C

Cm

BWa

b

a

ARR2A

KRMMC

1

iUfJtc

t

i

U

0d

Resuscitation 82 (2011) 415–418

Contents lists available at ScienceDirect

Resuscitation

journa l homepage: www.e lsev ier .com/ locate / resusc i ta t ion

linical paper

linical emergencies and outcomes in patients admitted to a surgical versusedical service�

abak Sarania,∗ , Emily Palilonisa , Seema Sonnada , Meredith Bergeya , Carrie Simsa , Jose L. Pascuala ,illiam Schweickertb

Department of Surgery, Division of Traumatology and Surgical Critical Care, University of Pennsylvania, United StatesDepartment of Medicine, Division of Pulmonary, Allergy and Critical Care, University of Pennsylvania, United States

r t i c l e i n f o

rticle history:eceived 15 July 2010eceived in revised form3 November 2010ccepted 8 December 2010

eywords:apid response systemsedical emergency teamortality

ardiac arrest

a b s t r a c t

Background: The merit of rapid response systems (RRSs) remains controversial. A tailored approach tospecific groups may increase the efficacy of these teams. The purpose of this study was to comparedifferences in triggers for RRS activation, interventions, and outcomes in patients on medical and surgicalservices.Methods: A retrospective review RRS events was performed. The incidence of out of ICU cardiac arrestsand hospital mortality were compared 2 years prior to and following RRS implementation. Call trigger,interventions, and disposition between medical and surgical patients were compared over a 15 monthperiod.Results: Out of ICU cardiac arrest was significantly more prevalent in the medical group both beforeand after implementation of RRS. The out of ICU cardiac arrest rate decreased 32% in the surgical group(p = 0.05) but hospital mortality did not change. Out of ICU cardiac arrest decreased 40% in the medicalgroup (p < 0.001) and hospital mortality decreased 25% (p < 0.001) following RRS implementation. Therewere 1082 RRS activations, 286 surgical and 796 medical. Surgical patients were more likely to have

received sedation within 24 h of evaluation (14% vs. 4%, p < 0.001). The majority of patients in both cohortswere discharged alive.Conclusion: Implementation of a RRS had greater impact on reduction of out of ICU cardiac arrest andmortality in medical inpatients. Triggers for activation and interventions were similar between groups;however, surgical patients demonstrated substantial risk for decompensation within the first 24 h fol-lowing operation. More research is needed to evaluate the disproportionate benefit observed between cohorts.

. Introduction

Hospital-based rapid response systems have been embracednternationally as a proactive approach to patient safety.1 In thenited States, even greater support stems from the Institute

or Healthcare Improvement’s 100,000 Lives Campaign and theoint Commission’s requirement that all hospitals have a means

o improve recognition and response to changes in a patient’sondition.2,3

Rapid response systems (RRSs) developed from the recognitionhat cardiac arrest in hospitalized patients is frequently preceded

� A Spanish translated version of the abstract of this article appears as Appendixn the final online version at doi:10.1016/j.resuscitation.2010.12.005.∗ Corresponding author at: 3400 Spruce Street, 5 Maloney, Philadelphia, PA 19104,nited States. Tel.: +1 215 662 7323; fax: +1 215 349 5917.

E-mail address: saranib@uphs.upenn.edu (B. Sarani).

300-9572/$ – see front matter © 2010 Elsevier Ireland Ltd. All rights reserved.oi:10.1016/j.resuscitation.2010.12.005

© 2010 Elsevier Ireland Ltd. All rights reserved.

by unrecognized or unattended signs of physiologic instability.4–6

These teams provide prompt evaluation, triage, and treatmentof patients with recognized deterioration. Multiple studies havedemonstrated that implementation of a RRS decreases the inci-dence of cardiac arrest in hospitalized patients outside of theICU.7–11 However, the absence of consistent hospital-wide mor-tality reduction has generated controversy regarding the merit ofthe intervention.12–14

Patients on specific services may derive differential benefit fromthe implementation of an RRS. A tailored RRS infrastructure forselected patient cohorts or settings may increase the efficacy ofthese teams. However, studies to date have not differentiatedpatient outcome based upon medical versus surgical service. Such

a differentiation may be relevant because hospitalized medicalpatients tend to be older and have more co-morbidities than sur-gical patients. Conversely, surgical patients may have a heightenedinflammatory response or risk opiate overdose, particularly in theearly post-operative period, and surgeons may be less available to

416 B. Sarani et al. / Resuscitation 82 (2011) 415–418

Table 1Criteria for rapid response activation.

• Respiratory© Rate <8 or >32 breaths/min© Oxygen saturation <85%© Acute increase in oxygen need by 50%© Dyspnea

• Cardiac© Rate <40 or >140 beats per minute© Systolic blood pressure <80 or >200 mmHg© Diastolic blood pressure >110 mmHg© New onset chest painsara

• Neurologic© Seizure© Acute change in mental status

• Miscellaneous© Uncontrolled bleeding

ascn

oisvtc(

2

RaaptttarionaR

pdotadRnpC

diivC

Table 2Overall incidence of cardiac arrest outside of the ICU before and after RRSimplementation.

2004–2006(before MET)

2006–2008(after MET)

Cardiac arrest 290 177Discharges 71,242 76,178Cardiac arrest/1000 discharges* 4.07 2.32

**

© Inability to contact house-officer after 2 pages© Nurse concern/discretion© Physician concern/discretion

ssess acutely ill patients due to activities in the operating room. Asuch, these two groups may manifest different signs of impendingritical illness, experience varying periods of vulnerability, and mayeed different urgent intervention(s) to re-establish homeostasis.

The purpose of this study was to evaluate the impact of a RRSn out of ICU cardiac arrest and hospital mortality in hospital-zed patients on the medical and surgical services. Additionally, weought to identify differences in triggers for RRS activation, inter-entions performed and outcomes in hospitalized patients in thesewo populations. We hypothesized that outcomes, emergent clini-al needs and interventions rendered by a medical emergency teamMET) differ based on the population served.

. Methods

The Hospital of the University of Pennsylvania implemented aRS in July 2006. In July 2007, the team was split into a medicalnd surgical arm. Members of the MET that are common to bothrms include: an intensive care unit (ICU) nurse, pharmacist, res-iratory therapist, and a member of the patients’ primary residenteam. During the day, an ICU attending physician or fellow fromhe medical or surgical intensive care unit respond to all calls fromheir respective services. Night and weekend MET activations arettended by covering residents from the respective services withemote intensivists who are available for consultation by phonef needed. The surgical MET (sMET) responds to all events thatccur in patients admitted to the departments of surgery, otorhi-olaryngology, orthopaedics, neurological surgery, and obstetricsnd gynecology. The cardiac surgical service is excluded from theRS. The medical MET (mMET) responds to all other patient groups.

Anyone may activate the team, including non-clinical hospitalersonnel and families/visitors. Formal activation criteria that areistributed to all hospital staff are listed in Table 1. For the purposesf this study, neurologic emergencies were defined as: mental sta-us change, loss of consciousness, seizure, suspected stroke, andcute agitation. Cardiovascular triggers were defined as: bradycar-ia, tachycardia, chest pain, hypotension, and severe hypertension.espiratory triggers for team activation were defined as: tachyp-ea, dyspnea, and hypoxemia. Data on each rapid response event isrospectively gathered and entered into the National Registry forardiopulmonary Resuscitation (NRCPR).

After obtaining IRB approval, a retrospective study of the RRS

atabase was performed from July 1, 2006 to June 30, 2008. The

ncidence of cardiac arrest by service was compared between twodentical two-year intervals: July 1, 2004 to June 30, 2006 (pre-RRS)ersus July 1, 2006 to June 30, 2008 (post-RRS implementation).ase-mix index, hospital discharge volume, and hospital mortal-

CMI 1.72 1.70

* p < 0.001.** CMI = case mix index, p = NS.

ity for the two time periods were obtained from the Hospital ofthe University of Pennsylvania Clinical Effectiveness and Qual-ity Improvement Committee. The case-mix index is the averagediagnosis-related group weight for all of the hospital’s Medicarevolume and serves as a means to compare patient compositionbetween time periods and across hospitals; furthermore, it can beused as a surrogate evaluation for severity of illness.15 DetailedRRS cases of adult patients admitted to the medical service or asub-specialty of internal medicine and surgical services as definedabove were queried between July 2007 and October 2008. Exclu-sion criteria included patients in the ICU, patients on the cardiacsurgical service, and non-hospitalized patients such as outpatients,staff, and visitors. Variables abstracted from the RRS databaseincluded: time of activation, call trigger, interventions performed,disposition following evaluation, whether the patient was in theemergency department or the ICU within 24 h prior to activa-tion, and whether the patient required conscious sedation or wasin the post-operative care unit within 24 h of activation. Manualchart review was used to determine the reason for deterioration inpatients who had undergone operation within 24 h of RRS activa-tion.

Statistical significance was assessed using the Mann WhitneyU test for continuous variables and chi-square test for categoricalvariables. The Fisher exact test was used to assess for differenceswhen the sample size for a particular event was less than 5. Signifi-cance was defined as p < 0.05. A power analysis was not performedbecause we found very little overall difference in any endpointtested other than probability for need to ICU transfer. It is thereforeunlikely that a larger dataset would generate clinically or statisti-cally relevant differences.

3. Results

Using chi-square analysis, we found that the overall incidenceof out of ICU cardiac arrests per 1000 discharges substan-tially decreased following RRS implementation (4.07 vs. 2.32arrests/1000 discharges; p < 0.001; Table 2) despite no significantchange in the case mix index. The incidence of cardiac arrest wasalmost three times higher on the medical service than the surgicalservice (5.01/1000 discharges vs. 1.69/1000 discharges, p < 0.001)prior to implementation of the RRS (Table 3). Both services appre-ciated a significant reduction in cardiac arrest rate following RRSimplementation, but the degree of reduction was significantlyhigher in the medical cohort (40% reduction) compared to the surgi-cal cohort (32% reduction, p < 0.001). After RRS implementation, themedical service continued to have a significantly higher incidenceof cardiac arrest than the surgical service (3.02/1000 dischargesvs. 1.16/1000 discharges, p < 0.001). Hospital mortality decreasedsignificantly on the medical service following implementation of

the RRS. The medical service experienced a 25% relative reduc-tion in mortality (4.29–3.23%, p < 0.001) following the start of RRSwhereas mortality did not change significantly on the surgical ser-vice (1.21–1.11%, p = ns).

B. Sarani et al. / Resuscitation 82 (2011) 415–418 417

Table 3Service-based incidence of cardiac arrest outside of the ICU before and after RRSimplementation.

2004–2006(before MET)

2006–2008(after MET)

Medical serviceCardiac arrest on medicine service 156 102Discharges on medicine service 31,101 33,721Cardiac arrest/1000 discharges* 5.01 3.02Mortality* 4.29% 3.23%

Surgical serviceCardiac arrest on surgical service 64 44Discharges on surgical service 37,769 37,992Cardiac arrest/1000 discharges** 1.69 1.16Mortality 1.21% 1.11%Cardiac arrest on other services 70 31

* p < 0.001.** p = 0.05.

Table 4Triggers for RRS activation by service.

Trigger sMET (%) mMET (%)

Respiratory 102 (36) 255 (32)Cardiac 176 (62) 519 (65)Neurologic 13 (5) 33 (4)RN concern 115 (40) 292 (37)

pS

7tgcsco

tvwvhviseoM

twsyts

TT

P

Table 6Interventions performed by the MET.

Intervention sMET N (%) mMET N (%)

Non-invasive ventilation* 3 (1) 28 (4)Intubation 43 (15) 126 (16)Antiarrhythmia agent 16 (6) 60 (7)Antibiotic 21 (7) 80 (10)Naloxone/flumazenil 20 (8) 42 (5)

Other 13 (5) 46 (6)

= NS for all parameters.ome patients had more than 1 trigger.

We reviewed 1082 inpatient RRS activations, 286 surgical and96 medical responses. There were no significant differences inhe triggers for RRS activation between the medical and surgicalroups as tested by the Mann Whitney U test (Table 4). The mostommon triggers for RRS activation were cardiac (mMET 65% andMET 62%), respiratory (mMET 32% and sMET 36%), and nurse con-ern/discretion (mMET 37% and sMET 40%). There were no familyr patient initiated activations.

We found significant differences between the two groups inime of day for RRS activation and use of sedation within 24 h of acti-ation using chi-square analysis (Table 5). More medical patientsere evaluated by the MET between 7 am and 7 pm (mMET 60%

s. sMET 49%, p < 0.001), and surgical patients were more likely toave received sedation within 24 h of MET activation (mMET 4%s. sMET 14%, p < 0.001). Although triggers for RRS activation var-ed, we found that 15/41 (37%) events that occurred within 24 h ofedation in the surgical group were attributed to complications ofxcessive opiate administration. Both groups had an 8% incidencef having been in the emergency department or ICU within 24 h ofET activation.We assessed interventions performed by the MET using either

he Fisher exact test or chi-square test. Overall, medical patientsere more likely to require an intervention (mMET 52% vs.

MET 42%, p = 0.003). However, analysis of individual interventionsielded only one significant difference: medical patients were fourimes more likely to undergo a trial of non-invasive positive pres-ure ventilation (mMET 4% vs. sMET 1%, p = 0.04, Table 6).

able 5ime and location preceding RRS activation.

sMET (%) mMET (%)

Time of MET activation 7 am–7 pm* 140 (49) 481 (60)ICU or emergency department within 24 h

prior to MET39 (8) 107 (8)

Sedation or PACU within 24 h prior to MET* 41 (14) 29 (4)

ACU = post-operative care unit.* p < 0.001.

Vasoactive agents 15 (6) 64 (7)Cardiopulmonary resuscitation 3 (1) 19 (2)

* p = 0.04.

We did not find statistically significant differences using chisquare analysis in disposition between the two groups followingevaluation by the MET, but there was a trend toward a higherneed for transfer to the ICU in the mMET group (mMET 63% vs.sMET 53%, p = 0.06). Additionally, relative to sMET patients, med-ical patients were significantly more likely to transfer to anotherarea of the hospital (mMET 65% ICU transfer and 5% other area;sMET 53% ICU transfer and 10% other area, p = 0.005). Areas to whichpatients were transferred other than the ICU included the operat-ing room, telemetry capable beds, and the cardiac catheterizationsuite. Although the vast majority of patients in both cohorts wereultimately discharged alive from the hospital, significantly moresurgical patients survived to discharge than medical patients (sMET83% vs. mMET 72%, p < 0.001).

4. Discussion

Formal rapid response systems have been implemented in hos-pitals throughout the United States; however, great heterogeneityof service structure exists. Our hospital uses a two-arm medicalemergency team system to evaluate patients based on admissionto a medical or surgical service. Potential benefits of this strategyinclude greater physician familiarity with the expected hospitalcourse, common causes for deterioration, and the patient’s pri-mary team. Conceptually, this may foster greater RRS acceptance,streamline evaluation, and develop more comprehensive commu-nication and planning to facilitate pre-emptive care. Additionally,it may facilitate and expedite administration of appropriate ther-apy. This study was thus performed to assess the relative benefit onoutcomes following RRS implementation when stratified by patientpopulation – medical versus surgical – and to evaluate whetherthese services truly differ in the need for and type of emergentintervention rendered by a MET.

The current study recreated the finding that RRS implementa-tion reduces the incidence of out of ICU cardiac arrest,14 both foranalyses of the total population and by service stratification. Inter-estingly, cardiac arrest event rates were substantially higher in themedical population both before and after RRS implementation, andRRS implementation was associated with a much more pronouncedreduction in events in the medical patients. Accordingly, hospitalmortality for medical patients was significantly reduced with theimplementation of the RRS, but no significant change occurred insurgical patients.

The reasons for the disproportionate benefit in medical versussurgical patients cannot be explained by the current dataset. How-ever, the finding is contrary to the notion that RRS is most potentin assisting the “absent surgeon” and surgical inpatients. Potentialexplanations for the greater incidence of cardiac arrest on the med-ical service may be patients’ advanced co-morbidities and acuity of

illness. Although there are no studies directly comparing acuity ofillness and outcome between medical and surgical patients outsideof the ICU, a retrospective study found that medical patients whorequire admission to an ICU have a higher risk of death than surgicalpatients.16 Alternatively, the greater relative reduction in cardiac

4 citatio

atmmsosbmpaa

tctT8itRpf

fesaoduifnoimacomeacv

gniasesg

5

e

1

1

1

1

1

1

1

18 B. Sarani et al. / Resus

rrest incidence and mortality in the medical cohort may be dueo greater or (hypothetically) earlier utilization of this resource by

edical than surgical services. In the period surrounding imple-entation of the RRS, the Department of Medicine instituted a

eries of didactic lectures focusing on patient safety and the rolef RRSs, potentially increasing utilization of the system. A previoustudy in our institution found that medical residents more stronglyelieved that RRS improve patient safety,17 and therefore may beore likely to activate the system early. If this were the case, a

rior observational study has demonstrated that both cardiac arrestnd mortality rates decrease as the time from instability to RRSctivation decreases.18

Mortality can be impacted upon in many ways and it is possiblehat concomitant changes in patient care and patient populationontributed to the noted decrease; however, it is also likely thathe reduction in cardiac arrest volume contributed to this finding.his conclusion is supported by the finding that 72% of medical and3% of surgical patients survived to discharge, thereby demonstrat-

ng that the RRS does not simply facilitate the transfer of patientso the ICU where they ultimately cardiac arrest and possibly die.ather, the RRS ideally provides immediate diagnostic and thera-eutic interventions that can rescue deteriorating patients at riskor cardiac arrest and death.

Another key finding of this study was that there were few dif-erences between the medical and surgical clinical emergenciesvaluated. Although the lack of difference may be a byproduct of ourmall sample size, a few differences did exist as exemplified by opi-te administration in the post-operative period. Fourteen percentf surgical patients evaluated by the sMET experienced physiologiceterioration within 24 h of sedation. Because the need for eval-ation by the sMET was spread evenly over a 24 h time period, it

s less likely that this is due to persistent anesthetic effect or dif-erences in physician availability to direct care. Although we didot find a single consistent cause for deterioration in this group,piate overdose accounted for over one-third of cases. This find-ng occurred despite patient-controlled analgesia as our preferred

eans of intravenous narcotic administration outside of the ICU,nd was due to high dosing of medication by the prescribing clini-ian in the majority of cases. Other causes, such as dosing by personsther than the patient him/herself, could not be assessed via theedical record. This further reinforces the need for a post-operative

valuation of the patient by a member of the surgical team, nursingttention to the risk of over-sedation, and need for clear communi-ation between physician and nursing teams when managing theseulnerable patients.

Although this is the first study to directly compare clinical emer-encies stratified by medical and surgical services, the retrospectiveature, single institution, and small sample size create several lim-

tations that should be explored in future studies. In particular, wedvocate a multi-institutional query of event rates and outcometratified by inpatient service. Given the lack of significant differ-nces in triggers and interventions performed in the two cohortstudied in our institution, the merits of the two-arm medical emer-ency team remain uncertain.

. Conclusion

In conclusion, this is the first study to evaluate the differentialffect of RRS implementation on cardiac arrest and hospital mor-

1

1

n 82 (2011) 415–418

tality rates when stratified by medical or surgical service. We foundthat the baseline incidence of out of ICU cardiac arrest and hospitalmortality were higher in patients on the medical service and thatimplementation of a RRS had a more pronounced impact on bothendpoints in the medical patient cohort (as opposed to the surgi-cal cohort). Despite this reduction, however, the medical servicecontinues to experience a higher incidence of cardiac arrest andmortality, and future studies are needed to determine the causefor these disproportionate effects and event rates. Furthermore, wefound that few differences exist between the activation triggers andclinical interventions performed for the two cohorts. Future patientsafety and research efforts should be aimed at evaluating whetherthese discordant benefits are reproducible and whether these find-ings should prompt novel RRS structuring to maximize benefit inparticularly vulnerable populations.

Conflict of interest statement

The authors do not have any financial, personal, or other con-flicts of interest with any materials related to this work.

References

1. Devita MA, Bellomo R, Hillman K, et al. Findings of the first consensus conferenceon medical emergency teams. Crit Care Med 2006;34:2463–78.

2. Berwick DM, Calkins DR, McCannon CJ, Hackbarth AD. The 100,000 lives cam-paign: setting a goal and a deadline for improving health care quality. JAMA2006;295:324–7.

3. Joint Commission. National Patient Safety Goals: Goal No 16 – Improve recog-nition and response to changes in a patient’s condition; 2009. Available from:http://www.jcrinc.com/2009-NPSGs-Goal-16 [accessed 01.09.10].

4. Hillman K, Parr M, Flabouris A, Bishop G, Stewart A. Redefining in-hospitalresuscitation: the concept of the medical emergency team. Resuscitation2001;48:105–10.

5. Hillman KM, Bristow PJ, Chey T, et al. Antecedents to hospital deaths. Intern MedJ 2001;31:343–8.

6. Schein RM, Hazday N, Pena M, Ruben BH, Sprung CL. Clinical antecedents toin-hospital cardiopulmonary arrest. Chest 1990;98:1388–92.

7. Buist MD, Moore GE, Bernard SA, Waxman BP, Anderson JN, Nguyen TV.Effects of a medical emergency team on reduction of incidence of and mor-tality from unexpected cardiac arrests in hospital: preliminary study. BMJ2002;324:387–90.

8. DeVita MA, Braithwaite RS, Mahidhara R, Stuart S, Foraida M, Simmons RL. Use ofmedical emergency team responses to reduce hospital cardiopulmonary arrests.Qual Saf Health Care 2004;13:251–4.

9. Galhotra S, DeVita MA, Simmons RL, Dew MA. Mature rapid response systemand potentially avoidable cardiopulmonary arrests in hospital. Qual Saf HealthCare 2007;16:260–5.

0. Jones D, Bellomo R, Bates S, et al. Long term effect of a medical emergency teamon cardiac arrests in a teaching hospital. Crit Care 2005;9:R808–15 [Epub 2005].

1. Offner PJ, Heit J, Roberts R. Implementation of a rapid response team decreasescardiac arrest outside of the intensive care unit. J Trauma 2007;62:1223–7 [dis-cussion 7–8].

2. Winters BD, Pham J, Pronovost PJ. Rapid response teams – walk, don’t run. JAMA2006;296:1645–7.

3. Winters BD, Pham JC, Hunt EA, Guallar E, Berenholtz S, Pronovost PJ. Rapidresponse systems: a systematic review. Crit Care Med 2007;35:1238–43.

4. Chan PS, Jain R, Nallmothu BK, Berg RA, Sasson C. Rapid response teams: asystematic review and meta-analysis. Arch Intern Med 2010;170:18–26.

5. Jencks SF, Dobson A. Refining case-mix adjustment. The research evidence. NEngl J Med 1987;317:679–86.

6. Ghelani D, Moran JL, Sloggett A, Leeson RJ, Peake SL. Long-term survival of inten-sive care and hospital patient cohorts compared with the general Australianpopulation: a relative survival approach. J Eval Clin Pract 2009;15:425–35.

7. Sarani B, Sonnad S, Bergey M, et al. Resident and RN perceptions of the impactof a medical emergency team on education and patient safety in an academicmedical center. Crit Care Med 2009;37:3091–6.

8. Sebat F, Musthafa AA, Johnson D, et al. Effect of a rapid response system forpatients in shock on time to treatment and mortality during 5 years. Crit CareMed 2007;35:2568–75.