Clinical Outcomes in Patients on Beta-Blocker Therapy Admitted WithWorsening Chronic Heart Failure

Wendy A. Gattis, PharmD, Christopher M. O’Connor, MD, Jeffrey D. Leimberger, PhD,G. Michael Felker, MD, Kirkwood F. Adams, Jr., MD, and Mihai Gheorghiade, MD

Beta blockers have been shown to reduce morbidity andmortality in patients with heart failure without evidenceof overt congestion. No data are available describingoutcomes of patients admitted with exacerbated chronicheart failure who are receiving � blockade at the time ofadmission. The purpose of this analysis was to evaluateclinical outcomes in patients from the Outcomes of theProspective Trial of Intravenous Milrinone for Exacerba-tions of Chronic Heart Failure (OPTIME-CHF) study whowere prescribed � blockers on admission comparedwith patients who were not prescribed � blockers atadmission. In all, 212 patients were treated with �blockers at admission and 737 patients were not. Base-line characteristics were similar between groups, exceptthat patients prescribed � blockers on admission hadslightly higher ejection fractions, fewer New York Heart

Association class IV symptoms, and lower heart rates.There was no difference in clinical events between pa-tients who were treated with � blockers at the time ofadmission and those who were not. Exploratory analy-ses suggested that patients whose �-blocker therapywas discontinued had a higher risk of adverse out-comes, particularly in the subset of patients randomizedto milrinone. The data from this nonrandom comparisonsuggest that continuation of pre-existing �-blocker ther-apy is not associated with an increased risk of adverseclinical events in patients admitted with worsening heartfailure. These results also suggest that caution should betaken when withdrawing � blockade in thispopulation. �2003 by Excerpta Medica, Inc.

(Am J Cardiol 2003;91:169–174)

Beta blockers are considered standard of care in thetreatment of chronic heart failure.1–4 However,

few data are available to evaluate the safety and effi-cacy of� blockers in patients with worsening symp-toms of congestion. Clinical outcomes are unknownfor patients treated with long-term�-blocker therapywho are admitted to the hospital for exacerbatedchronic heart failure. It remains unproven whethersuch patients tolerate continued�-blocker therapy orif �-blocker therapy should be decreased or tempo-rarily discontinued.

METHODSOutcomes of a Prospective Trial of Intravenous

Milrinone for Exacerbations of Chronic Heart Failure(OPTIME-CHF) was a multicenter, randomized, dou-ble-blind controlled trial of milrinone versus placeboin patients hospitalized for exacerbated chronic heartfailure.5 The present analysis was conducted using theOPTIME-CHF database to determine clinical out-comes in patients treated with� blockers versus thosenot treated with� blockers at the time of hospital

admission. Secondary objectives were to explore clin-ical outcomes for patients treated with� blockerscompared with those not treated with� blockerswithin each OPTIME-CHF treatment group (milri-none vs placebo), and to conduct exploratory analysesamong the subgroups of patients who discontinued�-blocker therapy during hospitalization comparedwith those in whom therapy was continuous.

The design and results of the OPTIME-CHF trialhave been published.5,6 Briefly, 949 patients with sys-tolic dysfunction and heart failure exacerbation wererandomized to receive 48 to 72 hours of intravenousmilrinone therapy (0.5�g/kg/min without a loadingdose) or placebo. Patients requiring inotropic therapyand those with evidence of active ischemia within theprior 3 months, severe renal impairment, or unstablearrhythmias were excluded. Background therapy wasleft to the discretion of the treating physician, butspecific recommendations for optimal pharmacother-apy based on current guidelines were included withthe study protocol. These guidelines included a rec-ommendation to avoid�-blockade initiation duringthe period of acute decompensation, but made nospecific recommendations regarding continuation of�blockade therapy in patients already receiving�-blocker therapy. The primary end point was dayshospitalized from cardiovascular causes within 60days of randomization.

The current analysis compared patients receiving�-blocker therapy at the time of the index hospitaliza-tion with those not receiving such therapy. End pointsfor this analysis were the same as those in the overall

From Duke University Medical Center/Duke Clinical Research Institute,Division of Cardiology, Durham, North Carolina; University of NorthCarolina, Chapel Hill, North Carolina; and Northwestern University,Division of Cardiology, Chicago, Illinois. The main OPTIME-CHF trialwas funded by Sanofi, Inc., New York, New York. This analysis wasfunded by the Duke Clinical Research Institute, Durham, NorthCarolina. Manuscript received June 25, 2002; revised manuscriptreceived and accepted August 28, 2002.

Address for reprints: Wendy Gattis, PharmD, Duke University Medi-cal Center/Duke Clinical Research Institute, 2400 North Pratt Street,Durham, North Carolina 27705. E-mail: wendy.gattis@duke.edu.

169©2003 by Excerpta Medica, Inc. All rights reserved. 0002-9149/03/$–see front matterThe American Journal of Cardiology Vol. 91 January 15, 2003 PII S0002-9149(02)03104-1

OPTIME-CHF study, including days hospitalized forcardiovascular causes within 60 days, death at 60days, the composite of death or hospitalization at 60days, and the composite of death or any serious ad-verse events at 60 days.

Continuous variables are presented as median val-ues with 25th and 75th percentiles. Categorical dataare presented as frequencies and percentages. Allanalyses were performed based on the intention-to-treat principle. Continuous variables were comparedbetween groups using the Wilcoxon signed-rank test,and categorical data were compared using the likeli-hood ratio chi-square test. Statistical modeling wasperformed using the Cox proportional-hazards model.The following variables were included in the model:age, gender, race, ischemic etiology, randomization tomilrinone, �-blocker use, New York Heart Associa-tion (NYHA) class, heart rate, systolic blood pressure,S3 gallop, history of angina, history of myocardialinfarction, history of arrhythmia, history of diabetes,history of chronic pulmonary disease, time from hos-pitalization to OPTIME-CHF randomization, history

of stroke, history of transient ischemic attack, auto-matic implantable cardiac defibrillator, coronary ar-tery bypass grafting, percutaneous coronary interven-tion, Batista procedure, left ventricular assist device,hepatic disease, cholesterol, hypertension, smoking,respiratory rate, hepatojugular reflux, rales, peripheraledema, ejection fraction, sodium, blood urea nitrogen,and serum creatinine.

RESULTSAt the time of admission, 212 patients were receiv-

ing � blockers (22%) and 737 were not. Baselinecharacteristics between the 2 groups are listed in Table1. Patients treated with a � blocker at the time ofadmission had slightly higher ejection fractions andfewer NYHA class IV symptoms than patients whodid not receive �-blocker therapy. As expected, heartrate was slightly lower in patients receiving �-blockertherapy. Patients receiving � blockers had a higherincidence of several important comorbidities includ-ing history of angina, arrhythmias, automatic implant-able cardiac defibrillators, or percutaneous coronary

TABLE 1 Baseline Characteristics

Characteristic� Blocker(n � 212)

No � Blocker(n � 737) p Value

Age (yrs) (25th, 75th) 69.3 (57.8, 75.6) 66.8 (55.1, 76)Men (%) 140 (66) 487 (66.1)Caucasian (%) 142 (67) 471 (63.9)African-American (%) 67 (31.6) 243 (33)Ischemic etiology (%) 120 (56.6) 365 (49.5) 0.069LVEF 25 (17.5, 31.5) 22.5 (18, 30) 0.006NYHA class (%)

I 0 1 (0.1)II 20 (9.5) 44 (6)III 104 (49.5) 327 (44.6)IV 86 (41) 361 (49.2) 0.03 (IV vs others)

JVP �6 cm (%) 139 (66.5) 497 (68.3)H JR (%) 77 (40.1) 274 (41.5)S3 (%) 117 (56) 431 (59.4)MR murmur (%) 102 (49.5) 39 (47.9)Heart rate 80 (68, 90) 85 (74, 97) 0.0001Systolic BP 120 (103.5, 135.5) 120 (106, 132)Mo since HF dx 33 (11, 74) 36 (12, 69)Hospitalizations prior 12 mo 2 (1, 3) 1 (0, 3)History of angina (%) 105 (49.5) 304 (41.4) 0.035History of MI (%) 113 (53.3) 344 (46.7) 0.089History of HTN (%) 144 (67.9) 497 (67.5)History of arrhythmia (%) 115 (54.2) 342 (46.4) 0.044History of AICD (%) 29 (13.7) 49 (6.6) 0.002History of CABG (%) 73 (34.4) 224 (30.4)History of PTCI (%) 57 (26.9) 117 (15.9) 0.001History of diabetes (%) 97 (45.8) 321 (43.6)History of chronic pulmonary

disease (%)43 (20.3) 176 (23.9)

ACE inhibitors on admission (%) 146 (68.9) 519 (70.4)ARBs on admission (%) 34 (16) 85 (11.5) 0.089Amiodarone on admission (%) 31 (14.6) 116 (15.7)IV home dobutamine before

randomization (%)2 (0.9) 2 (0.3)

Baseline HF score 6 (5, 7) 6 (5, 7)Baseline visual analog score 45 (25, 60) 42 (25, 51)

*The p values provided only where p �0.1.ACE � angiotensin-converting enzyme; AICD � automatic implantable cardiometer-defibrillator; ARB � angiotensin receptor blocker; BP � blood pressure; CABG

� coronary artery bypass graft surgery; HF � heart failure; HF dx � heart failure diagnosis; HJR � hepatojugular reflex; HTN � hypertension; JVP � jugular venouspressure; LVEF � left ventricular ejection fraction; MI � myocardial infarction; PTCI � percutaneous transluminal coronary intervention.

170 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 91 JANUARY 15, 2003

intervention. Differences in baseline characteristicswere included in the adjusted model.

Most patients were treated with carvedilol (mediandose [25th, 75th] 12.5 mg/day) and metoprolol (me-dian dose [25th, 75th] dose 50 mg/day). Beta-blockerdoses did not change significantly between admissionand discharge. The unadjusted clinical end point dataare listed in Table 2. There was no significant differ-ence in the primary end point of days hospitalized forcardiovascular causes within 60 days of randomiza-tion between patients receiving � blockers and thosenot receiving � blockers at index admission (6 vs 7days, p � 0.28). Sixty-day mortality, the composite ofmortality and rehospitalization, death, any serious ad-verse event, or arrhythmia did not differ based on theuse of � blockade on admission.

No treatment interactions between randomizedtreatment (milrinone or placebo) and � blockers weredetected (Table 2). In patients receiving � blockers onadmission, no difference was observed in the primaryend point of days hospitalized for cardiovascularcauses within 60 days regardless of randomization tomilrinone or placebo. In addition, there were no dif-ferences in 60-day mortality, the composite of deathor rehospitalization, or the composite of death or anyserious adverse event. Similar results were seen in thegroup not taking � blockers on admission. The smallnumbers in each of the subgroups limit the ability tomake definitive conclusions about the presence orabsence of a treatment interaction between � blockersand milrinone therapy.

The findings of the adjusted Cox proportional-hazards model were consistent with those of the un-adjusted analysis. No statistical evidence of harm wasobserved in patients receiving � blockers at admis-sion, suggesting that patients taking pre-existing �blockade may be safely continued in this setting. The

relative risks (RR) and 95% confidence intervals (CIs)were as follows: death at 60 days, RR 1.32 (95% CI0.79 to 2.2, p � 0.29); death plus rehospitalization at60 days, RR 1.09 (95% CI 0.77 to 1.54, p � 0.64);death or any serious adverse event, RR 0.99 (95% CI0.71 to 1.39, p � 0.98).

We conducted an exploratory hypothesis generat-ing analysis to determine clinical outcomes in patientswho were taking � blockers at the time of admissionbut were subsequently discontinued during hospital-ization. First, we reviewed baseline characteristicsamong the 3 categories to determine if patients whose� blockade was discontinued were more severely illthan those whose � blockade was continued. Therewas no evidence from baseline characteristics to sug-gest that patients whose � blockers were discontinuedduring hospitalization were more severely ill than theother patient groups (Table 3). Although the numbersare small and the statistical power is limited, weobserved that patients who were taking � blockers atthe time of admission but discontinued taking thembefore discharge appeared to have a higher mortalityrate at 60 days than patients who either remained on �blockers at the time of admission or who were notreceiving �-blocker therapy (Table 4). Similar find-ings were observed when this analysis was furthersubcategorized by placebo or milrinone randomiza-tion. The event rates were higher in patients whose�-blocker therapy was discontinued, particularly inthose randomized to milrinone therapy (Table 5).

DISCUSSIONTo date, limited data are available describing clin-

ical outcomes of patients admitted for worsening heartfailure who are receiving �-blocker therapy. The datathat have confirmed the role of � blockers in themanagement of chronic heart failure offer minimal

TABLE 2 Clinical End Points (unadjusted analysis)

� Blocker(n � 212)

No � Blocker(n � 737) p Value

Median (25th, 75th) d deceased or hospitalizedfor CV cause at 60 d

6 (4, 12) 7 (4, 14) 0.28

Death during hospitalization (%) 5 (2.4) 24 (3.3) 0.49Death at 60 d (%) 21 (10.0) 69 (9.5) 0.85Death � rehospitalization at 60 d (%) 74 (35.1) 256 (35.2) 0.97Death or any SAE (%) 89 (42.2) 320 (44) 0.65Atrial fibrillation (%) 4 (1.9) 25 (3.4) 0.24Ventricular fibrillation (%) 5 (2.4) 6 (0.8) 0.08Ventricular tachycardia (%) 6 (2.8) 13 (1.8) 0.40Any arrhythmia (%) 11 (5.2) 40 (5.4) 0.89

Placebo(n � 105)

Milrinone(n � 107)

Placebo(n � 367)

Milrinone(n � 370) p Value

Median (25th, 75th) d hospitalized forCV cause at 60 d*

6 (4, 12) 6 (4, 12) 7 (4, 14) 6 (4, 13) 0.77

Death at 60 d (%)* 10 (9.6) 11 (10.3) 31 (8.6) 38 (10.4) 0.93Death � rehospitalization at 60 d (%)† 36 (34.6) 38 (35.5) 128 (35.6) 128 (34.9) 0.69Death or any SAE (%)† 39 (37.5) 50 (46.7) 159 (44.2) 161 (43.8) 0.15

*Cox proportional-hazards model.†Logistic regression model.CV � cardiovascular; SAE � serious adverse event.

HEART FAILURE/BETA BLOCKERS IN WORSENING HEART FAILURE 171

insight into this issue because patients enrolled inthose trials were without evidence of congestion. TheUS Carvedilol, Metoprolol CR/XL Randomized Inter-vention Trial in Congestive Heart Failure (MERIT-HF), and Cardiac Insufficiency Bisoprolol Study II(CIBIS-II) studies were conducted in patients withmild to moderate heart failure, confirmed by the an-nual placebo mortality rates of 7.8%, 11%, and 13.2%,

respectively.1–3 Among patients enrolled in these stud-ies, 8% experienced NYHA class IV symptoms atbaseline. The CarvedilOl ProspEctive RaNdomIzedCUmulative Survival (COPERNICUS) trial enrolledthe most advanced heart failure population to date,with an annual placebo mortality of 19.7% and allpatients presenting with NYHA class III or IV symp-toms.4 A small proportion of the total enrolled patients

TABLE 3 Baseline Characteristics by Modification of Beta-blocker Therapy

Characteristics

� Blocker atAdmission &

Discharge (n � 165)

� Blocker atAdmission Not at

Discharge (n � 47)No � Blocker at

Admission (n � 737) p Value*

Age (yrs) 65.9 68.2 64.9 0.29Male (%) 112 (68) 28 (60) 487 (66) 0.58Caucasian (%) 113 (68) 29 (62) 471 (64) 0.49Ischemic etiology (%) 97 (59) 23 (49) 365 (50) 0.10LVEF (25th, 75th) 25 (18, 33) 27 (18, 31) 23 (18, 30) 0.02NYHA class

I 0 (0) 0 (0) 1 (1)II 18 (11) 2 (4) 44 (6) 0.07III 78 (48) 26 (55) 327 (45)IV 67 (41) 19 (40) 361 (49)

JVP � 6 cm (%) 108 (71) 31 (72) 497 (73) 0.79HJR (%) 61 (41) 16 (36) 274 (41) 0.8S3 (%) 92 (56) 25 (56) 431 (59) 0.68MR murmur (%) 82 (51) 20 (44) 339 (48) 0.69Heart rate 80 (68, 90) 79 (71, 88) 85 (74, 97) 0.0001Systolic BP 120 (103, 136) 122 (107, 134) 120 (106, 132) 0.77Mo since HF dx 31 (11, 70) 37 (14, 74) 36 (12, 69) 0.88Hospitalizations prior 12 mo 2 (0, 3) 2 (1, 4) 1 (0, 3) 0.37History of angina (%) 89 (54) 16 (34) 304 (41) 0.0001History of MI (%) 92 (56) 21 (45) 344 (47) 0.10History of HTN (%) hypertension 109 (66) 35 (74) 497 (68) 0.54History of arrhythmia (%) 88 (53) 27 (57) 342 (46) 0.17History of AICD (%) 22 (13) 7 (15) 49 (7) 0.01History of CABG (%) 60 (36) 13 (28) 224 (30) 0.29History of PTCI (%) 48 (29) 9 (19) 117 (16) 0.0001History of diabetes (%) 76 (46) 21 (45) 321 (44) 0.84History of chronic pulmonary

disease (%)37 (22) 6 (13) 176 (24) 0.17

ACE inhibitors on admission (%) 114 (69) 32 (68) 519 (70) 0.90ARB on admission (%) 27 (16) 7 (15) 85 (12) 0.29Amiodarone on admission (%) 25 (15) 6 (13) 116 (16) 0.85IV home dobutamine before

randomization (%)2 (1) 0 (0) 2 (1) 0.31

HF score 6 (5, 7) 6 (5, 8) 6 (5, 7) 0.13Baseline visual analog score 40 (25, 60) 50 (25, 65) 42 (25, 51) 0.34

*Likelihood ratio (chi-square).MR � mitral regurgitation; other abbreviations as in Tables 1 and 2.

TABLE 4 Clinical End Points by Modification of Beta-blocker Therapy

Characteristics

� Blocker atAdmission &

Discharge (n � 165)

� Blocker atAdmission Not at

Discharge (n � 47)No � Blocker at

Admission (n � 737)

Median (25th, 75th) d deceased orhospitalized for CV cause at 60 d

6.0 (4.0, 12.0) 7.0 (4.0, 16.0) 7.0 (4.0, 14.0)

(95% CI) (5, 7) (5, 10) (6, 7)Death at 60 d (%) 13 (7.9) 8 (17.0) 69 (9.5)

(95% CI) (4.3, 13.2) (7.7, 30.8) (7.5, 11.9)Death � rehospitalization at 60 d (%) 57 (34.8) 17 (36.2) 256 (35.2)

(95% CI) (27.5, 42.6) (22.7, 51.5) (31.7, 38.8)Death/any SAE (%) 69 (42.1) 20 (42.6) 320 (44.0)

(95% CI) (34.4, 50) (28.3, 57.8) (40.3, 47.7)

CI � confidence interval; other abbreviations as in Table 2.

172 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 91 JANUARY 15, 2003

were randomized as hospital in-patients, providedthey were without evidence of volume overload.

This observational, nonrandom analysis from theOPTIME-CHF trial suggests that continuing, pre-ex-isting �-blocker therapy in patients admitted for ex-acerbated chronic heart failure is not associated withan increased risk of adverse clinical events within 60days. These data also suggest that the presence of �blockers do not appear to blunt the effect of therapiesused for worsening heart failure or prolong hospital-ization, because the number of days hospitalized wasnot different between �-blocker– and non–�-blocker–treated patients. These findings are consistent with asimilar analysis conducted from the Flolan Interna-tional Randomized Survival Trial (FIRST) database.Clinical outcomes related to �-blocker use at random-ization in the FIRST trial were explored in a retro-spective analysis.7 No excess risk was observed forpatients treated with � blockers at the time of hospitalpresentation in FIRST. This analysis from OPTIME-CHF lacks statistical power to determine definitivelythe effects of background �-blocker therapy on clini-cal outcomes, because patients were not randomizedto a � blocker, the population was small, and thefollow-up time period was short. However, these datado suggest that patients prescribed a � blocker onadmission are not at a higher risk of adverse clinicaloutcomes, and, importantly, it provides evidence thatmaintaining pre-existing �-blocker therapy duringhospitalization for worsening heart failure can be donesafely.

This analysis also provides hypothesis-generatingdata to suggest that withdrawing � blockade in pa-tients with exacerbated chronic heart failure may beassociated with harm, particularly in patients receiv-ing milrinone. Patients who discontinued �-blockertherapy may have had an apparent increase in mortal-ity because they were severely ill and would have diedanyway; however, the baseline characteristics do notappear to reflect a higher risk population. Alterna-tively, the observed higher mortality rate could besecondary to �-blocker withdrawal. The potential foradrenergic surge after �-blocker withdrawal is likelyto be harmful in this population. Prospective, con-

trolled, clinical trials with adequate populations forstatistical power are needed to fully explore this issueand answer these questions.

The appropriate adjustment of long-term �-blockertherapy during episodes of worsening chronic heartfailure is a clinical question of increasing importanceas acceptance of �-blocker therapy for heart failurebecomes more widespread. Several potential ap-proaches to this clinical situation have been empiri-cally adopted, including continuation of �-blockertherapy unchanged, decreasing �-blocker dosage, ordiscontinuing �-blocker therapy altogether. The OP-TIME-CHF study provides a unique database of pa-tients admitted for decompensated chronic heart fail-ure in which to begin to examine this important clin-ical question. The data suggest that patients treatedwith �-blocker therapy before hospital admission forworsening heart failure can be safely maintained onsuch therapy, and hospital admission alone does notmandate decreasing or discontinuing this therapy. Thedecision to modify pre-existing �-blocker therapyshould be made after considering several factors, in-cluding individual patient presentation, duration of�-blocker therapy before admission, �-blocker dose,and specific �-blocker drug selection. Patients re-cently initiated to �-blocker therapy may be consid-ered differently from those who have been maintainedon stable �-blocker therapy for several weeks ormonths before admission. A patient receiving long-term therapy may be more likely to continue�-blocker treatment successfully during hospital ad-mission as opposed to patients recently initiated to �blockers. For patients who are hospitalized on pre-existing �-blocker therapy with agents other thanthose with proven mortality benefits from large, ran-domized, controlled trials, the hospital setting may beideal for discontinuing the agent lacking evidence andinitiating therapy with a drug shown to improve out-comes (carvedilol, extended release [ER/CR/XL]metoprolol succinate, or bisoprolol).

Several limitations should be considered when in-terpreting this analysis. First, patients were not ran-domized to �-blocker treatment. Second, this was anobservational analysis, and not a prespecified sub-

TABLE 5 Clinical End Points by Modification of Beta-blocker Therapy and OPTIME Randomization

Outcome

� Blocker on Admission andDischarge

� Blocker on Admission ButNot Discharge No � Blocker on Admission

Placebo(n � 79)

Milrinone(n � 86)

Placebo(n � 26)

Milrinone(n � 21)

Placebo(n � 367)

Milrinone(n � 370)

Median (25th, 75th) ddeceased or hospitalizedfor CV cause at 60 d

6.0 (4.0, 13.0) 6.0 (4.0, 11.0) 6.5 (4.0, 9.0) 8.0 (5.0, 25.0) 7.0 (4.0, 14.0) 6.0 (4.0, 13.0)

(95% CI) (5, 9) (5, 7) (4, 9) (5, 29) (6, 8) (6, 7)Death at 60 d (%) 8 (10.3) 5 (5.8) 2 (7.7) 6 (28.6) 31 (8.6) 38 (10.4)

(95% CI) (4.5, 19.2) (1.9, 13.1) (10, 25.1) (11.3, 52.2) (5.9, 12) (7.4, 13.9)Death � rehosp. at 60 d (%) 28 (35.9) 29 (33.7) 8 (30.8) 9 (42.9) 128 (35.6) 128 (34.9)

(95% CI) (25.3, 47.6) (23.9, 44.7) (14.3, 51.8) (21.8, 66) (30.6, 40.7) (30, 40)Death/any SAE at 60 d (%) 31 (39.7) 38 (44.2) 8 (30.8) 12 (57.1) 159 (44.2) 161 (43.8)

(95% CI) (28.8, 51.5) (33.5, 55.3) (14.3, 51.8) (34, 78.2) (39, 49.5) (38.6, 49)

Abbreviation as in Tables 2 and 4.

HEART FAILURE/BETA BLOCKERS IN WORSENING HEART FAILURE 173

group analysis of the main trial. Third, the length of�-blocker therapy before admission is not known, andthe duration of �-blocker therapy beyond the time ofdischarge was not captured. Thus, changes in therapythat could have affected the 60-day time point remainundetected. As with all observational analyses, otherfactors and confounding variables may have beenpresent in the population that were either not capturedor accounted for in the OPTIME-CHF database oradjusted model. Finally, selection bias could haveplayed a role, because patients receiving � blockers athospital admission may have represented a patientpopulation more likely to tolerate the therapy.

Despite these limitations, these data are importantand suggest that patients admitted to the hospital withexacerbated chronic heart failure may be maintainedon �-blocker therapy without excess risk. It does notappear from these data that � blockade blunts orinterferes with other heart failure therapies. It is pos-sible that withdrawal of � blockade is associated witha higher risk of adverse outcome. If � blockers arediscontinued, these data suggest that inotropic therapyshould be avoided, because patients whose � blockerswere withdrawn and who received milrinone had the

highest risk of adverse clinical events. More data areneeded from prospective randomized trials to deter-mine the best approach for treating these patients.

1. Packer M, Bristow MR, Cohn JN, Colucci WS, Fowler MB, Gilbert EM,Shusterman NH. The effect of carvedilol on morbidity and mortality in patientswith chronic heart failure. US Carvedilol Heart Failure Study Group. N EnglJ Med 1996;334:1349–1355.2. MERIT-HF Study Group. Effect of metoprolol CR/XL in chronic heart failure:Metoprolol CR/XL Randomized Intervention Trial in Congestive Heart Failure(MERIT-HF). Lancet 1999;353:2001–2007.3. CIBIS-II Investigators and Committee. The Cardiac Insufficiency BisoprololStudy II (CIBIS-II): a randomised trial. Lancet 1999;353:9–13.4. Packer M, Coats AJS, Fowler MB, Katus HA, Krum H, Mohacsi P, RouleauJL, Tendera M, Castaigne A, Roecker EB, Schultz MK, Demets DL. Effect ofcarvedilol on survival in severe chronic heart failure. N Engl J Med 2001;344:1651–1658.5. Cuffe MS, Califf RM, Adams KF, Bourge RC, Colucci W, Massie B,O’Connor CM, Pina I, Quigg R, Silver M, et al. Rationale and design of theOPTIME CHF trial: outcomes of a prospective trial of intravenous milrinone forexacerbations of chronic heart failure. Am Heart J 2000;139:15–22.6. Cuffe MS, Califf RM, Adams KF, Benza R, Bourge R, Colucci WS, MassieBM, O’Connor CM, Pina I, Quigg R, Silver MA, Gheorghiade M. Short-termintravenous milrinone for acute exacerbation of chronic heart failure: a random-ized controlled trial. JAMA 2002;287:1541–1547.7. O’Connor CM, Gattis WA, Zannad F, McNulty SE, Gheorghiade M, AdamsKF, Califf RM, McKenna WJ, Soler-Soler J, Swedberg K. Beta-blocker therapyin advanced heart failure: clinical characteristics and long-term outcomes. EurJ Heart Failure 1999;1:81–88.

174 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 91 JANUARY 15, 2003