ORIGINAL CONTRIBUTION
Predicting Mortality Among Patients Hospitalized for Heart Failure
Derivation and Validation of a Clinical Model
Context A predictive model of mortality in heart failure may be useful for clinicians to improve communication with and care of hospitalized patients.
Objectives To identify predictors of mortality and to develop and to validate a model using information available at hospital presentation.
Design, Setting, and Participants Retrospective study of 4031 community-based patients presenting with heart failure at multiple hospitals in Ontario, Canada (2624 patients in the derivation cohort from 1999-2001 and 1407 patients in the validation cohort from 1997-1999), who had been identitifed as part of the Enhanced Feedback for Effective Cardiac Treatment (EFFECT) study.
Main Outcome Measures All-cause 30-day and 1-year mortality.
Results The mortality rates for the derivation cohort and validation cohort, respectively, were 8.9% and 8.2% in hospital, 10.7% and 10.4% at 30 days, and 32.9% and 30.5% at 1 year. Multivariate predictors of mortality at both 30 days and 1 year included older age, lower systolic blood pressure, higher respiratory rate, higher urea nitrogen level (all P<.001), and hyponatremia (P<.01). Comorbid conditions associated with mortality included cerebrovascular disease (30-day mortality odds ratio [OR], 1.43; 95% confidence interval [CI], 1.03-1.98; P=.03), chronic obstructive pulmonary disease (OR, 1.66; 95% CI, 1.22-2.27; P=.002), hepatic cirrhosis (OR, 3.22; 95% CI, 1.08-9.65; P=.04), dementia (OR, 2.54; 95% CI, 1.77-3.65; P<.001), and cancer (OR, 1.86; 95% CI, 1.28-2.70; P=.001). A risk index stratified the risk of death and identified low- and high-risk individuals. Patients with very low-risk scores (<60) had a mortality rate of 0.4% at 30 days and 7.8% at 1 year. Patients with very high-risk scores (>150) had a mortality rate of 59.0% at 30 days and 78.8% at 1 year. Patients with higher 1 -year risk scores had reduced survival at all times up to 1 year (log-rank, P<.001). For the derivation cohort, the area under the receiver operating characteristic curve for the model was 0.80 for 30-day mortality and 0.77 for 1 -year mortality. Predicted mortality rates in the validation cohort closely matched observed rates across the entire spectrum of risk.
Conclusions Among community-based heart failure patients, factors identifiable within hours of hospital presentation predicted mortality risk at 30 days and 1 year. The externally validated predictive index may assist clinicians in estimating heart failure mortality risk and in providing quantitative guidance for decision making in heart failure care.
Douglas S. Lee, MD Peter C. Austin, PhD Jean L. Rouleau, MD
Peter P. Liu, MD_
David Naimark, MD
Jack V. Tu, MD, PhD_
HEART FAILURE IS A CONDI-tion with an adverse prognosis; 1-year mortality rates in population-based studies have been reported to be 35% to 40%.1-5 Important prognostic factors have been identified among clinical trial enrollees. However, factors that predict mortality in the community setting may differ.6 Although heart failure is a common, serious condition treated by both generalist and specialist physicians, few methods exist to help quantatitively estimate prognosis. As a result, clinicians must rely on published mortality rates from clinical trials or other studies, in which patient populations may differ from those encountered in clinical practice.
Knowledge of mortality predictors can be used to generate predictive models that can aid clinicians' decision making, in particular by identifying patients who are at high or low risk of death.711 Such risk-assessment methods have been developed for acute myocardial infarction but not for heart failure.71213 Risk prediction models may be used in patient counseling to initiate the discussion about end-of-life issues and also maybe used for quality-of-care out-
JAMA. 2003;290:2581-2587
Author Affiliations: Departments of Health Policy, Management, and Evaluation/Clinical Epidemiology (Drs Lee, Austin, and Tu) and Public Health Sciences (Drs Austin and Tu), Division of Cardiology, University Health Network-Toronto General Hospital and Mt Sinai Hospital (Drs Rouleau and Liu), Heart and Stroke/Richard Le-war Centre of Excellence and Toronto General Hospital (Dr Liu), University of Toronto, Toronto, Ontario;
www.jama.com
Institute for Clinical Evaluative Sciences, Toronto, Ontario (Drs Lee, Austin, and Tu); and Department of Medicine, Sunnybrook and Women's College Health Science Centre, Toronto, Ontario (Drs Naimark and Tu). Corresponding Author and Reprints: Jack V. Tu, MD, PhD, Institute for Clinical Evaluative Sciences, Room G-106,2075 Bayview Ave, Toronto, Ontario, Canada M4N 3M5 (e-mail: tu@ices.on.ca).
©2003 American Medical Association. All rights reserved. (Reprinted) JAMA, November 19, 2003—Vol 290, No. 19 2581
Downloaded from jama.ama-assn.org by guest on March 8, 2012
PREDICTORS OF HEART FAILURE MORTALITY
come assessment. Patients at low risk could be potentially discharged from the hospital early, whereas those at high risk could benefit from intensive or specialized care units.
Previous work in heart failure risk determination focused primarily on risk adjustment in elderly Medicare populations.1416 These prior models are complex and arguably difficult for clinicians to use in routine practice. Furthermore, transportability of previous models may be limited because the variables that comprise these models are not routinely collected.1516 Our objective was to develop and externally validate a method to predict mortality risk in heart failure patients based on information routinely available to clinicians at hospital presentation such as demographic features, vital signs, and patient comorbid conditions. We hypothesized that the model could effectively stratify the risk of death among heart failure patients at both 30 days and 1 year.
METHODS
Patients
We identified newly admitted patients with a primary diagnosis of heart failure (International Classification of Diseases, Ninth Revision, Clinical Modification code 428) using the Canadian Institutes of Health Information hospital discharge abstract as described previously.1718 Of the patients identified, we further refined the cohort by only including patients with a clinical heart failure presentation who met modified Framingham heart failure criteria.319 21 We excluded patients who developed heart failure after admission (ie, in-hospital complication), patients transferred from another acute care facility, those aged 105 years or older, nonresidents, and those with an invalid health card number.
Using these methods, 2624 patients from 34 hospitals in Ontario, Canada, from April 1,1999, to March 31, 2001, were identified as part of the Enhanced Feedback for Effective Cardiac Treatment (EFFECT) study. The hospitals included teaching hospitals and community-based institutions from
both rural and urban locales. This sample comprised the mortality model derivation cohort. Similarly, we identified a cohort of 1407 heart failure patients from 14 different hospitals in Ontario from a prior period (April 1,1997, to March 31, 1999), which comprised an independent model validation cohort. Hospitals included in this study had a minimum yearly volume of more than 100 heart failure patient admissions during the years of sampling. From these hospitals, patients were sampled at random for abstraction of clinical data and subsequent follow-up. Institutional review board approval was obtained from each participating institution prior to the study.
Data Collection and Variable Definitions
Potential candidate variables were identified based on review of the literature, clinical relevance, and routine availability in the initial hours of hospital presentation.14"16,22"25 Variables selected for abstraction were further guided by the consensus of a Canadian expert panel of heart failure specialists.26 The potential candidate variables were either presentation features (eg, vital signs) or other data abstractable from the clinical record up to the first 24 hours of hospital presentation (eg, laboratory values, preexisting comorbid conditions) and were classified as demographic characteristics, presenting clinical and laboratory features, or preexisting comorbid conditions. Comorbidity data were subcat-egorized according to the disease moieties of the Charlson comorbidity index.24 These included cancer, dementia, diabetes mellitus, chronic obstructive pulmonary disease, cerebrovascular disease, peripheral vascular disease, cirrhotic liver disease, prior myocardial infarction, and renal indices (serum blood urea nitrogen and creatinine concentrations). Hyponatremia and hypokalemia were defined by the lower limit of the normal biochemical range. We also collected information when available on left ventricular function via echocardiography, radionuclide angiography, or cardiac catheterization.
Data abstraction from hospital records was conducted by highly experienced cardiology nurse abstractors using a computerized instrument with preprogrammed range checks. Reliability for abstraction of categorical variables was high. For prior myocardial infarction, reliability was 0.94 using crude agreement and k = 0.88; chronic obstructive pulmonary disease, 0.92 and k = 0.80; cancer, 0.97 and k = 0.89; and dementia, 0.97 and k = 0.82.27 Deaths occurring up to 1 year after hospital admission were identified by linkages with the Registered Persons Database, using the patient's encrypted health card number. The primary model outcomes of 30-day and 1-year mortality were used to eliminate biases related to the decision to discharge and its potential relationship with mortality.
Analysis
Candidate variables that were associated with 30-day and 1-year mortality on univariate analysis (P^.20) were included as potential covariates in a multiple logistic regression model.28 Variable selection in multivariable modeling was based on clinical and statistical significance.29 We examined the strength and shape of the relationships of continuous variables with the log odds of death using cubic spline plots.30 These functions were used to develop and refine the multivariable regression models as used previously.13 Discrimination of the model was assessed by the area under the receiver operating characteristic (ROC) curve31 and calibration was assessed using the Hosmer and Lemeshow x2 statistic (P>.05 for all models).28 Models for 30-day and 1-year mortality were assessed for possible overfit using linear shrinkage estimators.30,32 A sensitivity analysis was also conducted to assess the robustness of the model after accounting for reduced left ventricular systolic function (ejection fraction <0.40).
Score-based prediction rules for mortality at 30 days and 1 year were developed from logistic regression models by using a regression coefficient-based scoring method.33,34 Integer scores were assigned by dividing risk-factor coeffi-
2582   JAMA, November 19, 2003—Vol 290, No. 19 (Reprinted) ©2003 American Medical Association. All rights reserved.
Downloaded from jama.ama-assn.org by guest on March 8, 2012
PREDICTORS OF HEART FAILURE MORTALITY
cients by the age coefficient and rounding up to the nearest unit for continuous variables and up to the nearest 5 points for midpoints of stratified continuous or categorical variables.35 The overall risk score was calculated by adding each component together. Mortality rates were assessed according to the numeric value of the 30-day and 1-year risk scores. To assess if the prognostic rank of the 1-year risk score was preserved over time, Kaplan-Meier survival curves were constructed and stratified according to 1-year risk score risk quintile and comparisons were performed using the log-rank test.8
We validated the 30-day and 1-year mortality models internally using the bootstrap in the derivation dataset by sampling with replacement for 200 iterations.36 38 We externally validated both regression models by assessing model performance in the validation cohort.
Analyses were conducted using SAS statistical software (Version 8.0; SAS Institute Inc, Cary, NC) and ROC curve analysis was performed using STATA statistical software (Version 7.0; STATA Corp, College Station, Tex). Bootstrap sampling, cubic spline analysis, and ROC curve optimism analysis were conducted using S-plus statistical software (Version 6.0, Insightful Corp, Seattle, Wash).
RESULTS
Description of the Derivation and Validation Cohorts
There were 2624 heart failure patients in the derivation cohort and 1407 patients in the external validation cohort (N=4031 patients). Age and sex distributions in the 2 cohorts were comparable (Table 1). Approximately half were women who were on average, older than men (mean [SD] age of women, 78.5 [10.5] years vs 74.1 [11.5] years for men; P<. 001). Summary statistics of presenting features (vital signs, initial laboratory values) and frequency of comorbid conditions are shown in Table 1. Left ventricular function evaluation was performed in 1618 (61.7%) of the derivation cohort compared with 692 (49.2%) of the validation cohort.
The mortality rates for the derivation cohort were 8.9% (234 deaths) in-hospital, 10.7% (282 deaths) at 30 days, and 32.9% (862 deaths) at 1 year; validation cohort, 8.2% (115 deaths), 10.4% (147 deaths), and 30.5% (429 deaths), respectively. Thus, the mortality rates at 1 year were comparable with those reported in other population-based studies.
Predictors of Mortality
Results of univariate analysis for all potential predictors are shown in Table 2. Multivariate models for 30-day and 1-year mortality are shown in table 3. There was no statistical evidence of overfit as demonstrated by linear shrinkage estimation in both multivariable models. Model predictors of both 30-day and
Characteristic
Age, mean (SD), y Women
Vital sign, mean (SD)
Blood pressure, mm Hg
Systolic_
Diastolic_
Heart rate, beats/min_
Respiratory rate, breaths/min
Oxygen saturation, %
Serum concentration, mean (SD) Hemoglobin, g/dL_
Leukocyte count, /mm2_
Sodium, mEq/L_
Potassium, mEq/L_
Creatinine, mg/dL [umol/L]
Urea nitrogen, mg/dL [mmol/L]
Glucose, mg/dL [mmol/L]
Left ventricular ejection fraction <0.40§
Comorbid condition
Prior myocardial infarction
Atrial fibrillation_
Diabetes mellitus
1-year mortality included age, systolic blood pressure, respiratory rate, hyponatremia, and urea nitrogen concentration. Although low-hemoglobin concentration was predictive of 1-year death, it was not associated with 30-day mortality. Comorbid conditions associated with mortality common to both models included cerebrovascular disease, dementia, chronic obstructive pulmonary disease, cirrhosis, and cancer. Although creatinine elevation was a significant predictor of death, it was correlated with urea nitrogen concentration (Pearson R=.67) and was no longer significant when urea nitrogen concentration was entered into the model (Table 3). Notably, although diabetes was associated with decreased mortality in univariate analysis (Table 2), there was no signifi-
Derivation Cohort           Validation Cohort (n = 2624)t_(n = 1407)i
76.3 (11.2)_75.3 (11.8)_
711 (50.5)_
148 (33)_148 (34)_
80 (20)_83 (20)_
94 (25)_97 (25)_
26 (7)_27 (8)_
92 (8)_91 (8)_
12.4 (2.1)_12.4 (2.1)_
10.2 (5.1)_10.5 (9.4)_
138 (5)_138 (5)_
4.3 (0.7)_4.2 (0.6)_
1.45 (0.96) [128 {85}] 1.49 (1.15) [132 {102}]
29.4 (19.3) [10.5 {6.9}]        27.2 (19.9) [9.7 {7.1}]
164 (81) [9.1 {4.5}]_171 (105) [9.5 {5.8}]
854 (52.8) 330 (47.7)
952 (36.3)_533 (37.9)
795 (30.3)_402 (28.6)
891 (34.0)_471 (33.5)
Chronic obstructive pulmonary disease	543 (20.7)	371 (26.4)
Cerebrovascular disease	446 (17.0)	310 (22.0)
Peripheral vascular disease	355 (13.5)	254 (18.1)
Peptic ulcer disease	155 (5.9)	155 (11.0)
Liver disease I	34 (1.3)	26 (1.9)
Dementia	225 (8.6)	118 (8.4)
Cancer	234 (8.9)	207 (14.7)
*Values are expressed as number (percentage) unless otherwise indicated t1999-2001. t1997-1999.
§Performed in 1618 (61.7%) of the derivation cohort compared with 692 (49.2%) of the validation cohort ||Mild liver disease or hepatic cirrhosis.
Table 1. Clinical Characteristics of the Derivation and Validation Heart Failure Cohorts*
1325 50.5
©2003 American Medical Association. All rights reserved. (Reprinted) JAMA, November 19, 2003—Vol 290, No. 19 2583
Downloaded from jama.ama-assn.org by guest on March 8, 2012
PREDICTORS OF HEART FAILURE MORTALITY
Table 2. Univariate Predictors of Mortality in Heart Failure Derivation Cohort (n = 2624)
30-Day Mortality 1 -Year Mortality
P P Variable OR (95% CI)     Value      OR (95% CI) Value
Age, y (per 10-unit increase) 1.83
Men 0.81 Vital sign
Systolic blood pressure, mm Hg 0.85 (per 10-unit increase)
Heart rate, beats/min 1.01 (per 10-unit increase)
Respiratory rate, breaths/min 1.13 (per 5-unit increase)
Oxygen saturation, % 0.99 (per 5-unit increase)
Serum concentration
Hemoglobin < 10.0 g/dL 1.73
Sodium < 136 mEq/L 1.69
Potassium <3.5 mEq/L 0.66
Creatinine >2.0 mg/dL [>177 umol/L] 2.47
Urea nitrogen, mg/dL 1.32 (per 10-unit increase)
Comorbid condition
Prior myocardial infarction 0.98
Atrial fibrillation_1.21
Diabetes mellitus 0.75
Chronic obstructive pulmonary 1.51 disease
Cerebrovascular disease 1.51
Peripheral vascular disease 1.21
Peptic ulcer disease 0.95
Mild liver disease 1.52
Cirrhosis 2.21
Dementia 3.77
Cancer 1.81
Abbreviations: CI, confidence interval; OR, odds ratio
Table 3. Multivariable Predictors of Mortality
;1.59-2.10) <.0Q1 1.69(1.55-1.84) <.0Q1 ;0.63-1.04)      .10     0.84(0.71-0.99) .03
p.81-0.88)    <.001    0.89(0.86-0.91) <.001
;0.96-1.06)      .76     0.99(0.96-1.02) .48
1.04-1.22)      .002    1.07(1.02-1.13) .009
;0.92-1.07)      .78      1.02(0.96-1.07) .59
;1.25-2.36) <.0Q1 2.07(1.65-2.60) <.0Q1
;1.30-2.20) <.0Q1 1.61 (1.34-1.94) <.0Q1
;0.38-1.08) .12 0.68(0.49-0.93) .02
;1.84-3.29) <.0Q1 2.90(2.33-3.63) <.0Q1
;1.26-1.39) <.001 1.37(1.30-1.44) <.001
;0.75-1.26) .86 1.22(1.03-1.44) .02
;0.93-1.57) .15 1.13(0.94-1.34) .19
;0.57-0.98) .04 0.81 (0.68-0.96) .02
;1.14-2.00) .004 1.30(1.07-1.58) .009
;1.11-2.02)      .007 1.47(1.19-1.81) <.0Q1
;0.85-1.69)      .28 1.38(1.09-1.73) .007
;0.54-1.58)      .86 1.17(0.83-1.63) .37
;0.23-5.67)      .59 1.76(0.56-5.31) .31
;0.73-5.54)      .12 3.45(1.53-8.24) .004
|2.71-5.19) <.0Q1 2.99(2.27-3.96) <.0Q1
;1.28-2.52) <.001 1.86(1.45-2.38) <.001
cant effect after adjustment for other factors in the multivariable model (P>.20 for both models). The univariate analysis suggested that women had greater risk of death than men; however, the sex disparities were not significant after adjustment for age (P>.40 for both models).
Risk Scores
Multivariable risk scores for prediction of both 30-day and 1-year mortality were calculated (Table 4). Both scores were normally distributed with mean (SD) scores of 91 (26) at 30 days and 102 (27) at 1 year. Risk categories were assigned in 30-point increments to correspond with 1 or 2 SDs of risk above or below average based on the SDs of the 30-day and 1-year scores. The magnitude of the scores had prognostic implications (Figure). Stratification by quintile of risk score revealed a gradation in risk of mortality. The 30-day mortality rate was 0.8% for quintile 1 (with a corresponding score <69); 3.6%, quintile 2 (score, 69-82); 6.3%, quintile 3 (score, 83-96); 12.6%, quintile 4 (score, 97-113); and 30.5%, quintile 5 (score, 114-195). Similarly, a graded increase in risk occurred with 1-year score quintiles. The 1-year mortality rate was 9.0% for quintile 1 (with a corresponding score <80); 17.7%, quintile 2 (score, 80-93); 29.1%, quintile 3 (score, 94-107); 42.1%, quintile 4 (score, 108-123); and 66.3%, quintile 5 (score, 124-198). Assignment of 1-year risk scores at baseline maintained their prognostic implications at all time points up to 1 year of follow-up on Kaplan-Meier analysis (P<.001 between-risk quintiles).
Model Validation
In the derivation set (n=2624), the area under the ROC curve was 0.80 for 30-day mortality and 0.77 for 1-year mortality. When the 30-day model was applied to in-hospital mortality in the derivation set, the area under the ROC curve was 0.82. In the external validation set (n=1407), the discriminative ability of the models were maintained with an area under the ROC curve of 0.79 for 30-day mortality and 0.76 for 1-year mortality. The bootstrap-
30-Day Model 1 -Year Model
Variable	I OR (95% CI)	P Value	OR (95% CI)	P Value
Age, y (per 10-unit increase)	1.70 (1.45-1.99)	<.001	1.61 (1.46-1.77)	<.001
Vital sign Systolic blood pressure, mm Hg (per 10-unit increase)	0.84 (0.80-0.88)	<.001	0.88 (0.85-0.90)	<.001
Respiratory rate, breaths/min (per 5-unit increase)	1.23 (1.12-1.36)	<.001	1.15 (1.08-1.24)	<.001
Serum concentration Sodium <136 mEq/L	1.53 (1.14-2.05)	.005	1.46 (1.19-1.80)	<.001
Hemoglobin <10.0 g/dL	NA	NA	1.37 (1.05-1.78)	.02
Urea nitrogen, mg/dL (per 10-unit increase)	1.55 (1.42-1.71)	<.001	1.49 (1.39-1.60)	<.001
Comorbid condition Cerebrovascular disease	1.43 (1.03-1.98)	.03	1.36 (1.08-1.71)	.01
Dementia	2.54 (1.77-3.65)	<.001	2.00 (1.47-2.72)	<.001
Chronic obstructive pulmonary disease	1.66 (1.22-2.27)	.002	1.41 (1.13-1.75)	.003
Hepatic cirrhosis	3.22 (1.08-9.65)	.04	5.80 (2.23-15.11)	<.001
Cancer	1.86 (1.28-2.70)	.001	1.85 (1.40-2.43)	<.001
Abbreviations: CI, confidence interval; NA, not applicable to 30-day model; OR, odds ratio				
2584  JAMA, November 19, 2003—Vol 290, No. 19 (Reprinted) ©2003 American Medical Association. All rights reserved.
Downloaded from jama.ama-assn.org by guest on March 8, 2012
PREDICTORS OF HEART FAILURE MORTALITY
corrected area under the ROC curve was 0.79 for the 30-day mortality model and 0.76 for the 1-year mortality model. Predicted and observed mortality rates in the validation cohort were in close agreement across the entire spectrum of risk (Figure).
Effect of Left Ventricular Systolic Function
At both 30 days and 1 year, the presence of left ventricular systolic dysfunction was associated with increased mortality risk when compared with patients without systolic dysfunction (adjusted OR, 1.98 [95% CI, 1.25-3.16] at 30 days [P = .004] and 1.55 [95% CI, 1.19-2.02] at 1 year [P = .001]). After adjustment for left ventricular systolic dysfunction, the coefficients of model covariates remained similar to models without such adjustment. There were no significant interactions between left ventricular systolic function and comorbid conditions. When we accounted for the presence of systolic dysfunction, the discriminative ability of the mortality models improved for in-hospital mortality (ROC curve area, 0.84), 30-day mortality (ROC curve area, 0.81), and 1-year mortality (ROC curve area, 0.78).
COMMENT
We found that a simple model using data available in the initial hours of hos-
pital presentation predicted mortality in hospitalized heart failure patients at 30 days and 1 year. Both mortality models included acute physiological parameters and chronic disease comor-
bidities. Important acute physiological variables including hyponatremia, respiratory rate, blood pressure, and selected comorbid conditions including dementia, cirrhosis, and cancer were as-
Table 4. Heart Failure Risk Scoring System*		
	No. of Points	
1 Variable	30-Day Scoref	1 -Year Scored
Age, y	+Age (in years)	+Age (in years)
Respiratory rate, min (minimal 20; maximum 45)§	+Rate (in breaths/min)	+Rate (in breaths/min)
Systolic blood pressure, mm Hg s180	-60	-50
160-179	-55	-45
140-159	-50	-40
120-139	-45	-35
100-119	-40	-30
90-99	-35	-25
<90	-30	-20
Urea nitrogen (maximum, 60 mg/dl_)§H	+l_evel (in mg/dL)	+l_evel (in mg/dL)
Sodium concentration <136 mEq/L	+10	+10
Cerebrovascular disease	+10	+10
Dementia	+20	+15
Chronic obstructive pulmonary disease	+10	+10
Hepatic cirrhosis	+25	+35
Cancer	+15	+15
Hemoglobin <10.0 g/dL (<100 g/L)	NA	+10
Abbreviation: NA, not applicable to 30-day model.
*An electronic version ot the risk scoring system is available at: http://www.ccort.ca/CHFriskmodel.asp.
fCalculated as age + respiratory rate + systolic blood pressure + urea nitrogen + sodium points + cerebrovascular disease points + dementia points + chronic obstructive pulmonary disease points + hepatic cirrhosis points + cancer points.
^Calculated as age + respiratory rate + systolic blood pressure + urea nitrogen + sodium points + cerebrovascular disease points + dementia points + chronic obstructive pulmonary disease points + hepatic cirrhosis points + cancer points + hemoglobin points.
§Values higher than maximum or lower than minimum are assigned the listed maximum or minimum values,
((Increases were protective in both mortality models. Points are subtracted tor higher blood pressure measurements,
TJMaximum value is equivalent to 21 mmol/L. Score calculated using value in mg/dL.
Figure. Mortality Rates Stratified by 30-Day and 1-Year Risk Scores
30-Day Mortality Rate
1-Year Mortality Rate
100-90-80-
* 7°-
% 60-
I 50-
ST 40-Q
° 30-20-10-0-
□ Derivation Cohort H Validation Cohort
0.4 0.6
3.4 4.2
<60
Risk Category Very Low
12.2
J
61-90 91-120 121-150 >150
Low Intermediate High Very High
30-Day Risk Score
□ Derivation Cohort H Validation Cohort
100-
90-
80-* 70-t 60-° 50-ffl 40-
30-
20-
10-
0-
<60
Risk Category Very Low
12.9
32.5 30.2
61-90 91-120 121-150 >150
Low Intermediate High Very High
1 -Year Risk Score
Score categories were assigned according to 30-point increments corresponding to unit SD increments above and below the intermediate range (91-120). Error bars indicate 95% confidence intervals for the mortality rates in each category.
©2003 American Medical Association. All rights reserved.
(Reprinted) JAMA, November 19, 2003—Vol 290, No. 19 2585
Downloaded from jama.ama-assn.org by guest on March 8, 2012
PREDICTORS OF HEART FAILURE MORTALITY
sociated with an increased risk of death. Abnormal renal function, as measured by blood urea nitrogen was also a strong predictor of death. The risk score provides a simple method to stratify a patient's risk of death at the time of initial hospital presentation into very low (^60 points), low (61-90 points), high (121-150 points), and very high (> 150 points) risk categories relative to an intermediate (91-120 points) risk group at average risk.
This study is consistent with prior investigations of heart failure mortality risk. The adverse prognostic impact of increasing comorbidity burden, using the Charlson comorbidity index,24 has been described.17 However, we found that a subset of comorbid conditions had the greatest independent impact on mortality in community-based patients, and the relative contributions of the comorbid conditions to mortality differed substantially in comparison with the Charlson comorbidity index.24 Evidence of respiratory distress was associated with mortality, which also concurs with findings from other studies39 and guidelines40 that advise hospitalization in the presence of increased respiratory rate. Systemic hypotension and shock have been identified by others as mortality predictors.39,41 However, our findings extend prior observations and concur with those of Goldberger et al,42 who also reported a similar pattern of blood pressure effect on mortality in patients presenting with acute pulmonary edema. Although higher blood pressure was protective in an acute clinical presentation, our findings should not be interpreted as suggesting that maintaining higher blood pressure long term by avoiding the use of drug therapies is beneficial. Rather, the effects observed may relate to an acute protective mechanism from factors influencing cardiac output or vascular tone, which have yet to be elucidated.
Our model was designed to be independent of left ventricular function information because these data may not be available in the early hours of hospital presentation. This design allows for broader utility because community-
based heart failure studies have found that many patients do not undergo left ventricular function evaluation during the hospital stay.43 Therefore, our study did not exclude heart failure patients without documented ventricular function. Exclusion of this subset could bias parameter estimates because patients who have had echocar-diographic assessment may have a better prognosis.20 Nonetheless, when we assessed the robustness of our models in the subset who had left ventricular function assessed, the estimated ORs were not materially altered.
The present model differs from other models used to predict heart failure mortality in a number of important ways. A unique difference is prediction of mortality at both early and later times. Two other models were designed to predict mortality at 30 days, but none were intended to predict mortality at 1 year.1415 Second, based on the number of variables required, it has greater parsimony than most other predictive risk models. Only the Time-Insensitive Predictive Instrument model is simpler because it contains only 4 variables23; however, the instrument was developed for hospital mortality assessment and did not include comorbid conditions. Finally, the variables in the current model can be easily obtained and are not dependent on specialized laboratory tests, making it less susceptible to problems arising from missing data.44
Several reports have alluded to the need for tools to quantitively estimate prognosis in heart failure patients 45,46 A recent international survey found that the treatment aim of symptom relief held greater importance to physicians for elderly patients, while delay of death was thought to be more important for younger patients.45 The study implied that if prognosis was poor, physicians should consider symptom relief as their primary aim. Previously, the best estimates of risk from heart failure were either from the average mortality rates across all patients or the physician's best guess of prognosis.46 However, physicians may underestimate or overesti-
mate prognosis in heart failure patients based on memorable clinical experiences 47 In contrast to anecdotal experience, the heart failure index is an objective stratification of mortality risk. The index could be used as a framework to discuss prognosis and provides evidence to support rational decision making about end-of-life care in heart failure patients who are at highest risk 48 50 It could also potentially be used as an aid in making decisions about patient disposition (ie, discharge, or admission to ward or intensive/cardiac care units).
Our study has a number of limitations. Because this was a retrospective study, our results are dependent on the accuracy of recorded data. Despite this limitation, however, the model covar-iates (eg, vital signs, laboratory results) are likely to be accurately documented in medical records.51 Another possible limitation is misclassification of comorbid conditions based on underreporting in the medical record because coexisting but undiagnosed conditions may not be recognized or because of variability in abstraction. However, this would result in true effects that are larger because it would tend to bias results toward the null. Finally, this study included hospitalized patients and may not have direct applicability to the ambulatory chronic heart failure population as examined previously.52,53
In conclusion, significant predictors of mortality in heart failure patients presenting to the hospital include age, presentation vital signs, routine biochemistry, and comorbid conditions. A simple model and corresponding risk index predicted risk of 30-day and 1-year mortality in a broad sample of hospitalized heart failure patients using data available at hospital presentation. The risk index provides estimates of risk that may assist clinicians in counseling patients and families and guides clinical decision making.
Author Contributions: Dr Lee had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Lee, Liu, Naimark, Tu
2586  JAMA, November 19, 2003—Vol 290, No. 19 (Reprinted) ©2003 American Medical Association. All rights reserved.
Downloaded from jama.ama-assn.org by guest on March 8, 2012
PREDICTORS OF HEART FAILURE MORTALITY
Acquisition of data: Lee, Tu Analysis and interpretation of data: Lee, Austin, Rouleau, Liu, Naimark Drafting of the manuscript: Lee. Critical revision of the manuscript for important intellectual content: Lee, Austin, Rouleau, Liu, Naimark, Tu.
Statistical expertise: Lee, Austin, Naimark. Obtained funding: Lee, Tu Study supervision: Rouleau, Liu, Tu Funding/Support: This study was supported byagrant from the Ontario Ministry of Health (Ontario Program for Optimal Therapeutics) and by a grant to the Canadian Cardiovascular Outcomes Research Team from the Canadian Institutes of Health Research and the Heart and Stroke Foundation. Dr Lee is a fellow of the Canadian Institutes of Health Research/Heart and Stroke Foundation of Canada. Dr Austin is a new investigator of the Canadian Institutes of Health Research. DrTu is supported by a Canada Research Chair in Health Services Research
Role of the Sponsor: The sponsors funded the collection and analysis of the data for this study, but the authors had complete independence in the collection, analysis, and interpretation of the study results. Acknowledgment: We gratefully acknowledge the assistance of Linda Donovan and the administrative support provided by Francine Duquette. We thank Yanyan Gong and Julie Wang for programming assistance.
REFERENCES
1. Ho KK, Anderson KM, Kannel WB, Grossman W, Levy D. Survival after the onset of congestive heart failure in Framingham Heart Study subjects. Circulation. 1993;88:107-115.
2. Zannad F, Braincon S, Juilliere Y, et al. Incidence, clinical and etiologic features, and outcomes of advanced chronic heart failure: the EPICAL study. J Am Coll Cardiol. 1999;33:734-742.
3. Senni M, Tribouilloy CM, Rodeheffer RJ, et al. Congestive heart failure in the community. Circulation. 1998;98:2282-2289.
4. Maclntyre K, Capewell S, Stewart S, et al. Evidence of improving prognosis in heart failure. Circulation. 2000;102:1126-1131.
5. Tu JV, Zhang H. Congestive heart failure outcomes in Ontario. In: Naylor CD, Slaughter PM, eds. Cardiovascular Health and Services in Ontario: An ICES Atlas. Toronto, Ontario: Institute for Clinical Evaluative Sciences; 1999:111-122.
6. SharpeN. Heartfailure in the community. ProgCar-diovascDis. 1998;41(1 suppl 1):73-76.
7. Tu JV, Austin PC, Walld R, Roos L, Agras J, McDonald KM. Development and validation of the Ontario acute myocardial infarction mortality prediction rules. J Am Coll Cardiol. 2001;37:992-997.
8. Tu JV, Jaglal SB, Naylor CD, et al. Multicenter validation of a risk index for mortality, intensive care unit stay, and overall hospital length of stay after cardiac surgery. Circulation. 1995;91:677-684.
9. Reilly BM, Evans AT, SchaiderJJ, etal. Impact of a clinical decision rule on hospital triage of patients with suspected acute cardiac ischemia in the emergency department. JAMA. 2002;288:342-350.
10. Laupacis A, Sekar N, Stiell IG. Clinical prediction rules: a review and suggested modifications of methodological standards. JAMA. 1997;277:488-494.
11. Tu JV, Mazer CD, Levinton C, Armstrong PW, Naylor CD. A predictive index for length of stay in the intensive care unit following cardiac surgery. CMAI. 1994;151:177-185.
12. Krumholz HM, Chen J, Wang Y, Radford MJ, Chen
YT, Marciniak TA. Comparing AMI mortality among hospitals in patients 65 years of age and older. Circulation. 1999;99:2986-2992.
13. Lee KL, Woodlief LH, Topol EJ, et al. Predictors of 30-day mortality in the era of reperfusion for acute myocardial infarction. Circulation. 1995;91:1659-1668.
14. Rosenthal GE, Baker DW, Norris DG, Way LE, Harper DL, Snow RJ. Relationships between in-hospital and 30-day standardized hospital mortality. Health Serv Res. 2000;34:1449-1468.
15. Keeler EB, Kahn KL, Draper D, et al. Changes in sickness at admission following the introduction of the prospective payment system. JAMA. 1990;264:1962-1968.
16. Daley J, Jencks S, Draper D, Lenhart G, Thomas N, Walker J. Predicting hospital-associated mortality for Medicare patients. JAMA. 1988;260:3617-3624.
17. Jong P, Vowinckel E, Liu PP, Gong Y, Tu JV. Prognosis and determinants of survival in patients newly hospitalized for heartfailure: a population-based study. Arch Intern Med. 2002;162:1689-1694.
18. International Classification of Diseases, Ninth Revision, Clinical Modification. Washington, DC: Public Health Service, US Dept of Health and Human Services; 1988.
19. Vasan RS, Larson MG, Benjamin EJ, Evans JC, Reiss CK, Levy D. Congestive heartfailure in subjects with normal versus reduced left ventricular ejection fraction: prevalence and mortality in a population-based cohort. I Am Coll Cardiol. 1999;33:1948-1955.
20. Senni M, Rodeheffer RJ, Tribouilloy CM, etal. Use of echocardiography in the management of congestive heart failure in the community. I Am Coll Cardiol. 1999;33:164-170.
21. McKee PA, Castelli WP, McNamara PM, Kannel WB. The natural history of congestive heart failure. N Engl I Med. 1971;285:1441-1446.
22. Polanczyk CA, Rohde LE, Philbin EA, Di SalvoTG. A new casemix adjustment index for hospital mortality among patients with congestive heart failure. Med Care. 1998;36:1489-1499.
23. Selker HP, Griffith JL, D'Agostino RB. A time-insensitive predictive instrument for acute hospital mortality due to congestive heart failure. Med Care. 1994; 32:1040-1052.
24. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation J Chronic Dis. 1987;40:373-383.
25. Lemeshow S, Teres D, Klar J, Avrunin JS, Gehl-bach SH, Rapoport J. Mortality probability models based on an international cohort of intensive care unit patients. JAMA. 1993;270:2478-2486.
26. Lee DS, Tran C, Flintoft VF, Grant FC, Liu PP, Tu JV. CCORT/CCS quality indicators for congestive heart failure care. Can I Cardiol. 2003;19:357-364.
27. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159-174.
28. Hosmer DW, Lemeshow S. Applied Logistic Regression. 2nd ed. New York, NY: John Wiley & Sons Inc; 2000.
29. Steyerberg EW, Eijkemans MJ, Harrell FE Jr, Habbema JD. Prognostic modelling with logistic regression analysis. Stat Med. 2000;19:1059-1079.
30. Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models. StatMed. 1996;15:361-387.
31. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143:29-36.
32. Van Houwelingen JC, Le Cessie S. Predictive value of statistical models. StatMed. 1990;9:1303-1325.
33. Tu JV, Naylor CD. Clinical prediction rules. J Clin Epidemiol. 1997;50:743-744.
34. Moons KG, Harrell FE, Steyerberg EW. Should scoring rules be based on odds ratios or regression coefficients? I Clin Epidemiol. 2002;55:1054-1055.
35. FineMJ, AubleTE, Yealy DM, etal. A prediction rule to identify low-risk patients with community-acquired pneumonia. N Engl 1 Med. 1997;336:243-250.
36. Efron B, Tibshirani RJ. An Introduction to the Bootstrap. New York, NY: Chapman & Hall/CRC Press; 1998.
37. Steyerberg EW, Eijkemans MJ, Harrell FE Jr, Habbema JD. Prognostic modeling with logistic regression analysis: in search of a sensible strategy in small data sets. Med Decis Making. 2001 ;21:45-56.
38. Steyerberg EW, Harrell FE Jr, Borsboom GJ, et al. Internal validation of predictive models. J Clin Epidemiol. 2001 ;54:774-781.
39. Chin MH, Goldman L. Correlates of early hospital readmission or death in patients with congestive heart failure. Am J Cardiol. 1997;79:1640-1644.
40. Konstam M, Dracup K, BakerD, etal. Clinical Practice Cuideline No. 11: Heart Failure: Evaluation and Care of Patients With Left-Ventricular Systolic Dysfunction. Rockville, Md: Agency for Health Care Policy and Research; 2001. AHCPR No. 94-0612.
41. Graff L, Orledge J, Radford MJ, et al. Correlation of the Agency for Health Care Policy and Research congestive heartfailure admission guideline with mortality. Ann EmergMed. 1999;34(4 pt 1):429-437.
42. Goldberger JJ, Peled HB, Stroh JA, Cohen MN, Frishman WH. Prognostic factors in acute pulmonary edema. Arch Intern Med. 1986;146:489-493.
43. Masoudi FA, Havranek EP, Smith G, et al. Gender, age, and heart failure with preserved left ventricular systolic function. J Am Coll Cardiol. 2003;41 217-223.
44. Poses RM, McClish DK, Smith WR, et al. Results of report cards for patients with congestive heart failure depend on the method used to adjust for severity. Ann Intern Med. 2000;133:10-20.
45. Cleland JG.Cohen-SolalA, AguilarJC, etal. Management of heart failure in primary care (the improvement of heartfailure programme): an international survey. Lancet. 2002;360:1631-1639.
46. Eichhorn EJ. Prognosis determination in heartfailure. Am J Med. 2001 ;110(suppl 7A):14S-36S.
47. Hanratty B, Hibbert D, MairF, etal. Doctors' perceptions of palliative care for heart failure: focus group study. BMJ. 2002;325:581-585.
48. Wenrich MD, Curtis JR, Shannon SE, etal. Communicating with dying patients within the spectrum of medical care from terminal diagnosis to death. Arch Intern Med. 2001 ;161:868-874.
49. David AS, Mary R, Neil S, James G. Deaths from heart failure in general practice: implications for palliative care. PalliatMed. 2002;16:495-498.
50. Fried TR, Bradley EH. What matters to seriously ill older persons in making end-of-life treatment decisions: a qualitative study. J Palliat Med. 2003;6: 237-244.
51. Southard P, Frankel P. Traumacare documentation. I Emerg Nurs. 1989;15:393-398.
52. Aaronson KD, Schwartz JS, Chen T, Wong K, Goin JE, Mancini DM. Development and prospective validation of a clinical index to predict survival in ambulatory patients referred for cardiac transplant evaluation. Circulation. 1997;95:2660-2667.
53. Campana C, Gavazzi A, Berzuini C, et al. Predictors of prognosis in patients awaiting heart transplantation. I Heart Lung Transplant. 1993;12:756-765.
©2003 American Medical Association. All rights reserved. (Reprinted) JAMA, November 19, 2003—Vol 290, No. 19 2587
Downloaded from jama.ama-assn.org by guest on March 8, 2012