0
Original Research: CRITICAL CARE |

Predicting Cardiac Arrest on the WardsIn-hospital Cardiac Arrest Prediction: A Nested Case-Control Study FREE TO VIEW

Matthew M. Churpek, MD, MPH; Trevor C. Yuen, BA; Michael T. Huber, BA; Seo Young Park, PhD; Jesse B. Hall, MD, FCCP; Dana P. Edelson, MD
Author and Funding Information

From the Section of Pulmonary and Critical Care (Drs Churpek and Hall), Section of Hospital Medicine (Mr Yuen and Dr Edelson), Pritzker School of Medicine (Mr Huber), and Department of Health Studies (Dr Park), University of Chicago, Chicago, IL.

Correspondence to: Dana Edelson, MD, Section of Hospital Medicine, University of Chicago, 5841 S Maryland Ave, MC 5000, Chicago, IL 60637; e-mail: dperes@uchicago.edu


Funding/Support: Dr Edelson is supported by a career development award from the National Heart, Lung, and Blood Institute [K23HL097157-01]. Dr Churpek is supported by a National Institutes of Health grant [T32HL07605].

Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (http://www.chestpubs.org/site/misc/reprints.xhtml).


© 2012 American College of Chest Physicians


Chest. 2012;141(5):1170-1176. doi:10.1378/chest.11-1301
Text Size: A A A
Published online

Background:  Current rapid response team activation criteria were not statistically derived using ward vital signs, and the best vital sign predictors of cardiac arrest (CA) have not been determined. In addition, it is unknown when vital signs begin to accurately detect this event prior to CA.

Methods:  We conducted a nested case-control study of 88 patients experiencing CA on the wards of a university hospital between November 2008 and January 2011, matched 1:4 to 352 control subjects residing on the same ward at the same time as the case CA. Vital signs and Modified Early Warning Scores (MEWS) were compared on admission and during the 48 h preceding CA.

Results:  Case patients were older (64 ± 16 years vs 58 ± 18 years; P = .002) and more likely to have had a prior ICU admission than control subjects (41% vs 24%; P = .001), but had similar admission MEWS (2.2 ± 1.3 vs 2.0 ± 1.3; P = .28). In the 48 h preceding CA, maximum MEWS was the best predictor (area under the receiver operating characteristic curve [AUC] 0.77; 95% CI, 0.71-0.82), followed by maximum respiratory rate (AUC 0.72; 95% CI, 0.65-0.78), maximum heart rate (AUC 0.68; 95% CI, 0.61-0.74), maximum pulse pressure index (AUC 0.61; 95% CI, 0.54-0.68), and minimum diastolic BP (AUC 0.60; 95% CI, 0.53-0.67). By 48 h prior to CA, the MEWS was higher in cases (P = .005), with increasing disparity leading up to the event.

Conclusions:  The MEWS was significantly different between patients experiencing CA and control patients by 48 h prior to the event, but includes poor predictors of CA such as temperature and omits significant predictors such as diastolic BP and pulse pressure index.

Figures in this Article

More than 150,000 adult in-hospital cardiac arrests (CAs) occur in the United States each year, with up to half occurring outside of intensive care settings.1,2 Mortality for these patients is > 80%, but there is evidence to suggest that some of these deaths may be preventable.1,3 Rapid response teams (RRTs) were created to decrease adverse outcomes by bringing critical care resources to deteriorating patients on the hospital wards. However, results have been mixed, with some studies demonstrating benefit but others showing no improvement in mortality and hospital-wide CA rates.46

A major cited weakness of the RRT is the failure to identify patients in sufficient time to intervene prior to CA.79 For example, in the Medical Emergency Response Intervention Trial (MERIT), the only multicenter randomized trial of RRTs, most patients were identified < 15 min before CA, death, or ICU transfer.6,7 Furthermore, the criteria used to activate the intervention had a sensitivity < 50%.6,7 Currently, there are > 50 different published criteria designed to detect physiologic decline and activate the RRT, and each use varying vital sign combinations.10,11 However, these criteria, including the frequently described Modified Early Warning Score (MEWS) (Fig 1),1214 are based on expert opinion rather than being derived using ward vital signs, and most have not been scientifically validated.

Figure Jump LinkFigure 1. Modified Early Warning Score (MEWS). Unresp = Unresponsive.Grahic Jump Location

It is unknown whether further optimization of vital sign selection and thresholds would improve rapid response systems or whether vital sign changes occur too late in the course of physiologic deterioration to alter outcomes. To address these gaps in knowledge, we investigated the accuracy of the MEWS and of vital signs routinely collected on the wards, including systolic and diastolic BP, pulse pressure, heart rate, respiratory rate, oxygen saturation, mental status, and temperature, for predicting CA on the wards. We also evaluated the MEWS at different time points prior to CA to determine when the vital signs of these patients begin to differ significantly from other patients on the ward.

Study Setting and Population

We conducted a retrospective nested case-control study at an academic, tertiary care hospital with approximately 500 inpatient beds grouped by clinical service. Our hospital has had an RRT in place since 2008 that is led by a critical care nurse and respiratory therapist with consultation from a hospitalist physician and/or pharmacist upon request. The RRT activation criteria include “tachypnea,” “tachycardia,” “hypotension,” and “staff worry,” but specific vital sign thresholds are not stated. This team is separate from the team that responds to a CA.

Cases were consecutive adult patients who experienced a CA, defined as the loss of a palpable pulse with attempted resuscitation, on the ward between November 1, 2008, and January 31, 2011. All patients residing on the same ward at the time (T0) of the case patient’s CA were eligible to be control subjects. Four control subjects were then selected for each case patient using a random number generator.

Case patients were identified using a prospectively collected and verified CA quality improvement database. Our process of data collection on patients who had a CA has been previously described.1517 Briefly, trained data abstractors are paged concurrently with the CA response team, and follow-up within 48 h of the event to download the transcripts for every resuscitation on our hospital wards and ICUs. Each transcript is then reviewed and cross-referenced with the medical record to verify that those classified as having a true CA did indeed have a loss of pulse with attempted resuscitation. Patients were excluded if they had no ward vital signs in the 48 h prior to T0. If a case patient had more than one CA on the ward during the hospitalization, only the first event was included.

The study protocol, consent, and data collection mechanisms were approved by the University of Chicago Institutional Review Board (no. 16995A). A waiver of consent was granted on the basis of minimal harm and general impracticability. Collection of patient information was designed to comply with the Health Insurance Portability and Accountability Act of 1996 (HIPAA) regulations.

Data Collection

Patient demographic data were obtained from administrative databases. Time and location stamped vital signs, including temperature, heart rate, BP, oxygen saturation, respiratory rate, and mental status, were obtained from the electronic medical record (EPIC) on admission and from 48 h prior to until 30 min before T0. Vital signs within 30 min of T0 were excluded because the goal was to predict CA early enough to potentially intervene. Pulse pressure (systolic BP − diastolic BP) and pulse pressure index (pulse pressure divided by systolic BP) were also calculated. For each vital sign, change from baseline was calculated by subtracting each patient’s closest value prior to 30 min before T0 by the admission value.

Mental status was collapsed from four drop-down menu fields in the electronic medical record (orientation, level of consciousness, motor response, and responsiveness) into one score (alert, responsive to voice, responsive to pain, and unresponsive [AVPU]).18 Because AVPU was not specifically available in our medical record, we recoded mental status as “alert” if “alert” was documented in either the level of consciousness or the orientation field. “Responsive to voice” was coded if “to voice” was documented in the responsiveness field or if “easy to arouse” was documented in the level of consciousness field. We recoded mental status as “responsive to pain” if “localizes” or “withdraws” was documented in the motor response field or if “to pain” was documented in the responsiveness field. Finally, “unresponsive” was coded if “comatose” or “does not arouse” was found in the level of consciousness field or if “unresponsive” was documented in the responsiveness field. For each time point, the patient’s best mental status was used for AVPU recoding.

A MEWS was calculated for each patient on admission using the first respiratory rate, heart rate, systolic BP, temperature, and mental status documented in the hospital regardless of location, and then every 8 h during the 48-h time period prior to T0 using the closest vital signs measured prior to that time point. The MEWS was used in this study because it is the most commonly cited multiparameter RRT activation criteria.1921 A normal value for mental status (ie, “alert”) was imputed for patients without any mental status documentation, similar to previous studies evaluating clinical scoring systems.22,23

Statistical Analysis

Baseline characteristics as well as admission and change from baseline vitals signs and MEWS were compared between case and control patients using χ2 tests for categorical data and t tests or Wilcoxon rank-sum tests for continuous data, as appropriate. The maximum and minimum recorded values of each vital sign and MEWS in the 48 h preceding T0 were compared by creating receiver operating characteristic (ROC) curves and then calculating the area under the ROC curve (AUC) using the trapezoidal rule. The sensitivities and specificities of different cutoff thresholds were investigated for predictors with an AUC of at least 0.6. CIs for the AUC, sensitivity, and specificity of vital signs and MEWS were calculated using bootstrapping, a nonparametric method that involved taking 1,000 samples of the data with replacement to obtain an empirical sampling distribution.24 All tests of significance used a 2-sided P < .05. Statistical analyses were completed using R (R Foundation for Statistical Computing) and Stata, version 11.1 (StataCorp).

Patient Characteristics

During the study period there were 55,121 hospital admissions, 436 index CAs (7.9 arrests per 1,000 admissions), and 462 RRT calls (8.4 calls per 1,000 admissions).

Eighty-nine patients experienced a CA on the ward during the study period. One patient had no ward vital signs prior to CA and was excluded. Eighty-eight case patients were matched to 352 control subjects. Patient demographic data are shown in Table 1. Seventy-three percent of admissions were medical, and 27% were surgical. Case patients were older (mean age 64 ± 16 years vs 58 ± 18 years; P = .002), had been hospitalized longer prior to T0, and were more likely to have had a prior RRT call (15% vs 2%; P < .001) or ICU admission (41% vs 24%; P = .001). For cases, RRT calls occurred a median of 3.9 h (interquartile range [IQR], 0.6-211.9) prior to CA, with seven (53.8%) occurring within 24 h of the arrest. Forty-three percent of case patients and 40% of control subjects were transferred to the ICU following their RRT call (P = .78). Survival to discharge was lower in case patients (31% vs 99%; P < .001).

Table Graphic Jump Location
Table 1 —Patient Characteristics

Data are shown as No. (%) unless otherwise specified. IQR = interquartile range; RRT = rapid response team.

Vital Sign Comparisons

Vital signs were similar on admission except case patients had lower diastolic BP (Table 1). The AUCs for different maximum and minimum vital signs and MEWS are shown in Table 2. Maximum MEWS had the highest AUC (0.77; 95% CI, 0.71-0.82), followed by maximum respiratory rate (0.72; 95% CI, 0.65-0.78), maximum heart rate (0.68; 95% CI, 0.61-0.74), maximum pulse pressure index (0.61; 95% CI, 0.54-0.68), and minimum diastolic BP (0.60; 95% CI, 0.53-0.67). Maximum pulse pressure performed similarly to pulse pressure index (data not shown). The AUCs for systolic BP, temperature, and oxygen saturation were not statistically significant. The sensitivities and specificities of different vital sign cutoff values are shown in Table 3.

Table Graphic Jump Location
Table 2 —AUC of Different Vital Signs and MEWS

Data shown are AUC (95% CI). AUC = area under the receiver operating characteristic curve; MEWS = Modified Early Warning Score.

Table Graphic Jump Location
Table 3 —Sensitivity and Specificity of Different Vital Sign and MEWS Cutoff Values

See Table 2 legend for expansion of abbreviations.

Vital Sign Change Over Time

From admission to 30 min prior to T0, heart rate and respiratory rate increased in case patients by 9.6% ± 27% (P = .002) and 9.3% ± 28% (P < .0001), respectively, without significant change in control subjects (Figure 2). Systolic BP fell in both groups, while diastolic BP decreased and pulse pressure index increased in control patients. However, these changes were not significantly different between the two groups. The mean MEWS was similar between case patients and control subjects on admission (2.2 ± 1.3 vs 2.0 ± 1.3; P = .28), but was significantly higher in case patients than control subjects at 48 h (2.1 ± 1.0 vs 1.6 ± 1.0; P = .005) and 24 h (2.3 ± 1.3 vs 1.5 ± 0.9; P < .001) prior to T0 (Figure 3).

Figure Jump LinkFigure 2. Change in individual vital signs from admission. For each vital sign, change from admission was calculated by subtracting each patient’s closest value prior to time of case patient cardiac arrest (T0) by the admission value. P values refer to case patient vs control subject change from baseline comparisons. An asterisk signifies that the change from baseline for the individual vital sign is statistically significant.Grahic Jump Location
Figure Jump LinkFigure 3. Change in MEWS over time. Vital signs on admission and in the 48 h prior to T0 were used to calculate MEWS scores, and comparisons were made between case and control mean scores at 8-h time points in the 48 h prior to T0. See Figure 1 and 2 legends for expansion of abbreviations.Grahic Jump Location

In this longitudinal nested case-control study, we demonstrated that patients who experience a CA on the ward have vital signs that are similar to other patients on admission but significantly different in the 48 h prior to the event. The most accurate individual predictors of CA were maximum respiratory rate, heart rate, pulse pressure index, and minimum diastolic BP. Our results have significant implications for the detection arm of the RRT because most activation criteria use poor predictors of ward CA such as systolic BP, temperature, hypoxia, and bradycardia.10,11,13 By identifying the best predictors of CA, our study provides direction regarding which predictors should be included in future derivation studies of RRT activation criteria. In addition, our findings suggest that for many patients there is ample time prior to CA to provide potentially life-saving interventions.

Importantly, we found that minimum diastolic BP and maximum pulse pressure index were significant predictors of CA. To our knowledge, no study has evaluated the accuracy of these variables for predicting CA on the ward. Although sepsis can cause an increasing pulse pressure and decreasing diastolic BP, the fact that these variables changed similarly between the two patient groups over time suggests other explanations are also possible. Interestingly, low diastolic BP has also been associated with increased levels of atherosclerosis and cardiovascular disease.25 Similarly, a high pulse pressure has been associated with adverse cardiovascular outcomes such as myocardial infarction, stroke, and increased cardiovascular mortality.2629 Some authors have suggested that pulse pressure index may be superior to pulse pressure, as the latter has no relationship to the absolute BP.28 Therefore, although systolic BP is commonly used in RRT activation criteria,10,11 incorporation of pulse pressure, pulse pressure index, or diastolic BP in place of systolic BP into the predictive model may be superior. Future studies should validate the utility and predictive ability of these variables for predicting CA and other serious adverse events. In addition, as pulse pressure is less intuitive than other vital signs and requires a calculation, automated derivation in the electronic medical record may be necessary for this predictor to be most effective.

Other investigators have also demonstrated that systolic BP is a poor predictor of clinical deterioration. In a prior case-control study in surgical patients, Cuthbertson et al30 found that there was no significant difference in mean systolic BP between groups in the 24 h prior to ICU transfer. Similarly, Fieselmann et al31 found that systolic BP was a poor predictor of CA in medical patients. In contrast, Cretikos and colleagues32 reported that a systolic BP reading ≤ 90 mm Hg had a sensitivity of 34% and specificity of 93% for a combined end point of ICU transfer, death, or CA.

Similar to our study, all three of the previous case-control studies found that respiratory rate was the best vital sign predictor of adverse outcomes on the ward.3032 This may be because many causes of CA on the ward, including pulmonary embolism, sepsis, myocardial infarction, and respiratory failure, are associated with tachypnea.2,3,33,34 Importantly, the RRT activation criteria in the MERIT study used a respiratory rate cutoff of > 36 breaths/min.6 This threshold in our study had a sensitivity of < 20% for detecting CA, similar to the 18% sensitivity for the combined end point reported by Cretikos and colleagues.32 One difficulty in choosing an ideal cutpoint is that respiratory rate is often inaccurately measured and poorly documented in hospitalized patients.8,35,36 Automated methods, such as transthoracic impedance, have been created to overcome human error, but questions remain regarding their accuracy.36 It is unknown how more accurate documentation of respiratory rate would alter its ability to accurately identify patients at risk for adverse outcomes.

Although reports of its implementation are widespread, the MEWS was not derived using ward vital signs.13,14,37,38 As illustrated in our study and suggested by other groups,7,13,39,40 it contains variables that are poor predictors of CA such as temperature, systolic BP, and bradycardia. One study derived and validated a risk prediction model using admission vital signs,39 but similar studies using ward vital signs are lacking. Given that we found the most predictive changes to occur after admission, a scoring system derived using ward vital signs would likely have improved accuracy for detecting clinical deterioration.

Our finding that the MEWS of case patients were similar to control subjects on admission but then differed significantly at 48 h prior to the event demonstrates that while some studies have suggested that vital signs measured in the ED can determine the need for higher level care,13 they may not reliably predict which patients will go on to have a ward CA. Also, our findings confirm those of previous case series demonstrating that patients begin to show abnormal physiology several hours prior to the event.3,33,34 This suggests that many patients can be identified prior to CA, enabling potentially life-saving interventions.

One of the strengths of our study is how control patients were identified. By selecting control subjects on the same ward at the same time as the matched case patient, we controlled for factors other than vital signs that may contribute to CA risk such as day of week and nurse staffing ratios. Previous studies selected control subjects admitted during a specified period of time that may not have coincided with their matched case patient admission.3032,41 In addition, other studies matched on characteristics other than the admission type or ward, such as age and sex, which makes control subjects less representative of the hospitalized population and therefore the results less useful in clinical practice.31,32

This study has several limitations. First, because this was a single-center study at a tertiary care hospital, the results may not be generalizable to some hospitals. Our survival-to-discharge rate of 31% for our CA patients is higher than the average reported by previous large studies.2,42 However, our hospital is actively involved in several cardiopulmonary resuscitation quality improvement projects that have previously been described,15,16 and this survival rate is within the range reported in the literature.4345 In addition, our goal was to investigate vital sign changes before CA, and postarrest survival was not a focus of the study. Second, our case-control design did not allow us to calculate positive and negative predictive values. Finally, when calculating the MEWS every 8 h, not all vital signs were simultaneously collected and therefore imputation was sometimes necessary. However, by imputing the previously collected value if a vital sign was missing, we simulated what clinicians often do in practice.

In conclusion, we have shown that vital signs correlate with CA on the ward. Although similar on admission, the MEWS differs between case and control patients at least 48 h prior to the event. Because many published RRT activation criteria use poor predictors of CA such as temperature, minimum heart rate, and minimum respiratory rate, further work into deriving and validating a CA prediction tool is essential. Although individual vital signs are modest predictors of CA, deriving a scoring system using combinations of vital signs would likely result in a more accurate system. The next step will involve deriving a CA risk prediction model that uses the predictors we identified in this study. This would allow maximum sensitivity while minimizing false positives, decreasing the unnecessary burden on the RRT and giving it the best chance to improve outcomes.

Author contributions: Drs Churpek and Edelson had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Dr Churpek: contributed to the design of the study, data analysis, and manuscript preparation.

Mr Yuen: contributed to data collection and revisions to the manuscript.

Mr Huber: contributed to design of the study, data collection, and revisions to the manuscript.

Dr Park: contributed to data analysis and revisions to the manuscript.

Dr Hall: contributed to the design of the study and revisions to the manuscript.

Dr Edelson: contributed to the design of the study, data analysis, and manuscript preparation.

Financial/nonfinancial disclosures: The authors have reported to CHEST the following conflicts of interest: Dr. Edelson has received research support, speaking honoraria, and consulting fees from Philips Healthcare (Andover, MA), and is on the advisory board of Sotera Wireless (San Diego, CA). Dr Hall has received royalties from McGraw-Hill and honoraria from the American Thoracic Society and the American College of Chest Physicians. Drs Churpek and Park and Messrs Huber and Yuen have reported that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

Role of sponsors: The sponsors had no role in the design of the study, the collection and analysis of the data, or in the preparation of the manuscript.

Other contributions: We thank David Meltzer, MD, PhD; J. P. Kress, MD; and John McConville, MD, for their thoughtful advice in the planning and execution of this study. In addition, we thank Donald Saner, MS, and Contessa Hsu, AAS, for assistance with data extraction and technical support and Nicole Babuskow, BS, for administrative support.

AUC

area under the receiver operating characteristic curve

AVPU

alert, responsive to voice, responsive to pain, and unresponsive

CA

cardiac arrest

MEWS

Modified Early Warning Score

ROC

receiver operating characteristic

RRT

rapid response team

T0

time of case patient cardiac arrest

Sandroni C, Nolan J, Cavallaro F, Antonelli M. In-hospital cardiac arrest: incidence, prognosis and possible measures to improve survival. Intensive Care Med. 2007;332:237-245. [CrossRef] [PubMed]
 
Peberdy MA, Kaye W, Ornato JP, et al. Cardiopulmonary resuscitation of adults in the hospital: a report of 14720 cardiac arrests from the National Registry of Cardiopulmonary Resuscitation. Resuscitation. 2003;583:297-308. [CrossRef] [PubMed]
 
Smith AF, Wood J. Can some in-hospital cardio-respiratory arrests be prevented? A prospective survey. Resuscitation. 1998;373:133-137. [CrossRef] [PubMed]
 
Chan PS, Khalid A, Longmore LS, Berg RA, Kosiborod M, Spertus JA. Hospital-wide code rates and mortality before and after implementation of a rapid response team. JAMA. 2008;30021:2506-2513. [CrossRef] [PubMed]
 
Chan PS, Jain R, Nallmothu BK, Berg RA, Sasson C. Rapid response teams: a systematic review and meta-analysis. Arch Intern Med. 2010;1701:18-26. [CrossRef] [PubMed]
 
Hillman K, Chen J, Cretikos M, et al; MERIT study investigators MERIT study investigators Introduction of the medical emergency team (MET) system: a cluster-randomised controlled trial. Lancet. 2005;3659477:2091-2097. [CrossRef] [PubMed]
 
Cuthbertson BH, Smith GB. A warning on early-warning scores! Br J Anaesth. 2007;986:704-706. [CrossRef] [PubMed]
 
Edelson DP. A weak link in the rapid response system. Arch Intern Med. 2010;1701:12-13. [CrossRef] [PubMed]
 
DeVita MA, Smith GB, Adam SK, et al. “Identifying the hospitalised patient in crisis”–a consensus conference on the afferent limb of rapid response systems. Resuscitation. 2010;814:375-382. [CrossRef] [PubMed]
 
Smith GB, Prytherch DR, Schmidt PE, Featherstone PI. Review and performance evaluation of aggregate weighted ‘track and trigger’ systems. Resuscitation. 2008;772:170-179. [CrossRef] [PubMed]
 
Smith GB, Prytherch DR, Schmidt PE, Featherstone PI, Higgins B. A review, and performance evaluation, of single-parameter “track and trigger” systems. Resuscitation. 2008;791:11-21. [CrossRef] [PubMed]
 
McBride J, Knight D, Piper J, Smith GB. Long-term effect of introducing an early warning score on respiratory rate charting on general wards. Resuscitation. 2005;651:41-44. [CrossRef] [PubMed]
 
Subbe CP, Kruger M, Rutherford P, Gemmel L. Validation of a modified Early Warning Score in medical admissions. QJM. 2001;9410:521-526. [CrossRef] [PubMed]
 
Gardner-Thorpe J, Love N, Wrightson J, Walsh S, Keeling N. The value of Modified Early Warning Score (MEWS) in surgical in-patients: a prospective observational study. Ann R Coll Surg Engl. 2006;886:571-575. [CrossRef] [PubMed]
 
Edelson DP, Litzinger B, Arora V, et al. Improving in-hospital cardiac arrest process and outcomes with performance debriefing. Arch Intern Med. 2008;16810:1063-1069. [CrossRef] [PubMed]
 
Weidman EK, Bell G, Walsh D, Small S, Edelson DP. Assessing the impact of immersive simulation on clinical performance during actual in-hospital cardiac arrest with CPR-sensing technology: a randomized feasibility study. Resuscitation. 2010;8111:1556-1561. [CrossRef] [PubMed]
 
Edelson DP, Robertson-Dick BJ, Yuen TC, et al. Safety and efficacy of defibrillator charging during ongoing chest compressions: a multi-center study. Resuscitation. 2010;8111:1521-1526. [CrossRef] [PubMed]
 
Kelly CA, Upex A, Bateman DN. Comparison of consciousness level assessment in the poisoned patient using the alert/verbal/painful/unresponsive scale and the Glasgow Coma Scale. Ann Emerg Med. 2004;442:108-113. [CrossRef] [PubMed]
 
Devita MA, Bellomo R, Hillman K, et al. Findings of the first consensus conference on medical emergency teams. [published correction appears inCrit Care Med.2006;34(12):3070]. Crit Care Med. 2006;349:2463-2478. [CrossRef] [PubMed]
 
Jones DA, DeVita MA, Bellomo R. Rapid-response teams. N Engl J Med. 2011;3652:139-146. [CrossRef] [PubMed]
 
Gao H, McDonnell A, Harrison DA, et al. Systematic review and evaluation of physiological track and trigger warning systems for identifying at-risk patients on the ward. Intensive Care Med. 2007;334:667-679. [CrossRef] [PubMed]
 
Knaus WA, Zimmerman JE, Wagner DP, Draper EA, Lawrence DE. APACHE-acute physiology and chronic health evaluation: a physiologically based classification system. Crit Care Med. 1981;98:591-597. [CrossRef] [PubMed]
 
Seymour CW, Kahn JM, Cooke CR, Watkins TR, Heckbert SR, Rea TD. Prediction of critical illness during out-of-hospital emergency care. JAMA. 2010;3047:747-754. [CrossRef] [PubMed]
 
Carpenter J, Bithell J. Bootstrap confidence intervals: when, which, what? A practical guide for medical statisticians. Stat Med. 2000;199:1141-1164. [CrossRef] [PubMed]
 
Bots ML, Witteman JC, Hofman A, de Jong PT, Grobbee DE. Low diastolic blood pressure and atherosclerosis in elderly subjects. The Rotterdam study. Arch Intern Med. 1996;1568:843-848. [CrossRef] [PubMed]
 
Millar JA, Lever AF. Implications of pulse pressure as a predictor of cardiac risk in patients with hypertension. Hypertension. 2000;365:907-911. [CrossRef] [PubMed]
 
Vaccarino V, Holford TR, Krumholz HM. Pulse pressure and risk for myocardial infarction and heart failure in the elderly. J Am Coll Cardiol. 2000;361:130-138. [CrossRef] [PubMed]
 
Peng-Lin Y, Yue-Chun L. Pulse pressure index (pulse pressure/systolic pressure) may be better than pulse pressure for assessment of cardiovascular outcomes. Med Hypotheses. 2009;726:729-731. [CrossRef] [PubMed]
 
Benetos A, Rudnichi A, Safar M, Guize L. Pulse pressure and cardiovascular mortality in normotensive and hypertensive subjects. Hypertension. 1998;323:560-564. [CrossRef] [PubMed]
 
Cuthbertson BH, Boroujerdi M, McKie L, Aucott L, Prescott G. Can physiological variables and early warning scoring systems allow early recognition of the deteriorating surgical patient? Crit Care Med. 2007;352:402-409. [CrossRef] [PubMed]
 
Fieselmann JF, Hendryx MS, Helms CM, Wakefield DS. Respiratory rate predicts cardiopulmonary arrest for internal medicine inpatients. J Gen Intern Med. 1993;87:354-360. [CrossRef] [PubMed]
 
Cretikos M, Chen J, Hillman K, Bellomo R, Finfer S, Flabouris A. MERIT study investigators MERIT study investigators The objective medical emergency team activation criteria: a case-control study. Resuscitation. 2007;731:62-72. [CrossRef] [PubMed]
 
Berlot G, Pangher A, Petrucci L, Bussani R, Lucangelo U. Anticipating events of in-hospital cardiac arrest. Eur J Emerg Med. 2004;111:24-28. [CrossRef] [PubMed]
 
Schein RM, Hazday N, Pena M, Ruben BH, Sprung CL. Clinical antecedents to in-hospital cardiopulmonary arrest. Chest. 1990;986:1388-1392. [CrossRef] [PubMed]
 
Cretikos MA, Bellomo R, Hillman K, Chen J, Finfer S, Flabouris A. Respiratory rate: the neglected vital sign. Med J Aust. 2008;18811:657-659. [PubMed]
 
Lovett PB, Buchwald JM, Stürmann K, Bijur P. The vexatious vital: neither clinical measurements by nurses nor an electronic monitor provides accurate measurements of respiratory rate in triage. Ann Emerg Med. 2005;451:68-76. [CrossRef] [PubMed]
 
Subbe CP, Davies RG, Williams E, Rutherford P, Gemmell L. Effect of introducing the Modified Early Warning score on clinical outcomes, cardio-pulmonary arrests and intensive care utilisation in acute medical admissions. Anaesthesia. 2003;588:797-802. [CrossRef] [PubMed]
 
Morgan RJ, Wright MM. In defence of early warning scores. Br J Anaesth. 2007;995:747-748. [CrossRef] [PubMed]
 
Duckitt RW, Buxton-Thomas R, Walker J, et al. Worthing physiological scoring system: derivation and validation of a physiological early-warning system for medical admissions. An observational, population-based single-centre study. Br J Anaesth. 2007;986:769-774. [CrossRef] [PubMed]
 
Hucker TR, Mitchell GP, Blake LD, et al. Identifying the sick: can biochemical measurements be used to aid decision making on presentation to the accident and emergency department. Br J Anaesth. 2005;946:735-741. [CrossRef] [PubMed]
 
Cuthbertson BH, Boroujerdi M, Prescott G. The use of combined physiological parameters in the early recognition of the deteriorating acute medical patient. [published correction appears inJ R Coll Physicians Edinb.2010;40(2):190]. J R Coll Physicians Edinb. 2010;401:19-25. [CrossRef] [PubMed]
 
Nadkarni VM, Larkin GL, Peberdy MA, et al; National Registry of Cardiopulmonary Resuscitation Investigators National Registry of Cardiopulmonary Resuscitation Investigators First documented rhythm and clinical outcome from in-hospital cardiac arrest among children and adults. JAMA. 2006;2951:50-57. [CrossRef] [PubMed]
 
Henderson SO, McClung CD, Sintuu C, Swadron SP. The presence of an Emergency Airway Response Team and its effects on in-hospital Code Blue. J Emerg Med. 2009;362:116-120. [CrossRef] [PubMed]
 
Herlitz J, Bång A, Aune S, Ekström L, Lundström G, Holmberg S. Characteristics and outcome among patients suffering in-hospital cardiac arrest in monitored and non-monitored areas. Resuscitation. 2001;482:125-135. [CrossRef] [PubMed]
 
Sandroni C, Ferro G, Santangelo S, et al. In-hospital cardiac arrest: survival depends mainly on the effectiveness of the emergency response. Resuscitation. 2004;623:291-297. [CrossRef] [PubMed]
 

Figures

Figure Jump LinkFigure 1. Modified Early Warning Score (MEWS). Unresp = Unresponsive.Grahic Jump Location
Figure Jump LinkFigure 2. Change in individual vital signs from admission. For each vital sign, change from admission was calculated by subtracting each patient’s closest value prior to time of case patient cardiac arrest (T0) by the admission value. P values refer to case patient vs control subject change from baseline comparisons. An asterisk signifies that the change from baseline for the individual vital sign is statistically significant.Grahic Jump Location
Figure Jump LinkFigure 3. Change in MEWS over time. Vital signs on admission and in the 48 h prior to T0 were used to calculate MEWS scores, and comparisons were made between case and control mean scores at 8-h time points in the 48 h prior to T0. See Figure 1 and 2 legends for expansion of abbreviations.Grahic Jump Location

Tables

Table Graphic Jump Location
Table 1 —Patient Characteristics

Data are shown as No. (%) unless otherwise specified. IQR = interquartile range; RRT = rapid response team.

Table Graphic Jump Location
Table 2 —AUC of Different Vital Signs and MEWS

Data shown are AUC (95% CI). AUC = area under the receiver operating characteristic curve; MEWS = Modified Early Warning Score.

Table Graphic Jump Location
Table 3 —Sensitivity and Specificity of Different Vital Sign and MEWS Cutoff Values

See Table 2 legend for expansion of abbreviations.

References

Sandroni C, Nolan J, Cavallaro F, Antonelli M. In-hospital cardiac arrest: incidence, prognosis and possible measures to improve survival. Intensive Care Med. 2007;332:237-245. [CrossRef] [PubMed]
 
Peberdy MA, Kaye W, Ornato JP, et al. Cardiopulmonary resuscitation of adults in the hospital: a report of 14720 cardiac arrests from the National Registry of Cardiopulmonary Resuscitation. Resuscitation. 2003;583:297-308. [CrossRef] [PubMed]
 
Smith AF, Wood J. Can some in-hospital cardio-respiratory arrests be prevented? A prospective survey. Resuscitation. 1998;373:133-137. [CrossRef] [PubMed]
 
Chan PS, Khalid A, Longmore LS, Berg RA, Kosiborod M, Spertus JA. Hospital-wide code rates and mortality before and after implementation of a rapid response team. JAMA. 2008;30021:2506-2513. [CrossRef] [PubMed]
 
Chan PS, Jain R, Nallmothu BK, Berg RA, Sasson C. Rapid response teams: a systematic review and meta-analysis. Arch Intern Med. 2010;1701:18-26. [CrossRef] [PubMed]
 
Hillman K, Chen J, Cretikos M, et al; MERIT study investigators MERIT study investigators Introduction of the medical emergency team (MET) system: a cluster-randomised controlled trial. Lancet. 2005;3659477:2091-2097. [CrossRef] [PubMed]
 
Cuthbertson BH, Smith GB. A warning on early-warning scores! Br J Anaesth. 2007;986:704-706. [CrossRef] [PubMed]
 
Edelson DP. A weak link in the rapid response system. Arch Intern Med. 2010;1701:12-13. [CrossRef] [PubMed]
 
DeVita MA, Smith GB, Adam SK, et al. “Identifying the hospitalised patient in crisis”–a consensus conference on the afferent limb of rapid response systems. Resuscitation. 2010;814:375-382. [CrossRef] [PubMed]
 
Smith GB, Prytherch DR, Schmidt PE, Featherstone PI. Review and performance evaluation of aggregate weighted ‘track and trigger’ systems. Resuscitation. 2008;772:170-179. [CrossRef] [PubMed]
 
Smith GB, Prytherch DR, Schmidt PE, Featherstone PI, Higgins B. A review, and performance evaluation, of single-parameter “track and trigger” systems. Resuscitation. 2008;791:11-21. [CrossRef] [PubMed]
 
McBride J, Knight D, Piper J, Smith GB. Long-term effect of introducing an early warning score on respiratory rate charting on general wards. Resuscitation. 2005;651:41-44. [CrossRef] [PubMed]
 
Subbe CP, Kruger M, Rutherford P, Gemmel L. Validation of a modified Early Warning Score in medical admissions. QJM. 2001;9410:521-526. [CrossRef] [PubMed]
 
Gardner-Thorpe J, Love N, Wrightson J, Walsh S, Keeling N. The value of Modified Early Warning Score (MEWS) in surgical in-patients: a prospective observational study. Ann R Coll Surg Engl. 2006;886:571-575. [CrossRef] [PubMed]
 
Edelson DP, Litzinger B, Arora V, et al. Improving in-hospital cardiac arrest process and outcomes with performance debriefing. Arch Intern Med. 2008;16810:1063-1069. [CrossRef] [PubMed]
 
Weidman EK, Bell G, Walsh D, Small S, Edelson DP. Assessing the impact of immersive simulation on clinical performance during actual in-hospital cardiac arrest with CPR-sensing technology: a randomized feasibility study. Resuscitation. 2010;8111:1556-1561. [CrossRef] [PubMed]
 
Edelson DP, Robertson-Dick BJ, Yuen TC, et al. Safety and efficacy of defibrillator charging during ongoing chest compressions: a multi-center study. Resuscitation. 2010;8111:1521-1526. [CrossRef] [PubMed]
 
Kelly CA, Upex A, Bateman DN. Comparison of consciousness level assessment in the poisoned patient using the alert/verbal/painful/unresponsive scale and the Glasgow Coma Scale. Ann Emerg Med. 2004;442:108-113. [CrossRef] [PubMed]
 
Devita MA, Bellomo R, Hillman K, et al. Findings of the first consensus conference on medical emergency teams. [published correction appears inCrit Care Med.2006;34(12):3070]. Crit Care Med. 2006;349:2463-2478. [CrossRef] [PubMed]
 
Jones DA, DeVita MA, Bellomo R. Rapid-response teams. N Engl J Med. 2011;3652:139-146. [CrossRef] [PubMed]
 
Gao H, McDonnell A, Harrison DA, et al. Systematic review and evaluation of physiological track and trigger warning systems for identifying at-risk patients on the ward. Intensive Care Med. 2007;334:667-679. [CrossRef] [PubMed]
 
Knaus WA, Zimmerman JE, Wagner DP, Draper EA, Lawrence DE. APACHE-acute physiology and chronic health evaluation: a physiologically based classification system. Crit Care Med. 1981;98:591-597. [CrossRef] [PubMed]
 
Seymour CW, Kahn JM, Cooke CR, Watkins TR, Heckbert SR, Rea TD. Prediction of critical illness during out-of-hospital emergency care. JAMA. 2010;3047:747-754. [CrossRef] [PubMed]
 
Carpenter J, Bithell J. Bootstrap confidence intervals: when, which, what? A practical guide for medical statisticians. Stat Med. 2000;199:1141-1164. [CrossRef] [PubMed]
 
Bots ML, Witteman JC, Hofman A, de Jong PT, Grobbee DE. Low diastolic blood pressure and atherosclerosis in elderly subjects. The Rotterdam study. Arch Intern Med. 1996;1568:843-848. [CrossRef] [PubMed]
 
Millar JA, Lever AF. Implications of pulse pressure as a predictor of cardiac risk in patients with hypertension. Hypertension. 2000;365:907-911. [CrossRef] [PubMed]
 
Vaccarino V, Holford TR, Krumholz HM. Pulse pressure and risk for myocardial infarction and heart failure in the elderly. J Am Coll Cardiol. 2000;361:130-138. [CrossRef] [PubMed]
 
Peng-Lin Y, Yue-Chun L. Pulse pressure index (pulse pressure/systolic pressure) may be better than pulse pressure for assessment of cardiovascular outcomes. Med Hypotheses. 2009;726:729-731. [CrossRef] [PubMed]
 
Benetos A, Rudnichi A, Safar M, Guize L. Pulse pressure and cardiovascular mortality in normotensive and hypertensive subjects. Hypertension. 1998;323:560-564. [CrossRef] [PubMed]
 
Cuthbertson BH, Boroujerdi M, McKie L, Aucott L, Prescott G. Can physiological variables and early warning scoring systems allow early recognition of the deteriorating surgical patient? Crit Care Med. 2007;352:402-409. [CrossRef] [PubMed]
 
Fieselmann JF, Hendryx MS, Helms CM, Wakefield DS. Respiratory rate predicts cardiopulmonary arrest for internal medicine inpatients. J Gen Intern Med. 1993;87:354-360. [CrossRef] [PubMed]
 
Cretikos M, Chen J, Hillman K, Bellomo R, Finfer S, Flabouris A. MERIT study investigators MERIT study investigators The objective medical emergency team activation criteria: a case-control study. Resuscitation. 2007;731:62-72. [CrossRef] [PubMed]
 
Berlot G, Pangher A, Petrucci L, Bussani R, Lucangelo U. Anticipating events of in-hospital cardiac arrest. Eur J Emerg Med. 2004;111:24-28. [CrossRef] [PubMed]
 
Schein RM, Hazday N, Pena M, Ruben BH, Sprung CL. Clinical antecedents to in-hospital cardiopulmonary arrest. Chest. 1990;986:1388-1392. [CrossRef] [PubMed]
 
Cretikos MA, Bellomo R, Hillman K, Chen J, Finfer S, Flabouris A. Respiratory rate: the neglected vital sign. Med J Aust. 2008;18811:657-659. [PubMed]
 
Lovett PB, Buchwald JM, Stürmann K, Bijur P. The vexatious vital: neither clinical measurements by nurses nor an electronic monitor provides accurate measurements of respiratory rate in triage. Ann Emerg Med. 2005;451:68-76. [CrossRef] [PubMed]
 
Subbe CP, Davies RG, Williams E, Rutherford P, Gemmell L. Effect of introducing the Modified Early Warning score on clinical outcomes, cardio-pulmonary arrests and intensive care utilisation in acute medical admissions. Anaesthesia. 2003;588:797-802. [CrossRef] [PubMed]
 
Morgan RJ, Wright MM. In defence of early warning scores. Br J Anaesth. 2007;995:747-748. [CrossRef] [PubMed]
 
Duckitt RW, Buxton-Thomas R, Walker J, et al. Worthing physiological scoring system: derivation and validation of a physiological early-warning system for medical admissions. An observational, population-based single-centre study. Br J Anaesth. 2007;986:769-774. [CrossRef] [PubMed]
 
Hucker TR, Mitchell GP, Blake LD, et al. Identifying the sick: can biochemical measurements be used to aid decision making on presentation to the accident and emergency department. Br J Anaesth. 2005;946:735-741. [CrossRef] [PubMed]
 
Cuthbertson BH, Boroujerdi M, Prescott G. The use of combined physiological parameters in the early recognition of the deteriorating acute medical patient. [published correction appears inJ R Coll Physicians Edinb.2010;40(2):190]. J R Coll Physicians Edinb. 2010;401:19-25. [CrossRef] [PubMed]
 
Nadkarni VM, Larkin GL, Peberdy MA, et al; National Registry of Cardiopulmonary Resuscitation Investigators National Registry of Cardiopulmonary Resuscitation Investigators First documented rhythm and clinical outcome from in-hospital cardiac arrest among children and adults. JAMA. 2006;2951:50-57. [CrossRef] [PubMed]
 
Henderson SO, McClung CD, Sintuu C, Swadron SP. The presence of an Emergency Airway Response Team and its effects on in-hospital Code Blue. J Emerg Med. 2009;362:116-120. [CrossRef] [PubMed]
 
Herlitz J, Bång A, Aune S, Ekström L, Lundström G, Holmberg S. Characteristics and outcome among patients suffering in-hospital cardiac arrest in monitored and non-monitored areas. Resuscitation. 2001;482:125-135. [CrossRef] [PubMed]
 
Sandroni C, Ferro G, Santangelo S, et al. In-hospital cardiac arrest: survival depends mainly on the effectiveness of the emergency response. Resuscitation. 2004;623:291-297. [CrossRef] [PubMed]
 
NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging & repositioning the boxes below.

Find Similar Articles
CHEST Journal Articles
PubMed Articles
Guidelines
Rapid response team. Health care protocol.
Institute for Clinical Systems Improvement
  • CHEST Journal
    Print ISSN: 0012-3692
    Online ISSN: 1931-3543