Shock Index-C: An Updated and Simple Risk-Stratifying Tool in ST-Segment Elevation Myocardial Infarction

Background: Shock index (heart rate/systolic blood pressure, SI) is a simple scale with prognostic value in patients with ST-segment elevation myocardial infarction (STEMI) undergoing percutaneous coronary intervention (PCI). The present study introduces an updated version of SI that includes renal function. Methods: A total of 1,851 consecutive patients with STEMI undergoing PCI were retrospectively included at Cardiac Care Unit in Guangdong Provincial People's Hospital and divided into two groups according to their admission time: derivation database (from January 2010 to December 2013, n = 1,145) and validation database (from January 2014 to April 2016, n = 706). Shock Index-C (SIC) was calculated as (SI × 100)–estimated CCr. Calibration was evaluated using the Hosmer-Lemeshow statistic. The predictive power of SIC was evaluated using receiver operating characteristic (ROC) curve analysis. Results: The predictive value and calibration of SIC for in-hospital death was excellent in derivation [area under the curve (AUC) = 0.877, p < 0.001; Hosmer-Lemeshow chi-square = 3.95, p = 0.861] and validation cohort (AUC = 0.868, p < 0.001; Hosmer-Lemeshow chi-square = 5.01, p = 0.756). SIC exhibited better predictive power for in-hospital events than SI (AUC: 0.874 vs. 0.759 for death; 0.837 vs. 0.651 for major adverse clinical events [MACEs]; 0.707 vs. 0.577 for contrast-induced acute kidney injury [CI-AKI]; and 0.732 vs. 0.590 for bleeding, all p < 0.001). Cumulative 1-year mortality was significantly higher in the upper SIC tertile (log-rank = 131.89, p < 0.001). Conclusion: SIC was an effective predictor of poor prognosis and may have potential as a novel and simple risk stratification tool for patients with STEMI undergoing PCI.


INTRODUCTION
Despite advanced evidence-based medical treatments and the widespread use of percutaneous coronary intervention (PCI), ST-segment elevation myocardial infarction (STEMI) is a leading cause of mortality worldwide (1). The in-hospital mortality rate for STEMI is 3-4% following PCI, and can reach up to 10% in any given year (2,3). In addition, increased risk of bleeding and acute kidney injury was observed in patients with STEMI, which was associated with poor outcomes (4,5). Many scales have been developed for early identification of patients with STEMI at high risk for adverse outcomes. Global Registry of Acute Coronary Events (GRACE) and Thrombolysis in Myocardial Infarction (TIMI) risk scales are guideline-recommended risk stratification tools for prediction of STEMI-related mortality (6). Mehran et al. (7) and Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the American College of Cardiology/American Heart Association Guidelines (CRUSADE) scores are used to predict contrastinduced acute kidney injury (CI-AKI) and bleeding, respectively (8). Each score predicts a different event, and collection of data and calculation of these four scores is time consuming in clinical practice. A simpler scoring system that can simultaneously assess the risk of mortality, bleeding, CI-AKI, and other STEMI-related adverse events is of critical importance.
Shock index (SI), defined as the ratio of heart rate to systolic blood pressure, is a simple risk-stratification tool used to evaluate patients with STEMI (9). However, the discriminatory ability of SI for short-and long-term adverse events is considered insufficient (10). In addition to heart rate and systolic blood pressure, renal function is an essential element of the GRACE, Mehran and CRUSADE scales (7,11,12), but the effect of addition of renal function to SI on prediction of poor prognosis has not been characterized. In this study, we developed a new model, Shock Index-C [SIC; (SI × 100)-estimated creatinine clearance rate (CCr)], and validated its predictive ability in patients with STEMI undergoing PCI. guidelines (6). The exclusion criteria were as follows: (1) Hospital stay <24 h; (2) malignant tumor; (3) concomitant aortic dissection; and (4) missing admission serum creatinine (Scr) data. After screening, 1,851 patients were included (Supplementary Figure 1). This study was approved by the Ethics Committee of our hospital, and the requirement for informed consent was waived due to the retrospective nature of the study.

Clinical Data and SIC
Clinical data were collected from the electronic case report form by one researcher and randomly confirmed by another researcher. Data collected included patient demographics, previous medical history, laboratory results, PCI procedural details, adverse events, and medical treatment. Vital signs including blood pressure and heart rate were obtained from data recorded at admission. SI was calculated using the following formula: heart rate (bpm)/systolic blood pressure (mmHg). Estimated creatinine clearance rate (CCr) was calculated with the published equations for Cockcroft-Gault: man: (140-age)/ Scr, woman:(140-age) /Scr × 0.85 (13). SIC was calculated using the following formula: (SI×100)-estimated CCr. GRACE, CRUSADE, Mehran, and TIMI risk scores were calculated using initial clinical history, electrocardiograms, laboratory values, and PCI procedural information collected at admission.

Follow-Up and Endpoints
All surviving in-hospital patients were followed-up through telephone interviews. We reviewed hospital readmission records and outpatient clinic interviews for possible events. The primary endpoints were in-hospital and 1-year mortality. In-hospital major adverse clinical events (MACEs) such as stroke, dialysis, acute heart failure, and target vessel revascularization (TVR) during hospitalization were used as a composite end point. In addition, CI-AKI and bleeding were recorded. Contrast-induced acute kidney injury was defined as elevation of Scr by 50%, or 0.3 mg/dL from baseline, within 48 h, according to the Kidney Disease: Improving Global Outcomes criteria (14).

Statistical Analysis
Normally distributed continuous data are presented as the mean ± standard deviation, and non-normally distributed continuous data are presented as the median and interquartile range, which were compared using independent sample t and nonparametric tests, respectively. Categorical variables are displayed as numbers and percentages, were compared using chi-square tests. The Hosmer-Lemeshow goodness-of-fit test was used to assess the goodness of fit of the final regression model. Receiver operating characteristic (ROC) curve analysis was conducted to determine the optimal cut-off levels of SIC for prediction of adverse events. Decision curve analysis was used to compare the predicting performance for in-hospital mortality by quantifying the net benefits. Area under the curve values were compared using the nonparametric approach described by DeLong et al. (15). In addition, net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were calculated. The performance of SIC in subgroups was further assessed using   ROC curves. Variables significantly associated with mortality in the univariate analysis were included in the multivariable analysis. Kaplan-Meier curves were generated and compared using the log-rank test. P < 0.05 were considered statistically significant. All analyses were conducted using SPSS software (version 22.0; SPSS Inc, Chicago, IL, USA).

Development and Validation of SIC
In derivation database, the significant factors except the elements of SI and CCr in the univariate logistic regression analysis for in-hospital mortality were included into multivariate model (Supplementary Table 1). Cardiac arrest before admission, CCr, LVEF, and IABP were risk factors for in-hospital death independently of SI. Renal function is an essential element of previous scores in STEMI patients. Therefore, we added CCr into SI to create a new variable, SIC. The chi-square statistic for calibration was 3.95 (p = 0.861, Figure 1), indicating good discriminatory power and goodness-of-fit. Receiver operating characteristics curve analysis was performed to determine the predictive power of SI, CCr, the combination of SI and CCr using a logistic model (linear combination predictor), and SIC, for in-hospital death in derivation database. SIC had an excellent predictive value for in-hospital mortality (AUC = 0.877, 95% CI: 0.833-0.921, p < 0.001). In addition, the predictive power was higher than SI (AUC: 0.877 vs. 0.723, p < 0.001; NRI = 37.0%, 95% CI: 17.4-56.6, p < 0.001; IDI = 7.5%, 95% CI: 4.2-10.8, p < 0.001; Figure 2) and relatively higher than CCr (AUC: 0.877 vs. 0.838, p = 0.058; NRI = 21.1%, 95% CI: −3.6-45.7, p = 0.094; IDI = 6.6%, 95% CI: 0.5-12.8, p = 0.034; Figure 2). The decision curves analysis showed that SIC had the highest overall net benefit compared with SI and CCr (Figure 3). The AUC values for SIC and the linear combination predictor were similar (AUC: 0.877 vs. 0.876, p = 0.781, Figure 2). SIC was selected due to ease of clinical use. The AUC of SIC for in-hospital mortality was 0.868 (95% CI: 0.803-0.934) and was calibrated with a Hosmer-Lemeshow chi-square statistic of 5.01 (p = 0.756, Figure 1) in validation database.

Subgroup Analysis
In addition, the SIC also exhibited good discrimination among subgroups in the overall population (Table 3), such as age, gender, and infarcted area.

DISCUSSION
This study was the first to combine SI and renal function for determination of prognosis of patients with STEMI undergoing PCI. This novel prognostic scale, termed SIC exhibited a good discriminatory power and goodness-of-fit in derivation and validation database, which had greater predictive accuracy than SI for in-hospital adverse events. The discriminatory capacity of SIC for in-hospital death was non-inferior to the GRACE scale, and SIC was a better predictor than TIMI risk score. In addition, SIC displayed modest discriminatory ability for identification of CI-AKI and bleeding. The cumulative 1-year mortality was significantly higher in patients with a high SIC. Use of SIC could provide prognostic information to aid in early rapid risk assessment of patients with STEMI undergoing PCI.
The mortality rate of patients with STEMI is high despite use of PCI and secondary preventive strategies. Many factors, including tachycardia, hypotension, and renal function have been shown to be independent predictors of early STEMI-related death (6). However, no single biomarker has been identified that can predict adverse events. Therefore, several risk scales have been established as prognostic indicators for STEMI.
The TIMI risk score was first developed for fibrinolyticeligible patients with STEMI, and its accuracy and clinical applicability were validated in patients with STEMI undergoing PCI (16,17). This score is recommended by contemporary guidelines for use in patients with STEMI, and is a simple bedside scoring tool that can effectively predict in-hospital mortality (18). However, TIMI is only a modest predictor of 1-year mortality (AUC = 0.73) (19). Current guidelines also recommend use of the GRACE scale as a prognostic indicator for patients with STEMI (6). The GRACE scale exhibits good discriminatory performance for short-and long-term outcomes (AUC ≥ 0.8) (20). In our study, GRACE score had higher predictive value than TIMI risk score, which agreed with the findings in a previous study (21). However, GRACE score is not routinely used in patients with STEMI because it uses a complex formula that requires computer-assisted calculation. Therefore, a simpler risk score with adequate predictive accuracy was needed for patients with STEMI.
SI was initially described in 1967, and is commonly used to assess the severity of shock in the clinical setting (22). SI has also been used for risk assessment of several conditions including trauma, stroke, and sepsis (23)(24)(25). In STEMI, sustained obstruction of blood supply results in myocardial necrosis, which can result in cardiac dysfunction (26). Epidemiological data showed that approximately 6% patients with STEMI also had cardiogenic shock (27). Cardiogenic shock was associated with significantly higher short-and long-term risk for mortality, as evidenced by an increase from 25 to 50% (28). Therefore, SI is considered a viable risk projection model for patients with STEMI. However, the predictive accuracy of SI is not adequate.
The AUC values for in-hospital and one-year mortality were 0.703 and 0.660, respectively (10,29). Renal dysfunction is believed to be a risk factor for patients with STEMI (1, 6). Cywinski et al. showed that estimated renal function was a better prognostic indicator than Scr (30). CCr by Cockcroft-Gault has adequate discriminatory ability, with an AUC > 0.8 for prediction of poor outcomes, which was better than other equations for glomerular filtration rate estimation in patients with acute coronary syndrome (31,32). In addition, renal function is an important element in the GRACE, Mehran, and CRUSADE scales (7,11,12). Therefore, we hypothesized that addition of CCr to SI could result in better predictive accuracy in patients with STEMI undergoing PCI. Our results showed that the predictive value of SIC for in-hospital death was equivalent to that of GRACE and better than that of TIMI risk score. In addition, SIC did better in predicting MACEs than these two scales. SIC is calculated using only 3 variables, which may result in greater use by clinicians.
In the present study, we also explored the discriminatory ability of SIC for CI-AKI and bleeding. These complications occur frequently in patients with STEMI undergoing PCI, and are associated with poor prognosis (4,5). The Mehran score consists of 8 variables and has been validated as an accurate predictor of CI-AKI (7,33). The procedural variables (contrast media volume) included in this score have limited its early application. In our study, although SIC was similar to Mehran score for prediction of CI-AKI, it could serve as a rapid and effective tool for early prediction of CI-AKI. Given that blood pressure, heart rate, and renal function contribute to bleeding, we compared the predictive abilities of SIC and CRUSADE (12). SIC had equivalent predictive value for bleeding as CRUSADE.
SIC presented in this study has several advantages. First, it included 3 risk factors that were easily collected and calculated. Second, although it has less variables, it has similar discriminatory ability with guidelines recommending riskstratified score. Third, it shared the same risk factors with previous scores for predicting CI-AKI and bleeding. By using these simple data (heart rate, systolic blood pressure, and CCr), SIC can be used to assess and stratify the risk for multiple events rapidly and precisely, at no additional cost or effort.

Study Limitation
Our study suffered from several limitations. First, this study was retrospective, of moderate-scale, and from a single center, and this score should be externally validated in a largescale multicenter study. Second, blood pressure was not invasively measured in catheterization room, which is more reliable than sphygmomanometer. However, it could represent the contemporary and real-world clinical practice. Third, the predictive value of SIC was only validated in patients with STEMI undergoing PCI, and caution should be used when using this scale to evaluate patients not undergoing PCI. Fourth, the proportion of female in this study was relatively small, the predictive ability should be validated in another female cohort of STEMI, despite the good discrimination power in the subgroup analysis.

CONCLUSION
In conclusion, we showed that SI was a good prognostic indicator in patients with STEMI undergoing PCI, but its discriminatory ability was insufficient. Addition of renal function to SI resulted in better predictive power and good calibration. SIC had similar predictive value for in-hospital death as GRACE score, and better discrimination power than TIMI risk scales. In addition, SIC showed modest predictive value for CI-AKI and bleeding. Higher SIC was an independent predictor for 1-year mortality. This indicator might provide prognostic information for early and rapid risk assessment of patients with STEMI undergoing PCI.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethics Committee of Guangdong Provincial People's Hospital. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.