a b s t r a c t

Objectives: This study aimed to make a comparison between classical Shock Index , modified shock index (MSI), and age shock index (Age SI) for predicting critical patients presenting to the emergency department (ED) with gastrointestinal bleeding (GIS).

Methods: The study, which was planned retrospectively, consisted of patients diagnosed with GIS bleeding at the ED admission. Triage time vital signs were used to calculate SI, MSI, and age SI. These results were compared with intensive care admission, endoscopic/colonoscopic (E/C) intervention, blood transfusion, and mortality criteria, which we define as adverse outcomes.

Results: The study included 151 patients. Seventy-nine (52.32%) of the patients had at least one adverse outcome. Of the 151 patients, 19 (12.58%) had ICU admission, 27 (17.88%) underwent endoscopic/colonoscopic (E/C) in- tervention, 68 (45.03%) received a blood transfusion, and 6 (3.97%) died. There was a significant difference be- tween patients who had no adverse outcome and those who had at least one adverse outcome in terms of SI, age SI, and MSI. We performed ROC curve analyses to evaluate the diagnostic performances of all indices for predicting adverse outcomes. AUC values for age SI was the highest (age SI AUC = 0.711, p < 0.001; SI AUC = 0.616; MSI AUC = 0.617). The performance of the age SI was significantly higher than the SI (p = 0.013) and the MSI (p = 0.024) for predicting adverse outcomes. The cut-off value for the age shock index was 45.12.

Conclusions: In patients with GIS bleeding, age SI, which can be easily calculated in triage, is more significant than SI and MSI for predicting the critical patient.

(C) 2021

1. Introduction

Patients with gastrointestinal bleeding presenting to the Emergency Department (ED) are often critically ill and have much morbidity and mortality associated. There are more the one million hospitalizations per year in the US for GI bleeding and with admissions from the ED for GI bleeding ranking first in numbers in the US, it is also one of the leading causes of ED admissions and hospitalizations in our country [1]. The clinical spectrum of patients may range from coffee ground vomitus to sudden hypotension and mortality. For this reason, guide- lines recommend risk stratification at the first presentation to discrimi- nate between high and low-risk patients. Making this distinction particularly during the first presentation in the ED is important in terms of the need for blood transfusion, the timing of endoscopy, hospi- talization decision (intensive care or ward), and close follow-up of the risky patient.

* Corresponding author at: Balikesir University Faculty of Medicine Altieylul, Balikesir 10100, Turkey.

E-mail address: [email protected] (S. Kocaoglu).

Various scoring systems, such as Glasgow Blatchford score (GBS), Rockall Score, AIM65, and ABC score, have been developed and validated in previous studies [2]. However, the complexity in the calculation of these scoring systems, the need for laboratory results, and even the need for endoscopic evaluation make it difficult to use in practice by emergency physicians. We think that a simpler and more memorable method that enables early detection of risky patients may benefit clini- cians. Blood pressure and pulse are routinely measured in ED to deter- mine the severity of patients. ‘Shock index’ (SI) is a simple parameter that is calculated by dividing the heart rate by systolic blood pressure (SBP) and can show the hemodynamic status better than the use of heart rate and SBP alone [3]. The normal range has been accepted as 0.5-0.7 [4]. “Modified Shock Index” (MSI) and “Age Shock Index” (age SI) are indices derived from shock index, which include mean arte- rial pressure (MAP) and age and have been recently used in the progno- sis of critical patients. MSI is calculated by pulse/MAP and provides an assessment of diastolic blood pressure , too, compared to the tra- ditional shock index. Some studies have shown that it is more effective in predicting mortality than pulse, SBP, DBP, or shock index alone [5]. It is known that physiological reserve decreases and that mortality is more pronounced in elderly patients. The ‘age SI’, developed accordingly,

https://doi.org/10.1016/j.ajem.2021.05.008

0735-6757/(C) 2021

is calculated by multiplying the patient age by SI. It has been shown to predict 48-h mortality, particularly in elderly patients, better compared to pulse, SBP, and SI [6].

SI, MSI and age SI are parameters that can be easily calculated in the early period during triage assessment in ED. We know from previous studies that shock index is used to estimate prognosis in pa- tients with hypovolemia, sepsis, myocardial infarction, and pneumo- nia [7-9]. There are conflicting studies in the literature regarding whether the shock index is useful in estimating prognosis in patients with GIS bleeding [3,8,10]. However, we found no studies comparing these three indices with each other in estimating prognosis in pa- tients with GIS bleeding. This study aimed to investigate the superi- ority of the traditional shock index, which can be easily calculated in triage during the first presentation to the ED due to GIS bleeding, vs. MSI and age SI.

1. Materials and methods

This study was conducted with 151 patients who were aged 18 and older and presented to the Emergency Department of Balikesir Univer- sity Faculty of Medicine between January 2019 and December 2020 with symptoms of GIS bleeding. For this retrospective study, approval of the Clinical Research Ethics Committee was obtained (Decision no: 2021/12). Patients who were diagnosed with GIS bleeding as a result of the examination done by the emergency physician and laboratory findings and whose diagnosis was confirmed by gastroenterology con- sultation were included in the study. The data were obtained from the hospital archive. Patients whose files could not be accessed or had miss- ing data were excluded from the study.

Vital signs (SBP, DBP, pulse) of the patients measured by the tri- age nurses of the ED were obtained from patient files. Based on these data, SI (pulse/SBP), MSI (pulse/MAP), and age SI (agexSI) scores were calculated. Also, patients’ symptoms at emergency ad- mission and the initial hemoglobin (Hb), hematocrit (Htc), urea, and creatinine values were examined. Patients were classified by discharge, clinic, and intensive care admission in terms of the out- come in the ED. Whether the patient received a blood transfusion during hospital stay was recorded. Procedures, such as emboliza- tion, band ligation, sclerotherapy, and hemoclips applications, performed in patients with active bleeding during endoscopy/colo- noscopy were defined as endoscopic/colonoscopic (E/C) interven- tions. The occurrence of one of the criteria in patients, such as intensive care admission, blood transfusion, intervention during endoscopy/colonoscopy, or in-hospital mortality, was considered as adverse outcomes.

In our study, we compared the relationships of SI, MSI, and age SI with intensive care, E/C treatment, blood transfusion, mortality, and total adverse outcome separately.

• 1. Statistical analysis

Shapiro-Wilk test was used to test the normality of variables. Contin- uous variables were presented as mean +- standard deviation values for the variables which were normally distributed and as median (mini- mum – maximum) values for the variables which were not normally distributed. Mann-Whitney’s U test was performed for comparing two independent groups. Categorical variables were expressed with counts and percentages. receiver operating characteristics (ROC) curve analy- sis was performed to evaluate and compare the performances of diag- nostic markers. Pairwise comparisons of the AUC values have been performed with the method of DeLong et al. (1988). The relationships between the variables were evaluated with the spearman correlation coefficient. The significance level was taken as ? = 0.05. Statistical anal- yses were performed with IBM SPSS Statistics version 22.0 (IBM Corp., USA) and MedCalc 12.3.0.0.

1. Results

The study included 151 patients, 125 (82.78%) of whom had upper GIS bleeding and 26 (17.22%) of whom had lower GIS bleeding. The me- dian age of 151 patients was 71 (minimum-maximum: 19-102) years, and 101 (66.89%) were male. Seventy-nine (52.32%) of the patients had at least one adverse outcome. Of the 151 patients, 19 (12.58%) were admitted to the ICU, 27 (17.88%) underwent E/C intervention, 68 (45.03%) received a blood transfusion, and 6 (3.97%) died. Other charac- teristics are summarized in Table 1.

There was a significant difference between patients with no adverse outcome vs. those who had at least one adverse outcome in terms of SI, age SI, and MSI. SI, age SI, and MSI were significantly higher in the group in which patients had at least one adverse outcome (Table 2).

We performed ROC curve analyses to evaluate the diagnostic perfor- mances of SI, age SI, and MSI for predicting adverse outcomes. Signifi- cant diagnostic performances were obtained for all three indices. AUC (area under the curve) values for age SI was the highest (AUC = 0.711, p < 0.001). We compared the AUC values of three indices. Pairwise comparisons of three AUC values have been performed to test the statistical significance of the difference between the AUCs with the method of DeLong et al. (1988). The performance of the age SI was significantly higher than SI (p = 0.013) and MSI (p = 0.024) for predicting adverse outcomes. But there was no significant difference between the performances of SI and MSI (p = 0.985) (Table 3, Fig. 1).

AUC values for SI and age SI were also statistically significant in predicting ICU admission. AUC value for SI was statistically significant in predicting E/C intervention. All three indices were significant predic- tors of blood transfusion. There were no statistically significant differ- ences among the three indices in predicting mortality (Table 3). The performance of age SI was significantly higher than SI (p = 0.011) and MSI (p = 0.034) for predicting blood transfusion, but there was no

Table 1

Baseline characteristics and outcome of patients (n = 151).

Characteristics	Descriptive statistics	n (%)
Age (years)b Gender (male) Bleeding	71 (19-102)	101 (66.89)

UGISB 125 (82.80)

LGISB 26 (17.20)

SBP (mmHg)^a 126.07 +- 27.17

DBP (mmHg)^b 70 (40-120)

HR (beats/min)^b 87 (52-141)

Symptom 138 (91.38)

Hematemesis	36 (26.10)
Melena	48 (34.80)
Hematochezia	28 (20.30)
Hematemesis and melena	26 (18.80)

HGB (g/dl)a HTC (%)b Blood urea (mg/dl)b Creatinine (mg/dl)b	10.00 +- 3.08 30 (0-78) 57 (0-293) 1 (0-6.30)
Clinical hospitalization		101 (66.88)
Adverse outcome		79 (52.32)
ICU admission	19 (12.58)
E/C intervention	27 (17.88)
Blood transfusion	68 (45.03)
Mortality SIb	0.70 (0.36-1.59)	6 (3.97)
Age SIb MSIb	48.45 (13.00-133.64) 0.99 (0.53-2.05)

Data presented as ^amean +- standard deviation, ^bmedian (minimum-maximum), or n (%) values.

UGISB: upper GIS bleeding, LGISB: lower GIS bleeding, SBP: systolic blood pressure, DBP: diastolic blood pressure, HR: heart rate, ICU: intensive care unit, SI: shock index, MSI: modified shock index.

*Adverse outcome denotes having at least one of these adverse outcomes: ICU admission,

E/C intervention, blood transfusion and in-hospital mortality.

Table 2

Comparison of the indices between the two groups.

Index	Adverse outcome		p-value
	Present (n = 79)	Absent (n = 72)
SI	0.72 (0.39-1.59)	0.66 (0.36-1.20)	0.013
Age SI	53.60 (26.82-133.64)	42.01 (13.00-80.40)	<0.001
MSI	1.02 (0.60-2.05)	0.95 (0.53-1.69)	0.015

Data given as median (minimum-maximum). SI: shock index, MSI: modified shock index.

Table 3

Diagnostic performance of the scoring indices.

Scoring index	AUC (95% CI)	p-value
Adverse outcome SI	0.616 (0.534-0.694)	0.011
Age SI	0.711 (0.631-0.781)	<0.001
MSI	0.617 (0.534-0.695)	0.011
ICU admission SI	0.643 (0.561-0.719)	0.026
Age SI	0.670 (0.589-0.745)	0.005
MSI	0.625 (0.543-0.703)	0.063
E/C intervention SI	0.620 (0.537-0.697)	0.031
Age SI	0.568 (0.485-0.648)	0.271
MSI	0.606 (0.523-0.684)	0.067
Blood transfusion SI	0.619 (0.537-0.697)	0.009
Age SI	0.712 (0.633-0.783)	<0.001
MSI	0.626 (0.544-0.703)	0.006
Mortality SI	0.582 (0.499-0.661)	0.587
Age SI	0.636 (0.553-0.712)	0.371
MSI	0.570 (0.487-0.650)	0.631

AUC: Area under the curve, CI: Confidence interval. Values in bold are significant results.

Fig. 1. Comparison of AUC values for scoring indices, SI, age SI, and MSI for predicting at least one adverse outcome (SI: shock index, MSI: modified shock index).

significant difference between SI and MSI (p = 0.727). No significant difference was found among SI, age SI, and MSI (SI-age SI: p = 0.473; SI-MSI: p = 0.606; MSI-age SI: p = 0.353) for predicting ICU admission. There was no significant difference among the performance of SI, age SI and MSI (SI-age SI: p = 0.337; SI-MSI: p = 0.610; MSI-age SI:

p = 0.541) for predicting E/C intervention, either.

optimal cut-off values for prediction of adverse outcomes were ob- tained according to the Youden J index. Corresponding accuracy,

sensitivity, specificity, positive predictive values, and negative predic- tive values are shown in Table 4.

There were significant correlations between the number of adverse outcomes and SI (r = 0.255, p = 0.002), age SI (r = 0.360, p < 0.001), and MSI (r = 0.246, p = 0.002). It was observed that as the number of adverse outcomes increased, SI, age SI, and MSI values also increased (Table 5).

1. Discussion

Distinguishing critical patients among those who present to the ED with GIS bleeding is of significance in terms of early intervention, close follow-up, and making intensive Care decisions. Since it is difficult and complex to calculate scores, such as GBS, AIMS 65, Rockall score, and recently developed ABC, in critical patient identification, their use in emergency practice has been limited. SI and its derivations, MSI and age SI, are easy and practical parameters that can be calculated while the patient is still in triage. SI has already been used effectively to distin- guish critical patients in Serious diseases, such as trauma, hypovolemia, myocardial infarction, and sepsis. However, there are conflicting studies about its effectiveness in estimating the prognosis of patients with GIS bleeding. For example, Rassameehiran et al. stated that shock index was greater than 0.7 in patients with upper GIS bleeding and was effec- tive in predicting adverse outcomes, such as the need for intensive care, blood transfusion, and endoscopic treatment [3]. Also, Nakasome et al. showed that shock index was associated with angiographic extravasa- tion in GIS bleeding [11]. Moreover, Horibe et al. (2016) developed a new 3-point scoring system including the shock index [12]. Contrary to all these findings, some studies in the literature argued that shock index was not a useful predictor in patients with GIS bleeding [8,10,13]. However, no studies investigating the use of age SI in patients with GIS bleeding and also comparing it with SI and MSI were found in the literature.

According to the literature, GIS bleeding is more common in males. In similar studies, the proportion of males was found as 68%, 58%, 63%, and 66%. In our study, 66.9% of the patients were male, too [8,10,13,14]. Saffouri et al. found an AUC value of SI 0.655 for the need for major transfusion (4 units of Erythrocyte suspension) in patients with upper GIS bleeding [10]. In our study, the AUC value for blood transfusion was 0.619 (p = 0.009). AUC values of MSI and age SI for blood transfu- sion were 0.626 (p = 0.006) and 0.712 (p < 0.001), respectively. Although the age shock index was significantly superior to the other two indices, all of the three indices were significant predictors of

blood transfusion.

Saffouri et al. found an AUC value of SI for the endoscopic treatment need in the same study as 0.606 [10]. Similarly, we found an AUC value of SI for E/C intervention of 0.620 (p < 0.031).

Jung et al. compared SI and MSI with scoring systems in patients with stable upper GIS bleeding. They found AUC values of SI and MSI for determining the need for intensive care as 0.544 and 0.552, respec- tively [13]. In our study, the AUC values of SI, MSI, and age SI determined for the need for intensive care were 0.643, 0.625, 0.670 (p = 0.026, p = 0.063, p = 0.005), respectively.

Semerci et al. found that the shock index in GIS bleeding did not pre- dict in-hospital mortality (AUC = 0.519, p = 0.72) [8]. Saafouri et al. found the AUC value of SI for the 30-day mortality of upper GIS bleeding as 0.611 [10]. Jung et al., in a similar study, found the AUC values of SI and MSI for in-hospital mortality as 0.601 and 0.585, respectively [13]. In our study, the AUC values of SI, MSI, and age SI for mortality were found as 0.582, 0.570, and 0.636, respectively. All of the three indices were found to be inadequate in predicting mortality, which was consis- tent with the literature.

In their study with 1233 patients with stable upper GIS bleeding, Jung et al. determined adverse outcomes in 165 (13.4%) patients. However, intensive care admission, recurrent gastric bleeding, and mortality were evaluated as adverse outcome criteria in this study.

Table 4

Cut-off values of the scoring indices for the prediction of adverse outcomes.

Scoring index	Cut-off	Accuracy (%)	Sensitivity (%) (95% CI)	Specificity (%) (95% CI)	PPV (%) (95% CI)	NPV (%) (95% CI)
SI	>0.67	58.94	68.35 (56.90-78.40)	52.78 (40.7-64.7)	61.40 (54.4-71.6)	60.30 (47.2-72.4)
Age SI	>45.12	57.61	75.95 (65.00-84.90)	62.50 (50.3-73.6)	69.00 (58.1-78.5)	70.30 (57.5-81.2)
MSI	>0.83	74.83	84.81 (75.00-91.90)	38.89 (27.6-51.1)	60.40 (50.6-69.5)	70.0 (53.5-83.4)

AUC: Area under the curve, CI: Confidence interval, PPV: positive predictive value, NPV: negative predictive value.

Table 5

Correlations between the number of adverse outcomes and indices.

Scoring index	r	p-value
SI	0.255	0.002
Age SI	0.360	<0.001
MSI	0.246	0.002

The AUC values of SI and MSI in the study for detecting adverse out- comes were 0.569 and 0.565, respectively, which were lower than GBS [13]. In our study, adverse outcomes were found in 79 (52.3%) of 151 patients. We think that this obvious difference was due to our different adverse outcome criteria. Our AUC values for adverse outcomes were 0.711, 0.616, and 0.617 for age SI, SI, and MSI, respec- tively. Age SI was found to be superior over classical SI and MSI in de- termining adverse outcomes.

When studies in the literature on age shock index not involving GIS were examined, it was found that age SI was superior to SI and MSI in the prediction of mortality. For example, Lee et al. showed that age SI was superior to classical SI and MSI in predicting hypo- tension after intubation in the ED. AUC values of age SI, SI, and MSI were determined as 0.676, 0.614, and 0.611, respectively (age SI-SI, p = 0.005; age SI-MSI, p = 0.006) [15]. In their study on 45,880 ge- riatric trauma patients, Kim et al. found that the Predictive power of age SI for in-hospital mortality was more significant compared to SI and MSI [16]. In their study conducted on 3375 patients in a Tertiary ED, Torabi et al. showed that age SI was superior to SI and MSI in predicting mortality. The ideal cut-off value for age SI was found to be 44.6, and the AUC value of SI was determined as 0.678 [17]. Yu et al. compared these three indices for the prediction of mortality in patients with acute myocardial infarction and found that age SI was more significant than SI and MSI. The cut-off value for age SI was calculated as 41, and its sensitivity and specificity were found as 59% and 72%, respectively [18].

In our study, it was observed that age SI was superior to SI and MSI in determining adverse outcomes in patients with GIS bleeding. In predicting adverse outcomes, the performance of age SI was observed to be more significant than SI (p = 0.013) and MSI (p = 0.024). The ideal cut-off value determined for age SI was 45.1, with 75.9% sensitiv- ity, 62.5% specificity, 69% PPD, and 70.3% NPD.

1. Conclusions

Unlike other complex scoring systems, we think that age SI can be used in emergency departments to distinguish critical patients with GIS bleeding since it can be calculated easily and is superior to classical SI and MSI. Its use alone or its inclusion in new scoring systems may be useful for detecting critical patients with GIS bleeding.

• 1. Limitations

Our study was retrospective and carried out in a single center; there- fore, we think that our results should be supported by prospective stud- ies that will be conducted with a larger patient population.

Authors’ contribution

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by SK and HBC. The first draft of the manuscript was written by SK and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Compliance with Ethical Standards

All procedures performed in the study were in accordance with the ethical standards of the institutional and/or national research commit- tee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Informed consent

This study has been designed retrospectively.

Funding

We have no funding resource for the study.

Declaration of Competing Interest

No conflict of interest.

References

1. Goralnick E, Meguerdichian DA. Gastrointestinal bleeding. In: Marx JA, Hockberger RS, Walls RM, et al, editors. Rosen’s Emergency Medicine: Concepts and Clinical Practice. 8th ed. Philadelphia, PA: Elsevier Saunders; 2014 chap 30.
2. Saltzman JR, Tabak YP, Hyett BH, Sun X, Travis AC, Johannes RS. A simple risk score accurately predicts in-hospital mortality, length of stay, and cost in acute upper GI bleeding. Gastrointest Endosc. 2011 Dec;74(6):1215-24.
3. Rassameehiran S, Teerakanok J, Suchartlikitwong S, Nugent K. Utility of the shock index for risk stratification in patients with Acute upper gastrointestinal bleeding. South Med J. 2017 Nov;110(11):738-43.
4. Otero R, Nguyen B. Soktaki hastaya yaklasim. Judith Tintinalli, et al. [cev.] Cem Ertan. Acil tip Kapsamli Bir Calisma klavuzu. Istanbul: Nobel Tip Kitapevi. 2013:65-172.
5. Singh A, Ali S, Agarwal A, Srivastava RN. Correlation of shock index and modified shock index with the outcome of adult trauma patients: a prospective study of 9860 patients. N Am J Med Sci. 2014 Sep;6(9):450-2.
6. Zarzaur BL, Croce MA, Fischer PE, Magnotti LJ, Fabian TC. New vitals after injury: shock index for the young and age x shock index for the old. J Surg Res. 2008 Jun 15;147(2):229-36.
7. Sankaran P, Kamath AV, Tariq SM, Ruffell H, Smith AC, Prentice P, et al. Are shock index and adjusted shock index useful in predicting mortality and length of stay in community-acquired pneumonia? Eur J Intern Med. 2011 Jun;22(3):282-5.
8. Semerci E, Durukan P, Yildirim S, Baykan N, Yakar S, Ipekten F. Gastrointestinal sistem kanamali hastalarda sok indeksi vehematokrit duzeylerinin mortalite uzerine etkisi. Akad Gastroenterol Dergisi. 2018;17(2):85-9.
9. Mutschler M, Nienaber U, Munzberg M, Wolfl C, Schoechl H, Paffrath T, et al. The shock index revisited – a fast guide to transfusion requirement? A retrospective analysis on 21,853 patients derived from the TraumaRegister DGU. Crit Care. 2013 Aug 12;17(4):R172.
10. Saffouri E, Blackwell C, Laursen SB, Laine L, Dalton HR, Ngu J, et al. The shock index is not accurate at predicting outcomes in patients with upper gastrointestinal bleed- ing. Aliment Pharmacol Ther. 2020 Jan;51(2):253-60.
11. Nakasone Y, Ikeda O, Yamashita Y, Kudoh K, Shigematsu Y, Harada K. Shock index correlates with extravasation on angiographs of gastrointestinal hemorrhage: a lo- gistics regression analysis. Cardiovasc Intervent Radiol. 2007 Sep-Oct;30(5):861-5.
12. Horibe M, Kaneko T, Yokogawa N, Yokota T, Okawa O, Nakatani Y, et al. A simple scoring system to assess the need for an endoscopic intervention in suspected upper gastrointestinal bleeding: a prospective cohort study. Dig Liver Dis. 2016 Oct;48(10):1180-6.
13. Jung DH, Ko BS, Kim YJ, Kim WY. Comparison of risk scores and shock index in he- modynamically stable patients presenting to the emergency department with nonvariceal upper gastrointestinal bleeding. Eur J Gastroenterol Hepatol. 2019 Jul; 31(7):781-5.
14. Robertson M, Majumdar A, Boyapati R, Chung W, Worland T, Terbah R, et al. Risk stratification in acute upper GI bleeding: comparison of the AIMS65 score with the Glasgow-Blatchford and Rockall scoring systems. Gastrointest Endosc. 2016 Jun;83 (6):1151-60.
15. Lee K, Jang JS, Kim J, Suh YJ. Age shock index, shock index, and modified shock index for predicting postintubation hypotension in the emergency department. Am J Emerg Med. 2020 May;38(5):911-5.
16. Kim SY, Hong KJ, Shin SD, Ro YS, Ahn KO, Kim YJ, et al. Validation of the shock index, modified shock index, and age shock index for predicting mortality of geriatric trauma patients in emergency departments. J Korean Med Sci. 2016 Dec;31(12):2026-32.
17. Torabi M, Moeinaddini S, Mirafzal A, Rastegari A, Sadeghkhani N. Shock index, mod- ified shock index, and age shock index for prediction of mortality in emergency se- verity index level 3. Am J Emerg Med. 2016 Nov;34(11):2079-83.
18. Yu T, Tian C, Song J, He D, Sun Z, Sun Z. Age shock index is superior to shock index and modified shock index for predicting long-term prognosis in acute myocardial infarction. Shock. 2017 Nov;48(5):545-50.