Skip to main content
Download PDF

Validation and derivation of short-term prognostic risk score in acute decompensated heart failure in China

BMC Cardiovascular Disorders volume 22, Article number: 307 (2022) Cite this article

Abstract

Background

Few prognostic risk scores (PRSs) have been routinely used in acute decompensated heart failure (ADHF). We, therefore, externally validated three published PRSs (3A3B, AHEAD, and OPTIME-CHF) and derived a new PRS to predict the short-term prognosis in ADHF.

Methods

A total of 4550 patients from the Heb-ADHF registry in China were randomly divided into the derivation and validation cohorts (3:2). Discrimination of each PRS was assessed by the area under the receiver operating characteristic curve (AUROC). Logistic regression was exploited to select the predictors and create the new PRS. The Hosmer–Lemeshow goodness-of-fit test was used to assess the calibration of the new PRS.

Results

The AUROCs of the 3A3B, AHEAD, and OPTIME-CHF score in the derivation cohort were 0.55 (95% CI 0.53–0.57), 0.54 (95% CI 0.53–0.56), and 0.56 (95% CI 0.54–0.57), respectively. After logistic regression analysis, the new PRS computed as 1 × (diastolic blood pressure < 80 mmHg) + 2 × (lymphocyte > 1.11 × 109/L) + 1 × (creatinine > 80 μmol/L) + 2 × (blood urea nitrogen > 21 mg/dL) + 1 × [BNP 500 to < 1500 pg/mL (NT-proBNP 2500 to < 7500 pg/mL)] or 3 × [BNP ≥ 1500 (NT-proBNP ≥ 7500) pg/mL] + 3 × (QRS fraction of electrocardiogram < 55%) + 4 × (ACEI/ARB not used) + 1 × (rhBNP used), with a better AUROC of 0.67 (95% CI 0.64–0.70) and a good calibration (Hosmer–Lemeshow χ2 = 3.366, P = 0.186). The results in validation cohort verified these findings.

Conclusions

The short-term prognostic values of 3A3B, AHEAD, and OPTIME-CHF score in ADHF patients were all poor, while the new PRS exhibited potential predictive ability. We demonstrated the QRS fraction of electrocardiogram as a novel predictor for the short-term outcomes of ADHF for the first time. Our findings might help to recognize high-risk ADHF patients.

Peer Review reports

Background

Acute decompensated heart failure (ADHF) is a prevalent clinical syndrome with a poor prognosis in internal medicine and emergency department, and the majority of ADHF patients require hospitalization for further treatment [1, 2]. Despite remarkable therapeutic advancements, the mortality and rehospitalization rates of ADHF remain high [3], most conspicuously during hospitalization and the early post-discharge period [4,5,6], dubbed “the vulnerable phase” for heart failure (HF) [7]. Data from China show that ADHF is the leading cause of hospitalization in patients > 65 years old, with hospitalization mortality of 3% and an around 50% short-term readmission rate [8].

It is critical to accurately predict prognosis in patients with ADHF during and after hospitalization to optimize treatment decisions, thereby reducing readmissions and deaths [9]. For these purposes, several prognostic risk scores (PRSs) in patients with HF have been developed [10,11,12,13]. However, most of these scores have not been thoroughly externally validated and the effective values of predictors are differ among races and regions [14, 15]. Moreover, current scoring systems paid less attention to outcomes in patients with ADHF during the vulnerable phase, and it is unclear whether they can be directly applied to ADHF patients.

Among the previously published PRSs, the 3A3B score is developed for the Japanese HF preserved left ventricular ejection fraction (LVEF) (HFpEF). From a total of 14 covariates, age, albumin, anemia, BMI, BNP or NT-proBNP, and BUN were selected as long-term prognostic variables [13]. The predictive value was confirmed in the external validation in Asian cohorts. The discrimination abilities were all excellent in both derivation and validation cohorts (c-index = 0.708). However, the accuracy of this score in predicting short-term prognosis in ADHF patients is uncertain.

The AHEAD score is a simple tool based on 5 comorbidities (Atrial fibrillation, Hemoglobin (anemia), Elderly, Abnormal renal parameters (creatinine), and Diabetes mellitus) used to predict the short and long term prognosis of hospitalized patients with acute HF (AHF) in the European population. It was derived from a prospective multicenter registry. The score was externally validated in the GREAT registry consisting of nine AHF cohorts from Italy, Spain, France, Argentina, Finland, Switzerland, the USA, Tunisia, and Austria [12]. Nevertheless, its prognostic efficacy in Asian patients with ADHF has not been proven.

The OPTIME-CHF score has been suggested as a way to predict the risk of 60-day mortality for hospitalized patients with decompensated HF. Using data from a cohort from the United States, 5 routine variables on admission, including age, SBP, sodium, New York Heart Association (NYHA) class IV, and BUN, were found to be independently associated with the prognosis. The discrimination of the current score was excellent (c-index = 0.77). The authors further developed a nomogram based on each factor for bedside application [10], which, however, was not externally validated.

Considering the research context (population and regions), as well as the availability of variables, we selected the 3A3B [13], AHEAD [12], and OPTIME-CHF [10] scores as the alternative PRSs. This study therefore aimed to validate whether the 3A3B, AHEAD, and OPTIME-CHF could accurately predict the composite of in-hospital all-cause mortality, 30-day all-cause readmission, or 30-day all-cause mortality after discharge in ADHF patients. Meanwhile, we derived a new PRS and compared it with the existing systems. We believe this will contribute to the prognostic risk assessment of ADHF patients.

Methods

Study population

The study population consisted of hospitalized patients with ADHF enrolled in the Hebei Acute Decompensated Heart Failure (Heb-ADHF) registry (ChiCTR-POC-17014020) database, a prospective, multicenter, open study designed in real-world to assess risk predictors of influencing comprehensive treatment and short-term prognosis in patients with ADHF. From March 2016 and December 2018, patients were consecutively recruited in 13 tertiary hospitals in Hebei Province, China.

ADHF was defined as new-onset AHF or decompensation of chronic HF (CHF) [16]. All of the patients in this study were hospitalized with ADHF. Patients were eligible for inclusion if they were: (1) age ≥ 18 years; (2) unplanned admission; (3) with typical ADHF symptoms or signs; and (4) brain natriuretic peptide (BNP) > 100 pg/mL or N-terminal pro-brain natriuretic peptide (NT-proBNP) > 300 pg/mL. Diagnosis and treatment of ADHF were determined by the cardiologist-physicians according to the clinical guideline [8]. Exclusion criteria were: (1) hospital stay < 24 h (refuse further treatment or request a referral to higher-level hospitals); (2) heart transplantation; (3) renal replacement therapy; (4) massive stroke; (5) concomitant terminal disease; or (6) patients lost to follow-up. All eligible subjects were randomly divided into the derivation cohort and the validation cohort at a ratio of 3:2. The derivation cohort was used to validate the published PRSs and establish the new model, whereas the validation cohort was used to validate the findings.

Data collection

We recorded the following clinical data on admission of each enrolled patient: (1) demographic characteristics; (2) clinical history; (3) physical examination; (4) laboratory tests; (5) chest radiography (6) electrocardiogram; (7) echocardiography; and (8) medical treatment. Variables analysis of the outcome events group revealed the following cut-offs: age, 65 year-old; body mass index (BMI), 24 kg/m2; length of hospital stay (LoHS), 10 days; systolic blood pressure (SBP), 130 mmHg; diastolic blood pressure (DBP), 80 mmHg; anemia, < 13.0 g/dL in men or < 12.0 g/dL in women; red blood cells (RBC), 4.15 × 1012/L; neutrophil, 4.37 × 109/L; lymphocyte, 1.11 × 109/L; platelet, 155 × 109/L; total cholesterol (TC) 3.6 mmol/L; low density lipoprotein cholesterol (LDL-C), 2.0 mmol/L; creatinine, 80 μmol/L; blood urea nitrogen (BUN), 21 mg/dL; aspartate aminotransferase (AST), 32 U/L; alanine aminotransferase (ALT), 59 U/L; QRS fraction, 55%; left ventricular ejection fraction (LVEF), 36%; left atrial diameter (LAD), 41 mm. Phase values were used for BNP (NT-proBNP), 100 to < 500 (300 to < 2500), 500 to < 1500 (2500 to < 7500), and ≥ 1500 (≥ 7500) pg/mL, respectively. QRS fraction is defined as the sum of the R-wave amplitudes of the standard 12 leads (ΣR) divided by the sum of the absolute values of the QRS wave amplitudes of the 12 leads (ΣQRS) in electrocardiogram, i.e. (ΣR/ ΣQRS) × 100%. All cut-offs were calculated based on the predictive power of each variable for the outcome event.

Study endpoint and follow-up

The primary endpoint was the composite of in-hospital all-cause mortality, 30-day all-cause readmission, or 30-day all-cause mortality after discharge. The follow-up programs for all patients were carried out according to the original plan of the Heb-ADHF study via medical records, physician’s office visits, or telephone interviews from admission to 30 days after discharge.

Statistical analysis

Continuous variables are presented as means ± standard deviations or medians with interquartile range whenever appropriate; categorical variables are presented as frequencies (n) and proportions (%). The few missing data are replaced by the expectation–maximization method. The Student t-test or Mann–Whitney U test was used to evaluate differences between groups for continuous variables and the Chi-square (χ2) test for categorical variables whenever appropriate. The discrimination of each PRS was evaluated by the area under the receiver operating characteristic curve (AUROC) with a 95% confidence interval (CI) to assess ability to predict the endpoint. To begin, univariate logistic regression analyses were used to identify risk factors associated with outcome events. Variables with P-values of < 0.2 in univariate analyses were entered into the multivariable analysis. Then, variables with P-values of < 0.05 in the multivariate logistic regression were considered as independent variables for the new PRS. Odds ratios (ORs) with 95% CIs were calculated for the logistic regression analysis. The relative significance of each predictor within the final PRS model was assessed by the value of the partial Wald χ2 statistic minus the degrees of freedom (df) of predictors (χ2-df). In the new PRS model, we assigned the scores (points) for each independent variables based on their value of (χ2-df): (χ2-df) ≤ 5.0 assigned as 1 point, 5.0 < (χ2-df) ≤ 10 assigned as 2 points, 10 < (χ2-df) ≤ 15 assigned as 3 points, and (χ2–df) > 15 as 4 points. The Hosmer–Lemeshow goodness-of-fit test was used to assess calibration. Using the Z test, we also compared the predictive accuracy of new PRS with that of the 3 previously published PRS by AUROCs. The correlation between different score points and different outcome events was analyzed by Mantel–Haenszel trend test and Pearson correlation test. Two-sided P < 0.05 was deemed statistically significant. Statistical analyses were performed using SPSS software (version 26.0) and the AUROC curves were conducted using the Medcalc software (version 20.0.3).

Results

Population characteristics

Patients including 43 discharged within 24 h, 1 with heart transplant, 11 with renal replacement, 2 with massive stroke, 4 with malignant tumors, and 135 lost to follow-up were excluded. Finally, a total of 4550 patients from the Heb-ADHF registry were enrolled in this study, who were randomly divided into the derivation cohort (2,745) and the validation cohort (1,805) at a ratio of 3:2. The incidence of the composite of outcomes was 13.9% (381/2745) in the derivation and 13.9% (249/1805) in the validation cohorts. For both cohorts, the age, male, BMI, smoking, drinking, HF duration, number of hospitalizations for heart failure (NH-HF), LoHS, SBP, DBP, heart rate, etiology, heart function, comorbidities, urine protein, laboratory blood tests, chest radiography, electrocardiogram, echocardiography, medical treatment, and risk of outcome events were comparable without significant difference (P > 0.05). The clinical characteristics of derivation and validation cohorts are summarized in Table 1.

Table 1 Comparison of clinical characteristics between the derivation and validation cohorts
Full size table

Validation of previous scores in the derivation cohort

The results showed that the discriminations of 3A3B, AHEAD, and OPTIME-CHF score were all poor on the composite of in-hospital all-cause mortality, 30-day all-cause readmission, or 30-day all-cause mortality after discharge in the derivation cohort, with AUROCs of 0.55 (95% CI 0.53–0.57), 0.54 (95% CI 0.53–0.56), and 0.56 (95% CI 0.54–0.57), respectively. Only the specificity of 3A3B score was acceptable (85.3%), but the sensitivity was unsatisfactory (22.0%).

Logistic regression analysis and the new risk score model

Logistic regression analysis was performed in the derivation cohort (Table 2). Following univariate analysis, 27 variables (P < 0.2) entered into the multivariable analysis: sex (male), BMI, LoHS, DBP, anemia, RBC, neutrophil, lymphocyte, platelet, TC, LDL-C, creatinine, BUN, AST, ALT, cTnI/T, BNP (NT-proBNP), QRS fraction, pulmonary congestion (X-ray), LVEF, LAD, ACEI/ARB, β-blocker, Tolvaptan, recombinant human brain natriuretic peptide (rhBNP), calcium channel blockers, and Statin. However, multivariate analysis revealed that only 8 variables (P < 0.05) [DBP, lymphocyte, creatinine, BUN, BNP (NT-proBNP), QRS fraction of electrocardiogram, ACEI/ARB, and rhBNP] maintained an independent correlation with the composite endpoint, which were included into the development of the final model.

Table 2 Logistic regression analysis of the composite outcomes in the derivation cohort
Full size table

We calculated the relative importance of each independent predictor using the value of (χ2-df) (Fig. 1). The score was then assigned based on the (χ2-df) value of each independent predictor (Table 2). The new scoring system was as follows:

Fig. 1
figure 1

Relative importance of individual predictors within the final risk model. The relative importance of each predictor was calculated from the Wald chi-square (χ2) minus the predictor’s degrees of freedom (df) (χ2-df)

Full size image

1 × (DBP < 80 mmHg) + 2 × (lymphocyte > 1.11 × 109/L) + 1 × (creatinine > 80 μmol/L) + 2 × (BUN > 21 mg/dL) + 1 × [BNP 500 to < 1500 pg/mL (NT-proBNP 2500 to < 7500 pg/mL)] or 3 × [BNP ≥ 1500 (NT-proBNP ≥ 7500) pg/mL] + 3 × (QRS fraction of electrocardiogram < 55%) + 4 × (ACEI/ARB not used) + 1 × (rhBNP used).

Discrimination, calibration, and comparison of the new risk score.

The discrimination ability of the new PRS model was adequate, with an AUROC of 0.67 (95%CI 0.64–0.70) for the derivation cohort (Fig. 2A), and a good calibration (the Hosmer–Lemeshow test χ2 = 3.366, P = 0.186). The discrimination was validated by the AUROC for the validation cohort [0.65 (95% CI 0.61–0.69)] (Fig. 2B), and the calibration was likewise good (the Hosmer–Lemeshow test χ2 = 9.751, P = 0.283). In addition, the AUROCs of the new PRS in both the derivation and validation cohorts were the highest compared with the 3A3B, AHEAD, and OPTIME-CHF score systems (Fig. 2), of which the differences were statistically significant (P < 0.001; Table 3).

Fig. 2
figure 2

The receiver operating characteristic curve (ROC) for scoring models in the derivation and validation cohort. A The ROC for scoring models in the derivation cohort; B The ROC for scoring models in the validation cohort

Full size image
Table 3 Comparison of AUROCs between the new risk score and previous risk scores for predicting outcomes
Full size table

The risk stratification of the new risk score model in the entire cohort.

The new PRS was then used to classify patients into 3 groups: 0–4 points, low-risk group; 5–8 points, medium-risk group; ≥ 9 points, high-risk group. The incidence of in-hospital all-cause mortality in low-, medium-, and high-risk groups was 1.86% (14/754), 4.07% (74/1820), and 5.57% (110/1976), respectively (Fig. 3A). The incidence of 30-day readmission in low-, medium-, and high-risk groups was 3.18% (24/754), 6.43% (117/1820), and 11.03% (218/1976), respectively (Fig. 3B). The incidence of 30-day all-cause mortality after discharge were 0.53% (4/754), 1.04% (19/1820), and 2.53% (50/1976), respectively (Fig. 3C) and the composite of outcome events were 5.57% (42/754), 11.54% (210/1820), and 19.13% (378/1976), respectively (Fig. 3D). Mantel–Haenszel test revealed a linear trend between the risk stratification and the incidence of outcomes events (χ2 = 38.14, P < 0.001). Pearson correlation test indicated that the incidence of outcome events increased with the increase of the risk stratification (R = 0.480, P < 0.001).

Fig. 3
figure 3

The incidence of outcome events for patients with different risk stratification in the entire cohort. The points of 0–4, 5–8, and ≥ 9 represent the low-, medium-, and high-risk stratification respectively. A The incidence of all-cause in-hospital mortality for patients with different risk stratification in the entire cohort; B The incidence of all-cause 30-day readmission for patients with different risk stratification in the entire cohort; C The incidence of all-cause 30-day mortality after discharge for patients with different risk stratification in the entire cohort; D The incidence of the compositive outcome events for patients with different risk stratification in the entire cohort

Full size image

The new risk score model without rhBNP

Considering that rhBNP may not be widely available or used in other countries, a score without this variable was provided. After multivariate analysis, anemia, cTnI/T, BNP (NT-proBNP), QRS fraction of electrocardiogram, and ACEI/ARB were included in the final scoreing system (Table 4), which was calculated as follows:

Table 4 Logistic regression analysis of the composite outcomes without rhBNP in the derivation cohort
Full size table

1 × Anemia + 2 × (cTnI/T elevated) + 1 × [BNP 500 to < 1500 pg/mL (NT-proBNP 2500 to < 7500 pg/mL)] or 2 × [BNP ≥ 1500 (NT-proBNP ≥ 7500) pg/mL] + 3 × (QRS fraction of electrocardiogram < 55%) + 1 × (ACEI/ARB not used).

The discrimination ability of the PRS model without rhBNP was suitable, whose AUROC for the derivation cohort was 0.66 (95%CI 0.63–0.70) with good calibration (the Hosmer–Lemeshow test χ2 = 10.270, P = 0.247).

Discussions

Using the data from the Heb-ADHF registry, we externally validated 3 previously published risk models (3A3B, AHEAD, OPTIME-CHF) to predict the composite of in-hospital all-cause mortality, 30-day all-cause readmission, or 30-day all-cause mortality after discharge in patients with ADHF. The current findings revealed that all the 3 models performed poorly. We also developed a new PRS based on clinical characteristics on admission. The new scoring system comprised a combination of variables, including DBP, lymphocyte, creatinine, BUN, BNP (NT-proBNP), QRS fraction, ACEI/ARB, and rhBNP. We found that the new PRS outperformed the 3A3B, AHEAD, and OPTIME-CHF score in terms of predicting the short-term prognosis. More importantly, we first demonstrated that the QRS fraction is an independent predictor for the short-term prognosis of patients with ADHF. In addition, it was interesting that low DBP, lymphocyte count and rhBNP used were positive predictors of poor short-term outcomes in patients with ADHF in our study.

The short-term risk of death or readmission for ADHF was 13.9% in our study, which was somehow lower than the previous literature [17]. The reason may be that 18.9% of our patients with milder conditions (858 cases of cardiac function in class II). This also resulted in huge direct and indirect economic losses [1, 8, 18, 19]. Short-term prognosis assessment for patients with ADHF remains one of the major challenges for clinicians. Previous studies have demonstrated that the risk score can effectively predict the adverse outcomes of HF. Nevertheless, validation in different cohorts is a crucial step in providing evidence for the score performance, which must be considered in the context of different studies [20, 21]. The availability of the included variables determines the success of any PRS [11]. Likewise, when validating the original score in an external population, it is necessary to assess the applicability and variable availability of the original score in the existing population. We have considered employing other well-known models for the validation in our population. For example, the ADHERE model [16], only provided an inconvenient formula for calculating the log odds of mortality and was somewhat antiquated; for the EHMRG30-ST [22], we lacked data on how patients arrived at the hospital and metoprolol use prior to admission. Besides the applicability and variables availability, we chose the 3A3B, AHEAD, and OPTIME-CHF score because of their study populations involved: the OPTIME-CHF model from the United States [10], the AHEAD from Europe [12], and the 3A3B model from Asian [13]. We attempted to locate one of the existing risk models from different regions to evaluate their short-term prediction abilities for ADHF patients. Unfortunately, it did not work out. Although our endpoint-positive group had higher scores than the endpoint-negative group, the 3A3B, AHEAD, and OPTIME-CHF score performed poorly in the population of Heb-ADHF. The differences in study populations and endpoints design could explain disparity in the results. In other words, directly applying of the previously published HF risk scores to the ADHF population is debatable. This implies that the development and implementation of risk scores for different populations, as well as the variables included in these scores, should be specific and targeted.

In our study, the new PRS got a better predictive value than the 3 previous scoring systems in predicting the composite outcomes. Furthermore, our findings suggested that there was a linear correlation between the new PRS and outcome events. The higher the risk stratification, the higher the incidence of adverse outcome events (P-for-trend < 0.001), namely high-risk stratification increased the incidence of in-hospital all-cause mortality, 30-day all-cause readmission, 30-day all-cause mortality after discharge, or the composite outcomes. Although the discrimination (AUROC) results were not deal, it still has potential predictive value. In particular, our PRS contained some specific predictors with independent predictive value for the endpoint, such as the DBP, lymphocyte, rhBNP, and QRS fraction.

“Risk scores are multivariate predictive models in which relative weights are assigned to each variable in order to calculate the probability that a specific event (e.g. death, rehospitalization) will occur in the future” [9]. Therefore, risk variables are the basic elements of PRS, and different variables may play different roles in different PRS. In general, multiple risk variables can coexist in the same patient [23]. Anemia [12, 13], BNP or NT-proBNP [24], BUN or creatinine [25, 26], and/or ACEI/ARB use [27] have all been identified as common variables affecting the prognosis of patients with ADHF. In our new PRS model, ACEI/ARB use and high BNP (NT-proBNP) level are the two variables with the highest weights [(χ2-df) value] (Fig. 1). Additionally, more studies have suggested that low SBP is closely related to adverse outcomes in ADHF patients [28, 29]. However, studies examining the relationship between DBP and the prognosis of ADHF patients are relatively rare. Our results indicated that low DBP (< 80 mmHg), not SBP, was an independent variable of short-term adverse outcomes in patients with ADHF. Similarly, many studies have shown that hemoglobin (anemia) [12, 13], red blood cell distribution width [30], although blood cells, and/or hematocrit [31] are associated with poor prognosis in patients with ADHF, while the lymphocyte is rarely employed as a predictor. In our new PRS, higher lymphocytes count as an independent predictor could predict short-term poor prognosis for ADHF patients, which was different from the recent published research that believed patients with AHF in lower lymphocyte count had a higher risk of mortality [32]. Our study population was ADHF patients, including both new-onset AHF and decompensated CHF. The reason is not yet clear. We limited the study endpoint to short-term follow-up and the composite events, while the previous literature solely studied all-cause death with no time constraint on follow-up. In addition, lymphocytes are immunological cells, ADHF inpatients had a higher lymphocyte count [33]. Variations in the study population and design may account for the differences. More studies are needed in the future to verify this conclusion and explore the mechanisms involved.

More intriguing and meaningful, rhBNP used and low QRS fraction were also predictors of related adverse events. It is known that rhBNP is a synthetic drug frequently used in the clinical treatment of patients with ADHF and recommended by some guidelines [8, 34]. rhBNP was approved by US Food and Drug Administration in 2001, which can reduce pre-and post-load through dilated veins and arteries; it also has certain effects on promoting sodium excretion, diuresis, and inhibiting renin–angiotensin–aldosterone system and sympathetic nervous system. The drug is safe for patients with ADHF and can significantly improve hemodynamics and dyspnea-related symptoms, which is recommended for the treatment of ADHF by some HF guidelines [8, 34]. Generally speaking, rhBNP has protective effects on the heart and kidney [8, 34, 35]. Studies on patients with ADHF have demonstrated that the use of rhBNP could reduce the 30-day mortality and readmission rate [36, 37]. However, the use of rhBNP in our study, on the other hand, increased the composite outcomes. The result was somewhat unexpected. We do not believe the truth is as it appears. Although the exact proportion of rhBNP used in ADHF patients is unknown, it is unquestionably higher in severe patients. The most likely reason that treatment with rhBNP increased the composite endpoints in our study was the proportion of critical patients included in our cohort. They may need more rhBNP than those with milder conditions. This phenomenon needs further exploration. It should also be noted that rhBNP is not widely used in several other countries and is not further recommended in the latest guidelines for heart failure [38, 39]. As a result, we provided a score without rhBNP use. Although the variable composition changed slightly, its discrimination and calibration remained suitable. Moreover, the QRS fraction was always an independent predictor of the composite endpoint.

The QRS fraction included in our PRS as a novel independent predictor for the endpoints was a rather novel finding of our study. Prior published QRS scoring systems have been proved to be related to left ventricular function [40,41,42], but their computational complexity prevents them from convenient clinical application. Different from previous QRS scores, our novel QRS fraction is simpler and easier to achieve, which is defined as (ΣR/ΣQRS) × 100%. As described above, QRS scoring systems were associated with left ventricular function [40,41,42] and left ventricular function has been shown to predict the development and prognosis of heart failure [43]. This may be the reason why the QRS fraction can predict outcomes of our cohorts. To the best of our knowledge, this is the first study to use this method to evaluate the prognosis in patients with ADHF. After correction for confounding variables, the low QRS fraction remained an independent predictor of the composite of in-hospital all-cause mortality, 30-day all-cause readmission, or 30-day all-cause mortality after discharge for patients with ADHF. Excitingly, our results indicate that a low QRS fraction (< 55%) has a higher weight in our new PRS (Fig. 1). The new PRS outperformed the 3A3B, AHEAD and OPTIME-CHF scores, which was largely due to the introduction of the QRS score. We believe the newly found QRS fraction is a simple-to-obtain and practical parameter to employ in daily clinical practice. This is an issue that needs to be further addressed and we believe that combining the QRS fraction in future PRS will demonstrate its real value in predicting the short-term prognosis of ADHF patients.

In fact, using PRSs for ADHF in clinical practice is far from enough. A single PRS is unlikely to apply to all ADHF populations. However, if effective and simple PRSs are actively applied in daily clinical work and incorporated into electronic health records like the application of CHA2DS2-VASc score in atrial fibrillation [44] or GRACE score in acute myocardial infarction [45], we believe that patients with ADHF will receive better treatment and have a better prognosis. To achieve this goal, clinicians and researchers still have a long way to go.

Study limitations

There exist several limitations in the present study. First, patients' socio-economic background may strongly influence treatment. Some diagnostic and therapeutic approaches such as the detection of interleukin-1 receptor-like 1 (ST2), the use of angiotensin receptor-neprilysin inhibitor (ARNI), and the use of implantable cardioverter defibrillator (ICD) or cardiac resynchronization therapy-defibrillator (CRT-D), etc. were less common in the our study population, which may have influenced the inclusion of valuable variables and the discrimination for our new PRS. In the future, it will be necessary to include more updated variables to improve predictive accuracy. Second, because our study was conducted in multi-centers, some only tested BNP, while others only tested NT-proBNP, which may have a certain impact on the results. Third, other than the 3A3B, AHEAD and OPTIME-CHF scores, we have considered comparing the predictive accuracy of the new PRS with other scores in our population. Unfortunately, we failed since our cohort lacked key variables contained in other scores. Fourth, the clinical data were collected for patients on admission without considering pre-hospital management and there was a small amount of data missing, which could have influenced the results. Besides, our PRS still needs external validation.

Conclusions

Collectively, the present study validated 3A3B, AHEAD, and OPTIME-CHF score in ADHF patients from the Heb-ADHF registry, but their discriminations were all poor to predict the composite of in-hospital all-cause mortality, 30-day all-cause readmission, or 30-day all-cause mortality after discharge. We proposed a new short-term PRS (including DBP, lymphocyte, creatinine, BUN, BNP (NT-proBNP), QRS fraction, ACEI/ARB, and rhBNP) that showed moderate predictive capability with the larger AUROC than the 3A3B, AHEAD, and OPTIME-CHF scores. The newly found QRS fraction is an easy to obtain and practical parameter employed in daily clinical practice. When the risk score reaches ≥ 9 points, special attention must be paid and clinical decision-making must be formulated and adjusted more actively. More new variables and larger multicenter studies are required to make our PRS become a practical and reliable tool for evaluating the short-term adverse events for ADHF patients.

Availability of data and materials

The authors confirm that the data supporting the findings of this study are available within the article. The datasets used and analysed during the current study are available from the corresponding author on reasonable request.

References

  1. Lepage Serge. Acute decompensated heart failure. Can J Cardiol. 2008;24:6B-8B. https://doi.org/10.1016/S0828-282X(08)71022-5.

    Article  PubMed  PubMed Central  Google Scholar 

  2. Garcia-Gutierrez S, Quintana JM, Antón-Ladislao A, Gallardo MS, Pulido E, Rilo I, et al. Creation and validation of the acute heart failure risk score: AHFRS. Intern Emerg Med. 2017;12(8):1197–206.

    Article  Google Scholar 

  3. Butt JH, Fosbøl EL, Gerds TA, Andersson C, McMurray JJV, Petrie MC, et al. Readmission and death in patients admitted with new-onset versus worsening of chronic heart failure: insights from a nationwide cohort. Eur J Heart Fail. 2020;22(10):1777–85.

    Article  Google Scholar 

  4. Mounsey LA, Chang PP, Sueta CA, Matsushita K, Russell SD, Caughey MC. In-Hospital and postdischarge mortality among patients with acute decompensated heart failure hospitalizations ending on the weekend versus weekday: the aric study community surveillance. J Am Heart Assoc. 2019;8(15): e011631.

    Article  Google Scholar 

  5. van Diepen S, Podder M, Hernandez AF, Westerhout CM, Armstong PW, McMurray JJ, et al. Acute decompensated heart failure patients admitted to critical care units: insights from ASCEND-HF. Int J Cardiol. 2014;177(3):840–6.

    Article  Google Scholar 

  6. Selim AM, Mazurek JA, Iqbal M, Wang D, Negassa A, Zolty R. Mortality and readmission rates in patients hospitalized for acute decompensated heart failure: a comparison between cardiology and general-medicine service outcomes in an underserved population. Clin Cardiol. 2015;38(3):131–8.

    Article  Google Scholar 

  7. Greene SJ, Fonarow GC, Vaduganathan M, Khan SS, Butler J, Gheorghiade M. The vulnerable phase after hospitalization for heart failure. Nat Rev Cardiol. 2015;12(4):220–9.

    Article  Google Scholar 

  8. Heart Failure Group of Chinese Society of Cardiology of Chinese Medical Association; Chinese Heart Failure Association of Chinese Medical Doctor Association; Editorial Board of Chinese Journal of Cardiology. [Chinese guidelines for the diagnosis and treatment of heart failure 2018]. Zhonghua Xin Xue Guan Bing Za Zhi. 2018;46(10):760–89. Chinese.

  9. Passantino A, Monitillo F, Iacoviello M, Scrutinio D. Predicting mortality in patients with acute heart failure: Role of risk scores. World J Cardiol. 2015;7(12):902–11.

    Article  Google Scholar 

  10. Felker GM, Leimberger JD, Califf RM, Cuffe MS, Massie BM, Adams KF Jr, et al. Risk stratification after hospitalization for decompensated heart failure. J Card Fail. 2004;10(6):460–6.

    Article  Google Scholar 

  11. Pocock SJ, Ariti CA, McMurray JJ, Maggioni A, Køber L, Squire IB, et al. Predicting survival in heart failure: a risk score based on 39 372 patients from 30 studies. Eur Heart J. 2013;34(19):1404–13.

    Article  Google Scholar 

  12. Spinar J, Jarkovsky J, Spinarova L, Mebazaa A, Gayat E, Vitovec J, et al. AHEAD score–long-term risk classification in acute heart failure. Int J Cardiol. 2016;202:21–6.

    Article  Google Scholar 

  13. Kasahara S, Sakata Y, Nochioka K, Tay WT, Claggett BL, Abe R, et al. The 3A3B score: the simple risk score for heart failure with preserved ejection fraction—a report from the CHART-2 study. Int J Cardiol. 2019;284:42–9.

    Article  Google Scholar 

  14. Tromp J, Richards AM, Tay WT, Teng THK, Yeo PSD, Sim D, et al. N-terminal pro-B-type natriuretic peptide and prognosis in Caucasian vs. Asian patients with heart failure: Ethnicity and NT-proBNP in heart failure. ESC Heart Failure. 2018;5(2):279–87. https://doi.org/10.1002/ehf2.12252.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Miles JA, Quispe R, Mehlman Y, Patel K, Buchwald CLV, You JY, Sokol S, Faillace RT. Racial differences and mortality risk in patients with heart failure and hyponatremia. PLOS ONE. 2019;14(6):e0218504. https://doi.org/10.1371/journal.pone.0218504.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  16. Fonarow GC. Risk stratification for in-hospital mortality in acutely decompensated heart failureclassification and regression tree analysis. JAMA. 2005;293(5):572. https://doi.org/10.1001/jama.293.5.572.

    Article  CAS  PubMed  Google Scholar 

  17. Lan T, Liao YH, Zhang J, Yang ZP, Xu GS, Zhu L, et al. Mortality and readmission rates after heart failure: a systematic review and meta-analysis. Ther Clin Risk Manag. 2021;17:1307–20.

    Article  Google Scholar 

  18. Liu S, Li Y, Zeng X, Wang H, Yin P, Wang L, et al. Burden of cardiovascular diseases in China, 1990–2016: findings from the 2016 global burden of disease study. JAMA Cardiol. 2019;4(4):342–52.

    Article  Google Scholar 

  19. Virani SS, Alonso A, Aparicio HJ, Benjamin EJ, Bittencourt MS, Callaway CW, et al. Heart disease and stroke statistics-2021 update: a report from the American heart association. Circulation. 2021;143(8):e254-743.

    Article  Google Scholar 

  20. Altman DG, Vergouwe Y, Royston P, Moons KG. Prognosis and prognostic research: validating a prognostic model. BMJ. 2009;338:b605.

    Article  Google Scholar 

  21. Moons KGM, Kengne AP, Grobbee DE, Royston P, Vergouwe Y, Altman DG, Woodward M. Risk prediction models: II. External validation, model updating, and impact assessment. Heart. 2012;98(9):691–8. https://doi.org/10.1136/heartjnl-2011-301247.

    Article  PubMed  Google Scholar 

  22. Lee DS, Lee JS, Schull MJ, Borgundvaag B, Edmonds ML, Ivankovic M, et al. Prospective validation of the emergency heart failure mortality risk grade for acute heart failure. Circulation. 2019;139(9):1146–56.

    Article  Google Scholar 

  23. Cohen-Solal A, Laribi S, Ishihara S, Vergaro G, Baudet M, Logeart D, Mebazaa A, et al. Prognostic markers of acute decompensated heart failure: the emerging roles of cardiac biomarkers and prognostic scores. Arch Cardiovasc Dis. 2015;108(1):64–74.

    Article  Google Scholar 

  24. Santaguida PL, Don-Wauchope AC, Oremus M, McKelvie R, Ali U, Hill SA, Balion C, et al. BNP and NT-proBNP as prognostic markers in persons with acute decompensated heart failure: a systematic review. Heart Fail Rev. 2014;19(4):453–70.

    Article  CAS  Google Scholar 

  25. Ueda T, Kawakami R, Sugawara Y, Okada S, Nishida T, Onoue K, et al. Worsening of renal function during 1 year after hospital discharge is a strong and independent predictor of all-cause mortality in acute decompensated heart failure. J Am Heart Assoc. 2014;3(6):e001174.

    Article  Google Scholar 

  26. Ren X, Qu W, Zhang L, Liu M, Gao X, Gao Y, et al. Role of blood urea nitrogen in predicting the post-discharge prognosis in elderly patients with acute decompensated heart failure. Sci Rep. 2018;8(1):13507.

    Article  Google Scholar 

  27. Yoshioka K, Matsue Y, Yamaguchi T, Kitai T, Kagiyama N, Okumura T, et al. Safety and prognostic impact of early treatment with angiotensin-converting enzyme inhibitors or angiotensin receptor blockers in patients with acute heart failure. Am J Cardiovasc Drugs. 2019;19(6):597–605.

    Article  CAS  Google Scholar 

  28. Kawase Y, Kadota K, Nakamura M, Tada T, Hata R, Miyawaki H, et al. Low systolic blood pressure on admission predicts mortality in patients with acute decompensated heart failure due to moderate to severe aortic stenosis. Circ J. 2014;78(10):2455–9.

    Article  Google Scholar 

  29. Formiga F, Moreno-Gonzalez R, Chivite D, Yun S, Ariza A, Corbella X. Lower admission blood pressure as an independent predictor of 1-year mortality in elderly patients experiencing a first hospitalization for acute heart failure. Hellenic J Cardiol. 2019;60(4):224–9.

    Article  Google Scholar 

  30. Turcato G, Zorzi E, Prati D, Ricci G, Bonora A, Zannoni M, et al. Early in-hospital variation of red blood cell distribution width predicts mortality in patients with acute heart failure. Int J Cardiol. 2017;243:306–10.

    Article  Google Scholar 

  31. Mo R, Yu LT, Tan HQ, Wang Y, Yang YM, et al. A new scoring system for predicting short-term outcomes in Chinese patients with critically-ill acute decompensated heart failure. BMC Cardiovasc Disord. 2021;21(1):228.

    Article  Google Scholar 

  32. Majmundar M, Kansara T, Park H, Ibarra G, Marta Lenik J, Shah P, et al. Absolute lymphocyte count as a predictor of mortality and readmission in heart failure hospitalization. Int J Cardiol Heart Vasc. 2022;39:100981.

    PubMed  PubMed Central  Google Scholar 

  33. Curran FM, Bhalraam U, Mohan M, Singh JS, Anker SD, Dickstein K, et al. Neutrophil-to-lymphocyte ratio and outcomes in patients with new-onset or worsening heart failure with reduced and preserved ejection fraction. ESC Heart Fail. 2021;8(4):3168–79.

    Article  Google Scholar 

  34. Ponikowski P, Voors AA, Anker SD, Bueno H, Cleland JGF, Coats AJS, et al. 2016 ESC guidelines for the diagnosis and treatment of acute and chronic heart failure: The task force for the diagnosis and treatment of acute and chronic heart failure of the european society of cardiology (ESC) developed with the special contribution of the heart failure association (HFA) of the ESC. Eur Heart J. 2016;37(27):2129–200.

    Article  Google Scholar 

  35. Wang Y, Gu X, Fan W, Fan Y, Li W, Fu X. Effects of recombinant human brain natriuretic peptide on renal function in patients with acute heart failure following myocardial infarction. Am J Transl Res. 2016;8(1):239–45.

    PubMed  PubMed Central  Google Scholar 

  36. Recombinant Human Brain Natriuretic Peptide Multicenter Clinical Study Group, Hu DY. [Efficacy and safety of intravenous recombinant human brain natriuretic peptide in patients with decompensated acute heart failure: a multicenter, randomized, open label, controlled study]. Zhonghua Xin Xue Guan Bing Za Zhi. 2011;39(4):305–8. Chinese.

  37. Feng X, Wu Y, Meng Y, Wei Y. Efficacy and safety of intravenous recombinant human brain natriuretic peptide in patients with severe heart failure: a prospective multicenter clinical study. Zhonghua Wei Zhong Bing Ji Jiu Yi Xue. 2017;29(6):520–4 (Chinese).

    PubMed  Google Scholar 

  38. McDonagh TA, Metra M, Adamo M, Gardner RS, Baumbach A, Böhm M, et al. 2021 ESC guidelines for the diagnosis and treatment of acute and chronic heart failure. Eur Heart J. 2021;42(36):3599–726.

    Article  CAS  Google Scholar 

  39. Heidenreich PA, Bozkurt B, Aguilar D, Allen LA, Byun JJ, Colvin MM, et al. 2022 AHA/ACC/HFSA guideline for the management of heart failure: a report of the American college of cardiology/American heart association joint committee on clinical practice guidelines. Circulation. 2022;145(18):e895-1032.

    PubMed  Google Scholar 

  40. Gitler B, Cooper JA. A QRS scoring system for assessing left ventricular function after myocardial infarction. N Engl J Med. 1982;306(23):1423–4.

    CAS  PubMed  Google Scholar 

  41. Seino Y, Staniloff HM, Shell WE, Mickle D, Shah PK, Vyden JK. Evaluation of a QRS scoring system in acute myocardial infarction: relation to infarct size, early stage left ventricular ejection fraction, and exercise performance. Am J Cardiol. 1983;52(1):37–42.

    Article  CAS  Google Scholar 

  42. Handler CE, Ellam SV, Maisey MN, Sowton E. Use of an exercise QRS score and radionuclide left ventricular ejection fraction in assessing prognosis after myocardial infarction. Eur Heart J. 1987;8(3):243–53.

    Article  CAS  Google Scholar 

  43. Kuznetsova T, Herbots L, Jin Y, Stolarz-Skrzypek K, Staessen JA. Systolic and diastolic left ventricular dysfunction: from risk factors to overt heart failure. Expert Rev Cardiovasc Ther. 2010;8(2):251–8.

    Article  Google Scholar 

  44. January CT, Wann LS, Calkins H, Chen LY, Cigarroa JE, Cleveland JC Jr, et al. 2019 AHA/ACC/HRS focused update of the 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American college of cardiology/American heart association task force on clinical practice guidelines and the heart rhythm society in collaboration with the society of thoracic surgeons. Circulation. 2019;140(2):e125–51.

    Article  Google Scholar 

  45. Collet JP, Thiele H, Barbato E, Barthélémy O, Bauersachs J, Bhatt DL, et al. 2020 ESC Guidelines for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation. Eur Heart J. 2021;42(14):1289–367.

    Article  Google Scholar 

Download references

Acknowledgements

We would like to acknowledge Dr. Chunpeng Ji, Dr. Yaqing Zhou, Dr. Ming Zhang, and Dr. Boqun Shi for their help and appreciate all the researchers involved in the study of Heb-ADHF. The authors wish to thank all patients taking part in the study. Thanks to the editor and all the peer reviewers for their opinions and suggestions.

Funding

This work was supported by the S&T Program of Hebei, China (grant number 16277738D).

Author information

Author notes

  1. Hong-Liang Zhao and Xiao-Li Gao authors contributed equally to this work

Authors and Affiliations

  1. First Division, Department of Cardiology, The Second Hospital of Hebei Medical University, Heping West Road No. 215, Shijiazhuang, 050000, Hebei province, China

    Hong-Liang Zhao, Xue Geng, Qian Wang & Wei Cui

  2. Department of Cardiology, The First Hospital of Hebei Medical University, Shijiazhuang, 050031, Hebei province, China

    Hong-Liang Zhao

  3. Department of Cardiology, Huabei Petroleum Administration Bureau General Hospital, Renqiu, 062552, Hebei Province, China

    Xiao-Li Gao & Ying-Hua Liu

  4. Department of Cardiology, Zhangjiakou First Hospital, Zhangjiakou, 075000, Hebei Province, China

    Sen-Lin Li

  5. Department of Cardiology, Baoding First Central Hospital, Baoding, 071000, Hebei Province, China

    Qi Zhang

  6. Department of Cardiology, Affiliated Hospital of Chengde Medical University, Chengde, 067000, Hebei Province, China

    Wei-Chao Shan

  7. Department of Cardiology, Hengshui People’s Hospital, Hengshui, 053000, Hebei Province, China

    Qun Zheng

  8. Department of Cardiology, Chengde Central Hospital, Chengde, 067024, Hebei Province, China

    Jiang Zhou

  9. Department of Cardiology, Qinhuangdao First Hospital, Qinhuangdao, 066099, Hebei Province, China

    Yong-Zheng Liu & Li Liu

  10. Department of Cardiology, Cangzhou Central Hospital, Cangzhou, 061011, Hebei Province, China

    Nan Guo

  11. Department of Cardiology, Handan Central Hospital, Handan, 056000, Hebei Province, China

    Hong-Sen Tian

  12. Department of Cardiology, Xingtai People’s Hospital, Xingtai, 054001, Hebei Province, China

    Qing-Min Wei

  13. Department of Cardiology, Shijiazhuang People’s Hospital, Shijiazhuang, 050011, Hebei Province, China

    Xi-Tian Hu

  14. Department of Cardiology, The 252nd Hospital of People’s Liberation Army, Baoding, 071000, Hebei Province, China

    Ying-Kai Cui

Authors
  1. Hong-Liang Zhao
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Xiao-Li Gao
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Ying-Hua Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Sen-Lin Li
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Qi Zhang
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Wei-Chao Shan
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Qun Zheng
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Jiang Zhou
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Yong-Zheng Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Li Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Nan Guo
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Hong-Sen Tian
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. Qing-Min Wei
    View author publications

    You can also search for this author in PubMed Google Scholar

  14. Xi-Tian Hu
    View author publications

    You can also search for this author in PubMed Google Scholar

  15. Ying-Kai Cui
    View author publications

    You can also search for this author in PubMed Google Scholar

  16. Xue Geng
    View author publications

    You can also search for this author in PubMed Google Scholar

  17. Qian Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  18. Wei Cui
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

HLZ and XLG, and WC designed the study, analyzed and interpreted the patient data, and were major contributors in writing the manuscript. XL, YHL, SLL, QZ, WCS, QZ, JZ, YZL, LL, NG, HST, QMW, XTH, YKC, XG, QW collected and assembled the patient data. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Wei Cui.

Ethics declarations

Ethics approval and consent to participate

All methods were performed in accordance with the ethical standards as laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. The study was approved by the ethics committee of the Second Hospital of Hebei Medical University (2015110). All patients signed informed consent to participate in Hebei Acute Decompensated Heart Failure (Heb-ADHF) registry study.

Consent for publication

Not applicable.

Competing interests

All the authors have no conflicts of interest and had access to the data and wrote the manuscript.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhao, HL., Gao, XL., Liu, YH. et al. Validation and derivation of short-term prognostic risk score in acute decompensated heart failure in China. BMC Cardiovasc Disord 22, 307 (2022). https://doi.org/10.1186/s12872-022-02743-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12872-022-02743-1

Keyword

BMC Cardiovascular Disorders

ISSN: 1471-2261

Contact us