Medicines optimization prior to discharge in patients admitted to hospital with heart failure

Patient population

OPERA-HF was a prospective observational study of adults, aged >18 years, admitted with a primary diagnosis of suspected HF to a hospital serving a local community of ~500 000 people in Kingston upon Hull and East Riding of Yorkshire, UK. Other eligibility criteria included treatment with a loop diuretic and at least one of the following as evidence of cardiac dysfunction: left ventricular ejection fraction (LVEF) < 40%, left atrial diameter > 4 cm, or N-terminal pro-B-type natriuretic peptide (NT-proBNP) > 400 ng/L for those in sinus rhythm or >1200 ng/L for those in atrial fibrillation (AF). Variable NT-proBNP cut-offs were used to ensure we enrolled patients whose primary problem was HF and not lone AF (the majority of whom have NT-proBNP levels > 400 ng/L). Patients who were unable or unwilling to give informed consent were not included in the study. Ethical approval was awarded by the South Yorkshire Research Ethics Committee (REC ref.: 12/YH/0344). Patients were enrolled over 50 months between October 2012 and January 2017. The study was conducted in accordance with the Declaration of Helsinki.

Data on demographics, symptoms, bed-side observations, blood results, echocardiography, and 12-lead electrocardiograms were collected on admission and blood results and bed-side observations on the day of, or just prior to, discharge.

HeFREF was defined as an LVEF < 40% or at least moderate left ventricular systolic dysfunction by semi-quantitative visual estimation. Only patients who were alive at discharge with complete data on renal function, systolic blood pressure (SBP), and heart rate (HR) prior to discharge were included in this analysis. Estimated glomerular filtration rate (eGFR) was calculated using the Modification of Diet in Renal Disease formula.

Patients who were discharged on a combination of morphine or diamorphine, midazolam, hyoscine butylbromide, and haloperidol¹¹ or who were not prescribed any of ACE-I or ARB, beta-blocker, or MRA and who died within 30 days of discharge were considered to have been discharged for end-of-life care and were excluded from eligibility analyses. Patients in whom the HF admission was deemed to be due to severe valve disease were also excluded from the eligibility analysis.

Polypharmacy was defined as prescription of ≥5 medication types per day not including ACE-I, ARB, beta-blocker, MRA, loop diuretics, or, for patients in AF, oral anticoagulants. Medications that could be stopped were (i) those with randomized controlled trial data demonstrating absence of benefit for HF with reduced ejection fraction (amlodipine,¹² statins,¹³ and oral iron¹⁴) and (ii) the potential to cause harm based on the American Heart Association list of medicines that have a detrimental effect in patients with HF. These include non-steroidal anti-inflammatory drugs, thiazolidinediones, dipeptidyl peptidase 4 inhibitors, doxazosin, and citalopram.¹⁵

Eligibility for sacubitril–valsartan, dapagliflozin, and empagliflozin was derived from the PIONEER-HF,¹⁶ DAPA-HF,¹⁷ and EMPEROR-Reduced¹⁸ trial entry criteria, respectively. Patients who were taking ACE-I or ARB, beta-blocker, or MRA were split into three groups (Steps 1–3) based on stages of up-titration derived from the ESC HF guidelines (Supporting Information, Table S1).⁶ Eligibility for initiation or up-titration of ACE-I, ARB, beta-blocker, or MRA was based on two different sets of criteria: (i) advice on when to be cautious outlined in the supplementary data appendix for the ESC HF guidelines 2021 (do not initiate or up-titrate ACE-I, ARB, or MRA if SBP < 90 mmHg, potassium > 5 mmol/L, eGFR < 30 mL/min/1.73 m², or serum creatinine concentrations > 221 mmol/L; and do not initiate or up-titrate beta-blocker if HR < 50 b.p.m. or SBP < 90 mmHg)¹⁹; and (ii) more conservative criteria using a higher SBP cut-off of <105 mmHg and less aggressive HR cut-offs (<70 b.p.m. for those in sinus rhythm and <80 b.p.m. for those in AF). Loop diuretic dose was defined as furosemide equivalents where 40 mg of furosemide ≈ 1 mg of bumetanide.

Outcome data were collected from the hospital's electronic records. Outcomes assessed were all-cause mortality, and all-cause mortality or first readmission for HF during the first year from discharge. Deaths during a first readmission were counted as deaths in the first-event composite outcome.

Statistical analysis

eGFR was calculated using the Modification of Diet in Renal Disease formula. Worsening renal function (WRF) was defined as an increase in serum creatinine of >26.5 μmol/L during admission.²⁰ Categorical data are presented as percentages, and continuous data are presented as median (25th and 75th centiles). Chi-squared tests were used to compare categorical variables. Independent samples t-test and one-way analysis of variance were used to compare normally distributed continuous variables. The Mann–Whitney U and Kruskal–Wallis tests were used to compare non-normally distributed continuous variables. Associations between variables and outcome were assessed using Kaplan–Meier curves and a Cox proportional hazard regression model, which included a small number of variables to avoid statistical overfitting chosen a priori (age, sex, eGFR, haemoglobin, and log[NT-proBNP]).

Prescribing patterns by age, sex, and renal function

Older patients were less likely to be taking disease-modifying medication. Women were less likely to be taking an MRA or higher dose loop diuretic (≥80 mg/day of furosemide equivalents) and more likely to be taking regular analgesia than men. Patients with an eGFR < 60 mL/min/1.73 m² were less likely to be taking an MRA and more likely to be taking a thiazide in addition to a loop diuretic, hypoglycaemic medicines, proton pump inhibitors, oral iron, or xanthine oxidase inhibitor than those with an eGFR ≥ 60 mL/min/1.73 m². However, there was no difference in the rate of prescription of ACE-I or ARB, or beta-blocker based on renal function (Table 1).

Despite great variation of medications prescribed on admission, most patients prescribed an ACE-I or ARB prior to discharge were taking either enalapril or ramipril, and the most patients prescribed a beta-blocker were taking either carvedilol or bisoprolol (Supporting Information, Figure S2).

Eligibility for sacubitril–valsartan or sodium-glucose co-transporter-2 inhibitor

Sixty-two per cent, 76%, and 71% of patients were potentially eligible to start on sacubitril–valsartan, dapagliflozin, and empagliflozin prior to discharge, respectively (Table 1). There was no difference in eligibility based on age, but women were more likely than men to be eligible for empagliflozin (Table 2). The most common reason for ineligibility was low SBP followed by low eGFR (Figure 2). No patient would have failed eligibility for trial inclusion based on their plasma NT-proBNP being insufficiently elevated.

Patients eligible for sacubitril–valsartan or dapagliflozin were less likely to have an LVEF < 30% or inpatient WRF than those deemed ineligible. Patients eligible for empagliflozin were more likely to have a diagnosis of HF prior to admission and less likely to have an LVEF < 30%, but there was no difference in the proportion of patients with WRF. Patients eligible for either class of medication had lower all-cause mortality rates in the following year than those deemed ineligible.

Medicines optimization

Using ESC guideline criteria, ACE-I or ARB could have been initiated in approximately half of patients not taking these medicines; using our more conservative criteria, this fell to about one-third. Using ESC guideline criteria, beta-blockers could have been initiated in most patients, but only one in four using our more conservative cut-offs. MRA could have been initiated in approximately half of patients regardless of which criteria were applied (Table 3 and Supporting Information, Figure S3A–C). Renal dysfunction, hyperkalaemia, cough, lung disease, and gynaecomastia were not common reasons for non-prescription of medications (Table 3).

From the total population, we identified 299 patients (66%) who should have had an ACE-I or ARB either initiated or up-titrated and 358 patients (79%) who should have had a beta-blocker initiated or up-titrated (Supporting Information, Figure S4). Using higher SBP and HR cut-offs, the proportion of patients in whom an ACE-I or ARB, or beta-blocker might have been initiated or up-titrated decreased to 45% and 24%, respectively (Table 3 and Supporting Information, Figures S3A–C and S5A,B).

Almost all patients (96%) were taking ≥5 medications. If ACE-I or ARB, beta-blocker, MRA, loop diuretic, and oral anticoagulation were subtracted, the proportion taking ≥5 medications fell to 68%, but 51% were taking medications either with no known benefit or with potential harm (Table 1). Although only a minority of patients was taking a medicine that might exacerbate HF (N = 23), even after adjusting for age, sex, renal function, haemoglobin, and NT-proBNP, these patients were at two-fold higher risk of 1 year all-cause mortality or HF hospitalization [hazard ratio = 2.05 (95% confidence interval 1.22–3.45); χ² = 7; P = 0.007] compared with those not taking such medication.

Outcome

All patients had at least 1 year follow-up during which time, 106 patients died (23%), and a further 118 were readmitted with HF (Supporting Information, Table S2). Patients prescribed an ACE-I or ARB prior to discharge were approximately half as likely to experience either endpoint, independent of age, sex, renal function, haemoglobin, log[NT-ProBNP], and other medication prescriptions (Supporting Information, Table S3). Findings were similar for beta-blockers with a similar trend for MRAs. Compared with those taking only a single agent, those taking a combination of ACE-I or ARB plus beta-blocker had a lower risk of death during the first year following discharge [hazard ratio = 0.51 (0.28–0.93); P = 0.03]. Those taking triple therapy were at even lower risk [hazard ratio = 0.34 (0.19–0.59); P < 0.001] (Figure 3).

Inpatient initiation and titration: can it be done?

Inpatient initiation and titration: should it be done?

Of those who survive to discharge, one in four patients will be readmitted and one in eight will die within a month of admission.^{9, 31} Quadruple therapy can reduce the risk of cardiovascular death or HF hospitalization by 62% compared with treatment with ACE-I or ARB plus beta-blocker alone,¹ and the time to onset of benefit of several HF medications is known to be <1 month.^2-4 As a result, guidelines recommend that patients are seen within 2 weeks of discharge by an HF specialist,⁶ but fewer than half of care providers can meet this target.⁹ HF specialist nurses are ideally placed to assess patients soon after discharge, but their availability is highly variable.³² Many patients remain on the doses of medication on which they were discharged until they are readmitted.³³ Hospital admission is thus a vital opportunity to start (and perhaps up-titrate) HF medications.

Clinical inertia in the face of apparent clinical stability is often cited as a reason for failure to prescribe or up-titrate life-prolonging medications in outpatients with HeFREF.³⁴ Similar mechanisms may be at play for inpatients. Access to specialist care is another important factor: patients under the care of a cardiologist are far more likely to be discharged on triple therapy than those who do not see an HF specialist as an inpatient.⁹ Ageism may also play a role; the proportion of patients prescribed life-prolonging medication decreases with advancing age.^{9, 29, 30}

Unwillingness to interfere with the medication regimen of a patient with HF is understandable; those who survive to discharge have often been treated with high doses of intravenous diuretic for more than a week.^{9, 30} Neither the patient nor clinician wishes the hospital stay to be longer than necessary: each day in hospital with HF costs the National Health Service in the United Kingdom ~£450,³⁵ and nosocomial infection (including COVID) is a common and often fatal complication.³⁶ There is a difficult balance to strike.

It is unrealistic, and potentially dangerous, to attempt to initiate and up-titrate all medications to target doses during an inpatient stay^{7, 8}; such a strategy has never been tested in a prospective trial. However, SGLT2Is do not require titration, and spironolactone potentially only requires one increment in dose from 25 mg/day. Titration of other disease-modifying agents is more complex. Hospital admission may be the best opportunity for initiation and pragmatic initiation and up-titration of quadruple therapy, while also providing a chance to assess tolerability. The selection and timing of device therapies also needs to be considered.⁶

Discontinuation and polypharmacy

Polypharmacy, and its consequences (such as non-adherence),³⁷ is an increasing problem for patients with HeFREF and those who care for them. Even excluding useful polypharmacy (medications that have prognostic or symptomatic benefit for HF or related conditions), we found that over half of patients met the definition for polypharmacy (≥5 medication types per day).¹⁰ Furthermore, we found that ~50% of patients were taking medications with no proven benefit or that were potentially harmful. We cannot infer that it would have been appropriate to stop such medications in all patients, but our data suggest that there is scope for discontinuing some as part of ‘medicines optimization’ prior to discharge.

Limitations

Although our cohort was enrolled prospectively, the limitations of post hoc analyses apply, and we cannot exclude confounding factors. For example, although we had data on some medications given during admission such as intravenous diuretics, nitrates, and inotropic agents, we did not have data on dose or duration of loop diuretics. Although we did not have comprehensive data on symptoms or signs of congestion on either admission or discharge to include in the final analysis, we assume that very few patients were discharged with a substantial amount of congestion as this is against guideline recommendations.⁶

Our data are from a single centre; thus, our conclusions might not be generalizable, particularly outside the United Kingdom. However, our data are similar to those published by the National Heart Failure Audit⁹ and in registries elsewhere in Europe and beyond.^{29, 30}

We also lacked more detailed echocardiographic assessment beyond classification of the severity of left ventricular systolic dysfunction. Treatments such as MRA, sacubitril–valsartan, and SGLT2I have only been investigated in patients with an LVEF ≤ 35%, but our population matched the definition of HeFREF in the most recent ESC HF guidelines, which advocate treatment with each of those medications for patients with LVEF < 40%.⁶

Finally, although observational studies can only demonstrate associations, failure to prescribe guideline-recommended therapies indicates a poor outcome, either because the medications are contraindicated or because someone has neglected to prescribe them.