The data for this study were derived from a prospective observational cohort study conducted in China (Trial registration: ChiCTR1800017204). Elderly patients aged 65 years and older, admitted to Beijing Hospital between September 2018 and April 2019, were recruited. The study received approval from the Ethics Committee of Beijing Hospital (No. 2018BJYYEC-121-02). The criteria for inclusion were as follows: (1) Individuals aged 65 years or older, hospitalized due to cardiovascular illnesses; (2) Voluntary participation in this study with signed informed consent. The exclusion criteria included: (1) Patients unable of completing the thorough geriatric assessment due to significant cognitive impairment, hearing loss, or other problems; (2) Patients with acute exacerbation of chronic obstructive pulmonary disease (COPD); (3) Patients with end-stage renal disease undergoing dialysis; (4) Refusal to sign informed consent. A total of 524 elderly patients from the cardiology ward were included in the study. The participant inclusion process is illustrated in Fig. 1.

In this study, baseline data were gathered from the electronic health records of the patients. This data included information such as demographic information (e.g., age, gender), hospitalization details, medical conditions, physical examination results, laboratory test values, and other relevant health information.

The IC score is a measure designed to assess healthy aging, with a range from 0 (representing the worst possible IC) to 10 (representing the highest possible IC). The score equally weighs five dimensions that are critical to an individual’s overall health and functional status. Five Dimensions were: cognition, vitality, sensory, psychological and locomotion. For each of these five dimensions, the IC score is stratified within a 0–2 range, based on the level of function impairment in each dimension: 0 = function significant loss, 1 = function decline, 2 = function stable. Followings are the detailed scoring criteria:

① Cognition (0–2 points): we applied mini-mental state examination (MMSE) to assess patients’ cognition dimension [10]. A MMSE score of 27–30, represents normal cognitive function, scoring 2 points in this dimension; a MMSE score of 10–26, represents mild to moderate cognitive impairment, scoring 1 point in this dimension; a MMSE score of 0–9, represents moderate to severe cognitive impairment, scoring 0 points in this dimension.

② Vitality (0–2 points): we applied the mini nutritional assessment - short form (MNA-SF) to assess patients’ vitality dimension [11]. A MNA-SF score of 12–14, represents normal nutritional status, scoring 2 points in this dimension; a MNA-SF score of 8–11, represents a risk of malnutrition, scoring 1 point in this dimension; a MNA-SF score of 0–7, represents malnutrition, scoring 0 points in this dimension.

③ Sensory (0–2 points): we applied two self-reported question to assess patients’ sensory dimension. “Do you have any difficulty with your hearing?”, response: yes = 1 point; no = 0 point. “Do you have any difficulty with your vision?”, response: yes = 1 point; no = 0 point. The combined scores from the two questions constitute the final score for this dimension.

④ Psychological (0–2 points): we concurrently applied the geriatric depression scale - 5 items (GDS-5) and the hospital anxiety and depression scale-anxiety subscale (HADS-A) to assess patients’ psychological dimension [12, 13]. A GDS-5 score of 0–1, represents no depression, scoring 1 point; a GDS-5 score of 2–5, represents a likelihood of depression, scoring 0 point. A HADS-A score of 0–7, represents no anxiety, scoring 1 point; a HADS-A score of 8–21, represents borderline or significant anxiety, scoring 0 point. The combined scores from the two scales constitute the final score for this dimension.

⑤ Locomotion (0–2 points): we applied the short physical performance battery (SPPB) to assess patients’ locomotion dimension [14]. A SPPB score of 10–12, represents robustness, scoring 2 points in this dimension; a SPPB score of 3–9, represents possible sarcopenia, scoring 1 point in this dimension; a SPPB score of 0–2, represents sarcopenia and cachexia, scoring 0 point in this dimension.

Age, creatinine, and ejection fraction score (ACEF score) [15]: The ACEF score, which stands for age, creatinine, and ejection Fraction, is a simplified risk stratification tool originally developed to predict operative mortality in patients undergoing elective cardiac surgery. It is calculated using the following formula: Age (years)/Left ventricular ejection fraction (LVEF) (%) + 1 (if serum creatinine $>$2.0 mg/dL).

Atrial fibrillation (AF)/Atrial flutter (AFL): AF and AFL are common supraventricular arrhythmias characterized by disorganized or rapid atrial electrical activity, leading to irregular or rapid ventricular response. Both persistent and paroxysmal AF/AFL were included in this study.

Cancer history: This variable is extracted from the discharge diagnoses of the participants. Individuals with current malignant neoplasms were not included in this study.

Coronary artery disease (CAD): CAD is a condition characterized by atherosclerotic narrowing or blockage of the coronary arteries. This variable is extracted from the discharge diagnoses of the participants.

Chronic kidney disease (CKD): The estimated glomerular filtration rate (eGFR) 60 mL/min/1.73 m² was defined as CKD, eGFR was calculated using the CKD Epidemiology Collaboration equation [16]. This study did not include patients with end-stage renal disease receiving hemodialysis treatment.

COPD: is a progressive respiratory disorder characterized by persistent airflow limitation that is not fully reversible. This variable is extracted from the discharge diagnoses of the participants. Participants with acute exacerbation of COPD were excluded from this study.

Diabetes: Diabetes is a chronic metabolic disorder characterized by hyperglycemia resulting from defects in insulin secretion, insulin action, or both. This variable is extracted from the discharge diagnoses of the participants. The specific types of diabetes were not differentiated in this study.

Heart failure (HF): HF is a clinical syndrome characterized by impaired cardiac function resulting in inadequate perfusion to meet the metabolic demands of the body. This variable is extracted from the discharge diagnoses of the participants.

Hypertension (HTN): HTN is a chronic medical condition characterized by persistently elevated arterial blood pressure. This variable is extracted from the discharge diagnoses of the participants.

Polypharmacy [17]: Polypharmacy was defined as the presence of $\ge$7 different drugs taken by a patient at discharge.

Smoking: Including participants who are current smokers or have a history of smoking.

The primary outcome of this study was the rate of all-cause mortality over a period of five years. Clinical follow-ups were routinely performed via phone annually. If unable to contact the patient or their family, the patient’s medical records were used to determine their survival status.

Continuous baseline variables were summarized using either the mean with standard deviation (SD) or the median with interquartile range (IQR), based on the data distribution. Categorical variables were described as frequencies and percentages. Student’s t-test was used to compare normally distributed continuous variables between groups, while the Mann–Whitney U test was employed for continuous variables that were not normally distributed. For categorical variables, the Chi-square test was utilized to compare data between groups. We used receiver operating characteristics (ROC) curve analysis to obtain the optimal cut-off point of IC score to predict 5-year all-cause mortality according to maximal Youden index. In this study, univariable and multivariable Cox proportional hazards analyses were conducted to identify factors associated with all-cause mortality over the 5-year period. Kaplan-Meier analysis was used to estimate the cumulative incidence of the primary endpoint, and log-rank tests to analyze for significant differences. In this study, variance inflation factor (VIF) values were computed for all predictors. A stepwise elimination procedure was applied, removing the variable with the highest VIF in each iteration until all remaining variables had VIF values below 5. To identify independent predictors of survival, Cox proportional hazards regression analysis was performed. A stepwise variable selection procedure based on Akaike’s Information Criterion (AIC) was applied using backward elimination. Starting from the full model, variables were sequentially removed to minimize the AIC value, aiming to achieve a best model. The final multivariable Cox model included covariates retained after stepwise selection. We also employed Cox proportional hazards regression models to conduct subgroup analyses on the association between IC score and 5-year all-cause mortality. Subgroups were classified by age, sex and comorbidities. To evaluate the predictive value of the IC score, we compared it with the ACEF score by assessing improvements in discriminative ability using the delta C-index, Integrated Discrimination Improvement (IDI), and Net Reclassification Index (NRI). In this study, all statistical tests were two-tailed, and a p value of 0.05 was considered statistically significant. All analyses were conducted using R software (version 4.2.2; R Core Team, Vienna, Austria).

Among 524 participants, the mean age was 75.2 $\pm$ 6.5 years old, and 51.7% were men. The median IC score was 8.0 (7.0, 9.0). The distribution of patients’ IC score and the score of the five IC dimensions are presented in Fig. 2, with the sensory and locomotion dimensions showing relatively high rates of function decline (1 point) or significant loss (0 point). The participants were divided into 2 groups (mortality group and survival group). Baseline characteristics comparison between two groups are shown in Table 1. Compared with the survival group at baseline, patients in the mortality group were older (81.1 $\pm$ 5.7 vs. 74.0 $\pm$ 6.0, p 0.01), showed a higher male proportion (61.6% vs. 49.8%, p = 0.04), had a lower IC [7.0 (6.0, 8.0) vs. 8.0 (7.0, 9.0), p 0.01], and a higher prevalence rates of AF or AFL (34.9% vs. 20.1, p 0.01), HF (44.2% vs. 11.2%, p 0.01), diabetes (48.8% vs. 33.1%, p 0.01) and CKD (19.8% vs. 4.3%, p 0.01). Moreover, patients in the mortality group had lower hemoglobin (Hb) value, lower ALB value, lower LVEF, higher uric acid value, higher ACEF score and higher n-terminal pro-B-type natriuretic peptide (NT-proBNP) value.

The 5-year survival status of all 524 participants was available, all-cause mortality occurred in 86 patients (16.5%). Univariable Cox regression analysis (Table 2) showed that IC score was a protect factor of all-cause mortality (HR: 0.62; 95% CI: 0.56–0.70, p 0.01). Additionally, age, AF/AFL, HTN, CKD, HF, diabetes, cancer history, polypharmacy, lower BMI, lower Hb values, lower ALB values, lower LVEF values, higher uric acid values and higher LogNT-proBNP values were statistically significant risk factors.

Variables were ultimately selected for inclusion in the multivariable Cox regression model based on initial criteria, which included variables with a p value 0.10 in the univariate Cox regression analysis. Stepwise regression was then performed, and the AIC was calculated to select the optimal model. The initial variables included in the model were IC score, age, male, Heart rate, BMI, AF/AFL, HTN, diabetes, cancer history, CKD, HF, polypharmacy, Hb, uric acid, ALB, LDL-C, LVEF, and NT-proBNP. The initial AIC was 930.29. After stepwise regression, the final model included eleven variables: IC score, age, male, HTN, HF, diabetes, CKD, BMI, uric acid, Hb and LVEF, with an AIC of 915.60 and a C-index of 0.844. After adjusting for covariates, multivariate Cox regression showed that IC score was independently associated with a lower hazard ratio of 5-year all-cause mortality (HR = 0.79, 95% CI: 0.69–0.92, p 0.01) (Table 3).

ROC analysis was used to evaluate to obtain the optimal cut-off point of IC score to predict 5-year all-cause mortality according to maximal Youden index (Fig. 3). IC score = 7 points (rounding 6.5) was the optimal cut-off point to predict 5-year all-cause mortality. Kaplan-Meier survival curve showed IC score 7 points has significantly higher all-cause mortality survival rate (log-rank p 0.01) within 5 year (Fig. 4).

We employed Cox regression analysis to investigate the impact of IC score $\ge$7 on 5-year all-cause mortality across various subgroups. As outlined in the Methods section, the subgroups were classified based on age, sex, HF, AF/AFL, CAD, HTN, CKD, diabetes, cancer history and polypharmacy status. The results indicated that a higher IC score was associated with a reduced risk of 5-year all-cause mortality in all subgroups (all p 0.05 for effects within each subgroup). No significant difference between groups was observed in subgroups (all p for interaction $>$0.05) (Fig. 5).

To demonstrate the clinical value of IC score, we first compared the predictive performance of two Cox proportional hazards models: Model 1, which included the ACEF score, and Model 2, which included both the ACEF score and IC score. The C-index for Model 1 (the ACEF score) was 0.726, with a 95% CI ranging from 0.644 to 0.807. In contrast, Model 2 (ACEF + IC score) demonstrated an improved C-index of 0.789, with a 95% CI between 0.700 and 0.878. The difference in C-index between the two models was 0.063 (95% CI: 0.026–0.100), which was statistically significant (p 0.01), indicating that the inclusion of the IC score enhanced the model’s discriminatory ability.

We calculated the differences in IDI and NRI between the ACEF score and the IC score. The results from the IDI analysis indicated that compared to Model 1 (the ACEF score), Model 3 (IC score) provided improvement in model discrimination: 0.022 (95% CI: 0.011–0.033, p 0.01). The results from the IDI analysis also showed that compared to Model 1 (the ACEF score), Model 3 (IC score) provided improvement in model discrimination: 0.126 (95% CI: 0.084–0.289, p 0.01) (Fig. 6). Compared to the ACEF score, the IC score demonstrated improved discriminatory performance, as evidenced by higher IDI and NRI values statistically significant (both p 0.01).