- Email: [email protected]
Routine Admission Electrocardiography in Emergency Department Patients
ORIGINAL CONTRIBUTION
Routine Admission Electrocardiography in Emergency Department Patients
From the Division of Pulmonary, Allergy, and Critical Care Medicine, Department of Medicine; and the Division of Emergency Medicine, University of Pittsburgh, Pittsburgh, Pennsylvania.
Jeffrey L Garland, MD Allan B Wolfson, MD
Receivedfor publication January 28, 1993. Revision received June 7, 1993. Acceptedfor publication June 17, 1993. Presented in part at the Annual Meeting of the Pennsylvania Chapter of the American College of Physicians in Pittsburgh, October 1991; and presented at the Societyfor Academic Emergency Medicine Annual Meeting in Toronto, Ontario, Canada, May 1992.
Study objectives: To determine whether routine performance of an ECGcould have been safely avoided in a subset of emergency department patients admitted to a university hospital. Design: Retrospective consecutive case series. Setting: University teaching hospital. Type of participants: All ED patients admitted to the medical service of the study hospital during a three-month period. Methods and interventions: Acceptable indications for an admission ECGwere prospectively developed. Charts of all patients were reviewed to determine whether any of these indications were present, whether an admission ECGwas performed, and whether an admission ECG resulted in a change in patient management or outcome. An ECGwas classified as routine when performed in the absence of documentation of any of these indications. No interventions were performed.
Measurements and main results: There were 636 ED admissions to the medical service during the study period. Of the 631 patients whose chart could be located, 384 (61%) had at least one indication for an ECGand all but one had an ECG performed. No indications for an admission ECGwere identified in the remaining 247 patients; of these, 202 (82%) had an ECG performed and 45 (18%) did not. Among the 202 who had a routine admission EC6,the test resulted in a change in management in only three (1.5%) (95% confidence interval [Oil, 0.3% to 4.3%) and affected patient outcome in none (95% Cl, 0% to 1.5%). Among the 45 without indications who did nothavea routine admission ECG,none experienced an identifiable adverse consequence during hospitalization (95% CI, 0% to 6.7%). Conclusion: The admission ECGcould have been avoided in an identifiable subset of ED patients admitted to the medical service of our hospital, with no adverse effect on patient outcome. This finding, if corroborated in other patient populations, suggests the potential for significant cost savings for the US health care system as a whole.
FEBRUARY 1994 23:2 ANNALS OF EMERGENCY MEDICINE
275
ROUTINE ECG
Garland& Wolfson
[Garland JL, Wolfson AB: Routine admission electrocardiography in emergencydepartment patients. Ann EmergMed February 1994;23:275-280.]
goal was to estimate the potential financial savings that might be realized by the health care system were such criteria to be applied successfully to a broader patient population.
INTRODUCTION
MATER ALS AND METHODS
Although it has long been standard practice in many institutions to order a battery of routine diagnostic tests when patients are admitted to the hospital, the diagnostic yield and cost-effectiveness of routine admission testing have been investigated to only a limited extent. Among those tests that have been studied, the routine admission chest radiograph has received the closest attention. Several investigators have shown that a chest radiograph need not be ordered routinely on all patients and that the decision to order this examination can be guided appropriately by such criteria as the patient's age and clinical presentation. 1-4 In contrast, only one published study has examined in detail the usefulness of the routine admission ECG. Moorman et al, investigating all admissions to the medical service, found that a routine admission ECG resulted in a change in outcome in only 0.3% of patients with neither a history nor symptoms or signs of a cardiac abnormality. 5 The present study focused on emergency department patients admitted to the medical service of a tertiary-care university teaching hospital. The goal was to develop a set of criteria that might be used to identify those individuals for whom an admission ECG would alter neither hospital management nor ultimate outcome. A second
All patients admitted from the ED to the medical service of the Presbyterian University Hospital of Pittsburgh from July 1 through September 30, 1989, were eligible for inclusion in the study. The decision to order or not to order an admission ECG was made by ED staff. Eligible patients were identified from the medical admission list compiled by the medical admitting resident in the ED. For patients admitted more than once during the study period, each admission was considered a separate "patient." Hospital admission records were retrieved when available and reviewed by one of the investigators. For patients whose charts could not be located, a further search was conducted to determine whether these individuals' names appeared in the files of the hospital's risk management office or in records of claims made against the hospital. The following were proposed prospectively as appropriate indications for obtaining an admission ECG: a history of coronary artery disease, arrhythmia, congestive heart failure, conduction disturbance, cor pulmonale, pericarditis, or cardiomyopathy; palpitations, syncope,
Table 1.
Figure.
Performance of admission ECG and clinical outcome in 636 consecutive patients admitted to the medical service from the ED
Proposed criteria for obtaining an admission ECG 636(Admissions)
History of: Coronary artery disease Arrhythmia Congestiveheart failure Conduction disturbance Cor pulmonale
Pericerditis Cardiomyopathy
Palpitations Syncope or coma Symptom complex suggestiveof angina or congestive heart failure Suspected cardiotoxic overdose Irregular pulse Pulse >120 or <60 Systolic blood pressure >200 or <90 Diastolic blood pressure >120 Serum potassium>5.7 or < 3.0
276
631(Chartfound)
5 (Chartnotfound)
247(Notindicated)
384(indicated)
,/
/',, 383(Indicative, done)
1(Indicative, notdone)
,,,
202 (Nat indicative, done)
,/
199 (No change in managementl
45 {Notindicative, not donel
',, ,,/ ",, 3 (Changein management}
3 (Nochangein outcome)
ANNALS OF EMERGENCY MEDICINE
0 (Nochangein outcome)
23:2
FEBRUARY 1994
ROUTINE ECG
Garland & Wolfson
coma, or symptom complex suggestive of angina or congestive heart failure; suspected cardiotoxic drug overdose; irregular pulse or pulse less than 60 or more than 120; systolic blood pressure more than 200 or less than 90; diastolic blood pressure more than 120; or serum potassium less than 3.0 or more than 5.7 (Table 1). Age per se was not an indication. With these criteria, each admission ECG that was performed was classified as either "indicated" or, if the patient did not have any of the defined set of indications, "routine." For patients who had a routine admission ECG performed, the chart was reviewed further to determine whether performance of the test had resulted in any identifiable change in patient management or outcome during the hospitalization. An ECG was determined to have caused a change in management if its performance resuited in any diagnostic or therapeutic intervention that would not otherwise have been performed. A change in outcome was defined as an alteration in the patient's condition that would not have occurred if the ECG had not been performed. For patients who did not have an admission ECG done, the chart was reviewed to determine whether the omission of the test had resulted in any identifiable adverse consequence.
Table 2.
Principal admission diagnosis in 247 patients without any of the indications in Table i No. of Patients Pneumonia Urinary tract infection AIDS Other source of fever Complication of malignancy COPD/asthma PE/DVT/phlebitis Gastrointestinal bleed/gastritis Liver disease Abdominal pain Cerebrovascular accident/transient ischemic accident Seizure Diabetes mellitus Sickle cell anemia Other anemia Rheumatologic disorder Failure to thrive Miscellaneous Total
FEBRUARY 1994 23:2 ANNALS OF EMERGENCY MEDICINE
13 15 16 47 23 10 5 14 9 7 5 6 4 20 5 6 6 36 247
RESULTS ED records listed 636 admissions to the medical service of Presbyterian University Hospital of Pittsburgh during the study period. There were five patients whose names appeared on the medical admission list but whose charts could not be located for review. No record of these individuals' cases could be found in the files of the hospital's risk management office or in lists of claims made against the hospital. Of the 631 patients on whom records were available, 384 (61%) had at least one indication for an ECG and all but one had an admission ECG performed. No indications for an admission ECG were identified in the remaining 247 patients; of these, 202 (82%) had an admission ECG performed and 45 (18%) did not (Figure). Among the 202 patients who had a routine admission ECG performed despite the absence of any of the proposed indications for the test, the admission ECG resulted in a change in management in only three (1.5%) (95% confidence interval [CI], 0.3% to 4.3%) and a change in outcome in none (95% CI, 0% to 1.5%). Of the 45 patients without indications who did not have a routine admission ECG, none (95% CI, 0% to 6.7%) experienced an identifiable adverse consequence during the course of their hospitalization. Thus, of the 247 patients without indications for an admission ECG, none experienced an identifiable change in outcome (95% CI, 0% to 1.2%). The principal admission diagnoses in these patients are shown in Table 2. The cases of the three patients for whom the admission ECG resulted in a change in management clearly merit careful review. The details of these cases are presented here and summarized in Table 3. CASE REPORTS Case 1 A 46-year-old man with a history of IV drug abuse presented to the ED with pleuritic chest pain, dyspnea, and fever. He was admitted because of the possibility of endocarditis and pulmonary embolism. The admission ECG revealed inverted T-waves in leads V 1 and V2. He was admitted to a nonmonitored bed, and the ECG was repeated three times and showed no change. No further cardiac evaluation was performed, and no adverse cardiac events were noted during the hospitalization. Case 2 A 50-year-old woman was admitted because of upper gastrointestinal bleeding. Admission ECG revealed inverted T-waves in leads V~ through V6, and the patient
277
ROUTINE ECG
Garland & Wolfson
was admitted to a monitored bed. Further ECGs showed no change. After acute myocardial infarction was ruled out, no further cardiac evaluation was performed. No adverse cardiac events were noted during the hospitalization. Case 3 A 76-year-old woman with chronic obstructive pulmonary disease and a history of laryngectomy for laryngeal cancer was admitted because of increasing dyspnea and purulent sputum. Admission ECG revealed inverted T-waves in leads V 1 through V6, as well as left ventricular hypertrophy. She was admitted to a monitored bed, where further ECGs showed no change and acute myocardial infarction was ruled out. No further cardiac evaluation was performed, and there were no adverse cardiac events during the hospital stay. Although the routine admission ECG resulted in a change in management in these three cases, patient outcome was not affected. DISCUSSION
We have shown that routine admission ECG could have been omitted safely in an identifiable subset of ED patients admitted to the general medical service of a tertiary-care university hospital, with minimal adverse effect on patient outcome. Although there have been a number of studies of the usefulness of the routine preoperative ECG, 6-8 only one specifically has addressed the value of the routine admission ECG in medical patients. Moorman et al prospectively studied 1,410 patients admitted to the general medical service .of a university hospital through the ED or outpatient clinics or by transfer from other hospitals; patients admitted to an intensive care setting were excluded. 5 Of the 1,410 admitted patients, 775 had no "cardiac abnormality" or other apparent indications for an admission ECG. Performance of an ECG in these 775 individuals resulted in a change in management in 26 (4.1%) and a change in management that was judged
likely to have improved outcome in only two (0.3%). It was noted that the ECG was helpful more often in patients more than 45 years old, but this association was of only borderline statistical significance when considered independently of the presence of identified cardiac abnormalities. In the present study, the subjects were all admitted to the medical service from the ED (patients admitted to an intensive care setting were included), whereas those admitted from outpatient clinics, electively, or by transfer from other institutions were not. Thus, the present study focused on decision making in ED patients who tended to be acutely ill and in whom the decision of whether to order an admission ECG had to be made in the urgent environment of a busy ED. To better reflect these decisionmaking issues, the acceptable indications for admission ECG were broader than those used by Moorman et al. Our criteria for performing an admission ECG did not include patient age per se. Although the three patients whose management was altered by a routine admission ECG were all more than 45 years old, there was no change in outcome for these individuals; thus, adding age to our list of indications would not have resulted in an identifiable benefit for any patient. One argument that has been advanced in defense of routine ECG testing is that it provides a baseline tracing that is likely to be useful in making management decisions in the future. Fesmire et al found that comparison of the ED ECG with a previous ECG provided important prognostic information for patients admitted with suspected myocardial infarction. 9 Lee et al found that the availability of a prior ECG appeared to be of benefit in avoiding inappropriate hospital and cardiac care unit admission for patients who presented with chest pain but who were found in retrospect not to have sustained an acute myocardial infarction, to In neither of these studies, however, is it clear that the prior ECGs were obtained
Table 3.
Clinical features in three patients for whom routine admission ECG resulted in a change in management
Patient
Age/Sex
1
46/M
2
50/F
3
76/F
278
Chief Complaint
Change in Management
ECG Finding
Change in Outcome
Pleuritic chest pain, fever, shortness of breath Upper gastrointestinal bleed
Inverted T-wave (V1-V2)
ECG × 3
None
Inverted T-wave (VcV s)
None
Chronic obstructive pulmonary disease, shortness of breath, sputum
Left ventricular hypertrophy, inverted T-wave (V~-Vs)
Monitor, EC6 x 3, enzyme x 3 Monitor, EC8 x 3, enzyme x 3
None
ANNALS OF EMERGENCYMEDICINE 23:2 FEBRUARY1994
ROUTINE ECG Garland & Wolfson
"routinely" or that they resulted in a change in patient outcome. More recently, Ziemba et al concluded that a baseline ECG was of benefit in making management decisions in elderly patients with a past history of heart disease who were being evaluated in the ED. 11 This conclusion does not conflict with the findings of the present study, in which these patients would have been considered to have indications for an admission ECG. The usefulness of a baseline ECG was addressed specifically by Rubenstein and Greenfield, who reviewed the records of 236 patients presenting to two EDs with a chief complaint of chest pain. 12 The authors concluded that although the availability of a baseline ECG might have prevented the unnecessary hospitalization of 11 of the 236 patients, in no instance would its absence have resuited in an adverse outcome. Further evidence against the value of a baseline ECG is provided by Hoffman and Igarashi, who in a study of 84 ED patients found that physicians never changed their minds in deciding whether to admit or discharge a patient with chest pain when the decision was made first with, and then without, a baseline ECG. 13 Moorman et al reached a similar conclusion, reporting that "in no instance was an admission ECG...indispensable to the care of patients who subsequently developed symptoms during the admission. ''5 Finally, in a recent review of the subject, Sox also concluded that a "baseline ECG is not likely to be useful in asymptomatic persons with no risk factors for cardiovascular disease. 'q4 In the present study, there was no instance in which the baseline admission ECG obtained in a patient without specific indications was later found to have been helpful in the patient's hospital management. Our findings thus lend further support to the conclusion that an ECG need not be ordered routinely on admission if the sole purpose is to provide a baseline against which to evaluate possible future events. Another justification for performing a routine admission ECG on every patient might be its potential value as a screening test to identify otherwise unrecognized heart disease. However, the usefulness of the ECG as a screening test that is simply occasioned by acute hospital admission is not supported by published studies. 15 A number of potential shortcomings of the present study should be noted. The number of patients is relatively small, and the study is subject to all of the weaknesses inherent in any retrospective investigation. In particular, our review may not have identified indications for an admission ECG that were in fact present but inadequately documented on the chart, leading
FEBRUARY 1994 23:2 ANNALS DE EMERGENCY MEDICINE
to an inappropriate classification of the ECG as "routine." Because our chart review was performed by the principal investigators, there was a potential for reviewer bias in judging whether indications for testing were present. Although this potential was minimized by the use of an explicit set of indications, the possibility of bias in judging any effects on management and outcome cannot be excluded completely. Moreover, our patients were followed up only for the duration of their hospitalization. It is conceivable that an abnormality detected on a routine admission ECG might have had a positive impact on a patient's long-term outcome (eg, by leading to the initiation of a cardiac evaluation that might forestall a future infarction). However, it is not clear that the likelihood of acquiring such information in a patient without specific indications at the time of admission would be great enough to justify the application of what is essentially a screening test in this group of patients. A more important limitation of the present study is that its conclusions may not be generalizable to other patient populations and practice settings. In particular, our proposed criteria for performing an admission ECG must be validated by other investigators. In addition, it should be emphasized that we have not developed a clinical prediction rule in this study. Because we did not attempt to estimate the percentage of indicated admission ECGs that changed treatment or outcome, our data do not permit us to estimate the sensitivity and specificity of the proposed criteria for ordering an admission ECG, only to report on their negative predictive value (100%; 95% CI, 98.8% to 100%) in our own particular patient population. Finally, there were five patient charts that could not be located for review. The absence of any record of these individuals' cases in hospital risk management files or claims records makes it unlikely that any of them experienced an adverse occurrence as a result of the omission of an admission ECG. It is not impossible, however, that a routine ECG was performed on one or more of these patients and may have prevented the occurrence of a subsequent adverse event. Nevertheless, we have demonstrated that a significant proportion of routine admission ECGs obtained in ED patients had little, if any, use in improving patient management or clinical outcome. Our proposed criteria have not been validated in other settings, nor are they proposed as a rigid guideline for test ordering. However, when used in combination with the physician's clinical judgment, they might be useful in identifying individuals for whom an admission ECG could safely be omitted.
279
R O U T I N E ECG Garland & Wolfson
In purely economic terms, assuming a charge of $105 per test, the present study points to the potential for savings of more than $20,000 in patient ECG charges alone during the three-month study period in this single hospital. This finding, if corroborated in patient populations other than our own, suggests the potential for significant cost savings for the US health care system as a whole. However, although it is tempting to contemplate the elimination of routine admission ECG from ED practice altogether, such a step must be judged premature. Even were our criteria found to have widespread applicability, it would still remain to be established that the extremely low risk of the routine ECG and its relatively low cost compared with the cost of other frequently performed tests would not be outweighed by the small likelihood that the outcome of even an occasional ED patient would be improved. Moreover, the findings of Fesmire et al9 and Lee et al E° indicate that the availability of a baseline ECG may help to avoid unnecessary hospitalization or cardiac care unit admission in a certain subset of patients, a factor that could offset, at least partially, the potential savings suggested in the present study.
10. Lee TH, CookEF,Weisberg MC, et al: Impact of the availability of a prior electrocardiogram on the triage of the patient with acute chest pain. J GenInternMed 1990;5:381-388. 11. ZiembaSE, Hubbell FA, Fine MJ, et ah Restingelectrocardiogramsas baselinetests: Impact on the managementof elderly patients. Am J Med 1991;91:576-583. 12. RubensteinLZ, Greenfield S: The baseline ECGin the evaluation of acute cardiaccomplaints. JAMA 1980;244:2536-2539.
13. Hoffman JR, Igarashi E: Influence of electrocardiographicfindings on admissiondecisions in patients with acute chest pain. Am J Med 1985;79:699-707. 14. Sox HC: The baselineelectrocardiogram.Am J Med 1991;91:573475. 15. Sox HC, GarberAM, LittenbergB: The resting electrocardiogramas a screeningtest: A clinical analysis.Ann InternMed 1989;111:489402.
We thankDrs DavidMacPhersonand JeromeRHoffmanfor theirthoughtfulcommentsand suggestions.
Reprint no. 47/1~2,~,4 Address for reprints: Jeffrey L Garland, MD 440 Scaife Hall University of Pittsburgh Pittsburgh, Pennsylvania 15261
CONCLUSION
The routine admission ECG could have been omitted safely in an identifiable subset of ED patients admitted to the medical service of a university teaching hospital. We have proposed a set of criteria that, if validated in other settings, may be useful to the clinician in identifying such individuals. Our findings suggest the potential for significant cost savings for the US health care system as a whole. REFERENCES 1. Sagel SS, EvensRG, ForrestJV, et ah Efficacy of routine screeningand lateral chest radiographsin a hospital-basedpopulation. N EnglJ Med 1974;291:1001-1004. 2. Fink DJ, Fang M, Wyle FA: Routine chest x-ray films in a veterans hospital. JAMA 1981;245:1056-1057. 3. Hubbell FA, Greenfield S, Tyler JL, et al: The impact of routine admissionchest x-ray films on patient care. N EnglJ Med 1985;312:209-213. 4. TapeTG, Mushlin AI: The utility of routine chest radiographs.Ann InternMed 1986;104:663670. 5. MoormanJR, Hlatky MA, EddyDM, et al: The yield of the routine admission electrocardiogram:A study in a general medical service.Ann InternMed1985;103:590-595. 6. GoldbergerAL, O'Konski M: Utility of the routine electrocardiogrambefore surgeryand on general hospital admission.Ann InternMed 1986;105:552-557. 7. Gold BS, Young ML, KinmanJL, et ah The utility of preoperativeelectrocardiogramsin the ambulatorysurgical patient. Arch InternMed1992;152:301-305. 8. PeraskosJA: Who needs a preoperativeelectrocardiogram?ArchInternMed1992;152:261263. 9. FesmireFM, PercyRF,Wears RL: Diagnosticand prognosticimportanceof comparingthe initial to the previouselectrocardiogramin patients admitted for suspectedacute myocardial infarction. SouthMed J 1991;84:841-846.
280
ANNALS OF EMERGENCY MEDICINE
23:2
FEBRUARY 1994
Routine Admission Electrocardiography in Emergency Department Patients
From the Division of Pulmonary, Allergy, and Critical Care Medicine, Department of Medicine; and the Division of Emergency Medicine, University of Pittsburgh, Pittsburgh, Pennsylvania.
Jeffrey L Garland, MD Allan B Wolfson, MD
Receivedfor publication January 28, 1993. Revision received June 7, 1993. Acceptedfor publication June 17, 1993. Presented in part at the Annual Meeting of the Pennsylvania Chapter of the American College of Physicians in Pittsburgh, October 1991; and presented at the Societyfor Academic Emergency Medicine Annual Meeting in Toronto, Ontario, Canada, May 1992.
Study objectives: To determine whether routine performance of an ECGcould have been safely avoided in a subset of emergency department patients admitted to a university hospital. Design: Retrospective consecutive case series. Setting: University teaching hospital. Type of participants: All ED patients admitted to the medical service of the study hospital during a three-month period. Methods and interventions: Acceptable indications for an admission ECGwere prospectively developed. Charts of all patients were reviewed to determine whether any of these indications were present, whether an admission ECGwas performed, and whether an admission ECG resulted in a change in patient management or outcome. An ECGwas classified as routine when performed in the absence of documentation of any of these indications. No interventions were performed.
Measurements and main results: There were 636 ED admissions to the medical service during the study period. Of the 631 patients whose chart could be located, 384 (61%) had at least one indication for an ECGand all but one had an ECG performed. No indications for an admission ECGwere identified in the remaining 247 patients; of these, 202 (82%) had an ECG performed and 45 (18%) did not. Among the 202 who had a routine admission EC6,the test resulted in a change in management in only three (1.5%) (95% confidence interval [Oil, 0.3% to 4.3%) and affected patient outcome in none (95% Cl, 0% to 1.5%). Among the 45 without indications who did nothavea routine admission ECG,none experienced an identifiable adverse consequence during hospitalization (95% CI, 0% to 6.7%). Conclusion: The admission ECGcould have been avoided in an identifiable subset of ED patients admitted to the medical service of our hospital, with no adverse effect on patient outcome. This finding, if corroborated in other patient populations, suggests the potential for significant cost savings for the US health care system as a whole.
FEBRUARY 1994 23:2 ANNALS OF EMERGENCY MEDICINE
275
ROUTINE ECG
Garland& Wolfson
[Garland JL, Wolfson AB: Routine admission electrocardiography in emergencydepartment patients. Ann EmergMed February 1994;23:275-280.]
goal was to estimate the potential financial savings that might be realized by the health care system were such criteria to be applied successfully to a broader patient population.
INTRODUCTION
MATER ALS AND METHODS
Although it has long been standard practice in many institutions to order a battery of routine diagnostic tests when patients are admitted to the hospital, the diagnostic yield and cost-effectiveness of routine admission testing have been investigated to only a limited extent. Among those tests that have been studied, the routine admission chest radiograph has received the closest attention. Several investigators have shown that a chest radiograph need not be ordered routinely on all patients and that the decision to order this examination can be guided appropriately by such criteria as the patient's age and clinical presentation. 1-4 In contrast, only one published study has examined in detail the usefulness of the routine admission ECG. Moorman et al, investigating all admissions to the medical service, found that a routine admission ECG resulted in a change in outcome in only 0.3% of patients with neither a history nor symptoms or signs of a cardiac abnormality. 5 The present study focused on emergency department patients admitted to the medical service of a tertiary-care university teaching hospital. The goal was to develop a set of criteria that might be used to identify those individuals for whom an admission ECG would alter neither hospital management nor ultimate outcome. A second
All patients admitted from the ED to the medical service of the Presbyterian University Hospital of Pittsburgh from July 1 through September 30, 1989, were eligible for inclusion in the study. The decision to order or not to order an admission ECG was made by ED staff. Eligible patients were identified from the medical admission list compiled by the medical admitting resident in the ED. For patients admitted more than once during the study period, each admission was considered a separate "patient." Hospital admission records were retrieved when available and reviewed by one of the investigators. For patients whose charts could not be located, a further search was conducted to determine whether these individuals' names appeared in the files of the hospital's risk management office or in records of claims made against the hospital. The following were proposed prospectively as appropriate indications for obtaining an admission ECG: a history of coronary artery disease, arrhythmia, congestive heart failure, conduction disturbance, cor pulmonale, pericarditis, or cardiomyopathy; palpitations, syncope,
Table 1.
Figure.
Performance of admission ECG and clinical outcome in 636 consecutive patients admitted to the medical service from the ED
Proposed criteria for obtaining an admission ECG 636(Admissions)
History of: Coronary artery disease Arrhythmia Congestiveheart failure Conduction disturbance Cor pulmonale
Pericerditis Cardiomyopathy
Palpitations Syncope or coma Symptom complex suggestiveof angina or congestive heart failure Suspected cardiotoxic overdose Irregular pulse Pulse >120 or <60 Systolic blood pressure >200 or <90 Diastolic blood pressure >120 Serum potassium>5.7 or < 3.0
276
631(Chartfound)
5 (Chartnotfound)
247(Notindicated)
384(indicated)
,/
/',, 383(Indicative, done)
1(Indicative, notdone)
,,,
202 (Nat indicative, done)
,/
199 (No change in managementl
45 {Notindicative, not donel
',, ,,/ ",, 3 (Changein management}
3 (Nochangein outcome)
ANNALS OF EMERGENCY MEDICINE
0 (Nochangein outcome)
23:2
FEBRUARY 1994
ROUTINE ECG
Garland & Wolfson
coma, or symptom complex suggestive of angina or congestive heart failure; suspected cardiotoxic drug overdose; irregular pulse or pulse less than 60 or more than 120; systolic blood pressure more than 200 or less than 90; diastolic blood pressure more than 120; or serum potassium less than 3.0 or more than 5.7 (Table 1). Age per se was not an indication. With these criteria, each admission ECG that was performed was classified as either "indicated" or, if the patient did not have any of the defined set of indications, "routine." For patients who had a routine admission ECG performed, the chart was reviewed further to determine whether performance of the test had resulted in any identifiable change in patient management or outcome during the hospitalization. An ECG was determined to have caused a change in management if its performance resuited in any diagnostic or therapeutic intervention that would not otherwise have been performed. A change in outcome was defined as an alteration in the patient's condition that would not have occurred if the ECG had not been performed. For patients who did not have an admission ECG done, the chart was reviewed to determine whether the omission of the test had resulted in any identifiable adverse consequence.
Table 2.
Principal admission diagnosis in 247 patients without any of the indications in Table i No. of Patients Pneumonia Urinary tract infection AIDS Other source of fever Complication of malignancy COPD/asthma PE/DVT/phlebitis Gastrointestinal bleed/gastritis Liver disease Abdominal pain Cerebrovascular accident/transient ischemic accident Seizure Diabetes mellitus Sickle cell anemia Other anemia Rheumatologic disorder Failure to thrive Miscellaneous Total
FEBRUARY 1994 23:2 ANNALS OF EMERGENCY MEDICINE
13 15 16 47 23 10 5 14 9 7 5 6 4 20 5 6 6 36 247
RESULTS ED records listed 636 admissions to the medical service of Presbyterian University Hospital of Pittsburgh during the study period. There were five patients whose names appeared on the medical admission list but whose charts could not be located for review. No record of these individuals' cases could be found in the files of the hospital's risk management office or in lists of claims made against the hospital. Of the 631 patients on whom records were available, 384 (61%) had at least one indication for an ECG and all but one had an admission ECG performed. No indications for an admission ECG were identified in the remaining 247 patients; of these, 202 (82%) had an admission ECG performed and 45 (18%) did not (Figure). Among the 202 patients who had a routine admission ECG performed despite the absence of any of the proposed indications for the test, the admission ECG resulted in a change in management in only three (1.5%) (95% confidence interval [CI], 0.3% to 4.3%) and a change in outcome in none (95% CI, 0% to 1.5%). Of the 45 patients without indications who did not have a routine admission ECG, none (95% CI, 0% to 6.7%) experienced an identifiable adverse consequence during the course of their hospitalization. Thus, of the 247 patients without indications for an admission ECG, none experienced an identifiable change in outcome (95% CI, 0% to 1.2%). The principal admission diagnoses in these patients are shown in Table 2. The cases of the three patients for whom the admission ECG resulted in a change in management clearly merit careful review. The details of these cases are presented here and summarized in Table 3. CASE REPORTS Case 1 A 46-year-old man with a history of IV drug abuse presented to the ED with pleuritic chest pain, dyspnea, and fever. He was admitted because of the possibility of endocarditis and pulmonary embolism. The admission ECG revealed inverted T-waves in leads V 1 and V2. He was admitted to a nonmonitored bed, and the ECG was repeated three times and showed no change. No further cardiac evaluation was performed, and no adverse cardiac events were noted during the hospitalization. Case 2 A 50-year-old woman was admitted because of upper gastrointestinal bleeding. Admission ECG revealed inverted T-waves in leads V~ through V6, and the patient
277
ROUTINE ECG
Garland & Wolfson
was admitted to a monitored bed. Further ECGs showed no change. After acute myocardial infarction was ruled out, no further cardiac evaluation was performed. No adverse cardiac events were noted during the hospitalization. Case 3 A 76-year-old woman with chronic obstructive pulmonary disease and a history of laryngectomy for laryngeal cancer was admitted because of increasing dyspnea and purulent sputum. Admission ECG revealed inverted T-waves in leads V 1 through V6, as well as left ventricular hypertrophy. She was admitted to a monitored bed, where further ECGs showed no change and acute myocardial infarction was ruled out. No further cardiac evaluation was performed, and there were no adverse cardiac events during the hospital stay. Although the routine admission ECG resulted in a change in management in these three cases, patient outcome was not affected. DISCUSSION
We have shown that routine admission ECG could have been omitted safely in an identifiable subset of ED patients admitted to the general medical service of a tertiary-care university hospital, with minimal adverse effect on patient outcome. Although there have been a number of studies of the usefulness of the routine preoperative ECG, 6-8 only one specifically has addressed the value of the routine admission ECG in medical patients. Moorman et al prospectively studied 1,410 patients admitted to the general medical service .of a university hospital through the ED or outpatient clinics or by transfer from other hospitals; patients admitted to an intensive care setting were excluded. 5 Of the 1,410 admitted patients, 775 had no "cardiac abnormality" or other apparent indications for an admission ECG. Performance of an ECG in these 775 individuals resulted in a change in management in 26 (4.1%) and a change in management that was judged
likely to have improved outcome in only two (0.3%). It was noted that the ECG was helpful more often in patients more than 45 years old, but this association was of only borderline statistical significance when considered independently of the presence of identified cardiac abnormalities. In the present study, the subjects were all admitted to the medical service from the ED (patients admitted to an intensive care setting were included), whereas those admitted from outpatient clinics, electively, or by transfer from other institutions were not. Thus, the present study focused on decision making in ED patients who tended to be acutely ill and in whom the decision of whether to order an admission ECG had to be made in the urgent environment of a busy ED. To better reflect these decisionmaking issues, the acceptable indications for admission ECG were broader than those used by Moorman et al. Our criteria for performing an admission ECG did not include patient age per se. Although the three patients whose management was altered by a routine admission ECG were all more than 45 years old, there was no change in outcome for these individuals; thus, adding age to our list of indications would not have resulted in an identifiable benefit for any patient. One argument that has been advanced in defense of routine ECG testing is that it provides a baseline tracing that is likely to be useful in making management decisions in the future. Fesmire et al found that comparison of the ED ECG with a previous ECG provided important prognostic information for patients admitted with suspected myocardial infarction. 9 Lee et al found that the availability of a prior ECG appeared to be of benefit in avoiding inappropriate hospital and cardiac care unit admission for patients who presented with chest pain but who were found in retrospect not to have sustained an acute myocardial infarction, to In neither of these studies, however, is it clear that the prior ECGs were obtained
Table 3.
Clinical features in three patients for whom routine admission ECG resulted in a change in management
Patient
Age/Sex
1
46/M
2
50/F
3
76/F
278
Chief Complaint
Change in Management
ECG Finding
Change in Outcome
Pleuritic chest pain, fever, shortness of breath Upper gastrointestinal bleed
Inverted T-wave (V1-V2)
ECG × 3
None
Inverted T-wave (VcV s)
None
Chronic obstructive pulmonary disease, shortness of breath, sputum
Left ventricular hypertrophy, inverted T-wave (V~-Vs)
Monitor, EC6 x 3, enzyme x 3 Monitor, EC8 x 3, enzyme x 3
None
ANNALS OF EMERGENCYMEDICINE 23:2 FEBRUARY1994
ROUTINE ECG Garland & Wolfson
"routinely" or that they resulted in a change in patient outcome. More recently, Ziemba et al concluded that a baseline ECG was of benefit in making management decisions in elderly patients with a past history of heart disease who were being evaluated in the ED. 11 This conclusion does not conflict with the findings of the present study, in which these patients would have been considered to have indications for an admission ECG. The usefulness of a baseline ECG was addressed specifically by Rubenstein and Greenfield, who reviewed the records of 236 patients presenting to two EDs with a chief complaint of chest pain. 12 The authors concluded that although the availability of a baseline ECG might have prevented the unnecessary hospitalization of 11 of the 236 patients, in no instance would its absence have resuited in an adverse outcome. Further evidence against the value of a baseline ECG is provided by Hoffman and Igarashi, who in a study of 84 ED patients found that physicians never changed their minds in deciding whether to admit or discharge a patient with chest pain when the decision was made first with, and then without, a baseline ECG. 13 Moorman et al reached a similar conclusion, reporting that "in no instance was an admission ECG...indispensable to the care of patients who subsequently developed symptoms during the admission. ''5 Finally, in a recent review of the subject, Sox also concluded that a "baseline ECG is not likely to be useful in asymptomatic persons with no risk factors for cardiovascular disease. 'q4 In the present study, there was no instance in which the baseline admission ECG obtained in a patient without specific indications was later found to have been helpful in the patient's hospital management. Our findings thus lend further support to the conclusion that an ECG need not be ordered routinely on admission if the sole purpose is to provide a baseline against which to evaluate possible future events. Another justification for performing a routine admission ECG on every patient might be its potential value as a screening test to identify otherwise unrecognized heart disease. However, the usefulness of the ECG as a screening test that is simply occasioned by acute hospital admission is not supported by published studies. 15 A number of potential shortcomings of the present study should be noted. The number of patients is relatively small, and the study is subject to all of the weaknesses inherent in any retrospective investigation. In particular, our review may not have identified indications for an admission ECG that were in fact present but inadequately documented on the chart, leading
FEBRUARY 1994 23:2 ANNALS DE EMERGENCY MEDICINE
to an inappropriate classification of the ECG as "routine." Because our chart review was performed by the principal investigators, there was a potential for reviewer bias in judging whether indications for testing were present. Although this potential was minimized by the use of an explicit set of indications, the possibility of bias in judging any effects on management and outcome cannot be excluded completely. Moreover, our patients were followed up only for the duration of their hospitalization. It is conceivable that an abnormality detected on a routine admission ECG might have had a positive impact on a patient's long-term outcome (eg, by leading to the initiation of a cardiac evaluation that might forestall a future infarction). However, it is not clear that the likelihood of acquiring such information in a patient without specific indications at the time of admission would be great enough to justify the application of what is essentially a screening test in this group of patients. A more important limitation of the present study is that its conclusions may not be generalizable to other patient populations and practice settings. In particular, our proposed criteria for performing an admission ECG must be validated by other investigators. In addition, it should be emphasized that we have not developed a clinical prediction rule in this study. Because we did not attempt to estimate the percentage of indicated admission ECGs that changed treatment or outcome, our data do not permit us to estimate the sensitivity and specificity of the proposed criteria for ordering an admission ECG, only to report on their negative predictive value (100%; 95% CI, 98.8% to 100%) in our own particular patient population. Finally, there were five patient charts that could not be located for review. The absence of any record of these individuals' cases in hospital risk management files or claims records makes it unlikely that any of them experienced an adverse occurrence as a result of the omission of an admission ECG. It is not impossible, however, that a routine ECG was performed on one or more of these patients and may have prevented the occurrence of a subsequent adverse event. Nevertheless, we have demonstrated that a significant proportion of routine admission ECGs obtained in ED patients had little, if any, use in improving patient management or clinical outcome. Our proposed criteria have not been validated in other settings, nor are they proposed as a rigid guideline for test ordering. However, when used in combination with the physician's clinical judgment, they might be useful in identifying individuals for whom an admission ECG could safely be omitted.
279
R O U T I N E ECG Garland & Wolfson
In purely economic terms, assuming a charge of $105 per test, the present study points to the potential for savings of more than $20,000 in patient ECG charges alone during the three-month study period in this single hospital. This finding, if corroborated in patient populations other than our own, suggests the potential for significant cost savings for the US health care system as a whole. However, although it is tempting to contemplate the elimination of routine admission ECG from ED practice altogether, such a step must be judged premature. Even were our criteria found to have widespread applicability, it would still remain to be established that the extremely low risk of the routine ECG and its relatively low cost compared with the cost of other frequently performed tests would not be outweighed by the small likelihood that the outcome of even an occasional ED patient would be improved. Moreover, the findings of Fesmire et al9 and Lee et al E° indicate that the availability of a baseline ECG may help to avoid unnecessary hospitalization or cardiac care unit admission in a certain subset of patients, a factor that could offset, at least partially, the potential savings suggested in the present study.
10. Lee TH, CookEF,Weisberg MC, et al: Impact of the availability of a prior electrocardiogram on the triage of the patient with acute chest pain. J GenInternMed 1990;5:381-388. 11. ZiembaSE, Hubbell FA, Fine MJ, et ah Restingelectrocardiogramsas baselinetests: Impact on the managementof elderly patients. Am J Med 1991;91:576-583. 12. RubensteinLZ, Greenfield S: The baseline ECGin the evaluation of acute cardiaccomplaints. JAMA 1980;244:2536-2539.
13. Hoffman JR, Igarashi E: Influence of electrocardiographicfindings on admissiondecisions in patients with acute chest pain. Am J Med 1985;79:699-707. 14. Sox HC: The baselineelectrocardiogram.Am J Med 1991;91:573475. 15. Sox HC, GarberAM, LittenbergB: The resting electrocardiogramas a screeningtest: A clinical analysis.Ann InternMed 1989;111:489402.
We thankDrs DavidMacPhersonand JeromeRHoffmanfor theirthoughtfulcommentsand suggestions.
Reprint no. 47/1~2,~,4 Address for reprints: Jeffrey L Garland, MD 440 Scaife Hall University of Pittsburgh Pittsburgh, Pennsylvania 15261
CONCLUSION
The routine admission ECG could have been omitted safely in an identifiable subset of ED patients admitted to the medical service of a university teaching hospital. We have proposed a set of criteria that, if validated in other settings, may be useful to the clinician in identifying such individuals. Our findings suggest the potential for significant cost savings for the US health care system as a whole. REFERENCES 1. Sagel SS, EvensRG, ForrestJV, et ah Efficacy of routine screeningand lateral chest radiographsin a hospital-basedpopulation. N EnglJ Med 1974;291:1001-1004. 2. Fink DJ, Fang M, Wyle FA: Routine chest x-ray films in a veterans hospital. JAMA 1981;245:1056-1057. 3. Hubbell FA, Greenfield S, Tyler JL, et al: The impact of routine admissionchest x-ray films on patient care. N EnglJ Med 1985;312:209-213. 4. TapeTG, Mushlin AI: The utility of routine chest radiographs.Ann InternMed 1986;104:663670. 5. MoormanJR, Hlatky MA, EddyDM, et al: The yield of the routine admission electrocardiogram:A study in a general medical service.Ann InternMed1985;103:590-595. 6. GoldbergerAL, O'Konski M: Utility of the routine electrocardiogrambefore surgeryand on general hospital admission.Ann InternMed 1986;105:552-557. 7. Gold BS, Young ML, KinmanJL, et ah The utility of preoperativeelectrocardiogramsin the ambulatorysurgical patient. Arch InternMed1992;152:301-305. 8. PeraskosJA: Who needs a preoperativeelectrocardiogram?ArchInternMed1992;152:261263. 9. FesmireFM, PercyRF,Wears RL: Diagnosticand prognosticimportanceof comparingthe initial to the previouselectrocardiogramin patients admitted for suspectedacute myocardial infarction. SouthMed J 1991;84:841-846.
280
ANNALS OF EMERGENCY MEDICINE
23:2
FEBRUARY 1994