Since March 11, 2020, the World Health Organization has declared coronavirus disease 2019 (COVID-19) caused by severe acute respiratory coronavirus 2 (SARS-CoV-2) as an international pandemic and public health emergency. Reference Tian, Hu and Lou1 The pandemic has caused serious conditions such as mortality and morbidity all over the world. Due to the great number of people harmed by the disease, overcapacity has occurred in hospitals and emergency departments (EDs). For this reason, there were insufficiencies in the provision of health care. Reference Assiri, Al-Tawfiq and Al-Rabeeah2,Reference Lu, Yu and Wang3 Therefore, it is necessary to quickly identify patients who may require critical care. Reference Remuzzi and Remuzzi4
Shock index (SI) first performed by Allgöwer and Buri in 1967 and used to ascertain the hypovolemia degree in hemorrhagic shock. Reference Allgöwer and Buri5 SI, which is described as the division of heart rate over systolic blood pressure, is a fast, noninvasive method. The value of SI is normally between 0.5 and 0.7 in healthy adults. In the literature, some studies reporting that high SI values predict mortality risk, furthermore critical care requirements. Reference King, Plewa and Buderer6–Reference Talmor, Jones and Rubinson8 In a study in which higher than 0.9 value of SI was considered as a severity indicator in triage, it was observed that hospitalization and intensive care requirements were high in patients with high SI value. Reference Rady, Smithline and Blake9 This suggests that SI can be a valuable parameter for early diagnosis and evaluation of critical illnesses in the ED, and can also be used as a prognostic predictor. This study aims to ascertain whether prehospital SI could be used to predict the requirement for intensive care unit (ICU) of these patients and their mortality status in patients visited to the hospital by ambulance in terms of COVID-19.
Methods
Study Design
This observational study was conducted in the ED of Kartal Dr. Lütfi Kırdar City Hospital, retrospectively, between November 1, 2020, and February 1, 2021. The institutional review board confirmed the analysis and declared a waiver of approval (Date:24.02.2021, Ethics Committee Ruling number: 514/196/25). Ambulance care in the Turkey is provided by national Emergency Medical Services (EMS). Patient transported by ambulance has been standardized appropriately to the “COVID-19 Diagnosis and Treatment Guide” published in Turkish by Ministry of Health. 10
Selection of Patients
The consecutive patients whose age of ≥18 y and transported to the hospital by ambulance with a suspected or confirmed COVID-19 in the prehospital setting were included in the study. Exclusion criteria for the study were: atrial fibrillation that interfered with blood pressure measurement, Reference Yu, Tian and Song11 patients transferred from other hospitals, prehospital cardiac arrest patients, patients whose reverse transcriptase polymerase chain reaction (RT-PCR) test negative, Reference Seyhan, Doğanay and Yılmaz12 patients who had deficiency in prehospital vital signs and patients who could not be followed-up.
Data Collection and Variables
The data, that consists of this study were obtained from the hospital records and the forms, which were filled in ambulance. These are covered age, sex, prehospital vital signs (body temperature [Temp], heart rate [HR], systolic blood pressure [SBP], diastolic blood pressure [DBP], respiratory rate [RR], blood oxygen saturation [spO2], body temperature [Temp]) and SI. The SI was assessed as the rate of HR to SBP (SI = HR/SBP). Reference Allgöwer and Buri5
Outcome Measures
The SI at first presentation was calculated in prehospital setting. If vital signs were measured more than once in the prehospital phase, we used the lowest values of shock index. The primary outcome is to ascertain the association between prehospital SI and 30-d mortality rate. For 30-d mortality outcome, grouping was done as survivor and nonsurvivor. The determination of the relationship between prehospital SI and the way patients leave the ED was the secondary outcome in this study. For the way of leaving the ED, grouping was made as outpatient (OP), inpatient unit (IU), and ICU.
Statistical Analysis
IBM SPSS Statistics 26 (SPSS) and MedCalc Version 19 software were used for statistical analysis. While the characteristics of the study population were reported, sustained data were signified as median and interquartile ranges (25th-75th), and certain data as frequency and percentage. Kolmogorov-Smirnov test was performed to examine the reason of distributions for age, vital parameters, and SI variables. Normality assumption was rejected for both outcome groups. Because the continuous variables did not conform to the normal distribution, the analysis was continued with the Mann-Whitney U- and Kruskal-Wallis H-tests (Table 1). The results were reported as median and interquartile ranges (Table 1).
Table 1. Descriptive statistics for age, vital parameters, and SI values of the groups
Abbreviations: OP, Outpatient; IU, Inpatient Unit; ICU, Intensive Care Unit; SBP, Systolic blood pressure; DBP, Diastolic blood pressure; HR, Heart rate; RR, Respiratory rate; spO2, Blood oxygen saturation; SI, Shock index; Temp, Body temperature.
* Fisher-Freeman-Halton exact test.
The statistically significant difference between the groups for the SI value made it necessary to determine which threshold value is more useful. Chi-squared and Fisher-Freeman-Halton tests were performed to analyze the statistical significance of SI thresholds of 0.7, 0.8, 0.9, and 1.0 for both endpoints (Table 2). Reference Freeman and Halton13 Statistically, significant differentiation was perceived between the groups for all 4 threshold values. The ROC analysis was made to compare the accuracy of these 4 threshold values; sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated (Table 3). To compare the estimation accuracy, both the area under the curve (AUC) and the Youden J index were measured (Table 3). Reference Zhou14 DeLong et al. method was used while performing ROC analysis. Reference DeLong, DeLong and Clarke-Pearson15 The mortality and ICU requirement prediction performances of different threshold values were demonstrated by decision curve analysis. A P value of less than 0.05 was accepted as statistically significant.
Table 2. SI thresholds and sex descriptives of the study population
Abbreviations: Sig, significance; op, outpatient; IU, inpatient unit; ICU, intensive care unit; SI, shock index.
* Fisher-Freeman-Halton exact test.
Table 3. Prognostic accuracy of different SI thresholds by ROC analysis
Abbreviations: \ SI, shock index; Sens., sensitivity; Spec., specificity; PPV, positive predictive value; NPV, negative predictive value; AUC, area under the curve; p (AUC), statistical significance for AUC; YJI, Youden’s J index.
* 95% Confidence interval.
Results
This study was conducted with data from 364 patients after the inclusion and exclusion criteria were applied. The survivor group was including a total of 254 patients, 110 in the nonsurvivor group, 131 in the OP group, 158 in the IU group, and 75 in the ICU group. In the study population, 196 patients were men and 168 patients were women. There was no significant difference according to both 30-d mortality and ED outcomes in terms of sex (Table 1).
The median age in the study population was 69 y (range, 55-80 y). A significant difference was observed for 30-d mortality and ED outcomes in terms of patient age. The median age of the nonsurvivor group was significantly higher than the survivor group. In ED outcome groups, median age was revealed as OP, IU, and ICU from younger to older (Table 1).
A significant difference among the groups for prehospital vital parameters (SBP, DBP, HR, RR, spO2), and SI in terms of both 30-d mortality and ED outcomes, while there was no significant difference for Temp (Tables 1 and 2).
AUCs of 0.7, 0.8, 0.9, and 1.0 threshold values of prehospital SI for ED outcome were calculated as 0.746, 0.769, 0.836, and 0.712, respectively, with the ROC analysis. AUC values for 30-d mortality outcome were calculated as 0.672, 0.674, 0.755, and 0.626, respectively (Table 3; Figure 1).
Figure 1. ROC curves of SI thresholds in mortality prediction.
Pairwise comparisons of SI thresholds for 30-d mortality, there was no statistically significance between 0.7 with 0.8 (P = 0.944), 0.7 with 1.0 (P = 0.103) and 0.8 with 1.0 (P = 0.075). There were statistically significant difference between the 0.7 with 0.9 (P < 0.001), 0.8 with 0.9 (P < 0.001), and 0.9 with 1.0 (P < 0.001) thresholds (Table 3).
In the pairwise comparisons of SI thresholds for ICU requirement, there was no statistical significance between 0.7 with 0.8 (P = 0.291), 0.7 with 1.0 (P = 0.302), and 0.8 with 1.0 (P = 0.069) thresholds while there were statistically significant difference between the 0.7 (P < 0.001) with 0.9, 0.8 (P < 0.001) with 0.9, and 0.9 with 1.0 (P < 0.001) thresholds (Table 3; Figure 2).
Figure 2. ROC curves of SI thresholds in ICU requirement prediction.
YJI values of 0.7, 0.8, 0.9, and 1.0 threshold values of SI in terms of ED outcome were calculated as 0.4919, 0.5379, 0.6715, 0.4246, respectively. YJI values for 30-d mortality outcome were calculated as 0.3442, 0.3472, 0.5104, and 0.2513, respectively (Table 3).
Discussion
The COVID-19 outbreak is putting unprecedented pressure on health-care systems to examine an unexpected number of patients. The triage of COVID-19 patients and the corresponding use of health-care resources is a continuous state of study. This study has been revealed that SI can predict the ICU requirements of patients and their mortality risks. Patients whose prehospital SI higher than 0.9 were more presumably to admission to the ICU. Similarly, patients with prehospital SI higher than 0.9 had a higher mortality rate.
Vital signs are objective indicators that can be obtained in prehospital triage. SI can be easily calculated using vital signs and can be used in prehospital triage where there is a limited time. While it has been used in patients with hemorrhagic shock or trauma, Reference Allgöwer and Buri5 it has recently been used as a prognostic predictor in critically ill patients and patients with sepsis. Reference Tseng and Nugent16–Reference Torabi, Mirafzal and Rastegari18 In a study conducted for the early diagnosis of sepsis, the negative predictive value of a shock index above 0.7 was found to be similar to systemic inflammatory response syndrome (SIRS). Reference Berger, Green and Horeczko19 In the same study, it was shown that a SI of higher than 1 is a specific predictor of both hyperlactatemia and 28-d mortality. Kenzaka et al. conducted a study in 206 patients who were visited to the ED due to sepsis. They reported that there was a significant relationship between the development of organ failure and SI. Reference Kenzaka, Okayama and Kuroki20
There is no consensus on the appropriate threshold value of SI (0.8, 0.9, or 1) that should be used clinically. Reference El-Menyar, Goyal and Tilley21–Reference Wang, Bae and Park23 In a study by McNab et al., the relationship between prehospital SI for trauma and hospital resource use and mortality risk was investigated. In the study, 19.7% of total patients (16.269) had a SI value higher than 0.9. The relative mortality risk was reported as 1.5% in the main area and 1.7% in the trauma center of those with prehospital SI higher than 0.9. This group also pointed an association between prehospital SI and ICU time of stay and blood product use. Reference McNab, Burns and Bhullar22 In a study, Jehan et al. examined the relationship between prehospital SI with mortality and transfusion requirement for patients with trauma. They reported an increase of 7.1% in mortality risk when prehospital SI was higher than 1. Reference Jehan, Con and McIntyre24 Doğanay et al. emphasized that a SI value higher than 0.93 may be a predictor for mortality risk in COVID-19 pneumonia patients. Reference Doğanay, Elkonca and Seyhan25 This study has been shown that a prehospital SI higher than 0.9 may be a predictor for determining mortality risk and ICU requirements.
This study has some limitations. Patients’ data were limited to retrospective and medical records. Therefore, our results in a limited number of patients are not able to reflect in general. Repeatedly, it should be known that the existence of fundamental comorbidities such as hypertension, diabetes mellitus, and coronary artery disease may suppress the predictive value of the SI. Reference Rau, Wu and Kuo26
Conclusion
Early triage of COVID-19 patients will ensure efficient use of health-care resources. Prehospital SI might be a useful, fast, and reliable triage parameter for predicting mortality risk and ICU requirements of adult COVID-19 patients. It was concluded that the threshold value of 0.9 for the shock index in our study population was superior to other threshold values in predicting the prognosis of COVID-19 patients.
Author Contributions
R.Ak.: conceptualization (lead), data curation (lead), formal analysis (lead), funding acquisition (equal), investigation (equal), and methodology (equal); F.D.: conceptualization (equal), data curation (equal), formal analysis (equal), funding acquisition (equal), investigation (equal), methodology (equal), project administration (equal), resources (equal), software (equal), supervision (equal), validation (equal), visualization (equal), writing-original draft (equal), and writing-review and editing (equal)
Funding statement
The author(s) received no financial support for the research, authorship, and/or publication of this article
Conflict(s) of interest
Authors declare that they have no conflicts of interest.
