This was a prospective, multicentre cohort study conducted in 32 ICUs in Spain. The size and medical activity of the participating centres varied significantly, representing the range of real-life variation in Spanish ICUs. The study protocol was developed with the support of the Spanish Society of Intensive Care Medicine and Coronary Units (SEMICYUC) and approved by the ethics committees at all participating hospitals.

From April to June 2017, all adult patients (≥ age 18) admitted to the ICUs at the participating centres for any acute condition were consecutively enrolled in the study. Patients younger than age 18 and those undergoing scheduled procedures with an ICU stay<24h were excluded. Demographic data, reason for admission, illness severity (APACHE II score), and Charlson comorbidity index from all patients were recorded in an online database.

Since this was a “real-life” study, the diagnosis of AHF was made by the treating physician according to his/her clinical judgment based on signs, symptoms and complementary testing. The lead investigator at each centre reviewed the clinical records to determine whether AHF was recorded as a final diagnosis at ICU discharge.

Patients were classified into three groups, as follows: no AHF, primary AHF (main reason for ICU admission), or secondary AHF (developed during the ICU stay). Patients classified with primary AHF could not be further classified into the secondary group.

Diagnostic delay was recorded as a dichotomous variable and defined as>12h between the onset of signs and symptoms and the diagnosis of AHF. The 12-h cut-off point was based on data reported in previous studies.11 Diagnostic delay was assessed by the main investigator at each centre based on a review of clinical data and complementary tests. Treatment was considered early when started within the first 24h after diagnosis.

Data on precipitating factors, clinical manifestations, diagnostic tests, and treatments for heart failure were recorded at the time of diagnosis in patients admitted to the ICU for AHF as well as those who developed AHF during the ICU stay, regardless of whether the condition was de novo or acute decompensation of chronic heart failure. Pulmonary oedema was defined by the presence of signs and symptoms of AHF, respiratory failure, and radiological evidence of pulmonary congestion. Cardiogenic shock was defined as the presence of signs and symptoms of AHF and evidence of hypoperfusion. Both conditions were not mutually exclusive.

Diagnosis of sepsis/septic shock was made according to the Sepsis-3 criteria.12 Fluid overload was recorded as a dichotomous variable according to the clinical judgement of the main investigator at each centre after reviewing fluid balance records. Weaning-induced pulmonary oedema was diagnosed by the treating physician according to clinical criteria.6 Stress cardiomyopathy was defined as an acute, transient systolic dysfunction due to mental or physical stress, including sepsis-induced myocardial depression, in accordance with current clinical guidelines.13 Left ventricular systolic and diastolic function were evaluated according to the current guidelines.14,15

Data verification was performed by a steering committee comprised of three senior intensive care specialists who randomly selected 20% of patients diagnosed with AHF from each participating centre during the study period. To ensure the reliability of the study data, these patients were revaluated, and a concordance analysis (Cohen's kappa) was performed to confirm the diagnosis of AHF. If the concordance analysis revealed an excessive lack of agreement (kappa<0.8) between the initial diagnosis and the reevaluated diagnosis, the data from that centre were excluded. Consequently, of the 34 ICUs that initially agreed to participate, 32 were included in the final analysis.

Due to the observational nature of the study, a formal calculation of the study sample size was not applicable.

Data are presented as numbers (%) for categorical data. Continuous variables are presented as means with 95% confidence intervals (CI) if normally distributed or medians with interquartile range (IQR) if non-normally distributed. Fisher's exact test was used to compare categorical data. For univariate comparisons of continuous variables, we used the Student's t, ANOVA, Mann–Whitney U, or Kruskal–Wallis tests as appropriate based on the data distribution. Statistical significance was set at p≤0.05.

Univariate analyses were performed to assess the demographic variables and clinical variables potentially associated with 30-day mortality. Independent factors were identified using Cox regression analysis. The following variables were entered into the model based on their clinical relevance in previous studies8,16,17: age and APACHE II score at ICU admission; presence of cardiogenic shock; administration of vasopressors or inotropes; left ventricular ejection fraction; need for invasive mechanical ventilation; and highly significant (p<0.005) variables from the univariate analysis. The IBM-SPSS statistical software program (v. 26) was used to perform all statistical analyses.

During the study period, 4400 patients were admitted to the 32 participating ICUs (Fig. 1). Of these patients, 70 did not meet the study inclusion criteria and were excluded, leaving a total of 4330 patients. Most participants (80%) were recruited from medical-surgical ICUs, followed by surgical ICUs (12%), and coronary units (8%). Of these 4330 patients, 627 (14.5%) were diagnosed with primary or secondary AHF, distributed as follows: primary AHF (n=319; 7.4%) and secondary AHF (n=308; 7.1%).

Figure 1.

Study flow chart. The flow chart shows the exclusion criteria. Acute heart failure (AHF) included patients with de novo heart failure and those with an acute decompensation of chronic heart failure. ICU: intensive care unit.

(0.12MB).

The baseline characteristics of the sample (overall and by AHF group) are shown in Table 1. Compared to the patients without AHF, those diagnosed with AHF (primary or secondary) were older (62 vs. 69 years), more severely unwell at admission (APACHE II score: 13 vs. 19) and had more comorbidities (Charlson comorbidity index: 2 vs. 4). There were no differences in baseline characteristics between the primary and secondary groups. Admission for non-ischemic heart disease (52.4%) was more common in the primary AHF group while sepsis (24.3%) was more common in the secondary AHF group.

Table 2 shows the following: clinical findings, underlying cardiac disease, precipitating factors, tests performed at the time of diagnosis, and early interventions in patients with AHF. Pulmonary oedema was the most common form of presentation (78.5%). Although ischemic heart disease was the most common underlying disease in both groups, it was more prevalent in the primary group (42% vs. 32.5%). By contrast, stress cardiomyopathy was more prevalent in the secondary group (1.9% vs. 12.7%).

The main precipitating factor was unknown in 51.7% of patients. The following precipitating factors were all more common in the secondary group: infection (18.2% vs. 29.5%), fluid overload (12.9% vs. 23.4%) and weaning-induced pulmonary oedema (0.3% vs. 8.1%).

Electrocardiogram, chest X-ray, and troponin measurement were performed more frequently in the primary group. Natriuretic peptide levels were measured in 34.6% of patients, with no significant differences between groups. Echocardiography was performed in 73.4% of patients, but more commonly in the primary group (79.3% vs. 67.2%). Determination of left ventricular ejection fraction was reported in 419 of 460 echocardiograms, which was lower in the primary group (42% vs. 45%). Diastolic function was reported in 16.3% of echocardiograms performed. Minimally invasive and invasive hemodynamic monitoring was used in a higher proportion of the secondary group (18.5% vs. 26.9%). Overall, diagnosis of AHF was delayed in 11.8% of patients, with no significant differences between the primary and secondary groups (10.3% vs.13.3%, respectively).

The most common drug for early management were diuretics, with no significant between-group differences. Nitrates (32.3% vs. 24%) and inotropes (30.4% vs. 22.4%) were used in a higher proportion of the primary AHF group. Vasopressors were prescribed in 38% of patients, with no differences between the primary and secondary groups (36.7% vs. 39.3%, respectively).

Coronary angiography was performed in a higher proportion of patients in the primary group (35.1% vs. 22.7%). Mechanical circulatory support was necessary in 2.9% of patients with AHF. The use non-invasive ventilation was higher in the primary group (40.4% vs. 27.9%). Invasive mechanical ventilation was used in 33.5% of patients, with no differences between the primary and secondary groups.

Patient outcomes are shown in Table 3. Compared to the patients without AHF, those diagnosed with AHF had longer mean ICU (6 vs. 9 days) and hospital (19 vs. 23 days) stays. Compared to the primary group, the secondary group had a longer hospital stay (19 vs. 27 days).

Patients with AHF had significantly greater risks of ICU mortality (odds ratio [OR] 2.07 [95% CI: 1.62–2.64]; p<0.001) and 30-day mortality (OR 2.45 [95% CI 1.93–3.11]; p<0.001) than patients without AHF.

Detailed results of the univariate analysis of all risk factors associated with 30-day mortality in the AHF group are shown in Table S1. On the Cox regression analysis, adjusted for confounding factors, the following variables were independently associated with 30-day mortality: increase in APACHE II, presence of cardiogenic shock, left ventricular ejection fraction, need for early inotropic therapy, and diagnostic delay (Table 4). Comparison of the impact of diagnostic delay on 30-day mortality in the two AHF groups showed a significant effect in the secondary group (OR: 6.82; [95% CI: 3.31–14.04]; p<0.001) (Fig. 2).

Figure 2.

Impact of diagnostic delay on 30-day mortality by acute heart failure group. 30-Day mortality rate by group depending on the presence (white bars) or not (grey bars) of diagnostic delay. The bars give the mean mortality rate and the lines 95% confidence intervals. The p value indicates the comparison between patients with and without diagnostic delay in each AHF subgroup. AHF: acute heart failure.

(0.09MB).

This study has several limitations. The first limitation is related to the pragmatic real-life/bedside study design in which diagnosis was based on the treating physician's clinical criteria (based on signs, symptoms, and complementary tests), with a focus on analysing clinical outcomes in routine clinical practice. Second, due to the observational nature of the study, it is important to underscore that our results only indicate an association rather than causality. Third, there may be a selection bias in the patients admitted to the ICU for AHF given the potential for differences in ICU admission criteria among the 32 participating hospitals. Fourth, diagnostic delay was registered as a dichotomous variable (present or not) and thus we do not know if longer delays are associated with worse prognosis, a question that should be explored in future studies. Finally, this study was conducted in Spain and our findings might not be applicable in other countries with different health care systems and policies.

The main strengths of our study include the prospective multicentre real-life design and the study population. This study design allowed us to analyse routine clinical practice and to identify areas of improvement in the future management of AHF patients. Finally, the selection of this specific patient population—a general cohort of critically ill patients observed throughout their ICU stay—has allowed us to identify a subpopulation (secondary AHF) not present in previous studies.