ORIGINAL INVESTIGATION

Factors Identified as Precipitating Hospital Admissions

for Heart Failure and Clinical Outcomes
Findings From OPTIMIZE-HF
Gregg C. Fonarow, MD; William T. Abraham, MD; Nancy M. Albert, RN, PhD; Wendy Gattis Stough, PharmD;
Mihai Gheorghiade, MD; Barry H. Greenberg, MD; Christopher M. O’Connor, MD; Karen Pieper, MS;
Jie Lena Sun, MS; Clyde W. Yancy, MD; James B. Young, MD; for the OPTIMIZE-HF Investigators and Hospitals

Background: Few studies have examined factors iden- process (15.3%), ischemia (14.7%), and arrhythmia
tified as contributing to heart failure (HF) hospitaliza- (13.5%) being most frequent. Pneumonia (odds ratio, 1.60),
tion, and, to our knowledge, none has explored their re- ischemia (1.20), and worsening renal function (1.48) were
lationship to length of stay and mortality. This study independently associated with higher in-hospital mortal-
evaluated the association between precipitating factors ity, whereas uncontrolled hypertension (0.74) was asso-
identified at the time of HF hospital admission and sub- ciated with lower in-hospital mortality. Ischemia (1.52)
sequent clinical outcomes. and worsening renal function (1.46) were associated with
a higher risk of follow-up mortality. Uncontrolled hyper-
Methods: During 2003 to 2004, 259 US hospitals in tension as a precipitating factor was associated with lower
OPTIMIZE-HF submitted data on 48 612 patients, with a postdischarge death/rehospitalization (hazard ratio, 0.71).
prespecified subgroup of at least 10% providing 60- to 90-
day follow-up data. Identifiable factors contributing to HF
Conclusions: Precipitating factors are frequently iden-
hospitalization were captured at admission and included
tified in patients hospitalized for HF and are associated
ischemia, arrhythmia, nonadherence to diet or medica-
tions, pneumonia/respiratory process, hypertension, and with clinical outcomes independent of other predictive
worsening renal function. Multivariate analyses were per- variables. Increased attention to these factors, many of
formed for length of stay, in-hospital mortality, 60- to 90- which are avoidable, is important in optimizing the man-
day follow-up mortality, and death/rehospitalization. agement of HF.

Results: Mean patient age was 73.1 years, 52% of pa- Trial Registration: clinicaltrials.gov Identifier:
tients were female, and mean ejection fraction was 39.0%. NCT00344513
Of 48 612 patients, 29 814 (61.3%) had 1 or more pre-
cipitating factors identified, with pneumonia/respiratory Arch Intern Med. 2008;168(8):847-854

H
EART FAILURE (HF) IS THE A number of factors have been identi-
leading cause of hospital- fied that may acutely exacerbate HF and
ization among US adults contribute to hospitalization for it. These
older than 65 years, and include arrhythmias, myocardial ische-
these hospitalizations mia, respiratory infection, uncontrolled hy-
contribute substantially to the high costs pertension, and nonadherence to medica-
of the disease. There are 3.6 million hos- tions and diet.3-9 However, relatively few
pitalizations with HF as the primary or a studies have examined the frequency at
secondary cause each year in the United which these factors are present among pa-
States.1,2 Hospitalizations for HF are also tients hospitalized for HF.5-9 Most avail-
associated with substantial morbidity and able data are limited by being obtained
Author Affiliations are listed at mortality; the likelihood of death and re- from relatively small numbers of patients
the end of this article. hospitalization is considerably greater than hospitalized at a single center or derived
Group Information: A list of for a comparable period of chronic but from observations of patients enrolled in
the Organized Program to stable HF.3,4 Understanding precipitants clinical trials, which have select enroll-
Initiate Lifesaving Treatment in that contribute to exacerbations of HF and ment criteria and closer monitoring than
Hospitalized Patients With
Heart Failure (OPTIMIZE-HF)
lead to HF admissions, particularly those under usual care settings.5-9 Analyses of
hospitals and investigators was that are avoidable, is of great importance large representative patient populations
published in JAMA. to clinicians and could favorably influ- from all regions of the country and all types
2007;297(1):61-70. ence HF disease management. of hospitals are critical in providing in-

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 8), APR 28, 2008 WWW.ARCHINTERNMED.COM
847

©2008 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by Okto Sofyan Hasan on 07/13/2022

 sight into the frequency of factors that precipitate HF hos- of the Web-based case report form. Data on factors contribut-
pitalizations. Furthermore, whether there is an associa- ing to exacerbation of the patients’ HF were prespecified and
tion between factors identified at the time of hospital collected as follows: ischemia/acute coronary syndromes (ACSs),
admission as contributing to HF exacerbation and sub- uncontrolled hypertension, pneumonia/respiratory process,
worsening renal function, arrhythmia, nonadherence–diet, non-
sequent clinical outcomes has not been previously stud-
adherence–medications, and other. More than 1 factor could
ied, to our knowledge. be selected if applicable. Admission staff, medical staff, or both
The Organized Program to Initiate Lifesaving Treat- recorded race/ethnicity, usually as the patient was registered.
ment in Hospitalized Patients With Heart Failure Previous studies in patients hospitalized for HF have sug-
(OPTIMIZE-HF) is a registry and performance- gested differences in characteristics and outcomes based on race/
improvement program for patients hospitalized for HF.10 ethnicity. The registry coordinating center was Outcome Sci-
The objectives of this analysis of OPTIMIZE-HF data were ences Inc (Cambridge, Massachusetts).
to determine the frequency at which various factors con-
tributing to HF hospitalization are identified and to im- STATISTICAL ANALYSIS
prove the understanding of whether and to what extent
these factors influence clinical outcomes, including hos- All statistical analyses were performed independently at the Duke
pital length of stay, in-hospital mortality, early postdis- Clinical Research Institute. The data are reported as mean and
charge mortality, and death/rehospitalization. standard deviation for continuous variables and as percentages
of nonmissing values for categorical variables. Patient character-
METHODS istics and evidence-based treatments at hospital discharge were
compared by Pearson ␹2 test for categorical variables and Wil-
coxon rank sum test for continuous variables. Multivariate mod-
OPTIMIZE-HF is a comprehensive hospital-based registry and els of in-hospital death, length of hospital stay, postdischarge mor-
process-of-case improvement program designed to provide op- tality, and postdischarge death or rehospitalization were developed
timal medical care and education to patients hospitalized for to be used for consistent covariate adjustment across all studies
HF. The OPTIMIZE-HF program has been described in detail as previously described.10-13 The types of models were logistic for
elsewhere10-13 and will be briefly summarized. in-hospital mortality, general linear modeling for length of stay,
Cox proportional hazards for postdischarge mortality, and logis-
PATIENT ELIGIBILITY tic for postdischarge mortality and rehospitalization (date of re-
hospitalization was not available for survival modeling). The
Hospital teams used HF case-ascertainment methods identical model-development process was similar for all 4 outcomes and
to those of The Joint Commission.14 Patients qualified for en- used stepwise and backward variable selection methods. The lin-
rollment if they were hospitalized for episodes of new or wors- earity assumption for continuous measures was evaluated by means
ening HF as the primary cause of admission or if significant of restricted cubic spline transformations. When needed, appro-
HF symptoms developed during hospitalization for another pri- priate transformations such as piecewise linear splines were ap-
mary diagnosis, with HF being the primary discharge diagno- plied. P⬍.05 was used for both entry and remaining in the model.
sis.10-13 Consecutive patients were enrolled irrespective of their The potential covariates were preselected, with 45 for in-
ventricular function, including systolic dysfunction docu- hospital mortality, 39 for length of stay, 19 for postdischarge mor-
mented by a left ventricular ejection fraction less than 40%, HF tality, and 70 for postdischarge mortality or rehospitalization
symptoms in the setting of preserved left ventricular systolic (posted at http://www.optimize-hf.org).10-13 To test the associa-
function (diastolic dysfunction HF), or HF without left ven- tion of precipitating factors and clinical outcomes in the final ad-
tricular function measurement.10-13 justed models, indicators for the presence or absence of each pre-
From March 1, 2003, to December 31, 2004, 48 612 patients cipitating factor were added to the previously developed models.
hospitalized at 259 centers in the United States were enrolled in Additional models were constructed to compare risk-adjusted out-
the OPTIMIZE-HF registry. All regions of the United States were comes for patients with no vs 1 or more precipitating factors iden-
represented, and institutions from community hospitals to large tified. SAS version 8.2 statistical software (SAS Institute Inc, Cary,
tertiary medical centers participated.10,11 A prespecified patient North Carolina) was used for all statistical analyses.
subgroup (targeted to be ⱖ10% of the total number) was fol-
lowed up for 60 to 90 days after discharge for the collection of RESULTS
outcomes data, as previously described.10 Sites had the option
of participating in the follow-up data collection, and the proto-
col was approved by each participating center’s institutional re- CLINICAL CHARACTERISTICS
view board or through use of a central institutional review board. OF PATIENTS HOSPITALIZED FOR HF
Ninety-one hospitals provided 60- to 90-day follow-up data, and
this cohort was demographically similar to patients in the over- The OPTIMIZE-HF enrolled a total of 48 612 patients hos-
all registry.10-13 Automated electronic data checks were used to pitalized for HF at 259 academic and community hospi-
prevent out-of-range entry or duplicate patients. A database au- tals of varying size from all regions of the United States.
dit was performed, on the basis of predetermined criteria, of a Mean patient age was 73.1 years; 52% of patients were
random sample of 5% of the first 10 000 patients verified against female and 74% were white (Table 1). Comorbidities
source documents.10,11 Written informed consent was obtained were frequent and included hypertension in 71% of pa-
before enrollment from patients who participated in the fol-
low-up data collection. Participants were screened for inclu-
tients, diabetes mellitus in 42%, and chronic obstruc-
sion before hospital discharge or identified from administrative tive pulmonary disease in 28%. The cause of HF was is-
databases subsequent to discharge. chemic in 46% of enrolled patients, and the mean left
The registry captured data on important characteristics ventricular ejection fraction was 39.0%. Of patients as-
(demographic, pathophysiologic, and clinical), treatment pat- sessed, 48.8% had documented left ventricular systolic
terns, and outcomes of patients hospitalized for HF by means dysfunction and 51.2% had HF with preserved systolic

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 8), APR 28, 2008 WWW.ARCHINTERNMED.COM
848

©2008 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by Okto Sofyan Hasan on 07/13/2022

 Table 1. Patient Characteristics and Comorbidities

Hospital Cohort Follow-up Cohort

(n = 48 612) (n = 5791)
Patient characteristics at admission
Age, mean (SD), y 73.1 (14.2) 72.0 (14.1)
Sex, No. (%) F 25 075 (51.6) 2826 (48.8)
Race, No. (%) 36 043 (74.1) 4526 (78.1)
White 8608 (17.7) 1044 (18.0)
Black 22 219 (45.7) 2435 (42.0)
Ischemic cause, No. (%) 11 135 (22.9) 1827 (31.5)
Hypertensive cause, No. (%) 7743 (15.9) 1049 (18.1)
Cigarette smoker within past year, No. (%) 5675 (11.7) 697 (12.0)
No known heart failure before admission, No. (%) 25 075 (51.6) 2826 (48.8)
LVSD, No. (%) a 20 118 (48.8) 2720 (53.2)
LVEF, mean (SD), % 39.0 (17.6) 36.9 (17)
Weight, mean (SD), kg 82.5 (26.4) 84.0 (26.1)
Blood pressure, mean (SD), mm Hg
Systolic 142.6 (33.2) 140.3 (32.8)
Diastolic 76.5 (19.9) 75.1 (19.7)
Heart rate, mean (SD), beats/min 86.6 (21.5) 85.7 (21.3)
BNP, median (25th-75th percentile), pg/mL 800.0 (403.0-1660.0) 813.0 (418.0-1710.0)
Troponin I, median (25th-75th percentile), ng/mL 0.10 (0.05-0.30) 0.20 (0.07-0.40)
Sodium, mean (SD), mEq/L 136.7 (11.1) 136.8 (9.2)
Serum creatinine, mean (SD), mg/dL 1.8 (1.8) 1.7 (1.4)
Hemoglobin, mean (SD), g/dL 12.1 (3.4) 12.2 (2.3)
Dyspnea at rest, No. (%) 21 279 (43.8) 2559 (44.2)
Dyspnea on exertion, No. (%) 29 856 (61.4) 3670 (63.4)
Rales, No. (%) 30 546 (64.0) 3522 (62.1)
Lower-extremity edema, No. (%) 30 710 (64.6) 3686 (65.1)
Jugular venous distention, No. (%) 13 725 (28.2) 1517 (26.1)
Orthopnea, No. (%) 13 298 (27.4) 2051 (35.4)
Patient comorbidities, No. (%)
Insulin-treated diabetes mellitus 8089 (16.6) 970 (16.8)
Non–insulin-treated diabetes mellitus 12 104 (24.9) 1499 (25.9)
Hypertension 34 479 (70.9) 4174 (72.1)
Atrial arrhythmia 14 970 (30.8) 1948 (33.6)
Ventricular arrhythmia 2681 (5.5) 473 (8.2)
Previous cerebrovascular accident or transient ischemic attack 7558 (15.5) 893 (15.4)
Hyperlipidemia 15 621 (32.1) 2300 (39.4)
Liver disease 791 (1.6) 126 (2.2)
Chronic renal insufficiency 9515 (19.6) 1186 (20.4)
Chronic obstructive pulmonary disease 13 395 (27.6) 1773 (30.6)
Peripheral vascular disease 6648 (13.6) 921 (15.9)
Anemia 8552 (17.6) 1289 (22.2)
Implantable cardioverter-defibrillator 2485 (5.1) 416 (7.2)
Cardiac resynchronization therapy 1599 (3.3) 277 (4.8)
Precipitating factors, No. (%)
Arrhythmia 7155 (13.5) 1090 (18.8)
Uncontrolled hypertension 5220 (10.7) 772 (13.3)
Ischemia/acute coronary syndromes 6552 (14.7) 1105 (19.1)
Worsening renal function 3304 (6.8) 509 (8.8)
Pneumonia/respiratory process 7426 (15.3) 1069 (18.5)
Nonadherence to medications 4309 (8.9) 602 (10.4)
Nonadherence to diet 2504 (5.2) 427 (7.4)
Other 6171 (12.7) 710 (12.3)
No. of precipitating factors
0 18 798 (38.7) 1818 (31.4)
1 20 504 (42.2) 2407 (41.6)
2 6599 (13.6) 1021 (17.6)
3 2050 (4.2) 384 (6.6)
ⱖ4 661 (1.4) 161 (2.8)

Abbreviations: BNP, B-type natriuretic peptide; LVEF, left ventricular ejection fraction; LVSD, left ventricular systolic dysfunction.
SI conversion factors: To convert creatinine to micromoles per liter, multiply by 88.4; hemoglobin to grams per liter, multiply by 10.0; sodium to millimoles per
liter, multiply by 1.0; and troponin I to micrograms per liter, multiply by 1.0.
a Denominators for the percentages are the numbers of patients in whom left ventricular function was assessed (41 267 and 5117, respectively).

function. The follow-up cohort included 5791 patients, One or more precipitating factors for HF admission were
whose characteristics were similar to those of the over- identified in 61.3% of patients. The frequencies of these in-
all registry (Table 1). dividual factors are shown in Table 1 for the hospital and

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 8), APR 28, 2008 WWW.ARCHINTERNMED.COM
849

©2008 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by Okto Sofyan Hasan on 07/13/2022

 10.0

9.0
8.0
8.0

7.0
5.8
Mortality, %

6.0

5.0 4.2 4.4

3.8 3.9
4.0

3.0
1.8 2.0
1.7
2.0

1.0

0
Total Cohort Ischemia /ACSs Arrhythmia Nonadherence Uncontrolled Nonadherence Pneumonia Worsening Other
to Diet Hypertension to Medications Renal Function

Figure. Unadjusted in-hospital mortality rates by precipitating factors for heart failure admission. ACSs indicates acute coronary syndromes.

Table 2. Precipitating Factors and Multivariate Risk-Adjusted In-Hospital Clinical Outcomes

In-Hospital Mortality

No. of Adjusted Length Adjusted Odds Ratio

Factor Patients of Stay Ratio P Value (95% Confidence Interval) P Value
Ischemia/acute coronary syndrome 7155 0.99 .22 1.20 (1.03-1.40) .02
Arrhythmia 6552 1.04 ⬍. 001 0.85 (0.71-1.01) .07
Nonadherence to diet 2504 0.96 .01 0.69 (0.48-1.00) .05
Uncontrolled hypertension 5220 0.96 ⬍. 001 0.74 (0.55-0.99) .04
Nonadherence to medications 4309 0.96 ⬍. 001 0.88 (0.67-1.17) .39
Pneumonia/respiratory process 7426 1.08 ⬍. 001 1.60 (1.38-1.85) ⬍.001
Worsening renal function 3304 1.09 ⬍. 001 1.48 (1.23-1.79) ⬍.001
Other 6171 0.99 .23 1.15 (0.97-1.36) .10

follow-up cohorts. The precipitating factors of pneumonia/ independently associated with in-hospital mortality in-
respiratory processes (15.3%), ischemia/ACSs (14.7%), ar- cluded patient age, admission systolic blood pressure and
rhythmia (13.5%), and uncontrolled hypertension (10.7%) heart rate, serum sodium and creatinine levels, and co-
were identified as the most common in the hospital co- morbidities such as chronic obstructive pulmonary dis-
hort. Nonadherence to medications was identified in 8.9% ease, liver disease, and peripheral vascular disease (data
and nonadherence to diet in 5.2% of patients. Two or more posted at http://www.optimize-hf.org). The median hos-
precipitating factors were identified in 9310 patients (19.2%) pital length of stay was 4.0 days (25th-75th interquar-
(Table 1). Specific precipitating factors were identified with tile range, 3.0-7.0) and mean length of stay was 6.4 days
greater frequency among the follow-up cohort. (SD, 85.2 days). Mean length of stay was shortest (5.1
days) in patients with uncontrolled hypertension and non-
IN-HOSPITAL OUTCOMES adherence to medications and longest in patients with
worsened renal function (6.9 days). In multivariate analy-
There were 1834 in-hospital deaths reported among sis, arrhythmia, pneumonia/respiratory process, and wors-
48 612 enrolled patients (3.8%). The in-hospital mortal- ening renal function were associated with significantly
ity rates (unadjusted) by precipitating factors are shown longer risk-adjusted length of stay, whereas uncon-
in the Figure. Patients in whom no precipitating factor trolled hypertension, nonadherence to diet, and nonad-
was identified were at modestly lower risk of in-hospital herence to medications were associated with shorter risk-
mortality than those with 1 or more precipitating fac- adjusted length of stay (Table 2).
tors risk (3.4% vs 4.0%; adjusted odds ratio, 0.88; 95%
confidence interval, 0.78-1.00; P = .046). Risk-adjusted 60- TO 90-DAY POSTDISCHARGE OUTCOMES
in-hospital mortality was significantly increased when ad-
mission was related to pneumonia/respiratory process, During the 60- to 90-day period after hospital dis-
ischemia/ACSs, or worsening renal function (Table 2). charge, the follow-up cohort experienced 465 deaths
Admission with uncontrolled hypertension or nonad- (8.3%), occurring a median of 42.0 days (25th-75th in-
herence to diet as a contributing factor was associated terquartile range, 24.0-66.0) after discharge. Rehospital-
with a lower risk-adjusted in-hospital mortality. There ization within the follow-up period occurred in 1715 pa-
were no significant interactions between the precipitat- tients (29.6%). The combined end point of death/
ing factors within the models. Other covariates that were rehospitalization was met in 36.0% of patients. The rates

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 8), APR 28, 2008 WWW.ARCHINTERNMED.COM
850

©2008 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by Okto Sofyan Hasan on 07/13/2022

 of death and death/rehospitalization for patients with each
of the precipitating factors are shown in Table 3. Pa- Table 3. Precipitating Factors and 60- to 90-Day Unadjusted
tients without a precipitating factor identified had risk- Postdischarge Outcomes
adjusted postdischarge mortality similar to that of pa-
No. of Follow-up Follow-up Death/
tients with 1 or more precipitating factors (6.9% vs 8.9%; Factor Patients Mortality (%) Rehospitalization (%)
adjusted odds ratio, 0.84; 95% confidence interval, 0.66-
Overall follow-up 5791 465/5616 (8.3) 2033/5610 (36.2)
1.07; P=.15). Risk-adjusted postdischarge mortality was cohort
significantly increased in patients whose admission was Ischemia/acute 1105 124/1043 (11.9) 395/1043 (37.9)
precipitated by ischemia/ACSs or worsening renal func- coronary syndrome
tion, independent of other predictive variables (Table 4; Arrhythmia 1090 87/1055 (8.3) 353/1053 (33.5)
additional data posted at http://www.optimize-hf.org). Nonadherence to diet 427 27/417 (6.5) 146/417 (35.0)
Among patients with uncontrolled hypertension, there Uncontrolled 772 35/749 (4.7) 209/747 (28.0)
hypertension
was significantly decreased risk-adjusted in-hospital mor- Nonadherence to 602 42/589 (7.1) 206/588 (35.0)
tality and significantly decreased mortality or readmis- medications
sion at 60 to 90 days after discharge. None of the other Pneumonia/respiratory 1069 108/1017 (10.6) 394/1015 (38.8)
precipitating factors were associated with significantly process
higher or lower postdischarge death/rehospitalization. Worsening 509 86/472 (18.2) 212/474 (44.7)
renal function
Other 710 73/686 (10.6) 251/688 (36.5)
COMMENT

The OPTIMIZE-HF has demonstrated that, among a large, blood pressure control with relatively short length of stay
representative population of patients admitted to the hos- and lower risk of adverse near-term outcomes.11
pital for HF, 1 or more exacerbating factors contributing Patient adherence to dietary restrictions and evidence-
to HF hospitalization were identified in most patients. The based medications is a cornerstone of HF disease manage-
contributing factors of pneumonia/respiratory processes ment, and nonadherence to medications has been associ-
(15.3%), ischemia (14.7%), arrhythmia (13.5%), and un- ated with increased risk of hospitalization and mortality
controlled hypertension (10.7%) were identified the most in outpatients with chronic HF.2,3,18 Patients with nonad-
frequently. Certain of these precipitating factors were in- herence to medications or diet are likely to be admitted with
dependently associated with worse clinical outcomes. Pneu- excessive sodium retention, which was the leading decom-
monia/respiratory processes and worsening renal func- pensation factor in 55% of 975 patients in a retrospective
tion as precipitating factors for hospitalization identified audit at 2 large midwestern medical centers.19 These pa-
patients at significantly increased risk of greater length of tients may more readily achieve compensation in re-
stay and in-hospital mortality. Ischemia and worsening re- sponse to salt restriction, adjustment of diuretics, and pro-
nal function as precipitating factors for admission were as- vision of medications during the inpatient hospitalization.
sociated with increased risk of mortality at 60 to 90 days It should be noted that patients with nonadherence to medi-
after discharge. Nonadherence to medications, nonadher- cations or diet as an admission precipitant were at high ad-
ence to diet, and uncontrolled hypertension each were as- justed risk of 60- to 90-day postdischarge mortality and
sociated with shorter stay and lower in-hospital mortality. death/rehospitalization similar to the overall HF popula-
These findings provide important insights into the fac- tion. Patients identified as nonadherent to medications
tors that contribute to admission for HF and their influ- would be expected to be counseled during the index hos-
ence on subsequent outcomes. pitalization regarding the importance of adherence to their
Worsening of renal function during hospitalization for medical regimen and thus may be less likely, at least in the
HF has previously been identified as being associated with short term, to repeat the medication nonadherence that pre-
worse outcomes.15,16 The OPTIMIZE-HF data further ex- cipitated a recent HF hospitalization. It should also be em-
tend those findings and demonstrate that HF admission phasized that the use of evidence-based HF medications
precipitated by worsening renal function is also associ- in eligible patients at discharge is strongly associated with
ated with significantly worse patient outcomes, both in- improved postdischarge outcomes.12
hospital and after discharge. The finding that pneumo- Previous studies have assessed the frequency at which
nia or another respiratory process as an admission precipitating factors are present among hospitalized pa-
precipitant is associated with increased mortality is con- tients with HF.5-9 A single-center study using retrospec-
sistent with previous studies of administrative data- tive chart review reported on causal factors among 435 pa-
bases showing that patients hospitalized for HF and tients hospitalized for HF.5 The most commonly identified
chronic obstructive pulmonary disease or pneumonia have factors for HF exacerbations leading to hospitalization in
higher mortality risk.17 Improved in-hospital and post- that study were acute chest pain in 33% of patients, res-
discharge prognosis among patients with HF admitted piratory tract infection in 16%, uncontrolled hyperten-
with highly elevated systolic blood pressure has been pre- sion in 15%, and nonadherence to medications in 15%.
viously reported and parallels the finding that uncon- An analysis of 328 HF hospitalizations from 161 unique
trolled hypertension as a precipitant of HF admission is patients referred to a hospital HF service for transplanta-
associated with better outcomes.11 Patients with decom- tion in Italy showed the most common potential caus-
pensation of HF resulting from uncontrolled hyperten- ative factors to be the presence of arrhythmias in 24% of
sion can usually be readily stabilized in the hospital with hospitalizations, infections in 23%, poor adherence in 15%,

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 8), APR 28, 2008 WWW.ARCHINTERNMED.COM
851

©2008 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by Okto Sofyan Hasan on 07/13/2022

 Table 4. Precipitating Factors and 60- to 90-Day Multivariate Risk-Adjusted Postdischarge Outcomes

Follow-up Mortality Follow-up Death/Rehospitalization

Adjusted Odds Ratio Adjusted Odds Ratio

Factor (95% Confidence Interval) P Value (95% Confidence Interval) P Value
Ischemia/acute coronary syndrome 1.52 (1.20-1.93) ⬍ .001 1.06 (0.90-1.25) .49
Arrhythmia 0.76 (0.57-1.02) .06 0.85 (0.72-1.01) .06
Nonadherence to diet 0.81 (0.51-1.29) .37 0.94 (0.73-1.21) .62
Uncontrolled hypertension 0.61 (0.40-0.95) .03 0.71 (0.58-0.88) .002
Nonadherence to medications 1.10 (0.75-1.61) .63 1.03 (0.82-1.29) .80
Pneumonia/respiratory process 1.25 (0.96-1.62) .10 1.02 (0.86-1.21) .80
Worsening renal function 1.46 (1.06-2.00) .02 1.01 (0.79-1.30) .91
Other 1.46 (1.09-1.96) .01 0.97 (0.80-1.18) .79

and angina in 14%.6 Study of 179 consecutive patients ad- comes independent of other prognostic variables. Un-
mitted to a teaching hospital in Germany identified di- derstanding whether and to what extent precipitants of
etary sodium excess in 43% of patients, nonadherence to HF hospitalization influence length of stay, mortality, and
medications in 24%, ischemia in 13%, and uncontrolled rehospitalization risk is important because this knowl-
hypertension in 8%.7 Another single-center study re- edge may help guide clinicians in designing more effec-
ported on precipitating factors leading to decompensa- tive management strategies for hospitalized patients with
tion of HF in 101 patients of low socioeconomic status using HF and to prevent HF hospitalizations.3,20
systematic patient interviews and medical record re- National HF guidelines recommend that patients hos-
view.8 The most common precipitating factors identified pitalized for HF undergo evaluation for precipitating fac-
were lack of adherence to a low-sodium diet, medica- tors and suggest that proper detection and treatment of
tions, or both in 64% of patients. Uncontrolled hyperten- precipitating factors is an important part of the manage-
sion was an identified cause in 44% and cardiac arrhyth- ment of acute decompensated HF.3 These recommenda-
mias in 29% of patients. A multicenter study of HF tions were level of evidence C, expert opinion only. These
precipitating factors involving 768 patients with systolic OPTIMIZE-HF data lend further support to these recom-
HF enrolled in the Randomized Evaluation of Strategies mendations and provide data demonstrating that certain
for Left Ventricular Dysfunction Pilot Study included a total precipitating factors are associated with clinical out-
of 323 episodes of worsening of HF in 180 patients, whether comes independent of other established prognostic fac-
resulting in an HF hospitalization or not, during 43 weeks tors. Patients identified as being at higher risk of adverse
of follow-up.9 Factors implicated in worsening of HF sta- outcomes may benefit from closer monitoring during hos-
tus in that study included nonadherence to salt restric- pitalization and more frequent follow-up after discharge.
tion (22% of patients); pulmonary infections (20%); use Several of these precipitating factors, including nonad-
of antiarrhythmic agents (15%), arrhythmias (13%), and herence to diet and medications, may be influenced by op-
calcium-channel blockers (13%); and inappropriate re- timizing patient education techniques and disease man-
ductions in HF therapy (10%). agement strategies.2,3,20 Because pneumonia/respiratory
Each of these studies had 1 or more major limitations, process was the most common precipitating factor and was
including retrospective nature of the data collection, lim- associated with worse outcomes, every effort should be
ited number of patients studied, involvement of a single made to prevent pneumonia in patients with HF, includ-
center, referral of patients to a specialty service or enroll- ing rigorous influenza and pneumococcal vaccination.3 Risk
ment of select patients in a double-blind trial of systolic of ischemia and ACSs may be reduced with antiplatelet
HF therapy, and inability to define whether the precipi- agents, statin therapy, and, possibly, revascularization in
tant was a cause or an effect of HF exacerbation.9 Identi- eligible patients.2,3 Disease management programs and treat-
fying precipitants of HF exacerbation within the context ment plans for patients with HF should include appropri-
of a clinical trial raises the issue of patient selection bias ate strategies for these concomitant conditions, and ex-
because these patients are more likely to adhere to medi- acerbation of these conditions should be avoided to the
cal advice and receive closer follow-up than in a usual care extent possible.2,3 Future studies should be designed to pro-
setting. Thus, the patients in these previous studies may spectively test interventions targeting these contributing
not represent the general hospitalized HF population, lim- factors in this high-risk HF patient population.
iting the applicability of the findings. This analysis of OPTIMIZE-HF data may be influ-
The strengths of the present study include that it was enced by several limitations. Precipitating factors of HF were
performed with the use of a systematic approach to iden- abstracted from documentation in the medical record and
tify factors contributing to HF hospitalization and was may have been underreported. It is known that patient in-
conducted in 259 US hospitals from all regions of the terview or clinician impression may be insufficient to de-
country with a well-defined cohort of patients.10-13 An- tect poor adherence to medications or diet.21 Follow-up data
other important feature of the present study is that it is were obtained in a subset of patients and were limited to
the first, to our knowledge, that assessed the relation- 60 to 90 days. Participation in OPTIMIZE-HF was volun-
ship between precipitating factors and clinical out- tary, and sites received payment to defray costs associated

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 8), APR 28, 2008 WWW.ARCHINTERNMED.COM
852

©2008 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by Okto Sofyan Hasan on 07/13/2022

 with data collection. These findings may not apply to hos- Author Contributions: Dr Fonarow had full access to all
pitals that differ from OPTIMIZE-HF hospitals in patient the data in the study and takes responsibility for the in-
characteristics or care patterns, although a recent study22 tegrity of the data and the accuracy of the data analysis.
suggests that Medicare patients enrolled in OPTIMIZE-HF Study concept and design: Fonarow, Abraham, Albert,
had demographics similar to those of Medicare patients hos- Stough, Greenberg, and O’Connor. Acquisition of data:
pitalized for HF in the nation as a whole. Given the over- Fonarow, Abraham, Albert, Greenberg, and Young. Analy-
all large number of patients observed, some differences, sis and interpretation of data: Fonarow, Abraham, Albert,
though statistically significant, may not be clinically rel- Gheorghiade, Greenberg, O’Connor, Pieper, Sun, Yancy,
evant. Also, despite multivariate analyses, we cannot ex- and Young. Drafting of the manuscript: Fonarow and
clude that residual measured and unmeasured confound- Greenberg. Critical revision of the manuscript for impor-
ing accounts for some of these observations. Despite these tant intellectual content: Fonarow, Abraham, Albert,
limitations, this analysis provides new insights into the fac- Stough, Gheorghiade, Greenberg, O’Connor, Pieper, Sun,
tors contributing to HF hospitalizations from a large rep- Yancy, and Young. Statistical analysis: Fonarow, Pieper,
resentative data set of patients hospitalized for HF from all and Sun. Obtained funding: Fonarow. Administrative, tech-
regions of the country and including patients with pre- nical, and material support: Fonarow and Abraham. Study
served systolic function and multiple comorbidities. supervision: Fonarow, Greenberg, and O’Connor.
Financial Disclosure: Dr Fonarow has received research
CONCLUSIONS grants from Amgen, Biosite, Bristol-Myers Squibb, Boston
Scientific/Guidant, GlaxoSmithKline, Medtronic, Merck &
Deterioration of clinical status leading to HF hospitaliza- Co, Pfizer, Sanofi-Aventis, Scios Inc, and the National In-
tions is frequently accompanied by identifiable factors that stitutes of Health (NIH). He is or has been on the speak-
contribute to decompensations beyond the underlying HF ers’ bureau or has received honoraria in the past 5 years
disease state. Among patients admitted for an HF hospi- from Amgen, AstraZeneca, Biosite, Bristol-Myers Squibb,
talization in OPTIMIZE-HF, close to two-thirds of pa- Boston Scientific/Guidant, GlaxoSmithKline, Kos,
tients had 1 or more precipitating factors identified. The Medtronic, Merck & Co, NitroMed, Pfizer, Sanofi-
most common factors identified as precipitants of HF ex- Aventis, Schering-Plough, Scios Inc, St Jude Medical, Tak-
acerbations necessitating hospitalization include arrhyth- eda, and Wyeth. He is or has been a consultant for Biosite,
mia, ischemia, pneumonia/respiratory process, and non- Bristol-Myers Squibb, Boston Scientific/Guidant,
adherence to diet and medications. Furthermore, factors GlaxoSmithKline, Medtronic, Merck & Co, NitroMed, Or-
contributing to HF admission identify patients at higher qis Medical, Pfizer, Sanofi-Aventis, Schering-Plough, Scios
and lower risk of in-hospital and postdischarge adverse out- Inc, and Wyeth. Dr Abraham has received research grants
comes, independent of other predictive variables. In- from Amgen, Biotronik, CHF Solutions, GlaxoSmithKline,
creased attention to these factors, many of which are avoid- Heart Failure Society of America, Medtronic, Myogen, the
able, is important in optimizing the management of HF. NIH, Orqis Medical, Otsuka Maryland Research Institute,
Paracor, and Scios Inc. He is or has been a consultant or
on the speakers’ bureau for Amgen, AstraZeneca, Boe-
Accepted for Publication: November 11, 2007. hringer-Ingelheim, CHF Solutions, GlaxoSmithKline,
Author Affiliations: Department of Medicine, UCLA [Uni- Guidant Corp, Medtronic, Merck & Co, Pfizer, ResMed,
versity of California, Los Angeles] Medical Center (Dr Respironics, Scios Inc, and St Jude Medical. He is on the
Fonarow); Division of Cardiology, Ohio State University, advisory board of CardioKine, CardioKinetix Inc, CHF So-
Columbus (Dr Abraham); George M. and Linda H. lutions, Department of Veterans Affairs Cooperative Stud-
Kaufman Center for Heart Failure (Dr Albert) and De- ies Program, Inovise, the NIH, and Savacor Inc. He has re-
partment of Cardiovascular Medicine, Heart Failure Sec- ceived honoraria from AstraZeneca, Boehringer-
tion (Dr Young), Cleveland Clinic Foundation, Cleve- Ingelheim, GlaxoSmithKline, Guidant Corp, Medtronic,
land, Ohio; Department of Medicine (Dr Stough) and Merck & Co, Pfizer, ResMed, Respironics, Scios Inc,
Division of Cardiology (Dr O’Connor), Duke University and St Jude Medical. Dr Albert is a consultant for
Medical Center, Durham, North Carolina; Department of GlaxoSmithKline and Medtronic. She is also on the speak-
Clinical Research, Campbell University School of Phar- ers’ bureau for GlaxoSmithKline, Medtronic, NitroMed, and
macy, Research Triangle Park, North Carolina (Dr Stough); Scios Inc and is employed by the Cleveland Clinic Foun-
Division of Cardiology, Feinberg School of Medicine, dation. Dr Stough has received research grants from Acte-
Northwestern University, Chicago, Illinois (Dr lion, GlaxoSmithKline, Medtronic, Otsuka, and Pfizer. She
Gheorghiade); Department of Medicine, University of Cali- is a consultant or on the speakers’ bureau for Abbott, As-
fornia San Diego Medical Center (Dr Greenberg); Duke traZeneca, GlaxoSmithKline, Medtronic, Novacardia,
Clinical Research Institute, Durham (Dr O’Connor and Mss Otsuka, Protein Design Labs, RenaMed, Sigma Tau, and
Pieper and Sun); and Department of Medicine, The Uni- Scios Inc. She has received honoraria from Abbott, Astra-
versity of Texas Southwestern Medical Center, Dallas (Dr Zeneca, GlaxoSmithKline, Medtronic, and Pfizer. Dr
Yancy). Dr Yancy is now with Baylor Heart and Vascular Gheorghiade has received research grants from the NIH,
Institute, Baylor University Medical Center, Dallas. Otsuka, Sigma Tau, Merck & Co, and Scios Inc. He is or
Correspondence: Gregg C. Fonarow, MD, Ahmanson- has been a consultant for Debbio Pharm, Errekappa Tera-
UCLA Cardiomyopathy Center, UCLA Medical Center, peutici, GlaxoSmithKline, PDL, and Medtronic. He has
10833 LeConte Ave, Room 47-123 CHS, Los Angeles, CA received honoraria from Abbott, AstraZeneca,
90095-1679 ([email protected]). GlaxoSmithKline, Medtronic, Otsuka, Protein Design Lab,

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 8), APR 28, 2008 WWW.ARCHINTERNMED.COM
853

©2008 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by Okto Sofyan Hasan on 07/13/2022

 Scios Inc, and Sigma Tau. Dr Greenberg has received re- tistics—2007 update: a report from the American Heart Association Statistics
Committee and Stroke Statistics Subcommittee [published correction appears
search grant support from Amgen, Cardiodynamics,
in Circulation. 2007; February 6 115(5):e172]. Circulation. 2007;115(5):e69-
GlaxoSmithKline, Millennium, Novacardia, Otsuka, Pfizer, e171. doi:10.1161/CIRCULATIONAHA.106.179918.
Sanofi-Aventis, and Titan. He is on the speaker’s bureau/ 2. Hunt SA, Abraham WT, Chin MH, et al; American College of Cardiology/
consultant for Amgen, AstraZeneca, GlaxoSmithKline, American Heart Association. ACC/AHA 2005 guideline update for the diagnosis
Guidant Corp, Medtronic, Merck & Co, NitroMed, Pfizer, and management of chronic heart failure in the adult: a report of the American
Remon Medical Technologies, and Scios Inc. He is an ad- College of Cardiology/American Heart Association Task Force on Practice Guide-
lines (Writing Committee to Update the 2001 Guidelines for the Evaluation and
visory board member for CHF Solutions, GlaxoSmithKline, Management of Heart Failure). J Am Coll Cardiol. 2001;38(7):2101-2113.
and NitroMed. He has received honoraria from AstraZen- 3. Heart Failure Society of America. HFSA 2006 comprehensive heart failure prac-
eca, GlaxoSmithKline, Medtronic, Merck & Co, Ni- tice guideline. J Card Fail. 2006;12(1):e1-e122. doi:10.1016/j.cardfail.2005.11.005.
troMed, Novartis, Pfizer, and Scios Inc. Dr O’Connor has 4. Gheorghiade M, Zannad F, Sopko G, et al; International Working Group on Acute
received research grant support from the NIH. He is on the Heart Failure Syndromes. Acute heart failure syndromes: current state and frame-
work for future research. Circulation. 2005;112(25):3958-3968.
speakers’ bureau and/or a consultant for Amgen, Astra-
5. Chin MH, Goldman L. Factors contributing to the hospitalization of patients with
Zeneca, Bristol-Myers Squibb, GlaxoSmithKline, Guidant congestive heart failure. Am J Public Health. 1997;87(4):643-648.
Corp, Medtronic, Merck & Co, NitroMed, Novartis, Otsuka, 6. Opasich C, Febo O, Riccardi G, et al. Concomitant factors of decompensation in
Pfizer, and Scios Inc. He has received honoraria from chronic heart failure. Am J Cardiol. 1996;78(3):354-357.
GlaxoSmithKline, Pfizer, and Otsuka. Mss Pieper and Sun 7. Michalsen A, Konig G, Thimme W. Preventable causative factors leading to hos-
are employees of the Duke Clinical Research Institute. Dr pital admission with decompensated heart failure. Heart. 1998;80(5):437-441.
8. Ghali JK, Kadakia S, Cooper R, Ferlinz J. Precipitating factors leading to decom-
Yancy has received research grants from Cardiodynamics, pensation of heart failure: traits among urban blacks. Arch Intern Med. 1988;
GlaxoSmithKline, Scios Inc, Medtronic, and NitroMed. He 148(9):2013-2016.
is also a consultant or on the speakers’ bureau for Astra- 9. Tsuyuki RT, McKelvie RS, Amold JMO, et al. Acute precipitants of congestive
Zeneca, Cardiodynamics, GlaxoSmithKline, Medtronic, Ni- heart failure exacerbations. Arch Intern Med. 2001;161(19):2337-2342.
troMed, Novartis, and Scios Inc. He is on the advisory board 10. Fonarow GC, Abraham WT, Albert NM, et al. Organized Program to Initiate Life-
saving Treatment in Hospitalized Patients with Heart Failure (OPTIMIZE-HF): ra-
for CHF Solutions, the Food and Drug Administration car-
tionale and design. Am Heart J. 2004;148(1):43-51.
diovascular device panel, and the NIH. He has received 11. Gheorghiade M, Abraham WT, Albert NM, et al; OPTIMIZE-HF Investigators and
honoraria from AstraZeneca, Cardiodynamics, Coordinators. Systolic blood pressure at admission, clinical characteristics, and
GlaxoSmithKline, Medtronic, Novartis, and Scios Inc. Dr outcomes in patients hospitalized with acute heart failure. JAMA. 2006;296
Young has received research grants from Abbott, Acorn, (18):2217-2226.
Amgen, Artesion Therapeutics, AstraZeneca, Biosite, 12. Fonarow GC, Abraham WT, Albert NM, et al; OPTIMIZE-HF Investigators and Hos-
pitals. Association between performance measures and clinical outcomes for pa-
GlaxoSmithKline, Guidant Corp, Medtronic, MicroMed, tients hospitalized with heart failure. JAMA. 2007;297(1):61-70.
the NIH, Scios Inc, Vasogen, and World Heart. He is a con- 13. Fonarow GC, Abraham WT, Albert NM, et al; OPTIMIZE-HF Investigators and Hos-
sultant for Abbott, Acorn, Amgen, Biomax Canada, Bi- pitals. Influence of a performance-improvement initiative on quality of care for
osite, Boehringer-Ingelheim, Bristol-Myers Squibb, Coth- patients hospitalized with heart failure: results of the Organized Program to Ini-
erix, Edwards Lifescience, GlaxoSmithKline, Guidant Corp, tiate Lifesaving Treatment in Hospitalized Patients with Heart Failure (OPTIMIZE-HF).
Arch Intern Med. 2007;167(14):1493-1502.
Medtronic, MicroMed, Novartis, Paracor, Proctor & 14. Joint Commission on Accreditation of Healthcare Organizations. Specification
Gamble, Protemix, Scios Inc, Sunshine, Thoratec, Trans- Manual for National Implementation of Hospital Core Measures. Oakbrook Ter-
world Medical Corp, Vasogen, Viacor, and World Heart. race, IL: Joint Commission on Accreditation of Healthcare Organizations; 2004.
Funding/Support: GlaxoSmithKline funded the 15. Forman DE, Butler J, Wang Y, et al. Incidence, predictors at admission, and im-
OPTIMIZE-HF registry under the guidance of the pact of worsening renal function among patients hospitalized with heart failure.
J Am Coll Cardiol. 2004;43(1):61-67.
OPTIMIZE-HF Steering Committee. 16. Gottlieb SS, Abraham W, Butler J, et al. The prognostic importance of different
Role of the Sponsor: GlaxoSmithKline was involved in definitions of worsening renal function in congestive heart failure. J Card Fail.
the design and conduct of the OPTIMIZE-HF registry and 2002;8(3):136-141.
funded data collection and management through Out- 17. Holguin F, Folch E, Redd SC, Mannino DM. Comorbidity and mortality in COPD-
come Inc, data management and statistical analyses through related hospitalizations in the United States, 1979 to 2001. Chest. 2005;128
(4):2005-2011.
the Duke Clinical Research Institute, and administrative
18. Granger BB, Swedberg K, Ekman I, et al; CHARM investigators. Adherence to
and material support by Accel Health. The sponsor was not candesartan and placebo and outcomes in chronic heart failure in the CHARM
involved in the management, analysis, or interpretation of programme: double-blind, randomized, controlled clinical trial. Lancet. 2005;
data or the preparation of the manuscript. GlaxoSmithKline 366(9502):2005-2011.
did review the manuscript before submission. 19. Bennett SJ, Huster GA, Baker SL, et al. Characterization of the precipitants of
hospitalization for heart failure decompensation. Am J Crit Care. 1998;7(3):
Additional Information: All data analyses were per-
168-174.
formed independently at the Duke Clinical Research In- 20. McAlister FA, Lawson FM, Teo KK, Armstrong PW. A systematic review of ran-
stitute by Mss Pieper and Sun, with the input of the domized trials of disease management programs in heart failure. Am J Med. 2001;
OPTIMIZE-HF Steering Committee. 110(5):378-384.
21. Haynes RB, Taylor DW, Sackett DL, Gibson ES, Bernholz CD, Mukherjee J.
Can simple clinical measurements detect patient noncompliance? Hypertension.
REFERENCES 1980;2(6):757-764.
22. Curtis LH, Hernandez AF, Greiner MA, et al. Validity of a national heart failure
1. Rosamond W, Flegal K, Friday G, et al; American Heart Association Statistics quality of care registry: comparison of Medicare patients in OPTIMIZE-HF vs.
Committee and Stroke Statistics Subcommittee. Heart disease and stroke sta- non-OPTIMIZE-HF hospitals [abstract]. Circulation. 2007;115(21):595.

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 8), APR 28, 2008 WWW.ARCHINTERNMED.COM
854

©2008 American Medical Association. All rights reserved.

Downloaded From: https://jamanetwork.com/ by Okto Sofyan Hasan on 07/13/2022