Emergency Department CK-MB: A Predictor of Ischemic Complications

17

CK-MB and Complications, Hoekstra et al.

Emergency Department CK-MB: A Predictor of Ischemic Complications James W. Hoekstra, MD, Jerris R. Hedges, MD, MS, W Brian Gibler, MQ R. Mike Rubison, PhD, Ronald A . Christensen, MD For the National Cooperative CK-MB Project Group*

ABSTRACT Objective: To demonstrate that a positive CK-MB in the emergency department (ED) predicts an increased risk for complications of myocardial ischemia in patients admitted to the hospital for evaluation of chest pain. Methods: 53 academic and community hospital EDs participated in this prospective observational cohort analysis of 5,120 patients with chest pain without ST-segment elevation on the initial ED 12-lead electrocardiogram. All patients were admitted for evaluation of chest pain in one of the participating hospitals as part of the National Cooperative CK-MB Project. Patients were stratified by whether or not they had an elevated CKMB level in the ED. CK-MB measurements were made on ED presentation and two hours later. Patient medical records were reviewed for inpatient diagnoses -myocardial infarction (MI) or other diagnosis -and for ischemic complications-cardiac-related death, recurrent or delayed in-hospita1 MI, significant ventricular arrhythmias, new conduction defects, congestive heart failure, and cardiogenic shock. Results: 369 (7.2%) of the 5,120 patients had MI. The proportion of patients with any complication in the MI group was 24%, while the complication rate in the non-MI group was 0.4%. In all patients, regardless of final diagnosis, the relative risk of any complication was 16.1 (95% CI 11.O-23.6) in those with apositive ED CK-MB versus negative ED CKMB patients. Similarly, the relative risk of death was 25.4 (95% CI 10.860.2) in positive ED CK-MB versus negative ED CK-MB patients. Conclusions: Multicenter data support the hypothesis that CK-MB measurements can help risk-stratify ED chest pain patients whose initial ECGs are without diagnostic ST-segment elevation. Acad. Emerg. Med. 1994; 1:17-28.

*Participating physician coordinators and project coordinators for the National Cooperative CK-MB Project are listed in Appendix A.

18

ACADEMIC EMERGENCY MEDICINE JAN/FEB 1994 VOL 1/NO 1

Patients who present to the emergency department (ED) with chest pain and electrocardiograms (ECGs) without ST-segment elevation are a difficult challenge for emergency physicians. The lack of objective data to diagnose myocardial infarction (MI) and predict ischemic complications in these patients at the time of initial ED evaluation has led to a conservative approach to their disposition by emergency physicians and cardiologists. Admission of patients at greatest risk for ischemic complications and MI to closely monitored hospital beds is believed to minimize ischemia-related cardiac morbidity and mortality.1.2 At present, two-thirds of the patients admitted to critical care units (CCU) for presumed cardiac chest pain do not develop MI, and one-third have no evidence of coronary artery disease.3-5 In order to use monitored beds appropriately, more objective criteria are needed to aid in the diagnosis and disposition of chest pain patients.

,

Chest pain patients who present to the ED with endorgan dysfunction, hemodynamic instability, congestive heart failure, cardiac arrhythmias, or ST-segment elevations on ECGs are easily admitted to the CCU setting by emergency physicians. Conversely, hemodynamically and rhythm stable patients who have chest pain and ECGs without ST-segment elevation can be a significant disposition problem. These patients are routinely admitted to “rule out” MI, but the decision to send the patient to a stepdown unit or telemetry bed versus the CCU is often problematic.6-8 The only generally accepted measure of risk stratification in hemodynamically and rhythm stable chest pain patients is acute ST-segment elevation on the E c G . 9 ~ 1 ~ Other ECG patterns such as old Q-waves, ST-segment depression, T-wave inversion, bundle-branch-block pat-

TABLE M1 Characteristics of the National Cooperative CK-MBProject Participants ............... ..........................................................

Hospital

1 2 3 4 5 6 7

8 9 10 11 12 13 14 15

16 17 18 19 20 21

22 23 24 25 26 27 28 29 30

Months in Study

9 7 29 10 21 14

Cases Entered

Staffed %MI* Beds

....................................................................

CCUt

Beds

ED$ Visits

COTHg

Hospital

16 29

123 27 105 197 225 43 24 28 210 87

23.6 0.0 33.3 5.1 10.7 25.6 12.5 25.0 17.6 11.5

239 462 525 380 462 364 529 239 258 231

29 24 85 40 24 22 36 14 21 22

25,943 40,496 44,883 40,589 43,346 36.429 59,385 15,934 30,359 29,200

No Yes No Yes No No Yes No No No

31 32 33 34 35 36 37 38 39

17 14 8 15 19 16 22 10 4 29

56 30 31 11 27 114 68 45 14 124

16.1 13.3 19.4 27.3 18.5 11.4 20.6 8.9 14.3 13.7

142 269 354 174 185 194 370 292 118 308

10 16 44 14 16 24 46 16 8 50

19,917 33,827 37,505 13,391 22,410 13.500 13,146 31,778 12,102 18.244

No No No No No No No No Yes

9 9 6

22 157 23 139 327 24 52 133 37 17

13.6 13.4 26.1 27.3 14.7 8.3 25.0 9.0 37.8 17.6

261 484 325 349 719 177 215 349 650 212

12 40 33 14 44 10 9 26 47 21

27,839 28,164 17,560 20,232 89,525 21,653 22,896 43,018 35,409 23.862

Yes Yes No No Yes No No Yes Yes No

5 5

11

14 18 6 19 22 6

......

*Percent of enrolled cases with myocardial infarction. ?Critical care unit (intensive and coronary care units). $Emergency department. $Council of Teaching Hospitals member.

No

....................................................

Months in Cases Study Entered

%MI*

Staffed CCUt Beds Beds

..........

ED$ Visits

COTHS

40

8 18 19 16 11 7 9 8 19 14

48 40 52 100 31 17 7 143 233 152

10.4 15.0 40.4 8.0 9.7 17.6 0.0 9.8 18.0 9.9

54 238 345 254 366 200 325 376 497 507

4 12 23 19 35 14 16 12 39 47

12,868 22,807 41,627 27,784 26,026 11,063 22.143 34.088 55,852 37,987

No No No No No No No No Yes Yes

41 42 43 44 45 46 47 48 49 50

4 30 29 5 12 13 8 19 11 23

16 352 400 118 116 42 268 11 162

56.3 16.8 16.5 0.0 11.9 11.2 9.5 10.1 0.0 5.6

255 203 295 274 243 532 241 220 335 138

24 14 21 22 28 60 13 16 18 12

26,224 20,259 21,477 45,991 54,900 30,869 10.111 35,645 22,944 22,307

No No Yes No No Yes No No No No

51 52 53

13

562 215 51

5.0 16.7 15.7

146 417 300

9 12 90

55,874 78,504 10,525

No Yes No

11

19

5

19

CK-MB and Complications, Hoekstra et al.

METHODS

TABLE M2 Inclusion and Exclusion Criteria for the Study Population

........... ........... .............

Study Design This multicenter prospective observational cohort analysis evaluated patients enrolled at 53 academic and community hospitals participating in the National Cooperative CK-MB Project Group. The months of patient enrollment varied with the institutions involved, and are listed in Table M1. The experimental protocol was approved by the institutional human subjects review boards at the individual hospitals when not exempted from review. Individual patient consent was obtained as required by the human subjects review boards.

. ....... . .. ... .. .., .. .... ... .. ..... .. . ... .. ...... . ...............

Inclusion criteria Age greater than or equal to 30 years Chief complaint of chest pain History consistent with myocardial ischemia/infarction Electrocardiogram obtained Hospital admission to evaluate chest pain Exclusion criteria History of chest trauma Chest radiograph that indicates cause of chest pain E.g., rib fractures, pneumothorax, pneumonia ECG consistent with acute myocardial infarction 1.e.. ST elevation of 2 0.1 mV or new Q-waves in two electrically contiguous leads Incomulete Datient data

Participants Patients presenting to one of the participating hospital EDs with the chief complaint of chest pain were considered for entry into the study. Patients 30 years of age or older with a history of chest pain or chest discomfort consistent with myocardial ischemia were considered eligible. An ECG was obtained as part of the initial ED evaluation. All patients were admitted to monitored beds in the hospital for evaluation of possible acute coronary artery syndrome. Exclusion criteria included a history of chest wall trauma or chest radiographic findings that accounted for the chest pain. Also excluded were patients who had obvious MI by ECG criteria (ST-segment elevation > 0.1 mV in two or more electrically contiguous leads or new Q-waves) such that they were ECG candidates for thrombolytic therapy. Patients with other ECG changes, including ST-segment depression, bundle-branch blocks, paced rhythms, or ventricular hypertrophy with strain, remained in the study. Patients who developed new ECG ST-segment changes or Q waves while in the hospital also remained in the study. Patients for whom data were incomplete (missing both CK-MB values or unobtainable hospital records) were excluded. Table M2 lists inclusion and exclusion criteria.

written consent was obtained (where required) and the patient was enrolled. Following enrollment, 10-ml venous blood samples were obtained using standard venipuncture techniques at initial ED evaluation and again two hours later. CKMB analysis was performed by one of the following monoclonal double antibody techniques: Hybritech ICON-QSR assay (Hybritech, Inc., San Diego, CA), the Corning Magic-Lite assay (Ciba-Corning, Medfield, MA), the Baxter Stratus assay (Baxter Healthcare Corp., Miami, FL), and the Abbott IMX assay (Abbott Labs, Chicago, IL). Based upon prior sensitivity analysis, the following enzyme thresholds were chosen to represent a positive CK-MB level: Hybritech ICON-QSR, > 7 ng/ml; Corning Magic-Lite, > 7.5 ng/ml; Baxter Stratus, > 5.6 ng/ml; Abbott IMX, > 6 ng/ml. A CK-MB level over the threshold at either the initial or the twohour blood draw was considered positive.

Experimental Protocol

Following initial ED presentation and evaluation by history, physical examination, and ECG, all enrolled patients were admitted to the hospital for evaluation of their chest pain, including serial CK-MB and ECG analysis. The treating emergency physicians were not blinded to the ED CK-MB results. Admission decisions were left to the discretion of the treating physicians.

Patients with chest pain involved in this study were evaluated by emergency physicians as dictated by the standards of care at their institutions. Initial evaluation included a thorough history, physical examination, and a presenting ECG. If the patient was deemed eligible for the study based on the inclusion and exclusion criteria,

Data gathered at the time of presentation including patient age and race, time of chest pain onset, time of ED presentation, times of initial and two-hour CK-MB sampling, initial and two-hour CK-MB levels, initial ECG reading, admission diagnosis, and hospital disposition (CCU, telemetry, regular bed).

20

ACADEMIC EMERGENCY MEDICINE JAN/FEB 1994 VOL 1/NO 1

Inpatient Follow-up Inpatient records were reviewed by study personnel to determine final diagnoses. If MI was listed as a hospital discharge diagnosis, the methods of diagnosis (serial CK-MB levels, ECG changes, angiographic findings, or a combination of methods) were documented. Hospital discharge diagnoses were used to classify MI versus non-MI status. Patients who initially were ruled out for MI but subsequently developed MI were categorized as MI patients and included in the “recurrent or delayed in-hospital MI” complication group. Potentially life-threatening cardiovascular complications that occurred during the hospital stay were documented. Complications sought were cardiac-related death, recurrent or delayed in-hospital MI, significant ventricular arrhythmias (ventricular fibrillation, ventricular tachycardia, or premature ventricular contractions requiring therapy), new conduction defects (second- and thirddegree heart blocks), congestive heart failure (new or exacerbations of old heart failure), and cardiogenic shock. Outcomes were defined at the discretion of study personnel at each study site. Local study personnel were aware of the hypothesis of a parallel study (i.e., did the ED CK-MB predict MI in admitted patients), but were not aware of the hypothesis of the current study (i.e., did the ED CK-MB predict in-hospital complications). Patients whose inpatient records were incomplete were excluded from the study.

Data Analysis

CK-MB results into positive and negative CK-MB groups. In addition, the patients were classified as positive or negative for MI by discharge diagnoses. Rates of ischemic complications were calculated for MI versus non-MI groups as well as for positive CK-MB versus negative CK-MB groups. Complication rates were compared using Pearson’s chi-square test or Fisher’s exact test.27The relative.risk (RR) ratios of complications were calculated for patients with positive CKMB versus negative CK-MB results at ED presentation, regardless of final diagnosis. These RR results were further stratified with regard to time from chest pain onset until ED CK-MB determination to determine whether patients who presented later in the course of their chest pain were more at risk for complications. Ninety-five percent confidence intervals (CIS) were calculated for all RR ratios.28 A significance level of < 0.05 was used throughout.

As a complementary analysis, logistic regression analysis (stepwise regression) was used to develop models predicting either any complication or death using the following independent variables: CK-MB value (abnormal within two hours of ED presentation), patient age group (25-48 years or > 48 years), patient sex, availability of a hospital catheterization lab, hospital affiliation with a medical school, availability of open-heart surgery capability, and annual number of visits to the hospital ED (I 55,852 or > 55,852). Adjusted odds ratios (OR) and 95% confidence intervals (95% CI) for significant independent variables were determined.29

The patients were classified according to their ED

terns, paced rhythms, and left ventricular hypertrophy with strain also are believed to be associated with greater risk for complications.9-15 These abnormal ECG patterns generally have been grouped with diagnostic ST-segment elevation. Hence, many physicians admit to the CCU patients whose ECG findings are abnormal but nondiagnostic. One study suggested that patients with these other ECG abnormalities do not develop inpatient complications to the extent seen with acute ST-segment elevation.16 Many of these patients may not require monitoring in a CCU setting. Conversely, given that up to 50% of patients who have MI will present without diagnostic ST-segment elevation,17-’9 more objective criteria are needed to risk-stratify patients with chest pain for guiding monitored bed use. Recently, the measurement of initial and serial ED serum creatine phosphokinase-MB fraction (CK-MB) has been used to detect MI in the Using monoclonal antibody techniques, serum CK-MB levels obtained within

three hours of initial ED presentation have been shown to be 80%sensitive and 94% specific for MI in patients whose ECGs are without ST-segment elevation or new Q-waves.25 While serum CK-MB may be useful in the early diagnosis of MI, its usefulness in the prediction of inpatient ischemic cardiovascular complications has not been studied adequately. A recent study from two teaching hospitals found that serum CK-MB levels obtained at zero and three hours after ED presentation were predictive of inpatient complications in patients presenting with chest pain,26 but the number of patients involved in the study was small and the study setting was uniform. The purpose of this study was to assess the ability of ED serum CK-MB levels to predict inpatient ischemic cardiovascular complications in a multicenter trial of stable chest pain patients without diagnostic ECG ST-segment elevation on presentation.

~

CK-MB and Complications, Hoeksrra

el

21

al.

RESULTS

TABLE 2 Inpatient Ischemic Complications, Regardless of Final Hospital Diagnosis*

In the 30 months of the study period, from July 1990 until December 15, 1992, 5,583 patients were enrolled in the study. Of these, 463 patients were subsequently excluded because of presenting ECGs positive for MI, leaving 5,120 patients who were admitted to the hospital for evaluation of chest pain without diagnostic ECG ST-segment elevation.

Of the 369 patients who had MI by discharge diagnosis, 230 (62%)had positive ED CK-MB results, 138 (37%)had negative ED CK-MB results, and one was excluded due to a missing CK-MB result (Table 1). Of all the patients who had MI, 90 (24%)developed at least one complication in the hospital. Of the patients who had MI and positive ED CK-MBs, 61 (26%) had at least one complication, while 169 (74%)did not. Of the patients who had MI but negative ED CK-MBs, 29 (21%)developed at least one complication, while 109 (79%)did not. For patients who had MI, the

2:

................................................................................................................

No.

(S)

Positive ED CK-MB patients (n = 488) Cardiac-related death Congestive heart failure Recurrent or delayed MI New conduction defect Ventricular arrhythmias Cardiogenic shock

19 41 7 17 11 9

(3.9) (8.4) (1.4) (3.5) (2.3) (1.8)

Negative ED CK-MB patients ( n = 4575) Cardiac-related death Congestive heart failure Recurrent or delayed MI New conduction defect Ventricular arrhythmias Cardiogenic shock

7 20 3 8 14 4

(0.2) (0.4) (0.1) (0.2) (0.3) (0.1)

‘Several patients suffered more than one complication, thus accounting for the discrepancy between these numbers and the patient numbers listed in Table 3.

TABLE 1 Patient Classification Diagram

...........................................................................................................

complication rates of positive versus negative CK-MB groups were not significantly different (p = 0.23, Pearson’s chi-square test).

All patients ( n = 5,583) MI(n = 778)

t

ECG ST-segment elevation ( n

=

409) EXCLUDE

ECG no ST-segment elevation ( n

I

=

369)

CK-MB missing ( n = 1) EXCLUDE CK-MB positive (n = 230) Any complication (n = 61) No complication (n

=

169)

CK-MB-negative ( n = 138) Any complication ( n = 29)

No complication ( n = 109) NonMI(n = 4,805)

t

ECG ST-segment elevation (n

=

54) EXCLUDE

ECG no ST-segment elevation (n = 4,751) CK-MB missing ( n = 56) EXCLUDE CK-MB positive (n = 258) Any complication ( n No complication ( n

=

=

6)

Of the 4,751 patients without MI by standard inpatient evaluation, 258 (5%) had false-positive ED CK-MB results, 4,437 (94%)had negative ED CK-MB results, and 56 (1 %) were eliminated due to missing CK-MB values (Table 1). Of all the patients without MI, 17 (0.4%)had at least one complication and 4,678 did not. Of non-MI patients who had false positive CK-MB results, six (2%)developed at least one complication and 252 (98%)did not. Of non-MI patients whose CK-MB results were negative, 11 (0.2%) developed at least one complication, while 4,426 (99.8%) did not. For non-MI patients, the complication rates for positive versus negative CK-MB groups were significantly different (p < 0.0001, Fisher’s exact test). Of the patients whose ED CK-MB results were positive, regardless of the final hospital diagnosis, 67 of 488 (13.7%)had at least one complication during hospitalization (Table 2). This compares with 40 (0.9%)of 4,575 patients whose ED CK-MB results were negative. The RR ratio of any complication with a positive ED CK-MB compared to anegative ED CK-MB was 16.1 (95%CI 11.023.6). The RR ratio for each complication for positive ED CK-MB versus negative ED CK-MB groups are provided in Table 3 .

252)

CK-MB negative (n = 4,437) Any complication ( n = 11)

No complication ( n = 4,426)

Of the patients with positive ED CK-MB results, regardless of their final diagnoses, 19 of 488 (4%) died during hospitalization, versus eight of 4,575 (0.2%)with negative ED CK-MB results. The RR ratio of death was 25.4 (95%CI 10.8-60.2) in patients with positive ED CK-

22

ACADEMIC EMERGENCY MEDICINE JAN/FEB 1994 VOL 1/NO 1

TABLE 3 Unadjusred Relative Risk (RR) Ratios of Ischemic Complications of Positive ED CK-MB versus Negative ED CK-MB Patients, Regardless of Final Hospital Diagnosis ~~

RR Ratio

95% CI

Congestive heart failure Recurrent of delayed MI New conduction defect Ventricular arrhythmias Cardiogenic shock Cardiac-related death

19.2 21.9 19.9 7.4 21.1 25.4

11.4-32.5 5.7-84.3 8.6-45.9 3.4-16.1 6.5-68.2 10.8-60.2

Any adverse outcome

16.1

11.0-23.6

Complication

MB versus those with negative ED CK-MB (Table 3). When the RR ratios of any complication and of death for positive ED CK-MB versus negative CK-MB groups are stratified by times from chest pain onset to the determination of CK-MB levels, the results do not vary significantly with time (Figs. 1 and 2). In the first multivariate model, predicting any ischemic complication, the following variables were statistically significant: CK-MB elevation (p = O.OOOl), hospital’s number of annual ED visits (p = 0.003), and patient’s age group (p = 0.0002). Because patient’s sex approached statistical significance (p = 0.11) and had face validity in the model, this variable was retained in the final model. The results for this model are shown in Table 4. In the second multivariate model, predicting death, CK-MB elevation (p = O.OOOl), hospital’s number of annual ED visits (p = 0.018), and hospital affiliation with a medical school (p = 0.012) were statistically significant. For consistency, patient’s age group (p = 0.08) and sex (p = 0.16) were retained in the final model. The results for this model also are shown in Table 4. Further use of these models demonstrated that the adjusted risk associated with a CK-MB elevation for cardiac-related death or any complication was independent of the presence of MI.

predictive of complications13 and new ST changes may indicate a higher-risk patient,30 these findings are not helpful for the majority of ED chest pain patients. The emergency physician commonly has few or no objective data to support admission-disposition decisions under these circumstances. Our data support a prior study26reporting that ED CKMB levels provide objective information for the prediction of ischemic complications in stable chest pain patients. The RRs of ischemic complications and cardiac death are significantly higher when at least one of the ED CK-MB levels is elevated. The majority of the ED CK-MB’s predictive value is due to its ability to identify MI in stable chest pain patients. Of the 107 patients who developed any complication in this study, 90 (84%)were diagnosed as having MI. Sixty-one (68%)of these patients were identified as having MI by an elevated CK-MB in the emergency department. This sensitivity estimate may be lower than those of earlier studies because presenting and two-hour levels rather than presenting and three-hour levels were used. 20*25 Patients who have MI are not, however, the only chest pain patients who are identified as being at risk for complications based on a positive ED CK-MB. In our non-MI group, the incidence of complications was higher with a positive ED CK-MB than with a negative CK-MB (p < 0.0001). The reason for this observation cannot be discerned from our data, but it may, in part, reflect failure to recognize MI with older non-immunochemical CK-MB assays. In this study, the final diagnosis of MI or non-MI was determined by the attending physician, not by strict definitions of CK-MB results or ECG changes. Not all patients whose CK-MB levels are elevated receive the final diagnosis of MI.20-31Earlier multicenter data suggest that an elevated CK-MB can be a marker for other disease states such as unstable angina and renal insufficiency (personal

L TABLE 4 Adjusted Odds Ratios for Multivariate Models: Any Complication of and Death of Positive ED CK-MB versus Negative ED CK-MB Patients, Regardless of Hospital Diagnosis

........,,...,.....,... ........,........,. ...... ...,.,..,,,..................,..,,,........................... ,

DISCUSSION

Adjusted

Variable

When faced with a stable chest pain patient who is presumed or suspected to have MI, the emergency physician is obligated to admit that patient to the most appropriate monitored bed to evaluate for MI or acute coronary insufficiency and to optimize care. The clinician must predict not only whether or not the patient has had MI or noninfarction myocardial ischemia, but whether or not that patient will develop an ischemic complication. Prior studies9-16 have addressed the use of the initial ECG as an objective tool to predict inpatient complications in chest pain patients. While acute ST-segment elevation has been shown to be

~~

Odds Ratio ~

Any complication CK-MB elevation Sex male ED annual visits 48 years Death CK-MB elevation Sex male ED annual visits 48 years Medical school affiliation- none

95% C1 for Odds Ratio ~~

18.2 1.4 2.8 6.8

12.0-27.7 0.9-2.1 1.4-5.4 2.5-18.7

27.3 1.8 11.6 6.1

11.3-65.8 0.8-4.1 1.5-88.3

0.3

0.1-0.8

0.8-45.6

~

~~~

~

~

~

~~

23

CK-MB and Complications, Hoekstru et al.

-

40

-

35

I -

30 FIGURE 1. Relative risk ratios of any complication in positive ED CK-MB patients versus negative ED CK-MB patients stratified by time intervals from chest pain onset until ED CK-MB obtained. (95%CI denoted by vertical bars)

-I

I

I

25 20 15 10 5

01 0-3 HR

I

3-6 HR

I

6-9 HR

I >9 HR

communication, W. B. Gibler). These disease states may be associated with higher incidences of ischemic complications. In a study by Ravkilde and associates,32 a subgroup of chest pain patients without MI who had elevated CK-MB levels during their hospital stays had a one-year mortality rate equal to that of patients who had MI.

The ability of the CK-MB to risk-stratify for ischemic complications remained significant after adjustment for ischemic complications using limited patient (age group and sex) and site-of-practice parameters. We also plotted the RR ratios of any complication and death with positive ED CK-MB versus times from chest pain onset to ED CKMB. We assumed, as prim studies have s h o ~ n , 3 3 , 3that ~ the risk of complications would increase as the time from chest pain onset to ED CK-MB determination increased. We did not find a significant difference in the risks of death or other complications with time. Two possible explanations exist. First, patients with large MIS are often identified earlier by CK-MB than are patients with small infarcts whose CK-MB levels may not be elevated until hours later. This phenomenon may place more complication-prone MI patients in the shorter time intervals. In addition, patients who stay away from the hospital longer may develop complications or ECG ST-segment elevations prior to arrival at the ED. This would eliminate them from the study population and reduce the proportion of study patients presenting late with complications.

While the predictive value of positive a ED CK-MB may be helpful in the disposition of the stable patient with chest pain to a CCU setting, the converse may not be true for negative ED CK-MB. Of the 4,575 patients with negative ED CK-MB results, 138 (3%) developed MI and 40 (0.9%)developed at least one complication. While these percentages appear small, one-third of the MIS and 37% of the complications in this study were not identified by an ED CK-MB. If these patients had been assumed to be at low risk for complications and placed in non-monitored beds, their morbidity and mortality might have been greater. Prior studies suggest that such patients do well in stepdown units or telemetry beds,6 but using CK-MB results to admit them to non-monifored bed settings would not be warranted based on the results of this study.

There are a number of limitations of this study that should be addressed. First, the definitions used in the study are subject to individual interpretation at the 53 study sites. Determinations of MI, significant ventricular arrhythmias, and exacerbation of congestive heart failure were at the discretion of the attending physicians. In general, complications that required treatment or caused morbidity or mortality were considered significant and documented. We accepted the standard of care and methods of diagnosis used at each study center. Second, because the patient charts were reviewed to determine complications, undocumented complications may have been missed in data gathering. Those patients with missing diagnoses, missing CKMB results, and incomplete data were eliminated from the study.

70 60

50 40 30 20 10 0 0-3 HR

3-6 HR

6-9HR

>9 HR

FIGURE 2. Relative risk ratios of death in positive ED CK-MB patients versus negative ED CK-MB patients stratified by time intervals from chest pain onset until ED CK-MB obtained. (95% CI denoted by vertical bars)

24

ACADEMIC EMERGENCY MEDICINE JAN/FEB 1994 VOL 1/NO 1

While ED CK-MB may predict increased risk for ischemic complications, prolonged observation and serial CKMB testing for periods longer than two hours are needed at this point to identify not only MI, but also the full risk for complications. Continuous ST-segment monitoring to identify ongoing ischemia, short-stay protocols to diagnose MI ,35-38 and exercise cardiac imaging with echocardiography or nuclear medicine scanning deserve further study to help in the identification and risk stratification of this difficult patient population.

FUTURE QUESTIONS Future investigations should address the relative role of ED CK-MB elevations in the setting of ECG abnormalities other than diagnostic ST-segment elevation and in the settings of other historical and clinical features. For example, does the presence of ST-segmentdepression or new crackles on examination predict ischemic complications better than CK-MB elevation? Is CK-MB elevation more predictive of ischemic complications in the setting of prior coronary artery disease? This study and others20.26JIJ2 have identified patients with elevated CK-MB levels who are considered not to have had MI. This patient population requires further analysis to ascertain the basis of the CK-MB elevations and their implications. Do these patients have subclinical infarctions? Do these patients have greater than normal proportions of CK-MB in cardiac muscle such that normal muscle breakdown is associated with CK-MB elevations? This study did not attempt to determine the costeffectiveness of CK-MB use. In this study, approximately 1.2 patients who developed complications were identified for every 100 patients tested. Despite this testing, 37% of patients who developed in-hospital complications were not identified with ED CK-MBs. The potential financial impact of using CK-MB results to guide in-hospital disposition decisions should be assessed. This study examined ischemic complications occurring at any time during hospitalization. We did not evaluate discharged patients for delayed complications. Further, we did not assess which patients developed complications within the first 48-72 hours. The study of Hedges et a1.26 suggests that considerably fewer complications will occur in the first 48 hours, and hence the yield of the CK-MB for ischemic complications during that period (when the CCU disposition decision is most crucial) also would be expected to be less. Future multicenter analyses should assess the relative risk of ischemic complications with an elevated CK-MB in the first 48 hours.

CONCLUSION We conclude that serum CK-MB values, measured immediately and two hours after initial ED presentation, are

predictive of in-hospital ischemic complications. Future investigations should address whether CK-MB levels can be used prospectively to aid in-hospital disposition of patients who have chest pain without diagnostic ECG STsegment elevation. i REFERENCES 1. Lee TH, Goldman L. The coronary care unit turns 25: historical

trends and future directions. Ann Intern Med. 1988; 108:887-94.

2. Wenger NK. Hellerstein HK, Blackburn H, Castronova SJ. Physician practice in the management of patients with uncomplicated infarction: changes in the past decade. Circulation. 1982; 65:421-7. 3. Bloom BS. Peterson OL. End results, cost, and productivity of coronary care units. N Engl J Med. 1973; 288:72-8. 4. Fuchs R, Scheidt S. Improved criteria for admission to cardiac care units. N Engl J Med. 1981; 246:203741.

5. Goldman L, Cook EE Brand DA, et al. A computer program to predict myocardial infarction in emergency department patients with chest pain. N Engl J Med. 1988; 318:797-803. 6. Fiebach NH, Cook EE Lee TH, et al. Outcomes in patients with myocardial infarction who are initially admitted to stepdown units: data from the multicenter chest pain study. Am J Med. 1990; 89:1520. 7. Lee TH, Cook EF. Weisberg M, Sargent RK, Wilson C , Goldman L. Acute chest pain in the emergency room: identification and examination of low-risk patients. Arch Intern Med. 1985; 145:65-9. 8. Fineberg HV, Scadden D, Goldman L. Care of patients with a low probability of acute myocardial infarction: cost-effectiveness of alternatives to coronary care unit admission. N Engl J Med. 1984; 310:1307-7. 9. Brush JE, Brand DA. Acampora D, et al. Use of the initial electrocardiogram to predict inhospital complications of acute myocardial infarction. N Engl J Med. 1985; 312:1137-41. 10. Stark ME, Vacek JL. The initial electrocardiogram during admission for acute myocardial infarction: use as a predictor of clinical course and facility utilization. Arch Intern Med. 1987; 147:843-6. 11. Slater DK, Hlartky MA, Mark DB,et al. Outcome in suspected acute

myocardial infarction with normal or minimally abnormal electrocardiographic findings. Am J Cardiol. 1987; 60:766-70. 12. Zalenski RJ, Sloan EP, Chen EH, et al. The emergency department ECG and immediately life-threatening complications in initially uncomplicated suspected myocardial ischemia. Ann Emerg Med. 1988; 17:221-6. 13. Fesmire FM, Percy RE Wears RL, et at. Risk stratification according to the initial ECG in patients with suspected acute myocardial infarction. Arch Intern Med. 1989; 149:1294-7. 14. Bell MR, Montarello JK, Steele PM. Does the emergency room electrocardiogram identify patients with suspected myocardial infarction who are at low risk of acute complications? Aust NZ J Med. 1990; 20~564-9. 15. Behar S, Schor S , Kariv I, Barell V, Modan B. Evaluation of the

electrocardiogram in the emergency room as a decision making tool. Chest. 1977; 71:486-91. 16. Yusuf S, Pearson M, Sterry H, et al. The entry ECG in the early diagnosis and prognostic stratification of patients with suspected

acute myocardial infarction. Eur Heart J. 1984; 5:690-6. 17. Short D. The earliest electrocardiographic evidence of acute myocardial infarction. Br Heart J. 1970; 32:6-15. 18. McQueen MJ. Holder D. El-Maraghi NRH. Assessment of the

25

CK-MB and Complications, Hoekstra et al.

accuracy of serial electrocardiograms in the diagnosis of myocardial infarction. Am Heart J. 1983; 105:258-61.

Epidemiologic Research: Principles and Quantitative Methods. New York: Van Nostrand Reinhold, 1982: 419-46.

19. Lee TH, Goldman L. Serum enzyme essays in the diagnosis of acute myocardial infarction. Ann Intern Med. 1986; 105:221-33.

30. Fesmire FM, Percy RE Wears RL, et al. Diagnostic and prognostic importance of comparing the initial to the previous electrocardiogram in patients admitted for suspected acute myocardial infarction. South Med J. 1991; 84:841-6.

20. Gibler WB, Lewis LM, Erb RE, et al. Early detection of acute myocardial infarction in patients presenting with chest pain and nondiagnostic ECGs: serial CKMB sampling in the emergency department. Ann Emerg Med. 1990; 19:1359-66. 21. Nowakowski JR. Use of cardiac enzymes in the evaluation of acute chest pain. Ann Emerg Med. 1986; 15:354-60. 22. Marin MM, Teishman SL. Use of rapid serial sampling of creatine kinase MB for very early detection of myocardial infarction in patients with acute chest pain. Am Heart J. 1992; 123:354-61. 23. Lee TH, Weisberg W, Cook EE et al. Evaluation of creatine kinase and creatine kinase-MB in diagnosing myocardial infarction: clinical impact in the emergency room. Arch Intern Med. 1987; 147:115-21. 24. Green GB, Hansen KN, Fleetwood D, et al. The potential utility of rapid CKMB assay in evaluating emergency department patients with possible myocardial infarction. Ann Emerg Med. 1991; 20:954-60. 25. Gibler WB, Young GR Hedges JR, et al. Acute myocardial infarction in chest pain patients with nondiagnostic ECGs: serial CKMB sampling in the emergency department. Ann Emerg Med. 1992; 21504-12. 26. Hedges JR, Young GP, Henkel GR, Gibler WB, Green TR, Swanson JR. Early CK-MB elevations predict ischemic events in stable chest pain patients. Acad Emerg Med. 1993; 1:9-16. 27. Systat Version 5.0, Systat, Inc. Evanston, Illinois.

3 1. Hedges JR, Young GP, Henkel GF, et al. Serial ECGs are less accurate

than serial CKMB results for the ED diagnosis of myocardial infarction. Ann Emerg Med. 1992; 21:1445-50. 32. Ravkilde J, Hansen AB, Horder M, Jorgenson PJ, Thygesen K. Risk stratification in suspected acute myocardial infarction based on a sensitive immunoassay for serum creatine kinase isoenzyme MB. Cardiology. 1992; 80:143-51. 33. Karlson BW, Herlitz J, Liljekvist JA, et al. Prognosis in unsuspected acute myocardial infarction in relation to delay time between onset of symptoms and arrival in hospital. Cardiology. 1991; 78:131-7. 34. Turi ZT, Stone PH, Muller JE, et al. Implications for acute intervention related to time of hospital arrival in acute myocardial infarction. Am J Cardiol. 1986; 58:203-9. 35. Quale J, Kimmelsteil C , Lipschik G , Schrem S. Use of sequential cardiac enzyme analysis in stratification of risk for myocardial infarction in patients with unstable angina. Arch Intern Med. 1988; 148:1277-9. 36. Gaspoz JM, Lee TH. Cook EF, Weisberg M, Goldman L. Outcome of patients who were admitted to a new short-stay unit to “rule out” myocardial infarction. Am J Cardiol. 1991; 68:145-9. 37. Lee TH, Juarez G , Cook EF,et al. Ruling out myocardial infarction: a prospective multicenter validation of a 12-hour strategy for patients at low risk. N Engl J Med. 1991; 324:1239-46.

28. Morris JA, Gardner MJ. Calculating confidence intervals for relative risks, odds ratios, and standardized ratios and rates. In Gardner and Altman (eds). Statistics with confidence: confidence intervals and statistical guidelines. Br Med J. 1989: 50-63.

38. Gibler WB, Walsh RA, Levy RC, Runyon JE! Rapid diagnostic and treatment center in the emergency department for patients with chest pain (abstr). Circulation. 1992; 86(Suppl 1):I-15.

29. Kleinbaum DG, Kupper LL, Morgenstern H. Modeling: theoretical considerations. In Kleinbaum DG, Kupper LL. Morgenstern H (eds).

........... ...... ...... . . .. ..... ... .......... .., ...........,. ........ ...,....,,,....,..............,, ...,..... .., ,

APPENDIX A

CK-MB Coordinators at the 53 Institutions That Participated in This Phase of the Study Hospital-Study Site

Physician Coordinator

Project Coordinator

Albany Memorial Hospital Albert Einstein Medical Center Butterworth Hospital Charleston Area Medical Center Community Hospital East Cox Medical Center-North Cox Medical Center-South Crittenton Hospital Eden Hospital Exeter Hospital Good Shepherd Medical Center HCA Rio Grande Regional Hospital Heartland Hospitals Huguley Hospital Humana Hospital -Audubon Humana Hospital-Brandon Humana Hospital University of Louisville Ingham Medical Center Iowa Lutheran Hospital

George M. Innes, Jr., MD Antionette Mangione, MD Jeffrey Jones, MD Jack Tolliver, M D Don B. Ziperman, MD Robert Beaton, MD Robert Beaton, MD Dr. R. Prasad David Anderson, MD Dr. Mark Josephs Dr. John Campbell Norman Rarnirez, MD York Silliman, MD Joseph LaBorde, MD Roy Gerrnano, MD; Edmond Hooker, MD R. Khant, MD Edmond A. Hooker, MD Earl Reisdorff, MD Larry J. Baker, DO

George M. Innes, Jr., MD Antionette Mangione, MD Torrey Nash, MD Ryan Henry, RN Ruth A. Giebel, MS, RN Cynthia Brown Cynthia Brown Joann Rivers, NM-CCU Jodi Akin Mary Palmer Janis Ashcraft Domingo H. Useda, MD B. Keller, MT (ASCP) Karen Cameron, RN Penny S . Schafhein, RN Susie Maslar, RN Peg Lansing Kevin Cavanagh, PhD Barb Yeazel, RN

26

ACADEMIC EMERGENCY MEDICINE JAN/FEB 1994 VOL 1/NO 1

~

-~

Hospital-Study Site

Physician Coordinator

Project Coordinator

Lenox Hill Hospital Louisihna State University Medical Center McKeesport Hospital Meadowlands Hospital and Medical Center Memorial Hospital of Laramie County Memorial Northwest Hospital Mercy Hospital Mercy Regional Medical Center Monongalia General Hospital Mt. Sinai Medical Center Mountainside Hospital Munroe Regional Medical Center Nashua Memorial Hospital Nazareth Hospital New Britain General Hospital Providence Hospital Providence Hospital Rhode Island Hospital St. Anthony’s Hospital Central St. Elizabeth Hospital Saint Joseph’s Hospital St. Joseph’s Hospital St. Louis University Hospital Sharon Regional Health System Shelby Medical Center South Louisiana Medical Center Spohn Hospital Sun Coast Hospital Taylor Hospital Tulsa Regional Medical Center Virginia Beach General Hospital Western Baptist Hospital Wetzel County Hospital York Hospital Emergency Department

Michael Marin, MD Thomas Arnold, MD Michael Ginsburg, MD Alvin Goldberg Dr. Ronald J. McKee Salah El Hafi, MD Mark Meengs , MD Dr. Arturo Garza-Gongora Robert Curtis, MD Dr. Arthur Diskin William Indruk, MD Donald Crowe, MD James High, MD William J. Warren, MD Louis Graff, MD Margaret Barron, MD Robert Swetnam, DO Robert Woolard, MD W. Peter Vellman, MD Dr. Paul Jordan Dr. William Knopf Robert Moravec, MD Lawrence Lewis, MD Dr. Sergio Segarra Dr. Thomas 0. Paul Thomas G. Ferguson, MD A.G. Wright, MD Paul Kudelko, DO Damian Cornacchia, DO Ernest Pickering, DO Dr. W. Andrew Dickinson Dr. Patrick J. Withrow David Hines, MD David R. Eitel, MD

Cindy Felski Sheila R. Dunn Rauf Chaudry, MD Cathy Egan, RN Leann Cooper Isabel Proskt5 Dennis Helmus Vicki Perales Martha Meyer Dana Sherman, ARNP William Indruk, MD Lou Pawlikowski Estelle MacPhail, RN, MS Dorothy E. Devlin Kathy Wnuk, RN Margaret Barron, MD Pat Howell Emily Connoly Ann Knightly Nancy Berger Jan Shaftel Judy Cusick, RN Laura Lasater, MS Donna Jovenall Myra S. Muse Bobby Denson Barbara Simon Pat Lucas, RN Harriett Gaudiosi Phyllis Cotham, BSN, RN Melody Bostian, RN Debbie Batts Taffer Dan Butler, MT, SC Cynthia S . Ogden

Further Thoughts from the Reviewers The data in the current study and in the companion paper by Hedges et a1.Z6 suggest that CK-MB levels obtained in the ED may help the clinician stratify “stable” chest pain patients with respect to the potential for ischemic complications. Because the authors of this large multicenter study did not collect data that characterized each patient’s cardiac history, examination, and ECG findings (aside from excluding patients with ST-segment elevation o r those judged by the treating physicians to be clinically “unstable”), it is not possible to determine the additive value of the CK-MB to clinical tools currently available. Further, clinical evaluations at the various sites may not be uniform, hence it may be preferable in future investigations to compare the CK-MB against other instruments used to risk-stratify patients for potential ischemia. Future investigations should compare the CK-MB against the ECG criteria of Brush and colleagues,9 the al-

gorithm of Goldman and associates,5 and the “acute cardiac ischemia time-insensitive predictive instrument” of Selker and coworkers (Med Care 1991; 29:610-27). While these instruments have all undergone prospective validation, comparisons between instruments and against CKMB or other cardiac enzyme markers have been limited. The study by Hoekstra and associates used the medical-record diagnoses for confirmation of both the diagnosis of MI and ischemic complications. More rigorous study definitions for ischemic complications and estimation of inter-rater reliability would strengthen this analysis. Further, because this study focused o n patients with the complaint of chest pain, the value of CK-MB elevations for risk stratification in patients who present with other anginal-equivalent syndromes is unknown. Finally, this study did not actually measure changes in

CK-MB and Complications, Hoekstra et at.

in-hospital disposition as a result of the CK-MB availability. While the data suggest that the CK-MB may be of some value for risk stratification for complications during hospitalization, this study does not tell us whether CK-MBs add to clinical evaluation and ECG analysis. Further, we do not know whether CK-MBs will help us identify those “stable” patients who will develop complications in the first 48 hours after presentation (the time interval during which ED

27

disposition to the CCU should be valuable). We do know that approximately 63% of the patients in this study who developed ischemic complications during hospitalization had an elevated CK-MB in the ED. How or whether the remaining 37% of such patients can be identified more accurately in the ED and whether such information will affect clinical decision making remain unclear.