RURUS SURYAWAN Score: A Novel Scoring System to Predict 30-Day Mortality for Acute Myocardial Infarction Undergoing Primary Percutaneous Coronary Intervention

1. Introduction

The mortality rate of acute myocardial infarction has significantly decreased in decade, due to the invention of primary percutaneous coronary intervention by Geoffrey Hartzler in the early 1980s [1]. The prognosis of acute myocardial infarction after invention of primary percutaneous coronary intervention is better [2]. For every time earlier, reperfusion treatment results in an improvement of mortality rate and primary percutaneous coronary intervention is still a treatment of choice for acute myocardial infarction [3]. Almost all reports showed a consequent reduction in mortality in acute myocardial infarction patients compared with thrombolysis, which was insufficient to assess the extent of the improvement [4]. Unfortunately, according to several reports, residual 30-day mortality among patients with acute myocardial infarction underwent primary percutaneous coronary intervention still high, ranges from 5–10%, especially in developing country [5]. Data conducted by Pramudyo et al (2022) revealed that Indonesia ranked fourth in the world for the highest mortality rate of acute myocardial infarction patients compared to 186 other countries [6]. 30-day mortality rate of acute myocardial infarction in Indonesia was approximately 10.6% for patients with acute myocardial infarction [6]. Still, this number is higher when compared to the mortality rate in other Asian countries (5%), European (4-8%) and American (6-9%) [7,8,9].

The high mortality rate of acute myocardial infarction in Indonesia has attracted much attention of cardiologist, and influenced by many factors, such as demographic factors, history of cardiovascular risk factors, clinical presentation, and the results of investigations and management in the hospital [10]. The high mortality rate in patients with acute myocardial infarction has also raised particular concerns for cardiologist to develop a method to accurately identify patients at high-risk so that awareness can be increased to reduce the risk of death [11,12]. Currently, there are several prognostic tool in the field of acute myocardial infarction : Global Registry of Acute Coronary Events (GRACE), Thrombolysis in Myocardial Infarction (TIMI), Platelet Glycoprotein IIb/IIIa in Unstable Angina: Receptor Suppression Using Integrilin (PURSUIT), Oxford Acute Severity of Illness Score (OASIS), Logistic Organ Dysfunction System (LODS) and Simplified Acute Physiology Score (SAPS II) [13,14,15,16,17,18].

The GRACE score was established to assess 30-day mortality and identify high-risk groups among patients with acute myocardial infarction globally [13]. For global use, GRACE score demonstrated a superior predictive power for the one-year mortality due to acute myocardial infarction compared with other scoring systems, such as the TIMI and PURSUIT risk scores [14,15]. The TIMI and GRACE risk scores were calculated using data from a Western Caucasian cohort with limited participation from an Asian cohort. Asian patients have been understudied [13,16].

OASIS was proposed in 2013 by machine-learning algorithms. This scoring system has better predictive models for ICU mortality with an area under the receiver operating characteristic curve of 0.88 and calibrated well [17]. Unfortunately, this sophisticated scoring systems require the collection of numerous physiologic measurements, making their applications in clinical practice difficult [17]. In China, a study conducted by Wang et al (2021) introduced LODS scoring system as a valuable scoring system for the prediction of 30-day mortality among intensive care patients diagnosed with ST-Elevation Myocardial Infarction (STEMI) [18]. LODS is an organ dysfunction-based scoring system and permits the calculation of predicted mortality based on the organ dysfunction score on the day of ICU admission [18]. With a more concise composition of local population and greater clinical benefit, LODS may be a better predictor of 30-day mortality for intensive care patients with STEMI in China [19].

Indonesia still does not have a local registry nor a scoring system for predictor of 30-day mortality among patients with acute myocardial infarction or acute coronary syndrome [19,20]. Thus, the aim of this study is to develop a novel scoring system to predict 30-day mortality of patients with firstly diagnosed acute myocardial infarction after primary percutaneous coronary intervention at Dr. Soetomo General Academic Hospital, Surabaya.

2. Materials and Methods

2.1. Study Population and Design

This retrospective study was conducted at Dr. Soetomo General Academic Hospital, Surabaya, as a tertiary referral hospital in Indonesia, from 2021 to 2024. The data collected included the baseline demographics, clinical characteristics, comorbidities, laboratory, echocardiography, angiography parameters and compared it with 30-day mortality of first episode of acute myocardial infarction underwent primary percutaneous coronary intervention during hospital admission. This study has been approved by the Health Research Ethics Committee of the Dr. Soetomo General Academic Hospital, Surabaya. This study was conducted in accordance with the Declaration of Helsinki. The patients in the database are anonymous.

2.2. Inclusion and Exclusion Criteria

The population of this study consisted of hospitalized patients with first episode of acute myocardial infarction underwent primary percutaneous coronary intervention at Dr. Soetomo General Academic Hospital, Surabaya. The inclusion criteria were: (a) adults aged above 18 years (b) admitted to the hospital for the first time with emergency causes or complications from first episode of acute myocardial infarction, (c) underwent primary percutaneous coronary intervention with total or subtotal occlusion coronary artery from cardiac catheterization, (d) willing to participate in the study by signing the informed consent form, (e) underwent anamnesis and had 50 ml blood taken for laboratory examination, (f) able to be followed up at least 1 months after hospitalization. The exclusion criteria were: (a) admitted for elective procedures, (b) presence of others unrelated disease such as primary valvular heart disease, congenital heart disease, liver cirrhosis, malignancy, tuberculosis or HIV/AIDS, (c) history of a mechanical circulatory support such as intra-aortic balloon pump or extracorporeal membrane oxygenation, e) special or vulnerable patients such as prisoners or homeless individuals, f) history of cardiac arrest before primary percutaneous coronary intervention.

2.3. Statistical Analysis

Statistical analyses were performed using IBM SPSS Statistics for Windows Operating System, version 26.0 software (IBM Corp., Armonk, N.Y., USA), STATA software, version 15.0 (Stata Corporation, College Station, Texas, USA) and R statistics 4.0.3 with the “glmnet” package (R Foundation for Statistical Computing, Vienna, Austria). Baseline characteristic such as gender, age, categories were described as a proportion if the data were categorical and described as mean or median. Clinical characteristics, comorbidities, laboratory, echocardiography, angiography parameters affecting 30-day mortality in acute myocardial infection after primary percutaneous coronary intervention were tested using the Multivariate Cox Regression test. A p-value below 0.05 was declared statistically significant. The scoring system was assessed to develop the score for each predictor variables contributing in 30-day mortality. To avoid the occurrence of intercorrelation among two or more independent variables, multicollinearity was assessed by examining tolerance and the Variance Inflation Factor (VIF). Primary outcome was 30-day mortality. The score was calculated using the beta coefficient and standard error. The predictive values of our novel risk score was assessed by receiver operating characteristics curve (ROC) analysis (using MedCalc Version 12.2.1; MedCalc Software, Mariakerke, Belgium), applying net reclassification and integrated discrimination improvement. Prognostic utility of the risk models for 30-day mortality has been assessed by deriving their C-statistics, using ROC curves. The final scores were classified based on the risk of acute myocardial infarction patients with high, moderate, and low mortality rates.

In the derivation set, we performed multivariable logistic regression model to determine predictors for 30-day mortality. We use the Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression algorithm in order to obtain a subset of predictor variables from the 20 candidate variables. The LASSO algorithm can select from the set of candidate variables that achieve greater importance once regularized. The LASSO algorithm finds the variables that contribute the least in the logistic regression model and forces them to have coefficients equal to zero. To check the stability in the selection of variables and in the calculation of the coefficients of the logistic regression, an analysis with random partitions of the derivation and validation sets was carried out. One hundred sets of derivation have been created randomly with the same number of records as those used in the temporal validation (n = 926). The LASSO variable selection methodology has been applied to each of these sets, and the corresponding logistic regression coefficients have been calculated. These results were compared with those obtained in the temporal validation.

3. Results

This retrospective study is evaluating the 30-day mortality from 1535 patients assigned to Dr. Soetomo General Academic Hospital with first episode of acute myocardial infarction underwent primary percutaneous coronary intervention during hospital admission. The Hosmer-Lemeshow statistic was 3.56_df8,(p = 0.795). The C statistic for the model in the test cohort was 0.85.

3.1. Baseline Characteristics

A total of 1714 patients assigned to Dr. Soetomo General Academic Hospital with first episode of acute myocardial infarction underwent primary percutaneous coronary intervention during hospital admission were screened. Patients without an ICU record or those with missing or incomplete echocardiography or laboratory data were excluded, and 1535 adult patients were included in this study. According to the grouping method of a previous study, all patients were randomly divided into a derivation (n=926, 60.33%) and a validation (n=609, 39.67%) group (Table 1).

3.2. Model Development

For the derivation group, Table 2 show the variables selected using regression analysis that was associated with 30-day mortality from univariate analysis. Statistically significant variables screened from the univariate analysis were included in the non-conditional binary multivariate logistic regression. A multivariate logistic regression identified resting heart rate, anion gap, lactate, calcium, BUN and diastolic blood pressure levels at the admission, and also the presence of CKD as the most significant mortality risk predictors. Heart rate at discharge was related to 1-year mortality after adjustment for these variables (hazard ratio 1.13; 95% confidence interval [CI], 1.03-1.24 per 10 beats per minute, P = 0.02). Among the variables screened using LASSO regression, in addition to the seven variables above, anemia and MCV were also identified as statistically significant variables using multivariate likelihood ratio (LR). For continuous variables entered into the multivariate LR model mentioned previously, potential non-linearity in the prediction of 30-day mortality was explored using a smoothing plot.

3.3. RURUS SURYAWAN as Novel Risk Score for 30-Days Mortality Among Acute Myocardial Infarction Patients Undergoing Primary Percutaneous Coronary Intervention

After obtaining results from univariate and multivariate logistic regression analysis, we developed a novel score “RURUS SURYAWAN” to predict 30-day mortality among first diagnosed acute myocardial infarction patients undergoing primary percutaneous coronary intervention.

Table 3. RURUS SURYAWAN score assignment.

Variables	Multivariate analysis			Point assigned
	OR	95% CI	P value
Resting heart rate (>110 bpm)	1.51	1.20 - 1.90	0.024	1
Underweight (BMI < 18kg/m2)	1.37	1.15 - 1.99	0.033	1
Respiratory rate (>28 / minutes)	1.18	1.03 - 1.34	0.039	1
Urine-output, first 24 hours admission (<0.5 ml / kg BW / hour)	2.56	1.79 - 4.44	<0.001	1
Saturation O2 peripheral (<90%)	1.35	1.21 - 1.49	0.026	1
Systolic blood pressure during admission (<90 mmHg)	2.31	1.67 - 4.02	<0.001	1
Urea nitrogen (> 50 mg/dL)	1.93	1.81 - 2.04	<0.001	1
Reduced ejection fraction. (baseline LVEF < 40%)	1.21	1.08 - 1.37	0.042	1
time delaYed of primary percutaneous coronary intervention (door-to-balloon time >2 hours)	1.89	1.56 - 2.37	0.002	1
Age (>70 years old)	1.55	1.23 - 2.08	0.021	1
Women	1.17	1.02 - 1.33	0.041	1
Anemia (Hb < 10 g/dL)	1.79	1.45 - 2.55	0.003	1
NT-proBNP (> 1500 pg/mL)	2.17	1.58 - 3.74	<0.001	1
Total score				13

Table 3. RURUS SURYAWAN score assignment.

Variables	Multivariate analysis			Point assigned
	OR	95% CI	P value
Resting heart rate (>110 bpm)	1.51	1.20 - 1.90	0.024	1
Underweight (BMI < 18kg/m2)	1.37	1.15 - 1.99	0.033	1
Respiratory rate (>28 / minutes)	1.18	1.03 - 1.34	0.039	1
Urine-output, first 24 hours admission (<0.5 ml / kg BW / hour)	2.56	1.79 - 4.44	<0.001	1
Saturation O2 peripheral (<90%)	1.35	1.21 - 1.49	0.026	1
Systolic blood pressure during admission (<90 mmHg)	2.31	1.67 - 4.02	<0.001	1
Urea nitrogen (> 50 mg/dL)	1.93	1.81 - 2.04	<0.001	1
Reduced ejection fraction. (baseline LVEF < 40%)	1.21	1.08 - 1.37	0.042	1
time delaYed of primary percutaneous coronary intervention (door-to-balloon time >2 hours)	1.89	1.56 - 2.37	0.002	1
Age (>70 years old)	1.55	1.23 - 2.08	0.021	1
Women	1.17	1.02 - 1.33	0.041	1
Anemia (Hb < 10 g/dL)	1.79	1.45 - 2.55	0.003	1
NT-proBNP (> 1500 pg/mL)	2.17	1.58 - 3.74	<0.001	1
Total score				13

The estimated multivariate logistic regression model with the 30-days mortality risk score demonstrated excellent calibration-in-the-large, with an intercept of 0.016. The difference in log-odds ratio between predictions and observed outcomes in the validation group was statistically significantly (P = 0.009). When the logistic regression model was recalibrated to the derivation group and applied to subjects in the validation group, it demonstrated good calibration-in-large and the calibration slope was significantly different from (P<0.001). The concordance between the observed probability of 30-days mortality and the predicted probability of 30-days mortality across the 13 vigintiles of variables in the validation sample is described in Figure 1. The Hosmer-Lemeshow X² statistic of the full model was 3.56 (P = 0.795)

When assigning a point score, each variables in the Table 3 gave 1 point assigned. Therefore, a total of 13 point could be assigned. The score ranged from 0 to 13 and mortality rates increased with higher score (Ptrend < .0001). After grouping patients into low-risk (0–3 points), intermediate-risk (4–6 points), high-risk (7–9 points), and very high-risk (10–13 points) in the classification groups (Table 3), most deaths occurred in the high-risk and very high-risk group. Mortality rates were <1%, 2-5%, 8-30% and >50% in low-risk, intermediate-risk, high-risk, and very high-risk groups, respectively (Ptrend < 0.001). This trend persisted when using multiple imputation or a conservative analytical approach as seen in the Figure 1.

Table 3. RURUS SURYAWAN score stratification.

RURUS SURYAWAN score	Stratification	Risk of 30-day mortality
0-3 points	Low risk	<1%
4-6 points	Intermediate risk	2-5%
7-9 points	High risk	8-30%
10-13 points	Very high risk	>50%

Table 3. RURUS SURYAWAN score stratification.

RURUS SURYAWAN score	Stratification	Risk of 30-day mortality
0-3 points	Low risk	<1%
4-6 points	Intermediate risk	2-5%
7-9 points	High risk	8-30%
10-13 points	Very high risk	>50%

The area under the ROC curve of the RURUS SURYAWAN score to predict 30-day mortality was was 0.944 (0.906–0.972) in the derivation set and 0.959 (0.921–0.983) in the validation set, with 94.6% sensitivity and 97.3% specificity (P < 0.001). The area under the ROC curve of the HAS-BLED score to predict 30-day mortality was 0.717 (0.680-0.752), with 85.1% sensitivity and 51.5% specificity (P < 0.001). The area under the ROC curve of TIMI score to predict 30-day mortality was 0.844 (0.813-0.871), with 91.0% sensitivity and 61.6% specificity (P < 0.001).

4. Discussion

We developed and validated RURUS SURYAWAN score, an improvisation from the previous risk model to predict 30-day mortality for firstly diagnosed acute myocardial infarction patients after primary percutaneous coronary intervention. RURUS SURYAWAN scoring system contains 15 variables with a score range from 0-20. The final parsimonious scoring systems doesn’t require any collection of sophisticated parameter, making their applications in clinical practice easier and quicker than others scoring system. This model includes resting heart rate (>110 bpm), underweight (BMI < 18kg/m²), respiratory rate (>28 / minutes), first 24 hours urine-output (<0.5 ml / kg BW / hour), peripheral O₂ saturation (<90%), systolic blood pressure during admission (<90 mmHg), urea nitrogen (> 50 mg/dL), reduced ejection fraction (< 40%), timing of primary percutaneous coronary intervention (door-to-balloon time >2 hours), age (>70 years old), sex (women), anemia (Hb < 10 g/dL), and elevated NT-proBNP (> 1500 pg/ml) as factors for risk adjustment. The model performed well in an independent validation cohort, as well as in various subgroups stratified by cardiac arrest and other clinical factors. A simplified integer score based on this model also performed well and can potentially serve as a foundation for prospective risk stratification at the point of care.

The RURUS SURYAWAN score is stratified into low-risk, intermediate-risk, and high-risk categories based on 13 variables as mentioned in the Table 3. The scores vary from “0 to 13-point scores.” As seen in the Table 4, a score of 0–3 indicates a patient is at “low risk” and advises consideration of routine or usual care after primary PCI and further investigations can be planned during outpatient department visits. Our study saw an 30-day mortality less than 1% for low-risk patients. Patients with a score of 4–6 are classified as “intermediate risk,” and they are usually considered safe but sometimes require monitoring with 2-5% risk of 30-day mortality. Patients with a score of 7-9 are classified as “high risk,” and they require strict clinical observation, including longer duration of intensive care. About 8-30% of high-risk patients had 30-day mortality in our study. Patients with a score of 10-13 are deemed “very high risk” with more than 50% risk of 30-day mortality and require early aggressive care with very close hemodynamic monitoring during 30-day stay.

In the literature, mostly comparable results were found when comparing RURUS SURYAWAN score with HEART score, TIMI score, ACUITY score, CRUSADE score, and GRACE score. In previous study, the AUC of the HEART score was 0.83 (95% CI: 0.81–0.85), the AUC of the TIMI score was 0.75 (95% CI: 0.72–0.77), the AUC of the ACUITY score was 0.72 (95% CI: 0.69–0.75), the AUC of the CRUSADE score was 0.64 (95% CI: 0.61–0.68), the AUC of the GRACE score was 0.78 (95% CI: 0.75–0.81); which are slightly lower than our AUC of RURUS SURYAWAN score was 0.944 (0.906–0.972) in the derivation set and 0.959 (0.921–0.983) in the validation set [22,23]. The TIMI risk score estimates mortality based on: age ≥65, ≥3 coronary artery disease (CAD) risk factors, known CAD (stenosis ≥50%), aspirin use in the past 7 days, severe angina (≥2 episodes in 24 hours), ST-segment deviation ≥0.5 mm, and positive cardiac markers. The GRACE risk score was calculated by the eight different baseline variables incorporated in the risk calculator: age, heart rate (beats per minute), systolic blood pressure, serum creatinine level, cardiac arrest at admission, ST-segment deviation on ECG, elevated cTn, and congestive heart failure (Killip class).

In our retrospective study, several predictors were associated with higher 30-day mortality. Relevant risk factors included vital signs. Vital signs are easy and inexpensive to measure, and when it is measured under controlled conditions, the results are highly reproducible. There are no technical or financial barriers to its incorporation during 30-day admission. Deterioration in vital sign due to acute myocardial infarction are known to indicate worsened prognosis. Thus, close and intense monitoring, early recognition of deterioration, and appropriate treatment to intervene vital signs during hospital admission may be beneficial to improve the prognosis of AMI patients. Cardiac arrest has been shown to be a strong important predictor of AMI mortality from several reports [24], nevertheless we decided not to include cardiac arrest since we faced a difficulty to adjust risk for these events, given their heterogeneity in clinical severity, and inclusion of these patients in hospital scorecards for percutaneous coronary intervention can result in unintended consequences to withhold aggressive treatment.

One of the most important predictor for mortality is resting heart rate. This study is in-line with study from Jabre, et al (2014) which revealed that elevated resting heart rate at the time of the acute myocardial infarction identifies patients at increased risk of all-cause and cardiovascular mortality [25]. Even though heart rate measured upon admission is dependent on the sympathetic activity due to the chest pain, stress, or anxiety, however Jabre, et al (2014) found that resting heart rate upon admission was still a strong prognostic marker of all-cause and cardiovascular mortality for resting heart rate > 110 bpm and cardiovascular mortality only for resting heart rate > 80 bpm during admission [25]. Indeed, resting heart rate might theoretically contribute to increased mortality by increasing myocardial oxygen consumption, thus worsening ischemia, increasing infarct size and stimulating the atherosclerotic progression and plaque instability [26].

Regarding blood urea nitrogen, according to Zhu et al (2022), blood urea nitrogen was robustly associated with increased short-term mortality in patients with acute myocardial infarction developing cardiogenic shock [27]. Horiuchi et al (2018) also found that blood urea nitrogen is a strong predictor for 30-day mortality among acute myocardial infarction patients [28].

Inclusion of ejection fraction obtained from transthoracic echocardiography may add significant prognostic information. Previous report from Bosch and Théroux (2005) revealed that adding baseline LVEF into the model improved the prediction of mortality (C statistic 0.73 vs 0.67) [29].

Individual patient data from four trials (CAPRICORN, EPHESUS, OPTIMAAL, and VALIANT), showed that LVEF is a ubiquitous risk marker associated with death regardless of type of AMI. Thus, recent clinical practice guidelines recommended initial measurement of LVEF for patients with STEMI and NSTEMI, because LVEF has both prognostic and therapeutic significance for acute myocardial infarction patients [30]. Unfortunately, significant variability in LVEF measurement rates across clinician remains.

Additionally, our study results showed that delayed of primary percutaneous coronary intervention poses a greater risk of 30-day mortality among patients with firstly diagnosed acute myocardial infarction patients. Data from Thailand PCI registry in 2006 showed the median door-to-device time for AMI patients was 122 minutes, and the overall mortality rate was 17.0% [31]. Existing studies have also revealed that the impact of short door-to-device time, particularly those smaller than 60 minutes, on patient mortality remains debatable [32]. Tsukui, et al (2020) observed that a longer duration of primary PCI >2 hours was significantly linked with 30-day mortality from all causes despite correcting for cardiogenic shock, arrhythmias, or kidney injury [33]. However, shortening door-to-device time <1 hour was not related with mortality from any cause in the multivariate Cox regression analysis [33]. Consistent with our study, AMI patients treated with primary PCI, door-to-device time longer than 120 minutes (2 hours) was related with 30-day mortality, but door-to-device time less than 60 minutes (1 hour) was not associated with survival benefit (OR 1.89, 95% CI: 1.56–2.37, P=0.002).

Age is one of the most important predictor of survival after acute myocardial infarction. Older patients have a greater risk of 30-days mortality due to recurrent myocardial infarction, heart failure, cardiogenic shock, atrioventricular block, and atrial fibrillation or flutter. In the angiographic substudy of GUSTO-III trial, older patients had greater risk for TIMI grade 0 flow and lower rates of TIMI grade 3 flow, more multivessel disease, and lower left ventricular ejection fractions [34]. Older patients had their own risk for all-cause mortality, due to their anatomic complexity, physiological vulnerability, age-related risks (including prevalent geriatric syndromes), multimorbidity, frailty, disability, cognitive decline, polypharmacy, heterogeneity in life expectancy and goals of care [35].

Women also tend to have significantly higher in-hospital mortality for acute myocardial infarction than men, according to our study. Since women are generally older at hospitalization and have higher comorbidity rate than men. Previous report has suggested that women have longer delays from symptom onset to hospitalization and, as a result, tend to have higher 30-day mortality rates for acute myocardial infarction than men [36]. Other study also showed women hospitalized for acute myocardial infarction died more frequently than men (9.3% vs. 6.1%, P < 0.01) [37].

Anemia is also recognized as strong, independent risk factors for mortality after acute myocardial infarction. The prevalence of anemia of 28% among acute myocardial infarction patients underwent primary percutaneous coronary intervention. Our study suggested a significant increase risk for 30-day mortality in acute myocardial infarction patients with anemia (Hb < 10 g/dL) once various confounding conditions were adjusted for. Previous study revealed that overall mortality in each group containing patients with anemia was higher than in the corresponding non-anemic groups [38]. From The Myocardial Ischemia National Audit Project (MINAP) registry, anemia is independently associated with 30-day (OR 1.28, 95% CI 1.22-1.35) and 1-year mortality (OR 1.31, 95% CI 1.27-1.35), with a reverse J-shaped relationship between hemoglobin levels and mortality outcomes [39]. In the setting of acute myocardial infarction, anemia might worsen ischemia by decreasing the oxygen delivery to the jeopardized myocardium and increase myocardial oxygen demand due to greater cardiac output to maintain adequate systemic oxygen delivery [40]. Also, patients with anemia are often underprescribed antiplatelet therapy due to bleeding concerns [41].

Lastly, NT-proBNP also has an incremental prognostic value for 30-day mortality over and beyond the TIMI risk score and the GRACE risk calculator. Data concerning optimal timing to measure NT-proBNP during acute myocardial infarction are limited. One of the aforementioned studies, which confirmed additional value of NT-proBNP, was based on measurements at 24–96 hours [42]. Other study revealed that NT-proBNP values upon admission were significantly higher in patients who died, compared to those who survived [43]. Thus, it is consistent with our findings that NT-proBNP had a good and comparable predictive value for 30-day mortality.

4.1. Strength

As far as we know, RURUS SURYAWAN score has the strongest AUC with better sensitivity and specificity to discriminate 30-day mortality in the in disease-specific cohorts among Asian patient with first episode of acute myocardial infarction underwent primary percutaneous coronary intervention during hospital admission. Our study used a combination of derivation and validation feature selection methods. Cross-validation and hyperparameter tuning improved model performance and reduced over-fitting risk. A pair-wise corrected resampled t-test was employed to compare model predictions. To ensure the current study’s reliability, all models were validated using untouched validation data that was not used for model construction. We expect our score to be further externally validated in different databases allowing additional comparison with other scores.

4.2. Limitation

There are certain limitations to our current study. Firstly, RURUS SURYAWAN score that we developed were for predicting 30-day mortality. The scores have not been validated for predicting mortality within different time frames. Secondly, it requires the collection of variables during admission, making their applications in clinical practice difficult. We attempted to mitigate this effect by using easily obtained variables in the emergency ward. On the other hand, we recognized that several missing variables could result in a skewed outcome. Another potential limitations of this study should be acknowledged. Since this research is based on a single-centre retrospective study, this study inevitably has a particular selection bias. It is difficult to control selection bias inside our database. Therefore, 30-day outcomes at our centre may not be generalizable to other centre. We expect that subsequent investigations conducted in the multicentre study will corroborate our findings.

5. Conclusions

Our results indicate that standard clinical characteristics routinely obtained during the initial medical evaluation of patients with firstly diagnosed acute myocardial infarction can be used to construct RURUS SURYAWAN score, a simple classification system to predict risk of 30-day mortality. This risk model for 30-day mortality could evaluate interventions to improve the outcomes of patients with acute myocardial infarction. Although it has achieved adequate internal validation, it must be externally validated.