Beta-Blocker Discontinuation in Acute Heart Failure Authors: Nima Alaeiilkhchi, Wade Thompson, Lisa M. McCarthy, William McDonald, Raha Eskandari, Yanrong Maggie Yang, Douglas M. Salzwedel, Parag Goyal Published: 2026 ABSTRACT BackgroundWhether beta-blockers should be continued during hospitalization for acute decompensated heart failure (ADHF) remains uncertain.ObjectivesThe objective of the study was to conduct a systematic review on continuation vs discontinuation of beta-blockers during ADHF hospitalization (PROSPERO CRD42024557727).MethodsWe searched MEDLINE, Embase, and CENTRAL from inception to September 11, 2025. We included randomized and observational studies comparing beta-blocker discontinuation with continuation during hospitalization for ADHF and grouped studies by timing of discontinuation. We conducted meta-analysis of observational studies using generic inverse-variance random-effects models, prioritizing adjusted estimates.ResultsThirteen studies met the eligibility criteria: 1 randomized trial (n = 147) and 12 cohort studies (n = 14,654). The associated risk of in-hospital mortality was uncertain for beta-blocker discontinuation vs continuation (pooled OR: 5.20; 95% CI: 0.24-113.06; I2 = 91%, 3 studies). Short-term discontinuation might be associated with higher risk of mortality up to 3 months (pooled HR: 2.30; 95% CI: 1.23-4.29; I2 = 0%, 2 studies). One study reported an adjusted estimate for long-term mortality (HR: 1.55; 95% CI: 0.69-3.51). The single randomized controlled trial showed an uncertain effect for in-hospital mortality (risk ratio: 1.77; 95% CI: 0.16-19.1) and short-term mortality (risk ratio: 0.88; 95% CI: 0.30-2.62). Findings on rehospitalization were inconsistent and did not favor either approach. All findings were low or very low certainty.ConclusionsAvailable evidence on beta-blocker discontinuation in ADHF is of very low certainty and findings of possible harm may reflect confounding by indication and failure to restart therapy rather than causal harm. Further high-quality randomized controlled trials and nonrandomized studies may resolve uncertainty. Acute decompensated heart failure (HF) (ADHF) is a leading cause of hospital admissions worldwide, contributing substantially to morbidity, mortality, and health care expenditures.1 In chronic HF, guideline-directed medical therapy typically includes beta-blockers because of their proven benefits in reducing mortality and hospitalizations.2,3 Beta-blockers are used in many patients with HF.4, 5, 6 When patients are hospitalized, clinicians are faced with the dilemma of continuing beta-blockers or discontinuing them due to concerns about blood pressure, heart rate, and cardiac output, which may be abnormal during an episode of ADHF.7 In real-world practice, beta-blocker therapy is frequently held during ADHF episodes—reported discontinuation rates are up to 70% in some cohorts.8,9 However, there are potential risks to discontinuing beta-blockers. Abrupt withdrawal of a beta-blocker can precipitate adverse rebound effects such as tachycardia,10,11 ischemia,12 arrhythmias,13 and sudden cardiac death.14 Over the longer term, patients whose beta-blockers are stopped in hospital may never have them reinitiated, thereby losing potential benefits of beta-blockade (especially in HF with reduced ejection fraction [EF] [HFrEF]15, 16, 17). However, continuing beta-blockers during acute decompensation might be deleterious in hemodynamically unstable patients.7,18 Current HF guidelines provide limited specific direction on beta-blocker management during ADHF. Observational studies suggest that abruptly withdrawing beta-blockers in ADHF may increase arrhythmogenic risk,19 whereas a 2015 systematic review found an association between beta-blocker discontinuation in ADHF and higher mortality risk.20 However, that review did not formally assess the certainty of evidence (eg via Grading of Recommendations Assessment, Development and Evaluation [GRADE]) and its search is now over a decade old.20 Therefore, a rigorous and up-to-date synthesis of available evidence is needed regarding the effect of beta-blocker discontinuation on clinical outcomes in ADHF. We conducted a systematic review on the benefits and harms of discontinuing beta-blocker therapy among adults hospitalized for ADHF. Methods Protocol and registration This review was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines.21 The protocol was registered prospectively (PROSPERO CRD42024557727). The registered protocol encompassed 2 related systematic reviews (1 focusing on ADHF and another on postmyocardial infarction/acute coronary syndromes); here we report the findings specific to the ADHF. This systematic review and meta-analysis used published, deidentified aggregate data; institutional review board approval and informed consent were not required. Eligibility criteria Eligible studies met the following criteria:•Randomized controlled trials (RCTs) or observational studies (cohort or case-control)•Enrolled adults (≥18 years old) hospitalized for ADHF•Patients were receiving beta-blocker therapy before or during hospitalization•The study compared beta-blocker discontinuation vs continuation during the acute hospitalization or at hospital discharge•Reported at least one clinical outcome of interest (eg, in-hospital mortality, short-term ≤3 months, long-term >3 months mortality, or rehospitalization) •Randomized controlled trials (RCTs) or observational studies (cohort or case-control) Randomized controlled trials (RCTs) or observational studies (cohort or case-control) •Enrolled adults (≥18 years old) hospitalized for ADHF Enrolled adults (≥18 years old) hospitalized for ADHF •Patients were receiving beta-blocker therapy before or during hospitalization Patients were receiving beta-blocker therapy before or during hospitalization •The study compared beta-blocker discontinuation vs continuation during the acute hospitalization or at hospital discharge The study compared beta-blocker discontinuation vs continuation during the acute hospitalization or at hospital discharge •Reported at least one clinical outcome of interest (eg, in-hospital mortality, short-term ≤3 months, long-term >3 months mortality, or rehospitalization) Reported at least one clinical outcome of interest (eg, in-hospital mortality, short-term ≤3 months, long-term >3 months mortality, or rehospitalization) Studies were excluded if they focused only on chronic HF without an acute decompensation episode, or if they were published only as conference abstracts without a full manuscript. No language limitation was applied. Information sources and search strategy A medical librarian (D.S.) performed the search. Ovid MEDLINE, Ovid Embase, and the Cochrane Central Register of Controlled Trials (CENTRAL) were searched from inception through September 11, 2025. We searched Epistemonikos, Google Scholar, and ClinicalTrials.gov for any relevant studies and reviewed reference lists of pertinent articles and prior reviews to identify further citations. The search strategy is provided in the Supplemental methods. Study selection References were imported into Covidence (Veritas Health Innovation) for deduplication and screening. Two reviewers independently screened titles and abstracts, as well as full text articles. Disagreements were resolved through discussion or by consulting a third reviewer. Data extraction Two reviewers independently extracted data from each included study using a standardized data collection form. Any discrepancies or errors in data extraction were resolved by consensus through re-examination of the original article. Extracted information included study characteristics (first author, year, design, setting, location, sample size) and patient demographics (mean age, sex, baseline left ventricular EF [LVEF]). We recorded how discontinuation was defined in the study, the timing of beta-blocker withdrawal or resumption, the number of patients in each group, and all reported outcomes. The outcomes of interest were in-hospital mortality, short-term mortality (within 3 months), long-term mortality (beyond 3 months), rehospitalization, and length of hospital stay. Risk of bias and quality assessment Two reviewers independently assessed the risk of bias for each included study. For RCTs, we used the Cochrane Risk of Bias 2.0.22 For observational (nonrandomized) studies, we used the Risk Of Bias In Non-randomized Studies of Interventions (ROBINS-I tool).23 Any differences in risk-of-bias judgments between the 2 reviewers were resolved by discussion and, if needed, by involving a third reviewer to reach consensus. We also evaluated the overall certainty of evidence for each outcome using the GRADE approach,24 considering factors such as study limitations, inconsistency, indirectness, imprecision, and publication bias. Per GRADE guidance, RCT evidence started as high certainty and observational evidence as low certainty and then were downgraded or upgraded based on the identified factors. Summary-of-finding tables were prepared to present the GRADE assessments. Data synthesis and analysis We prespecified outcomes by timing: in-hospital; short-term (≤3 months postdischarge); or long-term (>3 months postdischarge). For nonrandomized studies, direct pairwise adjusted effect estimates were prioritized for quantitative synthesis. Time-to-event mortality outcomes were synthesized using HRs when available; adjusted ORs were synthesized separately and served as supportive analyses when HRs were unavailable. We did not combine HRs, ORs, and risk ratios (RRs) within the same meta-analysis. The most fully adjusted direct estimate from each study was extracted for the primary adjusted synthesis. When adjusted direct estimates were unavailable or when only indirect/nonpairwise adjusted contrasts were reported, crude event-count syntheses using RRs were undertaken as secondary analyses. For continuous outcomes (such as length of stay [LOS]), we pooled mean differences (MDs) with 95% CIs. If a study reported continuous data as medians with IQRs or ranges, we converted these to means and SDs using established methods recommended in the Cochrane Handbook.25 Adjusted estimates were pooled on the log scale using generic inverse-variance random-effects models. For crude dichotomous outcomes, we calculated log RRs and corresponding SEs from raw event counts. Random-effects models used restricted maximum likelihood to estimate between-study variance (tau2). CIs for pooled effects used the Hartung-Knapp-Sidik-Jonkman method for syntheses with 3 or more studies. When only 2 studies were available, Wald-type CIs were used because Hartung-Knapp-Sidik-Jonkman can be unstable with very small meta-analyses. We report tau2, I2, and 95% prediction intervals for pooled models with 3 or more studies and interpreted prediction intervals cautiously because most analyses included few studies. We conducted subgroup analyses by LVEF (HFrEF-only vs mixed/unreported populations) and by exposure definition (in-hospital withholding vs discharge withdrawal) when data permitted, along with leave-one-out and intensive care unit/exposure-restriction sensitivity analyses. We used crude outcome data for subgroup and sensitivity analyses as these analyses were not possible using adjusted estimates. Meta-regression was not performed because all meta-analyses included fewer than 10 studies. Funnel plots were visually inspected for pooled observational analyses with at least 3 studies and interpreted cautiously because all analyses included fewer than 10 studies. Studies with zero events in both comparison groups were excluded from dichotomous meta-analyses. All meta-analyses, forest plots, and funnel plots were generated in R (R Foundation for Statistical Computing) using the meta package. Results Study selection and characteristics The search yielded 4,103 unique records. After title/abstract screening, and full-text review of 108 articles, 13 studies met the inclusion criteria (Supplemental Figure 1). These included 1 open-label randomized controlled trial (B-CONVINCED, n = 78 discontinuation and n = 69 continuation) and 12 cohort studies (n = 2,817 discontinuation and n = 11,837 continuation). Studies were conducted between 2003 and 2019 in Europe, North America, and the Middle East. Across the included studies, the mean patient age was 55 to 79 years, and most participants were males (Table 1). Four of 13 studies specifically enrolled patients with HFrEF (Table 1). The mean/median LVEF was reported in 6 studies and was <40% in all these studies. Four studies provided proportions of participants with different LVEF (most patients had LVEF <40% in these studies, see Table 1); 1 study did not report LVEF data for beta-blocker continuers/discontinuers separately but in the overall cohort the LVEF was 43% whereas another study reported that 89% of patients were hospitalized with left ventricular systolic dysfunction but did not clearly report LVEF; finally, 1 study did not report LVEF at all (Table 1). Studies examined outcomes among patients who were receiving beta-blockers before admission and had them discontinued in hospital during hospitalization compared to those who continued beta-blockers. The study population, analytic approach, and main findings are summarized in the Central Illustration.Table 1Characteristics of Included StudiesFirst Author, Year (Study)Design and SettingPopulationIntervention and ComparatorOutcomes and Follow-UpBB Discontinued/Continued, nOverall Risk-of-Bias (ROBINS-I or RoB-2)Abi Khalil et al, 2017a (GULF-CARE)26Prospective cohort study, 7 Middle-East countriesPatients with ADHF; analysis among those on BB with LVEF <40%Mean age 61 y (SD 14), 74% maleMean LVEF 27% in chronic HF groupBB discontinued vs continued during hospitalizationIn-hospital mortality and length of stay, plus 3- and 12-mo mortality, rehospitalization for HF and respective length of stay92/926SeriousAbi Khalil et al, 2017b9Retrospective cohort study of a 22-y registry of ADHF in Doha, QatarPatients admitted for ADHF; among those 1,242 on BB at admission, 26.5% had BB discontinuedMean age 64 y (SD 11), 68% male68% of patients had LVEF <40%BB discontinued vs continued during hospitalizationIn-hospital mortality, stroke/TIA, AF and length of stay in hospital and CCU329/913SeriousBöhm et al, (SURVIVE post hoc)27Retrospective cohort analysis of data from a multinational RCT (SURVIVE)Critically ill with acutely decompensated HF needing IV inotropesMean age 64 years (SD 12), 75% maleMean LVEF 24%Not on BB at discharge vs on BB31-day and 180-day mortality40/549SeriousBrouse and Hafezi, 200628Retrospective cohortPatients admitted for severe decompensated heart failureMean age 69 y (SD 20), 98% maleLVEF <30%: 58% of continuation group and 48% of discontinuation groupLVEF 30%-50%: 31% of continuation group, 38% of discontinuation groupDiscontinued vs same dose of chronic BB during admission ADHFIn-hospital new arrhythmias, length of stay at the hospital, mortality at 12 mo, median rate of readmission within 1 y21/26SeriousButler et al, 2006 (ESCAPE post hoc)6Retrospective cohort analysis of patients in an RCT: ESCAPEHospitalized advanced HF patients who were on BB preadmissionMean age 55 y (SD 14), 74% maleMean LVEF 27% for continuation group and 26% for discontinuation groupPreadmission BB discontinued vs continued in-hospital180-day mortality or rehospitalization54/209SeriousFonarow et al, 2008 (OPTIMIZE-HF)29Prospective cohort of OPTIMIZE-HF registry across academic and community hospitals in United StatesDecompensated HF admissions from a large U.S. registryMean age 69 y (SD 13), 64% maleMean LVEF 24%BB withdrawn at discharge vs continued at dischargeLength of stay during initial hospitalization and 60-90-d mortality/readmission79/1,350SeriousGattis et al, 200330Retrospective cohort analyses of a multicenter RCT called OPTIMIZE-CHFPatients hospitalized for exacerbated chronic HFMean age 68 years (SD 13), 66% maleMedian LVEF 25%Preadmission BB discontinued vs continued in-hospital60-day mortality, rehospitalization and adverse events47/165SeriousJondeau et al, 2009 (B-CONVINCED)31Randomized controlled open-labeled noninferiority trial in various cardiology centers in FrancePatients with LVEF <40% hospitalized for acute decompensation on stable BB for >1 moMean age 72 y (SD 12), 65% maleMean LVEF 32% for continuation group and 31% for discontinuation groupRandomized to discontinued vs continue BBMortality and length of stay during the initial hospitalization. General well-being and dyspnea on days 3 and 8. Mortality/readmissions at 3 mos78/69Some concernsMetra et al, 2007 (COMET post hoc)32Retrospective cohort analysis of a multicenter RCT of COMETPatients with symptomatic chronic HF with LVEF ≤35% in COMET hospitalized for decompensationMean age 64 y (SD 11), 77% maleMean LVEF 26%BB discontinued or continued after an HF hospitalization1-5-y mortality61/529SeriousMiro et al, 2016 (BETAWIN-AHF)8Retrospective cohort analysis of prospectively collected registry data from 35 Spanish EDsPatients with chronic HF on BB who present to ED with acute decompensationMean age 78 y (SD 10), 45% male62% had reduced EF; mean LVEF 36% among HFrEF patientsBB discontinued vs continuedIn-hospital all-cause mortality, 30-d all-cause or ED reconsultation for HF and prolonged hospitalization1,460/530SeriousOrso et al, 200933Prospective cohort analysis of an Italian SurveyPatients with worsening HF who were admitted to cardiology wardsMean age 71 y (SD 11), 68% male89% had LV systolic dysfunction (LVEF reporting unclear)BB discontinued vs continuedAll-cause in-hospital mortality (primary), plus arrhythmias, acute coronary syndrome (ACS), stroke, length of stay in the hospital and ICU141/362SeriousTran et al, 2018 (ARIC surveillance)34Based on retrospective data from the community surveillance cohort of the ARIC studyPatients hospitalized for HFCharacteristics not reported for BB continuers or discontinuers separately; in overall cohort:Mean age 75 y, 47% maleMean LVEF 43%BB discontinued vs continuedMortality at 1 y305/3,055SeriousZhou et al, 202535Retrospective cohort analysis of intensive care database at Beth Israel Deaconess Medical Center in Massachusetts.ICU patients with a diagnosis of AHF.Discontinuation group: median age 79 (IQR: 66-85), 42% femaleContinuation group: median age 77 (IQR: 67-85), 46% femaleLVEF not reportedBB discontinued vs continuedAll-cause in-hospital mortality (defined as any death occurring during the hospital stay up to 48 h postdischarge).188/3,223SeriousThis table details study design, setting, sample size, and key characteristics and outcomes.ADHF = acute decompensated heart failure; AF = atrial fibrillation; BB = beta-blocker; CCU = coronary care unit; ED = emergency department; EF = ejection fraction; HF = heart failure; HFrEF = heart failure with reduced ejection fraction; ICU = intensive care unit; IV = intravenous; LVEF = left ventricular ejection fraction; RCT = randomized controlled trial; RoB-2 = Cochrane Risk of Bias 2.0; ROBINS-1 = Risk Of Bias In Non-randomized Studies – of Intervention; TIA = transient ischemic attack.Central IllustrationBeta-Blocker Discontinuation in Acute Heart FailureFor adults hospitalized with ADHF, discontinuation of beta-blockers might be associated with higher postdischarge mortality up to 3 months though we are very uncertain about these findings. Findings on in-hospital mortality and long-term mortality were uncertain. RCT = randomized controlled trial Table 1Characteristics of Included StudiesFirst Author, Year (Study)Design and SettingPopulationIntervention and ComparatorOutcomes and Follow-UpBB Discontinued/Continued, nOverall Risk-of-Bias (ROBINS-I or RoB-2)Abi Khalil et al, 2017a (GULF-CARE)26Prospective cohort study, 7 Middle-East countriesPatients with ADHF; analysis among those on BB with LVEF <40%Mean age 61 y (SD 14), 74% maleMean LVEF 27% in chronic HF groupBB discontinued vs continued during hospitalizationIn-hospital mortality and length of stay, plus 3- and 12-mo mortality, rehospitalization for HF and respective length of stay92/926SeriousAbi Khalil et al, 2017b9Retrospective cohort study of a 22-y registry of ADHF in Doha, QatarPatients admitted for ADHF; among those 1,242 on BB at admission, 26.5% had BB discontinuedMean age 64 y (SD 11), 68% male68% of patients had LVEF <40%BB discontinued vs continued during hospitalizationIn-hospital mortality, stroke/TIA, AF and length of stay in hospital and CCU329/913SeriousBöhm et al, (SURVIVE post hoc)27Retrospective cohort analysis of data from a multinational RCT (SURVIVE)Critically ill with acutely decompensated HF needing IV inotropesMean age 64 years (SD 12), 75% maleMean LVEF 24%Not on BB at discharge vs on BB31-day and 180-day mortality40/549SeriousBrouse and Hafezi, 200628Retrospective cohortPatients admitted for severe decompensated heart failureMean age 69 y (SD 20), 98% maleLVEF <30%: 58% of continuation group and 48% of discontinuation groupLVEF 30%-50%: 31% of continuation group, 38% of discontinuation groupDiscontinued vs same dose of chronic BB during admission ADHFIn-hospital new arrhythmias, length of stay at the hospital, mortality at 12 mo, median rate of readmission within 1 y21/26SeriousButler et al, 2006 (ESCAPE post hoc)6Retrospective cohort analysis of patients in an RCT: ESCAPEHospitalized advanced HF patients who were on BB preadmissionMean age 55 y (SD 14), 74% maleMean LVEF 27% for continuation group and 26% for discontinuation groupPreadmission BB discontinued vs continued in-hospital180-day mortality or rehospitalization54/209SeriousFonarow et al, 2008 (OPTIMIZE-HF)29Prospective cohort of OPTIMIZE-HF registry across academic and community hospitals in United StatesDecompensated HF admissions from a large U.S. registryMean age 69 y (SD 13), 64% maleMean LVEF 24%BB withdrawn at discharge vs continued at dischargeLength of stay during initial hospitalization and 60-90-d mortality/readmission79/1,350SeriousGattis et al, 200330Retrospective cohort analyses of a multicenter RCT called OPTIMIZE-CHFPatients hospitalized for exacerbated chronic HFMean age 68 years (SD 13), 66% maleMedian LVEF 25%Preadmission BB discontinued vs continued in-hospital60-day mortality, rehospitalization and adverse events47/165SeriousJondeau et al, 2009 (B-CONVINCED)31Randomized controlled open-labeled noninferiority trial in various cardiology centers in FrancePatients with LVEF <40% hospitalized for acute decompensation on stable BB for >1 moMean age 72 y (SD 12), 65% maleMean LVEF 32% for continuation group and 31% for discontinuation groupRandomized to discontinued vs continue BBMortality and length of stay during the initial hospitalization. General well-being and dyspnea on days 3 and 8. Mortality/readmissions at 3 mos78/69Some concernsMetra et al, 2007 (COMET post hoc)32Retrospective cohort analysis of a multicenter RCT of COMETPatients with symptomatic chronic HF with LVEF ≤35% in COMET hospitalized for decompensationMean age 64 y (SD 11), 77% maleMean LVEF 26%BB discontinued or continued after an HF hospitalization1-5-y mortality61/529SeriousMiro et al, 2016 (BETAWIN-AHF)8Retrospective cohort analysis of prospectively collected registry data from 35 Spanish EDsPatients with chronic HF on BB who present to ED with acute decompensationMean age 78 y (SD 10), 45% male62% had reduced EF; mean LVEF 36% among HFrEF patientsBB discontinued vs continuedIn-hospital all-cause mortality, 30-d all-cause or ED reconsultation for HF and prolonged hospitalization1,460/530SeriousOrso et al, 200933Prospective cohort analysis of an Italian SurveyPatients with worsening HF who were admitted to cardiology wardsMean age 71 y (SD 11), 68% male89% had LV systolic dysfunction (LVEF reporting unclear)BB discontinued vs continuedAll-cause in-hospital mortality (primary), plus arrhythmias, acute coronary syndrome (ACS), stroke, length of stay in the hospital and ICU141/362SeriousTran et al, 2018 (ARIC surveillance)34Based on retrospective data from the community surveillance cohort of the ARIC studyPatients hospitalized for HFCharacteristics not reported for BB continuers or discontinuers separately; in overall cohort:Mean age 75 y, 47% maleMean LVEF 43%BB discontinued vs continuedMortality at 1 y305/3,055SeriousZhou et al, 202535Retrospective cohort analysis of intensive care database at Beth Israel Deaconess Medical Center in Massachusetts.ICU patients with a diagnosis of AHF.Discontinuation group: median age 79 (IQR: 66-85), 42% femaleContinuation group: median age 77 (IQR: 67-85), 46% femaleLVEF not reportedBB discontinued vs continuedAll-cause in-hospital mortality (defined as any death occurring during the hospital stay up to 48 h postdischarge).188/3,223SeriousThis table details study design, setting, sample size, and key characteristics and outcomes.ADHF = acute decompensated heart failure; AF = atrial fibrillation; BB = beta-blocker; CCU = coronary care unit; ED = emergency department; EF = ejection fraction; HF = heart failure; HFrEF = heart failure with reduced ejection fraction; ICU = intensive care unit; IV = intravenous; LVEF = left ventricular ejection fraction; RCT = randomized controlled trial; RoB-2 = Cochrane Risk of Bias 2.0; ROBINS-1 = Risk Of Bias In Non-randomized Studies – of Intervention; TIA = transient ischemic attack. Characteristics of Included Studies This table details study design, setting, sample size, and key characteristics and outcomes. ADHF = acute decompensated heart failure; AF = atrial fibrillation; BB = beta-blocker; CCU = coronary care unit; ED = emergency department; EF = ejection fraction; HF = heart failure; HFrEF = heart failure with reduced ejection fraction; ICU = intensive care unit; IV = intravenous; LVEF = left ventricular ejection fraction; RCT = randomized controlled trial; RoB-2 = Cochrane Risk of Bias 2.0; ROBINS-1 = Risk Of Bias In Non-randomized Studies – of Intervention; TIA = transient ischemic attack. Beta-Blocker Discontinuation in Acute Heart Failure For adults hospitalized with ADHF, discontinuation of beta-blockers might be associated with higher postdischarge mortality up to 3 months though we are very uncertain about these findings. Findings on in-hospital mortality and long-term mortality were uncertain. RCT = randomized controlled trial Risk of bias and certainty of evidence The 12 observational studies were judged to be at serious risk of bias on ROBINS-I, with confounding by indication being the most important limitation (patients with hypotension, shock, bradycardia, or need for inotropes may have been more likely to have beta-blockers withheld and to experience adverse outcomes) (Supplemental Figure 2). The RCT was judged to be at high risk of bias due to missing outcome data, open-label design, and small sample size. Using GRADE, certainty of evidence was low or very low for all outcomes (Supplemental Table 1). Given few eligible studies, inspection of funnel plots did not reveal any clear patterns. In-hospital mortality Five observational studies8,9,26,33,35 reported all-cause in-hospital mortality (Table 2). No studies reported time-to-event analyses. Three studies8,26,33 reported adjusted pairwise estimates (reporting OR) comparing beta-blocker discontinuation with continuation and the association with all-cause mortality (Supplemental Table 2). Pooling these adjusted ORs yielded OR 5.20 (95% CI: 0.24-113.06; I2 = 91%) (Figure 1A), indicating extreme imprecision and substantial heterogeneity. One additional observational study35 reported an adjusted comparison against a “not treated” reference group rather than continuation and was therefore not eligible for the adjusted pairwise meta-analysis (Supplemental Table 2). Secondary crude event-count pooling of 4 studies8,26,33,35 was directionally similar but highly heterogeneous (RR, 5.31; 95% CI: 0.93-30.41; I2 = 94%) (Supplemental Figure 3A). In the RCT,31 in-hospital mortality was 2.6% (2/78) with discontinuation vs 1.4% (1/69) with continuation (RR, 1.77; 95% CI: 0.16-19.1). Subgroup and sensitivity analyses generated imprecise estimates (Supplemental Table 3). The certainty of evidence was very low for both the observational studies and the RCT (Supplemental Table 1).Table 2All-Cause Mortality Results by TimeframeFirst Author, YearTimeframeDiscontinuation, n/N (%)Continuation, n/N (%)Estimate (95% CI) Discontinued vs ContinuedaIn-hospital Abi Khalil et al, 2017a37/92 (40.2%)15/926 (1.6%)Unadjusted RR: 24.8 (95% CI: 14.2-43.5)Adjusted OR: 20.0 (95% CI: 8.93-45.4) Abi Khalil et al, 2017b44/329 (13.4%)0/913 (0)Unadjusted RR not computable Jondeau et al, 2009 (RCT)2/78 (2.6%)1/69 (1.4%)Unadjusted RR: 1.77 (95% CI: 0.16-19.09) Miro et al, 201681/1,460 (5.5%)16/530 (3%)Unadjusted OR: 1.89 (95% CI: 1.09-3.26)Adjusted OR: 1.77 (95% CI: 1.01-3.09)Unadjusted RR: 1.84 (95% CI: 1.09-3.11) Orso et al, 200917/141 (12.1%)10/362 (2.8%)Unadjusted RR: 4.36 (95% CI: 2.05-9.30)Adjusted OR: 4.20 (95% CI: 1.59-11.1) Zhou et al, 202589/188 (47.3%)377/3,223 (11.7%)Unadjusted RR: 4.04 (95% CI: 3.39-4.84)Short term (≤3 mo) Abi Khalil et al, 2017a3 mo9/50 (18.0%)77/896 (8.6%)Unadjusted RR: 2.09 (95% CI: 1.12-3.93)Adjusted OR: 1.95 (95% CI: 0.87-4.33) Böhm et al, 201131 days4/40 (11%)11/549 (2%)Adjusted HR: 2.05 (95% CI: 0.36-11.63)Unadjusted RR: 4.99 (95% CI: 1.66-15.0) Fonarow et al, 200860-90 d19/79 (24.4%)117/1,350 (8.7%)Adjusted HR: 2.34 (95% CI: 1.20-4.55)Unadjusted RR: 2.78 (95% CI: 1.81-4.26) Gattis et al, 200360 d8/47 (17%)13/165 (7.9%)Unadjusted RR: 2.16 (95% CI: 0.95-4.90) Jondeau et al, 2009 (RCT)3 mo6/78 (8%)6/69 (9%)Unadjusted RR: 0.88 (95% CI: 0.30-2.62) Miro et al, 201630 d127/1,460 (8.7%)24/530 (4.5%)Unadjusted OR: 2.01 (95% CI: 1.28-3.15)Adjusted OR: 1.93 (95% CI: 1.22-3.06)Unadjusted RR: 1.92 (95% CI: 1.26-2.94)Long term (>3 mo) Abi Khalil et al, 2017a12 mo11/45 (24.4%)128/835 (15.3%)Unadjusted RR: 1.59 (95% CI: 0.93-2.73) Böhm et al 2011180 d12/40 (29%)71/549 (13%)Adjusted HR: 1.55 (95% CI: 0.69-3.51)Unadjusted RR: 2.32 (95% CI: 1.38-3.90) Brouse and Hafezi, 20066 mo6/21 (29%)1/26 (3.8%)Unadjusted RR: 7.43 (95% CI: 0.97-57.0) Brouse and Hafezi, 20061 y9/21 (43%)6/26 (23%)Unadjusted RR: 1.86 (95% CI: 0.79-4.38) Metra et al, 20071 y23/61 (37.7%)137/529 (25.9%)Unadjusted RR: 1.46 (95% CI: 1.02-2.07) Metra et al, 20075 y58/61 (95%)502/529 (95%)Unadjusted RR: 1.00 (95% CI: 0.94-1.06)RR = risk ratio; other abbreviation as in Table 1.aUnadjusted relative estimates were calculated when not provided by study authors.Figure 1Meta-Analyses for MortalityRandom-effects meta-analyses of direct pairwise adjusted comparative estimates from observational studies comparing beta-blocker discontinuation/withdrawal with continuation in adults hospitalized with ADHF. (A) In-hospital mortality (adjusted OR). (B) Short-term (≤3 months) mortality using adjusted HRs from time-to-event analyses. (C) Short-term (≤3 months) mortality using adjusted ORs as a supportive analysis. HKSJ = Hartung-Knapp-Sidik-Jonkman. Table 2All-Cause Mortality Results by TimeframeFirst Author, YearTimeframeDiscontinuation, n/N (%)Continuation, n/N (%)Estimate (95% CI) Discontinued vs ContinuedaIn-hospital Abi Khalil et al, 2017a37/92 (40.2%)15/926 (1.6%)Unadjusted RR: 24.8 (95% CI: 14.2-43.5)Adjusted OR: 20.0 (95% CI: 8.93-45.4) Abi Khalil et al, 2017b44/329 (13.4%)0/913 (0)Unadjusted RR not computable Jondeau et al, 2009 (RCT)2/78 (2.6%)1/69 (1.4%)Unadjusted RR: 1.77 (95% CI: 0.16-19.09) Miro et al, 201681/1,460 (5.5%)16/530 (3%)Unadjusted OR: 1.89 (95% CI: 1.09-3.26)Adjusted OR: 1.77 (95% CI: 1.01-3.09)Unadjusted RR: 1.84 (95% CI: 1.09-3.11) Orso et al, 200917/141 (12.1%)10/362 (2.8%)Unadjusted RR: 4.36 (95% CI: 2.05-9.30)Adjusted OR: 4.20 (95% CI: 1.59-11.1) Zhou et al, 202589/188 (47.3%)377/3,223 (11.7%)Unadjusted RR: 4.04 (95% CI: 3.39-4.84)Short term (≤3 mo) Abi Khalil et al, 2017a3 mo9/50 (18.0%)77/896 (8.6%)Unadjusted RR: 2.09 (95% CI: 1.12-3.93)Adjusted OR: 1.95 (95% CI: 0.87-4.33) Böhm et al, 201131 days4/40 (11%)11/549 (2%)Adjusted HR: 2.05 (95% CI: 0.36-11.63)Unadjusted RR: 4.99 (95% CI: 1.66-15.0) Fonarow et al, 200860-90 d19/79 (24.4%)117/1,350 (8.7%)Adjusted HR: 2.34 (95% CI: 1.20-4.55)Unadjusted RR: 2.78 (95% CI: 1.81-4.26) Gattis et al, 200360 d8/47 (17%)13/165 (7.9%)Unadjusted RR: 2.16 (95% CI: 0.95-4.90) Jondeau et al, 2009 (RCT)3 mo6/78 (8%)6/69 (9%)Unadjusted RR: 0.88 (95% CI: 0.30-2.62) Miro et al, 201630 d127/1,460 (8.7%)24/530 (4.5%)Unadjusted OR: 2.01 (95% CI: 1.28-3.15)Adjusted OR: 1.93 (95% CI: 1.22-3.06)Unadjusted RR: 1.92 (95% CI: 1.26-2.94)Long term (>3 mo) Abi Khalil et al, 2017a12 mo11/45 (24.4%)128/835 (15.3%)Unadjusted RR: 1.59 (95% CI: 0.93-2.73) Böhm et al 2011180 d12/40 (29%)71/549 (13%)Adjusted HR: 1.55 (95% CI: 0.69-3.51)Unadjusted RR: 2.32 (95% CI: 1.38-3.90) Brouse and Hafezi, 20066 mo6/21 (29%)1/26 (3.8%)Unadjusted RR: 7.43 (95% CI: 0.97-57.0) Brouse and Hafezi, 20061 y9/21 (43%)6/26 (23%)Unadjusted RR: 1.86 (95% CI: 0.79-4.38) Metra et al, 20071 y23/61 (37.7%)137/529 (25.9%)Unadjusted RR: 1.46 (95% CI: 1.02-2.07) Metra et al, 20075 y58/61 (95%)502/529 (95%)Unadjusted RR: 1.00 (95% CI: 0.94-1.06)RR = risk ratio; other abbreviation as in Table 1.aUnadjusted relative estimates were calculated when not provided by study authors. All-Cause Mortality Results by Timeframe RR = risk ratio; other abbreviation as in Table 1. Unadjusted relative estimates were calculated when not provided by study authors. Meta-Analyses for Mortality Random-effects meta-analyses of direct pairwise adjusted comparative estimates from observational studies comparing beta-blocker discontinuation/withdrawal with continuation in adults hospitalized with ADHF. (A) In-hospital mortality (adjusted OR). (B) Short-term (≤3 months) mortality using adjusted HRs from time-to-event analyses. (C) Short-term (≤3 months) mortality using adjusted ORs as a supportive analysis. HKSJ = Hartung-Knapp-Sidik-Jonkman. Short-term mortality (≤3 months) Five observational studies8,26,27,29,30 reported mortality outcomes within 3 months after the index hospital discharge (Table 2). Two studies27,29 reported adjusted time-to-event estimates (Supplemental Table 2) and formed the primary short-term mortality analysis. Pooling these adjusted HRs showed higher short-term mortality associated with discontinuation vs continuation (HR: 2.30; 95% CI: 1.23-4.29; I2 = 0%) (Figure 1B). Two additional studies8,26 reported adjusted ORs and yielded a directionally consistent pooled estimate (OR: 1.93; 95% CI: 1.30-2.88; I2 = 0%) (Figure 1C). Secondary crude event-count pooling of 5 studies8,26,27,29,30 was also similar in direction and magnitude (pooled RR: 2.35; 95% CI: 1.69-3.27; I2 = 0%) (Supplemental Figure 3B). The RCT31 showed no difference in mortality at 3 months: 6/78 (7.7%) in the discontinuation arm vs 6/69 (8.7%) in the continuation arm (RR: 0.88; 95% CI: 0.30-2.62) (Table 2). Notably, 76% of patients randomized to discontinuation were restarted on beta-blockers by 3 months compared with 90% in the continuation group. Subgroup analyses with crude data were directionally similar when restricted to HFrEF studies and when stratified by exposure definition (Supplemental Table 3). The certainty of evidence was very low for the observational studies and very low for the RCT (Supplemental Table 1). Long-term mortality (>3 months) Four observational studies26, 27, 28,32 evaluated mortality beyond 3 months (from 6 months up to 1 year) (Table 2). Only 1 study27 reported an adjusted estimate for long-term mortality (see Supplemental Table 4), precluding meta-analysis. This study27 found no association between beta-blocker discontinuation and long-term mortality compared with continuation (HR: 1.55; 95% CI: 0.69-3.51). Secondary analysis of crude outcome data from 4 studies26, 27, 28,32 suggested possible higher long-term mortality associated with beta-blocker discontinuation (pooled RR: 1.68; 95% CI: 1.19-2.36; I2 = 0%) (Supplemental Figure 3C). Subgroup analysis using crude outcome data restricting to 3 studies26,27,32 with HFrEF populations found a pooled RR of 1.67 (95% CI: 0.93-3.02) associated with beta-blocker discontinuation compared to continuation (Supplemental Table 3). The certainty of evidence was rated as very low for the observational studies (Supplemental Table 1). Length of hospital stay Five observational studies9,26,28,29,33 reported index-hospital LOS (Supplemental Table 4). One study (Abi Khalil et al9) was excluded from the quantitative synthesis because it reported medians with extremely wide ranges (1-451 days), precluding reliable conversions to a mean and SD. Based on meta-analysis of 4 studies,26,28,29,33 beta-blocker discontinuation was associated with a longer LOS compared with continuation (mean difference 1.45 days; 95% CI: 0.05-2.86; I2 = 0%) (Figure 2). The corresponding prediction interval crossed no difference (Supplemental Table 5). The RCT reported similar mean stays in the groups (discontinuation: 10.4 vs continuation: 11.5 days; P = 0.2). The certainty of evidence was very low for observational studies and low for the RCT (Supplemental Table 1).Figure 2Meta-Analysis for Length of StayRandom-effects meta-analysis of observational studies comparing index-hospital length of stay (days) between beta-blocker discontinuation/withdrawal and continuation in adults hospitalized with ADHF. Positive mean differences indicate longer stay with discontinuation/withdrawal. Squares indicate study-specific mean differences with 95% confidence intervals; diamonds indicate pooled effects. Abbreviation as in Figure 1. Meta-Analysis for Length of Stay Random-effects meta-analysis of observational studies comparing index-hospital length of stay (days) between beta-blocker discontinuation/withdrawal and continuation in adults hospitalized with ADHF. Positive mean differences indicate longer stay with discontinuation/withdrawal. Squares indicate study-specific mean differences with 95% confidence intervals; diamonds indicate pooled effects. Abbreviation as in Figure 1. Rehospitalization Definitions of rehospitalization varied across studies (eg, all-cause, HF-specific, cardiovascular in general, arrhythmia-related; and follow-up intervals ranged from 60 days to 1 year), which precluded any quantitative synthesis. In general, the observational studies26,29,30 showed readmission rates might not differ between patients who discontinued vs continued beta-blockers but we are very uncertain about this given differences in the indication for rehospitalization and follow-up duration (Table 3). Abi Khalil et al,26 Brouse and Hafezi,28 and Jondeau et al31 reported nonsignificantly higher readmission rates in the discontinuation group, whereas Fonarow et al29 and Gattis30 noted a nonsignificant reduction in 60 to 90-day rehospitalization rates with discontinuation (Table 3). The certainty of evidence was very low for the observational studies and low for the RCT.Table 3Rehospitalization ResultsFirst Author, YearTimeframeIndication for RehospitalizationDiscontinuation, n/N (%)Continuation, n/N (%)Estimate (95% CI)a Discontinued vs ContinuedShort Term (≤3 mo) Abi Khalil et al, 2017a3 moHeart failure15/41 (36.6%)204/818 (24.9%)Unadjusted RR: 1.47 (95% CI: 0.96-2.23) Fonarow et al, 200860-90 dAll-cause19/79 (24.1%)435/1,350 (32.2%)Unadjusted RR: 0.75 (95% CI: 0.50-1.11) Gattis et al, 200360 dCardiovascular causes9/47 (19.1%)44/165 (26.7%)Unadjusted RR: 0.72 (95% CI: 0.38-1.36)Long term (>3 mo) Abi Khalil et al, 2017a12 moHeart failure17/34 (50%)316/707 (44.7%)Unadjusted RR: 1.12 (95% CI: 0.79-1.58) Brouse and Hafezi, 20061 yHeart failureMedian 1.9 per yearIQR: (1-8)Median 1.6 per year IQR: (0.9-5.3)NAJondeau et al, 2009 (RCT)More than 3 moArrhythmia3/78 (4%)2/69 (3%)Unadjusted RR: 1.33 (95% CI: 0.23-7.71)More than 3 moHeart failure24/78 (32%)15/69 (22%)Unadjusted RR: 1.42 (95% CI: 0.81-2.47)More than 3 moAll-cause36/78 (47%)27/69 (40%)Unadjusted RR: 1.18 (95% CI: 0.81-1.72)Abbreviations as in Table 2.aRRs were calculated when not provided by authors. Table 3Rehospitalization ResultsFirst Author, YearTimeframeIndication for RehospitalizationDiscontinuation, n/N (%)Continuation, n/N (%)Estimate (95% CI)a Discontinued vs ContinuedShort Term (≤3 mo) Abi Khalil et al, 2017a3 moHeart failure15/41 (36.6%)204/818 (24.9%)Unadjusted RR: 1.47 (95% CI: 0.96-2.23) Fonarow et al, 200860-90 dAll-cause19/79 (24.1%)435/1,350 (32.2%)Unadjusted RR: 0.75 (95% CI: 0.50-1.11) Gattis et al, 200360 dCardiovascular causes9/47 (19.1%)44/165 (26.7%)Unadjusted RR: 0.72 (95% CI: 0.38-1.36)Long term (>3 mo) Abi Khalil et al, 2017a12 moHeart failure17/34 (50%)316/707 (44.7%)Unadjusted RR: 1.12 (95% CI: 0.79-1.58) Brouse and Hafezi, 20061 yHeart failureMedian 1.9 per yearIQR: (1-8)Median 1.6 per year IQR: (0.9-5.3)NAJondeau et al, 2009 (RCT)More than 3 moArrhythmia3/78 (4%)2/69 (3%)Unadjusted RR: 1.33 (95% CI: 0.23-7.71)More than 3 moHeart failure24/78 (32%)15/69 (22%)Unadjusted RR: 1.42 (95% CI: 0.81-2.47)More than 3 moAll-cause36/78 (47%)27/69 (40%)Unadjusted RR: 1.18 (95% CI: 0.81-1.72)Abbreviations as in Table 2.aRRs were calculated when not provided by authors. Rehospitalization Results Abbreviations as in Table 2. RRs were calculated when not provided by authors. Discussion Our systematic review identified 12 cohort studies and 1 RCT that examined clinical outcomes associated with discontinuation of beta-blockers during hospitalization for ADHF. Most evidence comes from individuals with LVEF <40%, thus our findings mainly apply to this population. Our findings suggested that beta-blocker discontinuation might be associated with an increased risk of short-term mortality compared with continuation. Findings on in-hospital mortality and long-term mortality were uncertain due to considerable imprecision and heterogeneity, with sensitivity analyses demonstrating fragility of the pooled estimate. Adjusted estimates and secondary syntheses of crude outcome data were generally directionally consistent, although crude syntheses are more vulnerable to confounding by indication and should be interpreted as supportive evidence only. The risk of rehospitalization following beta-blocker discontinuation was unclear due to differences in reporting. Nevertheless, available studies consistently suggested there might be no meaningful difference in rehospitalization rates between discontinuation and continuation. Beta-blocker discontinuation might be associated with a slightly longer index hospital LOS in observational studies, but this modest increase may not be clinically significant given the wide CI and the lack of a difference in LOS in the RCT. Given predominantly observational evidence and concerns around risk of bias, our confidence in our findings is very low. In particular, our findings may reflect confounding by indication (high-risk phenotypes having beta-blockers stopped) or reflect failure to restart therapy rather than causal harm from discontinuation of beta-blockers in ADHF. Although a prior systematic review20 similarly concluded that withdrawal may worsen outcomes, it predated several large registry analyses now included in our synthesis. By incorporating newer observational data, performing quality assessment, subgroup and sensitivity analyses, and distinguishing time-specific mortality windows, our review provides the most current and methodologically rigorous summary to date. Although our findings are very low certainty, and we cannot infer causality, findings generally support for the 2022 American Heart Association/American College of Cardiology/Heart Failure Society of America guideline on HF management and the 2021 European Society of Cardiology guidance indicating that patients experience worse outcomes when guideline-directed therapies (such as beta-blockers) are withheld in ADHF3,36 and recommend continuation in the absence of shock or severe hypotension.3,36 Nevertheless, real-world practice does not mirror these recommendations.8,9 Our findings reinforce practical points already embedded in guidelines,3,36 which highlight that individualizing therapy is important. For example, continuation of beta-blockers may be reasonable to avoid potential rebound sympathetic activation37 and/or loss of antiarrhythmic benefit,19,38 whereas temporary discontinuation is generally appropriate in cases of hypotension, bradycardia, and/or cardiogenic shock.39, 40, 41 Calculating the risk-benefit ratio of continuing (vs discontinuing) beta-blocker can be challenging. Although hypotension and bradycardia are clinically evident and easy to recognize, cardiogenic shock can present in subtle ways. Physicians must make difficult in-the-moment decisions about beta-blocker therapy for patients hospitalized for ADHF. Prior studies confirm that clinicians frequently opt to discontinue beta-blockers in this scenario.8 Based on our findings, it was unclear whether possible harm is a result of beta-blocker discontinuation during hospitalization or because beta-blockers may not be restarted after hospitalization. Our study could prompt the development of interventions focused on examining rates of inappropriately discontinued beta-blockers, and potentially strategies to reinitiate beta-blockers where appropriate either later in the hospitalization or soon-after hospital discharge. This stresses the importance of continuity of care with care transitions. Whether promoting beta-blocker dose reduction (instead of complete discontinuation) improves outcomes may be worth investigating formally, as observational data would indicate that this may be reasonable.42 Deprescribing has emerged as a useful tool for clinicians to improve outcomes in patients.43 The potential benefits of deprescribing beta-blockers in select scenarios have become an area of interest. Given the very low certainty of the available evidence, further research is needed to clarify uncertainties in evidence. The conduct of a large, pragmatic RCT with the following key design elements would be ideal: 1) broad inclusion criteria spanning but also differentiating between LVEF phenotypes; 2) randomization for situations where there is equipoise with withdrawal; 3) ascertainment of and stratification by hemodynamic status; and 4) inclusion of patient-centered outcomes (mortality, arrhythmia, and quality of life) with standardized outcome definitions (eg, time-to-first HF readmission). Until such trials are completed, high-quality observational studies using advanced causal-inference methods (eg, target-trial emulation) may add to the evidence base, particularly those that can avoid or mitigate confounding by indication and misclassification, which is a common challenge in observational studies of medication discontinuation. Strengths and limitations Strengths of this review include an a priori registered protocol, extensive searches, ROBINS-I and Cochrane Risk of Bias 2.0 appraisal, Hartung-Knapp random-effects pooling, and transparent GRADE assessments. We conducted several analyses to examine robustness of our findings, including providing parallel summaries of adjusted estimates as well as subgroup analyses by LVEF and exposure definition. The limitations in our review mirror limitations of observational studies on medication discontinuation, notably potential for residual confounding and confounding by indication (in particular, beta-blockers being discontinued because people had worse health or clinical status). There was minimal information on why a beta-blocker was being discontinued, along with potential for misclassification in exposure status. Many included studies did not report or stratify outcomes by LVEF, limiting our ability to assess whether the impact of beta-blocker withdrawal differs between HF with preserved EF or HFrEF. When we only analyzed studies among patients with HFrEF, results were similar to our main analysis; however, there were few data available for HFpEF. Overall, we cannot draw any conclusions around whether the effects of discontinuation differ according to EF. Moreover, we could not determine whether there is any difference based on exposure definition (ie, temporary inpatient holds vs permanent discontinuation at discharge). A short-term interruption with prompt reinitiation of beta-blockers may carry less risk than complete cessation; however, most studies did not report whether or when beta-blockers were restarted after the acute phase. We also lacked detailed data on background guideline-directed medical therapy in the observational studies. It was generally unclear whether patients who continued beta-blockers were receiving more comprehensive HF therapy over time than those who discontinued. Any imbalance in concurrent therapies could introduce confounding. Although 1 RCT was available, the event rates were low, limiting the ability to draw meaningful conclusions around in-hospital mortality due to imprecision. In this RCT, most patients randomized to discontinuation resumed beta-blockers after hospitalization, which also limited the ability to compare longer term clinical outcomes following discontinuation. Conclusions Our review provides a contemporary, rigorous, and up-to-date synthesis of available evidence related to discontinuation of beta-blockers during ADHF. Available evidence is of very low certainty and comes primarily from observational studies, precluding firm conclusions around the effects of beta-blocker discontinuation in ADHF. We found imprecision in estimates for in-hospital mortality and long-term mortality, and found beta-blocker discontinuation may be associated with higher risk of short-term mortality up to 3 months. However, findings of possible harm may reflect confounding by indication and failure to restart therapy, rather than causal harm from discontinuation during ADHF. Until additional higher certainty data become available, this synthesis of evidence can be combined with clinical judgment to guide decision-making in practice.PerspectivesCOMPETENCY IN MEDICAL KNOWLEDGE: Based on primarily observational studies of adults hospitalized for ADHF, the effects of beta-blocker discontinuation are not clear. There may be an association between discontinuation and mortality up to 3 months after discharge; however, confounding by indication is possible and causal effects remain uncertain.TRANSLATIONAL OUTLOOK: Further clinical trials and well-designed observational studies that capture reasons for discontinuation and restart timing are needed to determine causal effects of beta-blocker discontinuation during ADHF. COMPETENCY IN MEDICAL KNOWLEDGE: Based on primarily observational studies of adults hospitalized for ADHF, the effects of beta-blocker discontinuation are not clear. There may be an association between discontinuation and mortality up to 3 months after discharge; however, confounding by indication is possible and causal effects remain uncertain. TRANSLATIONAL OUTLOOK: Further clinical trials and well-designed observational studies that capture reasons for discontinuation and restart timing are needed to determine causal effects of beta-blocker discontinuation during ADHF. Funding support and author disclosures This study was supported by Dr Goyal’s career development award (10.13039/100000049National Institute on Aging grant K76AG064428). Dr Thompson is supported by a salary award from Michael Smith Health Research BC. The funder had no role in conduct of the study or preparation of the manuscript. Dr Goyal is supported by National Institute on Aging grants K76AG064428, R21AG077092, R01AG085420, R01AG088522, and R01AG091005. Dr Goyal was a member of the Junior Investigator Intensive Program of the US Deprescribing Research Network and is supported by a US Deprescribing Research Pilot Grant funded by the National Institute on Aging (R24AG064025). Dr Goyal has received consulting fees from Alnylam Pharmaceuticals, Axon therapies, Bayer HealthCare Pharmaceuticals, Eidos/BridgeBio, and Pfizer; and has received personal fees for medicolegal consulting and expert testimony related to heart failure. Dr Thompson is supported by a salary award from Michael Smith Health Research BC. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.