This was a quasi-experimental design study carried out in a single center (the cardiology department of Santa Maria University Hospital, Lisbon, Portugal).

The study population consisted of 50 consecutive patients admitted for acute heart failure (AHF), defined as new-onset AHF or decompensated chronic heart failure, to the general cardiology ward (index hospitalization), who were discharged after the implementation of a protocol-based follow-up program (beginning in April 2016). The diagnosis of HF was established according to the ESC guidelines, through the identification of symptoms and/or signs of HF caused by a structural and/or functional cardiac abnormality, resulting in reduced cardiac output and/or elevated intracardiac pressures.19

The control group consisted of patients selected from a cohort hospitalized for AHF in the same cardiology ward immediately before the beginning of the program (from October 2014 to April 2016). Patients from both cohorts (study group and control group) were classified according to New York Heart Association functional class (NYHA) at discharge (NYHA I vs. NYHA II vs. NYHA III or IV), left ventricular ejection fraction (LVEF) (tertiles), and age (tertiles). For each of the patients in the study group a patient from the control cohort with similar scores in each of the three variables was randomly selected.

Demographic, clinical, laboratory, echocardiographic, electrocardiographic and therapeutic data regarding the index hospitalization and the follow-up period after discharge were collected for the study group and for all patients who constituted the control cohort. Quality of life was assessed at discharge and at six-month follow-up using the validated Portuguese version of the Kansas City Cardiomyopathy Questionnaire (KCCQ).20

The main differentiating intervention in the protocol-based follow-up program was consultations by a cardiologist at days 7-10 and the first, third, sixth and 12th month after discharge (and additionally, whenever considered necessary) with pre-specified procedures. These included:

• (a)

  Clinical assessment aimed at identifying signs or symptoms of HF decompensation, residual congestion or low cardiac output;

• (b)

  Laboratory assessment, including monitoring of plasma N-terminal pro-brain natriuretic peptide (NT-proBNP) level, end-organ dysfunction, and development of common comorbidities in HF patients (diabetes, chronic pulmonary disease, dyslipidemia, thyroid dysfunction, anemia, iron deficiency)19;

• (c)

  Electrocardiogram at every visit, and transthoracic echocardiogram between the third and sixth months and every 12 months of follow-up; when considered necessary, cardiac magnetic resonance imaging study was requested;

• (d)

  Assessment of adherence and tolerance to therapy;

• (e)

  Individualized titration of therapy in accordance with the ESC guidelines19;

• (f)

  Patient education regarding self-care, lifestyle modifications, and management of HF decompensation.19

The primary outcome was all-cause readmission. HF readmission, death and the composite endpoint of all-cause readmission or death were secondary outcomes.

In the structured follow-up program group, changes in quality of life parameters was also considered a secondary outcome.

Also, in the study group, LVEF change was assessed in the subgroup of patients with LVEF <50%, and the prescription rate of neurohormonal antagonists and changes in their respective doses during follow-up was assessed in the subgroup of patients with HF with reduced LVEF (HFrEF) (LVEF <40%).

Assuming that the estimated annual rate of all-cause readmission would be 65% in the control group and 35% in the study group (based on HF populational studies10,13 and HF post-discharge programs studies,15 respectively), it was estimated that 42 patients would need to be followed in each group for 12 months to provide the study with a power of 80% to detect a significant relative reduction in the risk of all-cause readmission in the follow-up program group, at an overall two-sided alpha level of 0.05.

The statistical analysis was performed using IBM SPSS® Statistics version 20 (Chicago, IL, USA). Categorical variables are reported as absolute number and percentage and continuous variables are reported as mean and standard deviation or median and interquartile range. The impact of inclusion in the structured follow-up program on readmission and mortality rates was assessed using Cox regression and Kaplan-Meier survival analysis. Wilcoxon's test was used to assess the impact of the follow-up program on quality of life, doses of neurohormonal antagonists and LVEF. Differences between the groups regarding demographic, clinical and therapeutic data were established using the Mann-Whitney, Student's t, chi-square, one-way ANOVA and Fisher's exact tests. p values of <0.05 were considered to indicate statistical significance.

The study was approved by the local ethics committee and by the national Data Protection Authority. Patient confidentiality was ensured through anonymization of the collected data. All study procedures were carried out in accordance with the ethical principles expressed in the 2013 revision of the Declaration of Helsinki.21

The first patient was enrolled in the protocol-based follow-up program in April 2016 and the 50th patient in November 2017. The mean follow-up was 11±5.3 months in the study group and 10.9±5.5 months in the control group (p=0.81).

Patients’ demographic and clinical characteristics at discharge are described in Table 1.

The mean age of the follow-up program group was 67.1±11.2 years and 38 patients (76%) were male. Most patients were in NYHA I (30%) or II (64%) at discharge; all patients were in NYHA III (52%) or IV (48%) on admission (index hospitalization). The median LVEF documented at discharge was 27% (19.8-35.3), and 43 (86%) patients had HFrEF. There were no significant differences between the two groups regarding age, NYHA or LVEF.

The most frequent HF etiology in both groups was idiopathic dilated cardiomyopathy (56% vs. 40%), followed by ischemic heart disease (28% vs. 28%). Overall, 70% of the patients had a history of hypertension, making it the most common comorbidity in both groups.

Median length of stay and median plasma NT-proBNP at discharge did not differ significantly between groups.

Compared to patients in the control group, those included in the structured follow-up program showed a significant reduction in all-cause readmissions (26% vs. 60%; hazard ratio [HR] 0.38 [0.2-0.73]; p=0.003) (Figure 1A). The RRR was 56.7% and the number needed to treat (NNT) was 2.91. A similar benefit was achieved in HF readmission (RRR: 64.4%; NNT: 3.45) (Figure 1B). Eighteen (36%) patients in the control group were hospitalized due to AHF. Implementation of the protocol-based follow-up program led to an HF readmission rate of 16% (eight patients) (HR 0.4 [0.17-0.92]; p=0.032).

Figure 1.

Kaplan-Meier survival curves for all-cause readmission (A), heart failure readmission (B), death (C), and death or admission (D).

(0.26MB).

Mortality was significantly lower in patients enrolled in the follow-up program (4% vs. 20%; HR 0.21 [0.05-0.96]; p=0.044). The RRR was 80% and the NNT was 6.25 (Figure 1C).

During follow-up two (4%) patients in the study group died, one due to right ventricular failure in the immediate postoperative period following cardiac surgery, and the other due to sudden death. The latter patient underwent ICD implantation during the index hospitalization and sudden death occurred in the first week after hospital discharge, before the first protocol-based follow-up visit. Autopsy was not performed and interrogation of the ICD showed no dysrhythmia or evidence of device dysfunction. In the control group, two (4%) patients died during HF hospitalization, and four (8%) during hospitalization due to other causes.

The secondary outcome of death or all-cause hospitalization was significantly less frequent in the follow-up program group (28% vs. 68%; HR 0.36 [0.19-0.67]; p=0.001), with an NNT of only 2.5 (Figure 1D).

In the study group, a significant improvement was observed in all KCCQ domains, especially in the overall summary scores for symptoms (67% vs. 89%, p<0.001) and quality of life (66% vs. 80%, p<0.001) (Table 2).

Parallel to the improvement in the KCCQ symptoms domain reported by the patients, there was a significant improvement in NYHA class documented by the cardiologist in the last follow-up visit compared to NYHA class at discharge. In the last clinical assessment performed most patients were in NYHA I (64% vs. 30%, p<0.001) (Figure 2).

Figure 2.

Changes in New York Heart Association (NYHA) functional class during follow-up.

(0.1MB).

As stated above, the majority of patients in both groups presented HFrEF. Table 3 describes ongoing therapy (neurohormonal antagonists and cardiac devices) at the time of the last clinical assessment.

The rate of prescription of beta-blockers, mineralocorticoid receptor antagonists (MRAs) and ivabradine did not differ between the two groups. However, a significantly higher rate of prescription of angiotensin-converting enzyme inhibitors (ACEIs)/angiotensin receptor blockers (ARBs) was observed in patients included in the follow-up program group compared to controls (100% vs. 85%, p=0.01). There were no significant differences regarding CRT or ICD implantation rates (Table 3).

Considering the whole population, there were no significant differences in diuretic therapy prescription between groups (44 vs. 47 patients, p=NS). All these patients were medicated with loop diuretics; four patients in the follow-up program group and six patients in the control group were under an association of a loop plus a thiazide diuretic (p=NS).

There were no differences in myocardial revascularization procedures, either percutaneous angioplasty (6% vs. 8%, p=NS), or coronary artery bypass graft surgery (2% in both groups, p=NS). Rates of aortic (6% vs. 8%, p=NS) and mitral (6% vs. 4%, p=NS) valve interventions (percutaneous or surgical) were also similar in both groups.

In the study group there was an effective optimization of treatment during follow-up, including up-titration of recommended drugs. Significant increases in the doses of ACEIs/ARBs (p=0.001), beta-blockers (p<0.001) and MRAs (p<0.001) (Figure 3A-C, respectively) were observed between admission and the last follow-up visit.

Figure 3.

Changes in the doses of angiotensin-converting enzyme inhibitors (ACEIs)/angiotensin receptor blockers (ARBs) (A), beta-blockers (B) and mineralocorticoid receptor antagonists (MRAs) (C) during follow-up.

(0.2MB).

In the subgroup of patients with HFrEF or HF with mid-range ejection fraction enrolled in the protocol-based follow-up program (n=49), there was a significant improvement in LVEF during follow-up (27% [19.8-35.3] vs. 39.5% [29.5-50]; p<0.001).

The data reported should be interpreted in the light of certain limitations, particularly the fact that this was not a randomized controlled study and that the sample size was small (the first 50 patients enrolled and followed by protocol). Besides, only patients admitted to the cardiology department were included, which naturally entails selection bias, resulting in a population with a higher proportion of patients with HFrEF and younger than those reported in studies that included patients admitted to internal medicine departments.24 Additionally, cost-effectiveness analysis was not performed, so it is not possible to determine whether the reduction in events during the follow-up period was accompanied by a reduction in costs attributed to HF. However, as significant reductions in the admission rate were obtained, this probably translated into reductions in costs related to the syndrome and to associated comorbidities. Finally, the study only included patients with a recent admission due to AHF, so it is not possible to assess the impact of this follow-up program on stable patients with chronic HF but with no previous recent HF-related hospital admissions. As reported, most adverse events in HF patients occur in the first year after hospital discharge.10,11,13 In fact, despite the small number of patients who completed more than 12 months of follow-up in our population, through the data obtained from the Kaplan-Meier survival curves (Figure 1A-D), it may be assumed that the incidence of major adverse events during the second year of follow-up would have been low, suggesting a reduced benefit of this type of program in stable patients, as previously suggested.38,39 However, we consider that more evidence is needed in order to establish the ideal duration of this type of program, according to patients’ clinical profile and disease progression.