You are here

  • Home
  • Archive
  • Volume 34, Issue 2
  • Clinical metrics in emergency medicine: the shock index and the probability of hospital admission and inpatient mortality

Article Text

Article menu

Download PDFPDF

Original article
Clinical metrics in emergency medicine: the shock index and the probability of hospital admission and inpatient mortality
  1. Kamna S Balhara1,
  2. Yu-Hsiang Hsieh2,
  3. Bachar Hamade2,
  4. Ryan Circh3,
  5. Gabor D Kelen2,
  6. Jamil D Bayram2
  1. 1Department of Emergency Medicine, University of Texas Health Science Center at San Antonio, San Antonio, Texas
  2. 2Department of Emergency Medicine, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA
  3. 3Department of Emergency Medicine, Baltimore Washington Medical Center, University of Maryland, Glen Burnie, Maryland, USA
  1. Correspondence to Dr Jamil D Bayram, Department of Emergency Medicine, Johns Hopkins University School of Medicine, 5801 Smith Ave. Davis Building, Suite 3220, Baltimore, MD 21209, USA; jbayram1{at}jhmi.edu

Abstract

Study objectives The shock index (SI), defined as the ratio of HR to systolic BP, has been studied as an alternative prognostic tool to traditional vital signs in specific disease states and subgroups of patients. However, literature regarding its utility in the general ED population is lacking. Our main objective was to determine the probability of admission and inpatient mortality based on the first measured SI at initial presentation in the general adult ED population in our tertiary care centre.

Methods A retrospective chart review of all adult patients (≥18 years old) presenting to the ED at our tertiary care centre over a 12-month period was conducted. Likelihood ratios (LRs) were calculated in order to determine the optimal SI cut-off for predicting hospital admission and inpatient mortality.

Results We reviewed 58 336 ED patient encounters occurring between 1 October 2012 and 30 September 2013. SI >1.2 was associated with a large increase in the likelihood of hospital admission, with a positive LR (+LR) of 11.69 (95% CI 9.50 to 14.39) and a moderate increase in the likelihood of inpatient mortality with a +LR of 5.82 (95% CI 4.31 to 7.85). SI >0.7 and >0.9, the traditional ‘normal’ cut-offs cited in the literature, were only associated with minimal to small increases in the likelihood of admission and inpatient mortality.

Conclusions In our single-centre study, the initial SI recorded in the ED shows promise as a clinical metric in the general adult ED population, increasing the probability of both hospital admission and inpatient mortality, specifically at a threshold of SI >1.2.

  • emergency department
  • acute care
  • clinical assessment

https://doi.org/10.1136/emermed-2015-205532

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Key messages

    What is already known on this subject?

    • The shock index has shown to be a valuable prognostic tool for specific patient populations.

    • There are no large studies on the shock index in the general ED population.

    What this study adds?

    • The traditional shock index values described in the literature may not be applicable in the general ED population.

    • A shock index value of >1.2 may predict hospital admission and inpatient mortality in the general ED population.

    • This can act to supplement the existing triage system, to allow faster resource allocation and more rapid patient disposition.

    Introduction

    In the initial evaluation of ED patients, systolic BP (SBP) and HR are used by healthcare providers to assess patients' haemodynamic status. These vital signs, however, have been shown to be normal in some critically ill patients, leading to a false sense of clinical security and potentially delaying immediate treatment and disposition to a high level of care.1 ,2 The shock index (SI), defined as HR divided by SBP, was first described in 1967 by Allgower and Burri.3 Historically, the normal range for healthy adults has been proposed to fall between 0.5 and 0.7, and up to 0.9 in some studies.3–8 The SI is postulated to have an inverse relationship to the left ventricular stroke volume and cardiac output, which have important clinical implications in assessing cardiac function.4 ,9 The SI has been mainly studied in trauma patients, with elevations in SI mostly seen in the setting of acute hypovolaemia and circulatory failure.10 Abnormally elevated SI is often associated with mortality and increased inpatient resource consumption, and the SI has been posited to be a valuable guide in diagnosing early hypovolaemia despite normal HR and SBP in specific patient populations.4–6 ,9 ,11 As such, it may represent an important diagnostic marker that could denote critical illness well before abnormalities in conventional vital signs are noted.

    In the general ED population, the SI may have a role in the early prediction of the likelihood for hospital admission and inpatient mortality, which could be very valuable in guiding accurate initial triage, resource allocation and decision-making regarding disposition and level of care in the ED. However, there is to date very limited literature that has attempted to either assess the utility of the SI in the general ED population or establish a threshold specific to the diverse population seen in the ED.

    Our main objective was to determine the probability of admission and inpatient mortality based on the first measured SI at initial presentation in the general adult ED population at a single large academic centre.

    Methods

    Study design and setting

    After obtaining Institutional Review Board approval, a retrospective chart review of all ED adult patients who presented over a 12-month period from 1 October 2012 to 30 September 2013 was conducted at our academic tertiary care centre. All patients aged 18 years or older presenting to the ED during our study period were included. We excluded patients who arrived in cardiopulmonary arrest, left before triage, left without being seen, were screened and then left, left against medical advice or whose first set of vital signs (SBP and HR) was incomplete or unrecorded. Using these criteria, a total of 58 633 charts were identified from the electronic medical record, HealthMatics ED (HMED) (Allscripts, Chicago, Illinois, USA). For each of these charts, the following electronically recorded data were exported to Microsoft Excel (Microsoft, Redmond, Washington, USA: 2003): first HR recorded at presentation, first BP recorded at presentation, triage acuity (based upon the Emergency Severity Index (ESI) with 1 being most urgent and 5 least urgent), age in years, gender, ethnicity (black, white or other), disposition (discharge from ED vs admission to hospital) and, for admitted patients, admission destination (critical care floor vs general floor) and inpatient mortality of any cause. The measurement of the initial SI was dependent on the accuracy of the vital signs obtained by medical staff. Of these 58 633 charts, the first set of vital signs of 1723 encounters had been recorded manually on paper charts and then scanned into the electronic medical record at the time of the visit. Of note, even for patients whose first vital signs were on paper charting, all other data points had been electronically recorded in the electronic medical record and were thus electronically abstracted. These 1723 charts were distributed among three abstractors who adhered to data abstraction best practices previously described in the literature.12 The data abstractors reviewed the scanned paper charts to manually record initial HR and BP in the standardised abstraction form using an Excel spreadsheet. The data abstraction team consisted of two second-year medical students and a second-year emergency medicine resident. The two medical students took online courses in how to use HMED and the resident was familiar with HMED from clinical use prior to data abstraction. The team was trained prior to initiation of abstraction by a study team member using a set of ‘practice’ medical records. Abstractors were aware of the study hypothesis but did not calculate the SI for any manually abstracted charts. During the period of data abstraction, which lasted approximately 1 month, study team members communicated frequently with each abstractor to monitor data abstraction and respond to any queries. After completion of data abstraction, a study team member reabstracted a random sample of 30 charts from each abstractor to ensure that the initial vital signs recorded on scanned paper charts had been accurately identified and recorded in the standardised abstraction form. No errors or deviation from abstraction protocol were noted. After completion of abstraction from paper charts and final review of electronically abstracted data by two study team members, a further 297 charts were excluded if initial vital signs were incompletely recorded on paper charts or if the chart belonged to an admitted patient who had subsequently left the hospital against medical advice after admission or whose outcome after admission was not recorded.

    Outcome measures and data analysis

    The two outcome measures were admission to inpatient care services and inpatient mortality. We chose admission to inpatient care services as an outcome because patient disposition is a key decision point in practice of emergency medicine and admission to the hospital could serve as a surrogate marker for increased patient acuity and increased use of resources. Admission to specific units, such as intensive care units (ICUs), was excluded as an outcome measure as it was not a uniform outcome—since criteria for admission and level of care provided differed by type of ICU—and because the transition between two electronic medical record systems during our study affected the coding needed to extract this particular data. Inpatient mortality, as opposed to 30-day mortality, was chosen as an outcome measure since the primary aim of our study is to establish the utility of the SI as it pertains to a one-time patient encounter in the ED and mortality during that visit itself is a more proximate outcome measure for SI at presentation. We opted not to use 30-day mortality as an outcome measure as 30-day mortality can be influenced by multiple confounders entirely unrelated to the ED visit for which the patient presented. We used the first recorded HR and SBPs to calculate the initial SI for each patient's encounter. SAS V.9.3 (SAS Institute, Cary, North Carolina, USA) was used for statistical analysis. We calculated the likelihood ratios (LRs), sensitivity and specificity for a broad range of thresholds of SI for each of the two outcomes of hospital admission and inpatient mortality. Positive LRs (+LRs) for each SI value for each of the two outcomes were also stratified by race, gender and age. The SI threshold of 0.5–0.7 was tested as a unit because this range is often cited in the literature as being ‘normal’.3–8 +LRs were used to examine the likelihood of occurrence of each of the two outcomes for any given SI. +LRs were highlighted for their statistical robustness in measuring and expressing diagnostic accuracy, and being independent of pretest probability. +LR values of >5 and 10 were considered to indicate moderate and large increases in the likelihood of the outcomes, respectively.13 Sensitivity and specificity values were reported for conventional purposes.

    Results

    Characteristics of study subjects

    We reviewed 58 336 discrete ED patient encounters that met the criteria during the study period. Most patients fell in the 45–54-year-old range (21.2%), 52.1% were female, 74.0% of patients were discharged and 6.7% were triaged as ESI level 1, while 60.0% were triaged as level 3 (table 1).

    View this table:
    Table 1

    Demographics, disposition and acuity level of 58 336 patient encounters

    Main results

    SI values between 0.5 and 0.7 had the lowest likelihood of hospital admission with +LR of 0.74 (95% CI 0.73 to 0.76), negative LR (−LR) of 1.27 (95% CI 1.25 to 1.28), sensitivity of 38% (95% CI 37% to 38%) and specificity of 49% (95% CI 49% to 50%). Similarly, SI values between 0.5 and 0.7 had the lowest likelihood of inpatient mortality with +LR of 0.58 (95% CI 0.46 to 0.74), −LR of 1.25 (95% CI 1.17 to 1.34), sensitivity of 22% (95% CI 17% to 28%) and specificity of 62% (95% CI 61% to 63%).

    With increasing values of SI >0.5–0.7, we noted increasing proportions of patients who were admitted to the hospital, and who died during hospitalisation (figure 1). As SI increased, +LRs, relative risks and specificity all increased for both outcomes, while the sensitivities decreased (tables 2 and 3).

    View this table:
    Table 2

    Admission by the shock index (SI)

    View this table:
    Table 3

    Inpatient mortality by the shock index (SI)

    Figure 1
    Figure 1

    Rate of admission and mortality by the shock index (SI) value.

    In our study population, an SI threshold of >1.0 demonstrated a moderate increase in likelihood of admission to the hospital with +LR of 5.63 (95% 5.15 to 6.16), and a large increase in that likelihood with +LR of 11.69 (95% 9.50 to 14.39) when SI >1.2—an SI of >1.2 demonstrated a moderate increase in likelihood of inpatient mortality with +LR of 5.82 (95% CI 4.31 to 7.85) (tables 2 and 3). The one value of SI above which the +LR for both outcomes showed at least a moderate increase (ie, +LR>5) was 1.2 (tables 2 and 3).

    For patients with SI >0.7, the +LRs for admission and inpatient mortality were 1.40 (95% CI 1.37 to 1.43) and 1.49 (95% CI 1.36 to 1.63), respectively. For patients with SI >0.9, the +LRs for admission and inpatient mortality were 3.34 (95% CI 3.16 to 3.53) and 2.58 (95% CI 2.21 to 3.01), respectively (tables 2 and 3).

    +LR for admission and inpatient mortality for each SI value stratified by race (black or non-black), gender (male or female) and age (18–44 years, 45–64 years, 65 years or older) are presented in tables 4 and 5, respectively.

    View this table:
    Table 4

    Admission by the shock index (SI), stratified by race, gender and age

    View this table:
    Table 5

    Inpatient mortality by the shock index (SI), stratified by race, gender and age

    In regards to the admission outcome, SI >1.2 yielded +LR >5 for every patient group. At SI >1.2, +LR for admission was higher for black patients (13.49) compared with non-black patients (8.68), was similar for male and female patients (11.79 and 11.57 respectively) and was lowest for those aged 65 years or older (7.25) compared with other age groups.

    In regards to the inpatient mortality outcome, increasing SI values generally yielded increasing +LR in all subgroups. For SI >1.2, +LR for mortality was similar for black and non-black patients (5.78 and 5.83), was higher in male patients compared with female patients (7.38 vs 3.99) and was highest for patients aged 18–44 years old (10.36) compared with other age groups.

    Discussion

    The SI has been studied as a prognostic metric in patients with ruptured ectopic pregnancy, myocardial infarction, pneumonia, pulmonary embolism, gastrointestinal haemorrhage, sepsis and trauma.10 ,14–24 Birkhahn et al15 ,16 reported that normal vital signs alone were poor predictors of ruptured ectopic pregnancy, and an SI >0.85 made the diagnosis of ruptured ectopic pregnancy 15 times more likely. Their findings were supported by Onah et al,17 who reported that an SI ≥0.9 had an OR of 4.5 (95% CI 1.8 to 11.6, p=0.002) for the diagnosis of ruptured ectopic pregnancy. In patients with ST-elevation myocardial infarction, Huang et al18 reported that an SI ≥0.7 at admission was associated with 1.6-fold increased risk of 7-day major adverse cardiovascular events (hazard ratio 1.63, 95% CI 1.36 to 1.95). In adult ED patients with suspected sepsis, Berger et al23 showed that SI ≥0.7 was the most sensitive screening test for hyperlactemia and 28-day mortality, and SI ≥1.0 was the most specific predictor of both outcomes. Sloan et al reported that trauma patients with an SI ≥1.0, 1.4 and 1.8 at any time point were 2.3, 2.7 and 3.1 times, respectively, more likely to die within 28 days than were patients with SI values below these cut-offs (p<0.001). In the same study, and after 120 min of resuscitation, patients with an SI ≥1.0 were 3.9 times more likely to die within 28 days (40% vs 15%, p<0.001).24 A recent retrospective study of 20 000 adult patients from the trauma registry demonstrated that an SI of ≥1.4 was associated with worst injury severity score, largest fluid requirements, greatest transfusion requirements and highest proportion of patients transfused. In fact, more than half of the patients with SI ≥1.4 in this study needed at least 10 units of blood before going to the ICU.10

    The SI has demonstrated its utility across a large variety of pathophysiological states, suggesting that it may be especially useful in a diverse, undifferentiated setting such as the ED. However, in the general ED population, literature on the utility of the SI is very scarce. To our knowledge, the only prior study that evaluated the SI in the general adult ED population was conducted by Rady et al,4 published in 1994. This study of 275 ED patients concluded that SI >0.9 was associated with higher triage priority, immediate treatment, total admissions and ICU admissions. However, this study was conducted with a limited number of patients and compared groups of patients using a single traditional SI threshold.

    In our study population with 58 336 patient encounters examining a range of possible thresholds, an initial SI >1.2 at presentation to the ED was associated with +LRs of 11.69 and 5.82 for hospital admission and inpatient mortality, respectively, which indicates a large increase in the post-test probability of hospital admission and moderate increase in post-test probability of inpatient mortality. These findings, if validated, may have significant implications on clinical decision-making in the ED. In the appropriate clinical setting, an initial SI >1.2 at presentation may indicate the need for rapid allocation of critical care resources, quicker disposition to admission status and earlier consideration of the need for intensive care inpatient unit.

    Upon subgroup analysis, the +LR for admission for SI >1.2 remained >5 for all subsets of patients. The +LR for admission was lowest (7.25) in those aged 65 years or older compared with +LRs of 12.52 and 13.41 in 18–44 year olds and 45–64 year olds, respectively. While still indicating a moderate post-test probability of admission, the lower +LR and wide CI in this elderly age group may reflect the fact that the decision to admit these patients is often influenced by many factors other than severity of clinical illness or haemodynamic compromise, such as dementia, degree of disability, high-risk comorbidities and ability to care for themselves. SI >1.2 was especially relevant in predicting admission in black patients, while no major differences were noted by gender. Higher SI values had higher +LR for mortality across all subgroups. +LR for SI and inpatient mortality was lower at all values for female patients compared with male patients. +LR of the SI for inpatient mortality was highest in patients aged 18–44 years, perhaps indicative of the fact that the hardening of arterioles and use of medications such as β-blockers or calcium channel blockers in patients aged >45 years of age may generate misleading SI values that mask the true severity of illness. A retrospective study of >180 000 patients from the National Trauma Data Bank lends credence to this postulation; patients over age 55 in that study had lower HRs and higher SBP at admission, leading to a lower-appearing SI despite greater mortality compared with younger patients.25

    The often cited SI cut-offs of >0.7 and >0.9 were associated with minimal to small increases in the likelihood of admission and inpatient mortality in our general ED adult population. However, it is important to note that these thresholds have not always been specifically quoted for the same outcomes as in our study, and they have not been broadly cited in the general ED patient population. Our findings highlight the importance of establishing a threshold specific to all comers presenting to the ED.

    Limitations

    Our study did have certain limitations. It is a retrospective, single-site study examining a metric that as yet has no ‘gold standard’. As such, our study is a preliminary one, aiming to identify potential thresholds, rather than validate them. A portion of the data abstraction was conducted manually by abstractors who were aware of the study hypothesis, theoretically introducing a risk of bias. However, <3% of vital sign data was manually abstracted, so any effect of bias should be minimal. Our retrospective study design and resources did not allow us to have a 30-day mortality as an outcome, nor to follow-up on the outcomes of the 110 patients that were discharged home from the ED after an initial SI >1.2. Further research is needed on this subgroup of patients to evaluate their risk factors, comorbidities and ED therapeutics that might have affected their outcome. In addition, the nature of our data set did not allow us to parse out inpatient admissions that were discharged within 24 hours or patients who were admitted for social reasons. Our study was conducted in a tertiary, inner-city hospital, which may have contributed to higher prevalence of critical illness. While we looked at outcomes in patients with elevated SI values, our study had few distinct data points on the lower end of the SI spectrum, which limited our interpretation of results at the lower end of the spectrum. This study design looked only at the initial presenting SI because the objective was to use the first set of presenting HR and SBP as an early predictor of admission status. Additionally, we were logistically unable to exclude patients on β-blockers, whose SI at presentation may be different from what is physiologically expected. Finally, our outcome measure of inpatient mortality could have been affected by the fact that eventual inpatient mortality may not have been a direct consequence of the admission diagnosis from the ED.

    Conclusions

    A patient's vital signs are often the very first data points recorded upon visiting the ED; the SI is thus a metric that is immediately and easily available to the ED provider at no additional cost. Our study demonstrates that the SI has prognostic utility for outcomes and disposition for the general ED population, and suggests that a different threshold than those traditionally established may be needed for the diverse patients presenting to the ED. Our findings, if further validated, may potentially help in accurate acuity designation in triage by formally supplementing the existing triage system, may allow for more rapid resource allocation within minutes of patient arrival and may hasten and facilitate patient disposition to higher acuity units in the appropriate clinical setting. Our preliminary findings lay the groundwork for further studies that could aim to establish SI thresholds that can reproducibly function as important predictive clinical metrics for the general adult ED patient population and assess their impact on enhancing triage, time to disposition and patient length of stay.

    References

    1. ↵
      1. Shoemaker WC
      . Relation of oxygen transport patterns to the pathophysiology and therapy of shock states. Intensive Care Med 1987;13:230–43. doi:10.1007/BF00265111
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Shippy CR,
      2. Appel PL,
      3. Shoemaker WC
      . Reliability of clinical monitoring to assess blood volume in critically ill patients. Crit Care Med 1984;12:107–12. doi:10.1097/00003246-198402000-00005
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Allgöwer M,
      2. Burri C
      . [“Shock index”]. Dtsch Med Wochenschr 1967;92:1947–50. doi:10.1055/s-0028-1106070
      OpenUrlCrossRefPubMed
    4. ↵
      1. Rady MY,
      2. Smithline HA,
      3. Blake H, et al
      . A comparison of the shock index and conventional vital signs to identify acute, critical illness in the emergency department. Ann Emerg Med 1994;24:685–90. doi:10.1016/S0196-0644(94)70279-9
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Vandromme MJ,
      2. Griffin RL,
      3. Kerby JD, et al
      . Identifying risk for massive transfusion in the relatively normotensive patient: utility of the prehospital shock index. J Trauma 2011;70:384–8; discussion 388–390. doi:10.1097/TA.0b013e3182095a0a
      OpenUrlCrossRefPubMed
    6. ↵
      1. McNab A,
      2. Burns B,
      3. Bhullar I, et al
      . An analysis of shock index as a correlate for outcomes in trauma by age group. Surgery 2013;154:384–7. doi:10.1016/j.surg.2013.05.007
      OpenUrl
    7. ↵
      1. McNab A,
      2. Burns B,
      3. Bhullar I, et al
      . A prehospital shock index for trauma correlates with measures of hospital resource use and mortality. Surgery 2012;152:473–6. doi:10.1016/j.surg.2012.07.010
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Singh A,
      2. Ali S,
      3. Agarwal A, et al
      . 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;6:450–2. doi:10.4103/1947-2714.141632
      OpenUrl
    9. ↵
      1. Rady MY,
      2. Nightingale P,
      3. Little RA, et al
      . Shock index: a re-evaluation in acute circulatory failure. Resuscitation 1992;23:227–34. doi:10.1016/0300-9572(92)90006-X
      OpenUrlCrossRefPubMed
    10. ↵
      1. Mutschler M,
      2. Nienaber U,
      3. Münzberg M, et al
      . The Shock Index revisited—a fast guide to transfusion requirement? A retrospective analysis on 21,853 patients derived from the TraumaRegister DGU. Critical Care 2013;17:R172. doi:10.1186/cc12851
      OpenUrlCrossRefPubMed
    11. ↵
      1. Rady MY,
      2. Rivers EP,
      3. Martin GB, et al
      . Continuous central venous oximetry and shock index in the emergency department: use in the evaluation of clinical shock. Am J Emerg Med 1992;10:538–41. doi:10.1016/0735-6757(92)90178-Z
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Worster A,
      2. Bledsoe RD,
      3. Cleve P, et al
      . Reassessing the methods of medical record review studies in emergency medicine research. Ann Emerg Med 2005;45:448–51. doi:10.1016/j.annemergmed.2004.11.021
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Grimes DA,
      2. Schulz KF
      . Refining clinical diagnosis with likelihood ratios. Lancet 2005;365:1500–5. doi:10.1016/S0140-6736(05)66422-7
      OpenUrlCrossRefPubMedWeb of Science
    14. ↵
      1. Wira CR,
      2. Francis MW,
      3. Bhat S, et al
      . The shock index as a predictor of vasopressor use in emergency department patients with severe sepsis. West J Emerg Med 2014;15:60–6. doi:10.5811/westjem.2013.7.18472
      OpenUrl
    15. ↵
      1. Birkhahn RH,
      2. Gaeta TJ,
      3. Bei R, et al
      . Shock index in the first trimester of pregnancy and its relationship to ruptured ectopic pregnancy. Acad Emerg Med 2002;9:115–19. doi:10.1111/j.1553-2712.2002.tb00226.x
      OpenUrlPubMed
    16. ↵
      1. Birkhahn RH,
      2. Gaeta TJ,
      3. Van Deusen SK, et al
      . The ability of traditional vital signs and shock index to identify ruptured ectopic pregnancy. Am J Obstet Gynecol 2003;189:1293–6. doi:10.1067/S0002-9378(03)00663-X
      OpenUrlPubMed
    17. ↵
      1. Onah HE,
      2. Oguanuo TC,
      3. Mgbor SO
      . An evaluation of the shock index in predicting ruptured ectopic pregnancy. J Obstet Gynaecol 2006;26:445–7. doi:10.1080/01443610600747314
      OpenUrlPubMed
    18. ↵
      1. Huang B,
      2. Yang Y,
      3. Zhu J, et al
      . Usefulness of the admission shock index for predicting short-term outcomes in patients with ST-segment elevation myocardial infarction. Am J Cardiol 2014;114:1315–21. doi:10.1016/j.amjcard.2014.07.062
      OpenUrlCrossRefPubMed
    19. ↵
      1. Nüllmann H,
      2. Pflug MA,
      3. Wesemann T, et al
      . External validation of the CURSI criteria (confusion, urea, respiratory rate and shock index) in adults hospitalised for community-acquired pneumonia. BMC Infect Dis 2014;14:39. doi:10.1186/1471-2334-14-39
      OpenUrl
    20. ↵
      1. Kilic T,
      2. Ermis H,
      3. Gülbas G, et al
      . Prognostic role of the simplified pulmonary embolism severity index and shock index in pulmonary embolism. Pol Arch Med Wewn 2014;124:678–87.
      OpenUrl
    21. ↵
      1. Sohn CH,
      2. Kim WY,
      3. Kim SR, et al
      . An increase in initial shock index is associated with the requirement for massive transfusion in emergency department patients with primary postpartum hemorrhage. Shock 2013;40:101–5. doi:10.1097/SHK.0b013e31829b1778
      OpenUrl
    22. ↵
      1. Nakasone Y,
      2. Ikeda O,
      3. Yamashita Y, et al
      . Shock index correlates with extravasation on angiographs of gastrointestinal hemorrhage: a logistics regression analysis. Cardiovasc Intervent Radiol 2007;30:861–5. doi:10.1007/s00270-007-9131-5
      OpenUrlCrossRefPubMedWeb of Science
    23. ↵
      1. Berger T,
      2. Green J,
      3. Horeczko T, et al
      . Shock index and early recognition of sepsis in the emergency department: pilot study. West J Emerg Med 2013;14:168–74. doi:10.5811/westjem.2012.8.11546
      OpenUrlCrossRefPubMed
    24. ↵
      1. Sloan EP,
      2. Koenigsberg M,
      3. Clark JM, et al
      . Shock Index and Prediction of Traumatic Hemorrhagic Shock 28-Day Mortality: data from the DCLHb Resuscitation Clinical Trials. West J Emerg Med 2014;15:795–802. doi:10.5811/westjem.2014.7.21304
      OpenUrl
    25. ↵
      1. Zarzaur BL,
      2. Croce MA,
      3. Magnotti LJ, et al
      . Identifying life-threatening shock in the older injured patient: an analysis of The National Trauma Data Bank. J Trauma 2010;68:1134–8. doi:10.1097/TA.0b013e3181d87488
      OpenUrlCrossRefPubMed

    Footnotes

    • Presented at the meeting for the Society of Academic Emergency Medicine, San Diego, 2015.

    • Contributors JDB conceived the study. KSB and JDB obtained Institutional Review Board approval. KSB and BH obtained and verified the data. JDB and KSB designed the study and Y-HH provided statistical analysis of the data. JDB, KSB, BH, and GDK interpreted the data analysis. KSB, BH, Y-HH, RC, GDK and JDB drafted various sections of the manuscript, and all authors contributed substantially to its revision. JDB takes responsibility for the paper as a whole.

    • Competing interests None declared.

    • Provenance and peer review Not commissioned; externally peer reviewed.