospectively
retrieved from hospital medical records.
Procedure
AHF patients were divided into two groups according to in-
hospital mortality: the alive group (n = 750) and the dead group
(n = 74).
We established two new scoring systems for AHF comparing
the two groups. First, we compared the APACHE II score between
the two groups using a univariate analysis. Factors associated with
significantly more points in the alive group were selected to con-
struct the new scoring system. The total score of the significant
factors in the alive group was defined as the Modified APACHE II
score. Second, we compared the data for APACHE II factors between
the two groups using a univariate analysis. We then scored the
total number of factors specific to survival discharge (giving one
point for each factor), defined as the APACHE-HF score. The cut-off
value for each factor to give one point was defined by the receiver-
operating characteristic (ROC) curves for the in-hospital mortality
of each factor.
We determined the scores (APACHE II, Modified APACHE II, and
APACHE-HF) in each patient based on data obtained at admission,
according to a previous report [2], and evaluated the sensitivity,
specificity, and positive and negative predictive value for differen-
tiating the alive and dead groups. ROC curves were calculated to
predict the optimal cut-off values.
Furthermore, the mid-term prognosis was evaluated in terms of
all-cause death and HF events defined as all-cause death or read-
mission due to HF. The patients were clinically followed up in a
routine outpatient clinic. For the patients followed up at other insti-
tutes, the final prognosis was determined via telephone contact.
The patients were assigned to another two groups according to
the cut-off values of the ROC curves for the APACHE II, Modified
APACHE II, and APACHE-HF scores. The survival rates were then
analyzed using Kaplan–Meier curves according to the APACHE II,
Modified APACHE II, and APACHE-HF scores. The prognostic value of
the APACHE-HF score in these groups compared with that observed
443
in the group with lowest point as the referent was assessed using
a Cox regression hazard model.
Statistical analysis
All data were statistically analyzed using the SPSS 20.0J software
program (SPSS Japan Institute, Tokyo, Japan). All numerical data
were expressed as the mean ± standard deviation or median (range
or 25–75% interquartile range) depending on normality. Unpaired
Student’s t-test or the Mann–Whitney U-test was used to compare
the two groups. Normality was assessed using the Shapiro–Wilk W-
test. Comparisons of all proportions were made using a chi-square
analysis. A p-value of less than 0.05 was considered to be statisti-
cally significant. ROC curves were calculated to predict the cut-off
values, and the sensitivity, specificity and area, under the ROC curve
(AUC) were determined. The survival rates were analyzed between
the groups assigned based on the cut-off values of the ROC curves
for each scoring system using Kaplan–Meier curves, and significant
differences were calculated using the log-rank test. A Cox regres-
sion analysis was performed to obtain the hazard ratios (HRs) for
90-day mortality and 90-day HF events. Subsequently, a multivari-
ate analysis was performed using the variables with a p-value of
<0.05 in the univariate analysis to examine their independent asso-
ciations with 90-day mortality and 90-day HF events. A p-value of
less than 0.05 was considered to be statistically significant.
Ethical concerns
The institutional review board at Chiba Hokusoh Hospital, Nip-
pon Medical School approved the study protocol.
Results
Patient characteristics
The relationship between the treatment, including respiratory
support and medications prescribed during the first 5 days, and
in-hospital mortality are shown in Table 1. The patient cohort
included 67.2% male subjects, with a median age of 74 years.
The systolic blood pressure (BP) values were significantly lower,
the number of NYHA class IV patients was significantly higher,
the serum hemoglobin levels were significantly lower, the serum
urinary acid levels were significantly higher, the serum BUN lev-
els were significantly higher, the serum C-reactive protein (CRP)
levels were significantly higher, and the serum BNP levels were
significantly higher in the dead group than in the alive group.
Definition of the Modified APACHE II and APACHE-HF scoring
systems
Regarding the APACHE II score, the following six factors were
significantly different between the alive group and the dead group:
mean BP, sodium, potassium, creatinine, age, and Glasgow Coma
Scale (GCS) (Table 2). We constructed a predictive scoring system
based on the significant APACHE II factors in the alive group (mean
BP, sodium, potassium, creatinine, age, and GCS; giving points
based on the APACHE II system), defined as the Modified APACHE
II score (Fig. 1B).
On the other hand, the following eight factors were sig-
nificantly different between the alive group and dead group:
mean BP, pulse, sodium, potassium, hematocrit, creatinine, age,
and GCS (Table 3). We constructed a predictive scoring sys-
tem based on the significant APACHE II factors in the alive
group (mean BP, pulse, sodium, potassium, creatinine, hemat-
ocrit, age, and GCS; giving one point based on each cut-off value),
defined as the APACHE-HF score (Fig. 1C). The cut-off values for
444 H. Okazaki et al. / Journal of Cardiology 64 (2014) 441–449

Table 1
Relationships between the patient characteristics and in-hospital survival.

Characteristic Total (n = 824) Dead group (n = 74) Alive group (n = 750) p-Value

Age (years) 74 (65–80) 74 (64–80) 76 (70–81) 0.041

Gender (male, %) 554 (67.2%) 44 (59.5%) 510 (68.0%) 0.153
Type (new onset, %) 544 (66.0%) 43 (58.1%) 501 (66.8%) 0.157
Etiology
Ischemic heart disease (yes, %) 554 (67.2%) 44 (59.5%) 510 (68.0%) 1.000
Cardiomyopathy (yes, %) 141 (17.1%) 14 (18.9%) 127 (16.9%) 0.630
Hypertensive heart disease (yes, %) 143 (17.4%) 7 (9.5%) 136 (18.1%) 0.076
Valvular (yes, %) 117 (21.5%) 20 (27.0%) 157 (20.9%) 0.236
Others (yes, %) 22 (2.7%) 2 (2.7%) 20 (2.7%) 1.000
Past medical history
Hypertension (yes, %) 612 (74.3%) 47 (63.5%) 565 (75.3%) 0.036
Diabetes mellitus (yes, %) 341 (41.4%) 32 (43.2%) 309 (41.2%) 0.805
Dyslipidemia (yes, %) 374 (45.4%) 25 (33.8%) 349 (46.5%) 0.038
Vital signs and status
SBP (mmHg) 160 (132–186) 131 (107–168) 162 (138–188) <0.001
SBP > 140 mmHg (yes, %) 559 (67.8%) 31 (41.9%) 528 (70.4%) <0.001
SBP 100–140 mmHg (yes, %) 232 (28.2%) 29 (39.2%) 203 (27.1%) 0.031
SBP < 100 mmHg (yes, %) 66 (8.0%) 13 (17.6%) 53 (7.1%) 0.005
Diastolic blood pressure (mmHg) 90 (70–100) 80 (62–90) 90 (72–100) <0.001
LVEF (%) 35.0 (25–46) 32 (22–45) 35 (24–46) 0.150
LVEF > 40% (%) 295 (36.0%) 23 (31.9%) 272 (36.4%) <0.001
NYHA (IV, %) 667 (80.9%) 66 (89.2%) 601 (80.1%) 0.003
Laboratory data
Total bilirubin (mg/dL) 0.6 (0.4–0.8) 0.7 (0.4–1.0) 0.6 (0.4–0.8) 0.166
Urinary acid (mg/dL) 6.8 (5.5–8.1) 7.7 (5.9–9.6) 6.7 (5.3–8.0) 0.001
BUN (mg/dL) 23.2 (17.9–33.1) 36.4 (21.5–48.7) 22.7 (17.5–31.8) <0.001
Hemoglobin (g/dL) 12.5 (10.7–14.5) 11.3 (10.3–14.2) 12.5 (10.8–14.5) 0.001
CRP (mg/dL) 0.56 (0.19–1.78) 1.44 (0.56–4.58) 0.52 (0.17–1.63) <0.001
BNP (pg/mL) 805 (415–1403) 1363 (927–1787) 753 (399–1312) <0.001
Respiratory support
Endotracheal intubation (yes, %) 209 (25.4%) 36 (48.6%) 173 (23.1%) <0.001
NPPV (yes, %) 310 (37.6%) 24 (32.4%) 286 (38.1%) 0.380
Medication (cases) during the first 5 days
Furosemide (yes, %) 781 (94.8%) 66 (89.2%) 715 (95.3%) 0.047
Nitroglycerin (yes, %) 594 (69.1%) 37 (50.0%) 557 (74.3%) <0.001
Nicorandil (yes, %) 95 (11.5%) 11 (14.9%) 84 (11.2%) 0.341
Carperitide (yes, %) 449 (54.5%) 45 (60.8%) 404 (53.9%) 0.272
Dopamine (yes, %) 238 (28.9.0%) 41 (55.4%) 197 (26.3%) <0.001
Dobutamine (yes, %) 117 (21.5%) 36 (48.6%) 141 (18.8%) <0.001
ACE-I/ARB (yes, %) 333 (40.4%) 17 (22.9%) 316 (42.1%) 0.001
␤-Blocker (yes, %) 196 (23.8%) 16 (21.6%) 180 (24.0%) 0.775
Spironolactone (yes, %) 296 (35.9%) 18 (24.3%) 278 (37.1%) 0.031
Outcome
ICU hospitalization (days) 5 (3–7) 7 (4–17) 5 (3–7) <0.001
Total hospitalization (days) 29 (18–49) 30 (13–82) 29 (18–47) 0.489

SBP, systolic blood pressure; LVEF, left ventricular ejection fraction measured on echocardiography; NYHA, New York Heart Association; BUN, blood urea nitrogen; CRP,
C-reactive protein; BNP, brain natriuretic peptide; NPPV, non-invasive positive pressure ventilation; ACE-I, angiotensin-converting enzyme inhibitor ARB, angiotensin II
receptor blocker; ICU, intensive care unit.
p-Value between the alive group and dead group determined according to unpaired Student’s t-test and Mann–Whitney U-test.

Table 2
Relationships between the APACHE II score and in-hospital mortality.

All (n = 825) Alive group (n = 750) Dead group (n = 75) p-Value

Total APS
Body temperature (◦ C) 0.39 ± 0.57 0.38 ± 0.55 0.51 ± 0.67 0.162
Mean blood pressure (mmHg) 1.51 ± 1.36 1.54 ± 1.36 1.22 ± 1.33 0.036
Pulse (beats/min) 1.53 ± 1.23 1.54 ± 1.22 1.39 ± 1.24 0.227
Respiratory rate (per min) 1.36 ± 1.20 1.36 ± 1.19 1.38 ± 1.27 0.920
A-aDO2 (FiO2 > 0.5) or PaO2 (FiO2 < 0.5) 0.91 ± 1.46 0.90 ± 1.45 1.00 ± 1.52 0.778
pH 1.59 ± 1.60 1.59 ± 1.60 1.58 ± 1.51 0.985
Sodium (mmol/L) 0.05 ± 0.35 0.04 ± 0.31 0.18 ± 0.60 0.001
Potassium (mmol/L) 0.22 ± 0.66 0.19 ± 0.61 0.46 ± 0.91 0.021
Hematocrit (%) 0.55 ± 0.90 0.54 ± 0.90 0.61 ± 0.88 0.511
Creatinine (mg/dL) 0.96 ± 1.35 0.90 ± 1.32 1.61 ± 1.51 <0.001
White blood cell (/m3 ) 0.16 ± 0.44 0.16 ± 0.44 0.14 ± 0.40 0.873
Age points 5.05 ± 1.26 5.01 ± 1.29 5.39 ± 0.90 0.027
Chronic health points 0.53 ± 1.54 0.51 ± 1.52 0.68 ± 1.63 0.386
Glasgow Coma Scale 0.94 ± 2.49 0.88 ± 2.40 1.59 ± 2.90 0.006

APS, acute physiology score.

p-Value between the alive group and dead group determined according to Mann–Whitney U-test.
 H. Okazaki et al. / Journal of Cardiology 64 (2014) 441–449 445

Table 3
Relationships between the APACHE II data and in-hospital mortality.

All (n = 824) Alive group (n = 750) Dead group (n = 74) p-Value

Total APS
Body temperature (◦ C) 36.3 (35.7–36.8) 36.2 (35.7–36.8) 36.4 (35.2–36.9) 0.750
Mean blood pressure (mmHg) 112.0 (94–130) 113 (95–131) 97 (77–114) <0.001
Pulse (beats/min) 114 (95–132) 116 (96–132) 105 (82–125) 0.008
Respiratory rate (per min) 30 (23–35) 30 (24–35) 30 (23–35) 0.411
PaO2 (mmHg) 86.7 (65.5–128.7) 85.8 (65.5–123.8) 96.1 (67.2–147.5) 0.136
pH 7.32 (7.20–7.42) 7.32 (7.19–7.42) 7.21 (7.19–7.40) 0.873
Sodium (mmol/L) 140 (137–142) 140 (138–142) 137 (135–141) 0.001
Potassium (mmol/L) 4.3 (3.9–4.7) 4.2 (3.9–4.7) 4.4 (3.9–5.2) 0.004
Hematocrit (%) 38.0 (33.2–43.7) 38.5 (33.7–44.0) 34.7 (30.7–39.7) 0.001
Creatinine (mg/dL) 1.16 (0.91–1.73) 1.13 (0.89–1.67) 1.63 (1.15–2.55) <0.001
White blood cell (/m3 ) 9775 (7600–12,653) 9795 (7645–12,668) 9625 (6890–12,353) 0.300
Age (years) 74 (65–80) 74 (64–80) 76 (70–81) 0.041
Chronic health points 0 (0–0) 0 (0–0) 0 (0–0) 0.386
Glasgow Coma Scale 15 (15–15) 15 (15–15) 15 (13–15) 0.006

APS, acute physiology score.

p-Value between the alive group and dead group determined according to Mann–Whitney U-test.

Fig. 2. ROC curves for each scoring system. The APACHE II system demonstrated an
optimal balance between sensitivity and specificity (47.3% and 67.3%; AUC = 0.588,
p = 0.012) at 17.5 points (black line), while the Modified APACHE II system demon-
strated an optimal balance between sensitivity and specificity (51.4% and 69.4%, AUC
0.590, p = 0.001) at 9.5 points (blue line) and the APACHE-HF system demonstrated
an optimal balance between sensitivity and specificity (87.8% and 63.9%, AUC 0.779,
p < 0.001) at 2.5 points (red line). ROC, receiver-operating characteristic; AUC, area
under the ROC curve; HF, heart failure.
each factor were defined by ROC curves as follows: mean BP
(91.5 mmHg, AUC = 0.678), pulse (110.5 beats/min, AUC = 0.594),
sodium (137.5 mmol/L, AUC = 0.613), potassium (4.85 mmol/L,
AUC = 0.601), hematocrit (36.95 mg/dL, AUC = 0.617), creatinine
(1.475 mg/dL, AUC = 0.676), age (71.5 years, AUC = 0.572), and GCS
(13.5, AUC = 0.567).
Based on these cut-off values, the patients were divided into low
or high groups for each scoring system. The Kaplan–Meier survival
curves showed that the prognosis, including all-cause death, did
not differ between the patients with an APACHE II score of ≤17 and
those with an APACHE II score of ≥18. Meanwhile, the prognosis
was significantly poorer in the patients with a Modified APACHE II
score of ≥10 than in those with a Modified APACHE II score of ≤9 and
in the patients with an APACHE-HF score of ≥3 than in those with
an APACHE II-HF score of ≤2 (Fig. 3A, C, and E). The Kaplan–Meier
survival curves showed that the prognosis, including HF events, was
significantly poorer in the patients with an APACHE II score of ≥18,
a Modified APACHE II score of ≥10, and an APACHE-HF score of ≥3
than in those with an APACHE II score of ≤17, a Modified APACHE II
of ≤9, and an APACHE-HF score of ≤2, respectively (Fig. 3B, D, and
F).
The patients were assigned to five groups based on the APACHE-
HF score [Group 1: point 0 (n = 70), Group 2: points 1 and 2 (n = 343),
Group 3: points 3 and 4 (n = 294), Group 4: points 5 and 6 (n = 106)
and Group 5: points 7 and 8 (n = 11)].
The multivariate Cox regression model identified belong-
ing to Group 3 (HR: 7.700, 95%CI 0.935–63.421), Group 4
(HR: 12.357, 95%CI 1.388–110.010), or Group 5 (HR: 18.361, 95%CI
1.478–228.145) as independent predictors of 90-day mortality
(Table 4). Furthermore, the multivariate Cox regression model iden-
tified belonging to Group 2 (HR: 3.253, 95%CI 1.173–9.019), Group
3 (HR: 5.298, 95%CI 1.842–15.238), Group 4 (HR: 6.201, 95%CI
2.040–18.846), or Group 5 (HR: 9.413, 95%CI 2.432–36.436) as inde-
pendent predictors of HF events during the 90-day follow-up period
(Table 4). The Kaplan–Meier curves revealed a poorer prognosis,
including all-cause death and HF events, in Group 5 and Group 4
than in the other groups, in Group 3 than in Group 2 or Group 1,
and in Group 2 than in Group 1 (Fig. 4).

Predictive value of the Modified APACHE II and APACHE-HF Discussion

scoring systems
The APACHE II demonstrated an optimal balance between sensi-
tivity and specificity (47.3% and 67.3%; AUC = 0.588) at 17.5 points,
selecting 9.5 points for the Modified APACHE II score as a cut-off
value (sensitivity 51.4%, specificity 69.4%, AUC 0.590). Meanwhile,
a cut-off value of 2.5 points for the APACHE-HF score produced
an optimal balance (sensitivity 87.8%, specificity 63.9%, AUC 0.779)
(Fig. 2). A higher optimal balance was observed in the APACHE-HF
score than in the APACHE II or Modified APACHE II scores.
In the present study, the APACHE II scoring system did not
exhibit an adequate AUC. Furthermore, the Modified APACHE II
scoring system was inadequate to predict the mid-term mortality.
The new scoring system named APACHE-HF, which comprised
a combination of parameters, including mean BP, pulse, sodium,
potassium, creatinine, hematocrit, age, and GCS, exhibited signifi-
cantly higher sensitivity and specificity with an adequate AUC and
could be used to predict adverse mid-term outcomes in patients
with AHF.
446 H. Okazaki et al. / Journal of Cardiology 64 (2014) 441–449

(A) APACHEII (B) APACHEII

1.0
All cause death 1.0
HF event
Cumulave survival

0.8 0.8

Cumulave event
P=0.091 P=0.151
0.6 0.6

0.4 0.4
APACHE II ≤ 17 (n=544) APACHE II ≤ 17 (n=544)

0.2 APACHE II ≥ 18 (n=280) 0.2 APACHE II ≥ 18 (n=280)

days days
0.0 0.0
0 30 60 90 0 30 60 90
No. at Risk No. at Risk
Low 544 507 487 466 Low 544 497 469 439
High 280 255 249 234 High 280 255 241 218

(C) Modified APACHE II (D) Modified APACHE II

All cause death HF event
1.0 1.0
Cumulave survival

Cumulave event

0.8 0.8

0.6
P<0.001 0.6
P<0.001

0.4 0.4
Modified APACHE II ≤ 9 (n=558) Modified APACHE II ≤ 9 (n=558)
0.2 0.2
Modified APACHE II ≥ 10 (n=266) Modified APACHE II ≥ 10 (n=266)
days days
0.0 0.0
0 30 60 90 0 30 60 90
No. at Risk No. at Risk
Low 558 518 503 487 Low 558 515 488 463
High 266 239 231 213 High 266 237 222 194

(E) APACHE-HF (F) APACHE-HF

All cause death HF event
1.0 1.0
Cumulave survival

Cumulave event

0.8 0.8

0.6 P<0.001 0.6 P<0.001

0.4 0.4 APACHE-HF ≤ 2 (n=413)

APACHE-HF ≤ 2 (n=413)
APACHE-HF ≥ 3 (n=411)
APACHE-HF ≥ 3 (n=411)
0.2 0.2

days days
0.0 0.0
0 30 60 90 0 30 60 90
No. at Risk No. at Risk
Low 413 390 382 387 Low 413 390 371 359
High 411 367 352 322 High 411 362 339 298
Fig. 3. Kaplan–Meier curves for each scoring system. (A) The prognosis, including all-cause death, did not differ between the patients with an APACHE II score of ≤17 and
those with an APACHE II score of ≥18. (B) The prognosis, including HF events, was significantly poorer in the patients with an APACHE II score of ≥18 than in those with an
APACHE II score of ≤17. (C) The prognosis, including all-cause death, was significantly poorer in the patients with a Modified APACHE II score of ≥10 than in those with a
Modified APACHE II score of ≤9. (D) The prognosis, including HF events, was significantly poorer in the patients with a Modified APACHE II score of ≥10 than in those with
a Modified APACHE II score of ≤9. (E) The prognosis, including all-cause death, was significantly poorer in the patients with an APACHE-HF score of ≥3 than in those with
an APACHE II-HF score of ≤2. (F) The prognosis, including HF events, was significantly poorer in the patients with an APACHE-HF score of ≥3 than in those with an APACHE
II-HF score of ≤2. HF, heart failure.
 H. Okazaki et al. / Journal of Cardiology 64 (2014) 441–449 447

Table 4
Cox regression analysis of the associations between 90-days cumulative mortality and events, and the clinical findings.

Univariate analysis Multivariate analysis

HR 95%CI p-Value HR 95%CI p-Value

90-days mortality
Points of APACHE-HF score
Group 1 (point 0) 1.000 1.000
Group 2 (point 1, point 2) 1.221 0.147–10.139 0.854 1.110 0.127–9.694 0.925
Group 3 (point 3, point 4) 7.603 1.038–55.694 0.046 7.700 0.935–63.421 0.058
Group 4 (point 5, point 6) 17.522 2.366–129.756 0.005 12.357 1.388–110.010 0.024
Group 5 (point 7, point 8) 20.637 2.147–198.405 0.009 18.361 1.478–228.145 0.024
Adjusting factors
APACHE II (per 1 point increase) 1.054 1.013–1.097 0.009 1.011 0.936–1.092 0.780
Modified APACHE II (per 1 point increase) 1.120 1.061–1.183 <0.001 0.992 0.879–1.120 0.900
LVEF (per 1% increase) 0.987 0.970–1.003 0.987
NYHA (class IV) 2.352 1.015–5.451 0.046 2.486 0.918–6.729 0.073
Total bilirubin (per 0.1 mg/dL increase) 1.011 0.997–1.026 0.120
BUN (per 1.0 mg/dL increase) 1.026 1.017–1.036 <0.001 1.015 1.002–1.029 0.024
Hemoglobin (per 1.0 g/dL increase) 0.861 0.786–0.945 0.002 1.107 0.974–1.258 0.118
BNP (per 10 pg/mL increase) 1.002 1.001–1.003 0.007 1.001 0.999–1.003 0.269

90-days HF Events
Points of APACHE-HF score
Group 1 (point 0) 1.000 1.000
Group 2 (point 1, point 2) 4.315 1.582–11.769 0.004 3.253 1.173–9.019 0.023
Group 3 (point 3, point 4) 7.047 2.597–19.125 <0.001 5.298 1.842–15.238 0.002
Group 4 (point 5, point 6) 10.723 3.870–29.717 <0.001 6.201 2.040–18.846 0.001
Group 5 (point 7, point 8) 14.126 4.135–48.261 <0.001 9.413 2.432–36.436 0.001
Adjusting factors
APACHE-II (per 1 point increase) 1.032 1.011–1.053 0.002 0.983 0.946–1.021 0.371
Modified APACHE II (per 1 point increase) 1.082 1.050–1.115 <0.001 1.039 0.978–1.103 0.220
LVEF (per 1% increase) 0.991 0.983–0.999 0.033 0.989 0.980–0.998 0.023
NYHA (class IV) 1.723 1.189–2.497 0.004 1.778 1.165–2.715 0.008
Total bilirubin (per 0.1 mg/dL increase) 1.086 0.996–1.184 0.061
BUN (per 1.0 mg/dL increase) 1.019 1.013–1.025 <0.001 1.013 1.005–1.020 0.001
Hemoglobin (per 1.0 g/dL increase) 0.925 0.883–0.968 0.001 1.021 0.954–1.092 0.553
BNP (per 10 pg/mL increase) 1.001 1.001–1.002 0.001 1.000 0.999–1.001 0.461

HR, hazard ratio; CI, confidence interval; HF, heart failure, LVEF, left ventricular ejection fraction measured on echocardiography; NYHA, New York Heart Association; BUN,
blood urea nitrogen; CRP, C-reactive protein; BNP, brain natriuretic peptide.

APACHE II scoring system and AHF
The APACHE II scoring system was first published in 1985, and it
has been demonstrated that an increased APACHE II score on admis-
sion can be used to predict a worse in-hospital mortality and poor
long-term prognosis in intensive care patients [2]. The availability
of this scoring system has also been reported in the setting of both
endogenous and various surgical diseases requiring intensive care,
such as respiratory disease, abdominal sepsis, including acute pan-
creatitis, trauma, and postoperative diseases [6–9]. However, this
scoring system has not been evaluated in AHF patients. Moreover,
the APACHE II scoring system involves many factors and is cum-
bersome to apply; therefore, clinicians tend to hesitate to calculate
the score in all patients. It is necessary to develop a scoring system
that can be more easily and immediately applied in the emergency
department.
In the present study, the APACHE II scoring system was found
to have an inadequate AUC, while the Modified APACHE II sco-
ring system could not be used to predict the mid-term mortality.
This might be due to inadequacies in the APACHE II point system.
The APACHE II scoring scale was originally developed for patients
with various diseases requiring care in the intensive care unit. In
fact, the majority of patients evaluated in the original study of
the APACHE II scoring system had non-cardiac diseases. Cardiac
patients comprised only 14.7% of cases, with HF patients accounting
for less than 1.5%. We therefore created a new scoring system suit-
able for HF patients based on a simple point method. This system
is defined as the APACHE-HF system, which exhibits significantly
higher sensitivity and specificity with an adequate AUC compared
to the APACHE II and Modified APACHE II systems. Based on the
results of the present study, a combination of parameters, including
mean BP, pulse, sodium, potassium, creatinine, hematocrit, age, and
GCS, may be used to predict adverse outcomes of AHF.
Predictive factors for AHF
In previous observational studies, various predictive factors for
detecting adverse outcomes in AHF patients have been identified.
Among these factors, we showed the mean BP, pulse, sodium, potas-
sium, creatinine, hematocrit, age, and GCS at the time of admission
to be predictive factors in AHF patients.
An age over 70 or 75 years has been reported to be associated
with a poor prognosis [12,14]. In addition, hypotension is one of the
most important determinants of the prognosis. Gheorghiade et al.
reported that a systolic pressure under 120 mmHg at the time of
admission was associated with a poor prognosis compared with a
systolic pressure over 120 mmHg [10,21]. It has also been reported
that a diastolic pressure under 100 mmHg or 90 mmHg on admis-
sion is associated with poor mortality [14,22]. According to a study
of mean arterial BP, a diastolic pressure under 90 mmHg is a strong
predictor of AHF [23]. We used the pulse rate as a substitute for
heart rate, which is a strong predictor of cardiovascular mortal-
ity and morbidity in the general population [24]. With respect to
AHF, Aaronson et al. found, in a study of 268 ambulatory patients
with advanced HF, that the resting heart rate was the best pre-
dictor of an adverse outcome [25,26]. In another example, Ishii
et al. showed that a heart rate over 113 beats/min on admission
is associated with a better cardiac event-free survival rate than a
heart rate under 112 beats/min [27]; these findings support our
results. Hyponatremia is the most common electrolyte abnormality
in patients with AHF. In several large clinical trials, hyponatremia
was demonstrated to predict a poor prognosis [14,20,28]. The
448 H. Okazaki et al. / Journal of Cardiology 64 (2014) 441–449

APACHE-HF depends on the degree of renal dysfunction [30,31]. For example, in

All cause death the data from the AHEAD registry, patients with a creatinine level
of >120 mmol/L on admission had a poor prognosis [15]. In another
1.0
report, a serum creatinine level over 1.5 mg/dL or 2.0 mg/dL on
survival

admission was found to be a predictor of mortality [22,23]. Because

0.8 patients with AHF and anemia present more frequently with sig-
nificant renal dysfunction, renal dysfunction as a predictor of AHF
0.6 may have relevance to anemia. No relationship between the serum
Cumulave

potassium level on admission and the prognosis of AHF has been

Group 1 Group 2
P < 0.001 reported to date; however, this phenomenon may be explained by
0.4
Group 3 Group 4 an association with renal dysfunction. The relationship between
consciousness disturbance on admission and AHF has not been ade-
0.2 Group 5 quately reported. Sicker AHF patients with shock and acidosis tend
days to exhibit consciousness disturbance on admission. The level of
0.0 consciousness on admission is therefore a possible factor indicating
0 30 60 90 the general condition of AHF patients.
No. at Risk No AHF scoring systems including these key factors for AHF have
been previously reported; however, the development of a scoring
Group 1 70 66 65 63
system that can be applied more easily and immediately in the
Group 2 343 324 317 315
emergency department would be useful in the clinical setting for
Group 3 294 271 262 242
treating AHF patients.
Group 4 106 86 81 72
Group 5 11 10 9 8
Study limitations

HF event There are several limitations associated with the present study
1.0 that should be considered when interpreting the results. First, the
original APACHE II scoring system was based on the most deranged
(worst) physiologic value observed during the initial 24-h time
Cumulave event

0.8
period assuming that all pertinent physiologic values were avail-
able. In the present study, we used vital signs and laboratory data
0.6 obtained on admission because AHF patients are in the most criti-
cal condition on admission. Therefore, this system may not reflect
0.4
Group 1 Group 2 the worst condition of AHF patients. Second, the population evalu-
P < 0.001
ated in our study was limited to only patients admitted to the ICU,
Group 3 Group 4
and AHF patients admitted to general wards were excluded from
0.2
Group 5 this study. However, the patients were seen by the “closed ICU” in
days our institute, and all physicians in “closed ICU” are cardiologists.
0.0 Therefore, the majority of AHF patients were admitted to the ICU.
0 30 60 90 For example, the rate of CS1, CS2, and CS 3 (67.8%, 28.1%, and 8%)
No. at Risk patients were not different to the previous ATTEND registry with
Group 1 70 66 65 62 hospitalized HF (49.5%, 42.4%, and 7.9%) [32]. The patients’ selection
Group 2 343 324 306 297 bias might be minimal in the present study. Third, we persisted with
Group 3 294 266 250 225 the evaluation of original APACHE II, therefore, the scoring system
Group 4 106 86 80 66 did not include the widely accepted factors such as left ventricular
Group 5 11 10 9 7 ejection fraction, NYHA class, blood urea nitrogen, hemoglobin, and
BNP. Further study will be needed to evaluate these factors for the
Fig. 4. Kaplan–Meier curves for the APACHE-HF scoring system. (A) The next new scoring system.
Kaplan–Meier curves revealed a poorer prognosis, including all-cause death, in
Group 5 and Group 4 than in the other groups, in Group 3 than in Group 2 or Group
1, and in Group 2 than in Group 1. (B) The Kaplan–Meier curves revealed a poorer
prognosis, including HF events, in Group 5 and Group 4 than in the other groups,
Conclusion
in Group 3 than in Group 2 or Group 1, and in Group 2 than in Group 1. HF, heart
failure. The APACHE II scoring system cannot be used to adequately pre-
dict the prognosis of patients with AHF. Our new scoring system
including mean BP, pulse, sodium, potassium, hematocrit, creati-
present study also demonstrated a correlation between hypona-
nine, age, and GCS was found to be effective in predicting adverse
tremia and high mortality or HF events, and the cut-off value for the
outcomes in AHF patients.
serum sodium level (Na 137 mmol/L.) was similar to the conven-
tional definition of hyponatremia (Na < 135 mmol/L). Although the
relationship between hematocrit and the prognosis of HF remains Disclosures
to be elucidated, anemia is a famous comorbidity of heart fail-
ure and has been reported to be a predictor of poor outcomes None declared.
in patients with chronic and acute HF [29]. Although most pre-
vious studies used the hemoglobin level as a marker of anemia,
the present study employed hematocrit because it is a constitu- Conflicts of interest
tive factor of the APACHE II score. Renal dysfunction is also well
known to be a strong predictor of AHF, and the prognosis of AHF None declared.
 H. Okazaki et al. / Journal of Cardiology 64 (2014) 441–449
Acknowledgments
We are grateful to the staff of the ICU and the medical records
office at Chiba Hokusoh Hospital, Nippon Medical School for their
valuable assistance in collecting the medical data.
References
[1] Knaus WA, Zimmerman JE, Wagner DP, Draper EA, Lawrence DE. APACHE-Acute
Physiology and Chronic Health Evaluation: a physiologically based classifica-
tion system. Crit Care Med 1981;9:591–7.
[2] Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of
disease classification system. Crit Care Med 1985;13:818–29.
[3] Knaus WA, Wagner DP, Draper EA, Zimmerman JE, Bergner M, Bastos PG, Sirio
CA, Murphy DJ, Lotring T, Damiano A, Harrell FE. The APACHE III prognostic
system. Risk prediction of hospital mortality for critically ill hospitalized adults.
Chest 1991;100:1619–36.
[4] Zimmerman JE, Kramer AA, McNair DS, Malila FM. Acute Physiology and
Chronic Health Evaluation (APACHE) IV: hospital mortality assessment for
today’s critically ill patients. Crit Care Med 2006;34:1297–310.
[5] Jacobs S, Chang RW, Lee B. One year’s experience with the APACHE II severity
of disease classification system in a general intensive care unit. Anaesthesia
1987;42:738–44.
[6] Larvin M, McMahon MJ. APACHE-II score for assessment and monitoring of
acute pancreatitis. Lancet 1989;2:201–5.
[7] Del Bufalo C, Morelli A, Bassein L, Fasano L, Quarta CC, Pacilli AM, Gunella G.
Severity scores in respiratory intensive care: APACHE II predicted mortality
better than SAPS II. Respir Care 1995;40:1042–7.
[8] Schein M. APACHE II score in abdominal sepsis. Arch Surg 1988;123:1288.
[9] Goldhill DR, Sumner A. APACHE II, data accuracy and outcome prediction.
Anaesthesia 1998;53:937–43.
[10] Shiba N, Shimokawa H. Chronic heart failure in Japan: implications of the
CHART studies. Vasc Health Risk Manag 2008;4:103–13.
[11] van der Meer P, Postmus D, Ponikowski P, Cleland JG, O’Connor CM, Cotter
G, Metra M, Davison BA, Givertz MM, Mansoor GA, Teerlink JR, Massie BM,
Hillege HL, Voors AA. The predictive value of short-term changes in hemoglobin
concentration in patients presenting with acute decompensated heart failure.
J Am Coll Cardiol 2013;61:1973–81.
[12] Spinar J, Parenica J, Vitovec J, Widimsky P, Linhart A, Fedorco M, Malek F, Cihalik
C, Spinarova L, Miklik R, Felsoci M, Bambuch M, Dusek L, Jarkovsky J. Base-
line characteristics and hospital mortality in the Acute Heart Failure Database
(AHEAD) Main registry. Crit Care 2011;15:R291.
[13] Hata N, Yokoyama S, Shinada T, Kobayashi N, Shirakabe A, Tomita K, Kitamura
M, Kurihara O, Takahashi Y. Acute kidney injury and outcomes in acute decom-
pensated heart failure: evaluation of the RIFLE criteria in an acutely ill heart
failure population. Eur J Heart Fail 2010;12:32–7.
[14] Ambrosy AP, Vaduganathan M, Huffman MD, Khan S, Kwasny MJ, Fought AJ,
Maggioni AP, Swedberg K, Konstam MA, Zannad F, Gheorghiade M, EVEREST
Trial Investigators. Clinical course and predictive value of liver function tests
in patients hospitalized for worsening heart failure with reduced ejection frac-
tion: an analysis of the EVEREST trial. Eur J Heart Fail 2012;14:302–11.
[15] Malek F, Ostadal P, Parenica J, Jarkovsky J, Vitovec J, Widimsky P, Linhart A,
Fedorco M, Coufal Z, Miklik R, Kruger A, Vondrakova D, Spinar J. Uric acid,
allopurinol therapy, and mortality in patients with acute heart failure—results
of the Acute HEart FAilure Database registry. J Crit Care 2012;27(737):
e11–24.
[16] Desmoulin F, Galinier M, Trouillet C, Berry M, Delmas C, Turkieh A, Massabuau
P, Taegtmeyer H, Smih F, Rouet P. Metabonomics analysis of plasma reveals the
lactate to cholesterol ratio as an independent prognostic factor of short-term
mortality in acute heart failure. PLoS ONE 2013;8:e60737.
449
[17] Mebazaa A, Gayat E, Lassus J, Meas T, Mueller C, Maggioni A, Peacock F, Spinar
J, Harjola VP, van Kimmenade R, Pathak A, Mueller T, Tavazzi L, Disomma S,
Metra M, et al. Association between elevated blood glucose and outcome in
acute heart failure: results from an international observational cohort. J Am
Coll Cardiol 2013;61:820–9.
[18] Sato N, Gheorghiade M, Kajimoto K, Munakata R, Minami Y, Mizuno M,
Aokage T, Asai K, Sakata Y, Yumino D, Mizuno K, Takano T, ATTEND Investi-
gators. Hyponatremia and in-hospital mortality in patients admitted for heart
failure (from the ATTEND registry). Am J Cardiol 2013;111:1019–25.
[19] Gheorghiade M, Zannad F, Sopko G, Klein L, Pina IL, Konstam MA, Massie
BM, Roland E, Targum S, Collins SP, Filippatos G, Tavazzi L. Acute heart fail-
ure syndromes: current state and framework for future research. Circulation
2005;112:3958–68.
[20] Nieminen MS, Harjola VP. Definition and epidemiology of acute heart failure
syndromes. Am J Cardiol 2005;96:5G–10G.
[21] Gheorghiade M, Abraham WT, Albert NM, Greenberg BH, O’Connor CM, She L,
Stough WG, Yancy CW, Young JB, Fonarow GC, OPTIMIZE-HF Investigators and
Coordinators. Systolic blood pressure at admission, clinical characteristics, and
outcomes in patients hospitalized with acute heart failure. J Am Med Assoc
2006;296:2217–26.
[22] Moleerergpoom W, Hengrussamee K, Piyayotai D, Jintapakorn W, Sukhum P,
Kunjara-Na-Ayudhya R, Suwanjutah T, Laothavorn P. Predictors of in-hospital
mortality in acute decompensated heart failure (Thai ADHERE). J Med Assoc
Thai 2013;96:157–64.
[23] Adamopoulos C, Zannad F, Fay R, Mebazaa A, Cohen-Solal A, Guize L, Juilliere
Y, Alla F. Ejection fraction and blood pressure are important and interactive
predictors of 4-week mortality in severe acute heart failure. Eur J Heart Fail
2007;9:935–41.
[24] Kannel WB, Kannel C, Paffenbarger Jr RS, Cupples LA. Heart rate and cardiovas-
cular mortality: the Framingham Study. Am Heart J 1987;113:1489–94.
[25] Aaronson KD, Schwartz JS, Chen TM, Wong KL, Goin JE, Mancini DM. Devel-
opment and prospective validation of a clinical index to predict survival in
ambulatory patients referred for cardiac transplant evaluation. Circulation
1997;95:2660–7.
[26] Lanza GA, Fox K, Crea F. Heart rate: a risk factor for cardiac diseases and out-
comes? Pathophysiology of cardiac diseases and the potential role of heart rate
slowing. Adv Cardiol 2006;43:1–16.
[27] Ishii S, Inomata T, Ikeda Y, Nabeta T, Iwamoto M, Watanabe I, Naruke T, Shi-
nagawa H, Koitabashi T, Nishii M, Takeuchi I, Izumi T. Clinical significance of
heart rate during acute decompensated heart failure to predict left ventricu-
lar reverse remodeling and prognosis in response to therapies in nonischemic
dilated cardiomyopathy. Heart Vessels 2014;29:88–96.
[28] Gheorghiade M, Abraham WT, Albert NM, Gattis Stough W, Greenberg BH,
O’Connor CM, She L, Yancy CW, Young J, Fonarow GC, OPTIMIZE-HF Inves-
tigators and Coordinators. Relationship between admission serum sodium
concentration and clinical outcomes in patients hospitalized for heart failure:
an analysis from the OPTIMIZE-HF registry. Eur Heart J 2007;28:980–8.
[29] von Haehling S, Schefold JC, Hodoscek LM, Doehner W, Mannaa M, Anker SD,
Lainscak M. Anaemia is an independent predictor of death in patients hospi-
talized for acute heart failure. Clin Res Cardiol 2010;99:107–13.
[30] Shirakabe A, Hata N, Kobayashi N, Shinada T, Tomita K, Tsurumi M, Matsushita
M, Okazaki H, Yamamoto Y, Yokoyama S, Asai K, Mizuno K. Long-term progno-
stic impact after acute kidney injury in patients with acute heart failure. Int
Heart J 2012;53:313–9.
[31] Shirakabe A, Hata N, Kobayashi N, Shinada T, Tomita K, Tsurumi M, Matsushita
M, Okazaki H, Yamamoto Y, Yokoyama S, Asai K, Mizuno K. Prognostic impact
of acute kidney injury in patients with acute decompensated heart failure. Circ
J 2013;77:687–96.
[32] Sato N, Kajimoto K, Keida T, Mizuno M, Minami Y, Yumino D, Asai K, Murai
K, Muanakata R, Aokage T, Sakata Y, Mizuno K, Takano T, TEND Investigators.
Clinical features and outcome in hospitalized heart failure in Japan (from the
ATTEND Registry). Circ J 2013;77:944–51.