ORIGINAL CONTRIBUTION
ECG, quality assurance; quality assurance, ECG
Quality Assurance in the Emergency Department: Evaluation of the ECG Review Process From the Department of Medicine,
Marie Kuhn, MD, FACEP
UCLA School of Medicine; and the
Marshall T Morgan, MD
UCLA Emergency Medicine Center,
Jerome R Hoffman, MD, FACEP
Los Angeles, California. Receivedfor publication January 5, 199l. Revision received July 15, 199]. Acceptedfor publication July 19, 1991.
Study objective: TOdetermine whether the review of emergency department ECGs by cardiologists contributes to the quality of patient care. Study design: We retrospectively analyzed ED ECGsand compared interpretations of the treating emergency physicians with those of the reviewing cardiologists. We then evaluated the ED care of patients with potentially significant ECG abnormalities that were not detected by the treating emergency physicians, as well as the care of patients whose ECGswere "flagged" by the reviewing cardiologists as needing follow-up. Setting:
University hospital.
Participants:
Four hundred consecutive ECGsobtained on ED
patients. Measurements and main results: Thirty-three of the 400 tracings had undetected potentially significant or critical ECG abnormalities; this adversely affected patient care in two cases, These two tracings were not flagged by the reviewing cardiologists. Thirteen ECGswere flagged by the cardiologists; patient care was not altered in any of these cases.
Conclusion: Review of ED ECGs by cardiologists did not affect patient care at our institution. [Kuhn M, Morgan MT, Hoffman JR: Quality assurance in the emergency department: Evaluation of the ECG review process. Ann Emerg Med January 1992;21:10-15.]
18/10
I
ANNALS OF EMERGENCYMEDICINE 21:1 JANUARY1992
ECG R E V I E W
Kuhn, Morgan & Hoffman
INTRODUCTION
As part of emergency department quality assurance programs, the Joint Commission on Accreditation of Heahheare Organizations (JCAHO) requires that ECCs performed in the emergency department be reviewed by "physicians with such privileges" and that their interpretations be made available to emergency physicians. 1 Our extensive literature search (Melvyl Medline and Orion searches involving all possibly related subject headings over ten years, cheek of references in all articles identified, and survey of all books on quality assurance in a university medical library) did not, however, reveal any studies that substantiate the clinical efficacy of such a quality assurance system. Several studies concerning the diagnosis of myocardial isehemia in the ED have reported that failure to appropriately interpret the EGG was a significant factor in the treating physician's failure to diagnose myocardial ischemia. 2-4 However, others have questioned whether ECG changes do, or should, play a major role in decision making in this setting. 5-6 One published review of a teaching institution's ED quality assurance program found that during one year (1982), 74 ECGs had been misinterpreted in their ED and that 28% of those eases required immediate call-back.7 However, their study, which was designed to analyze their entire quality assurance program and did not specifically address the efficacy of their EGG review proeess, left a number of questions unanswered. The frequency with which errors in EGG interpretation occurred is unclear, as they did not include the total number of ECGs reviewed during that year. Also, the clinical significance of the ECG misinterpretations was unclear as they failed to note the criteria by which they concluded that an ECG had been misinterpreted or that urgent patient call-back was indicated. Nor did they note, beyond their decision to call the patient back, whether their system affected patient outcomes. Thus, there does not appear to be an examination of the ECG review process analogous to the study of the radiograph review process produced by O'Leary et al in which the efficacy of their quality assurance system was determined by its impact on patient outcomes. 8 We therefore decided to retrospectively review the ECGs taken in our ED to assess whether our quality assurance program succeeds in capturing all tracings with undetected, potentially significant or life-threatening abnormalities, as well as whether the identification of such tracings by the reviewing cardiologists affects patient care.
JANUARY 1992
21:1 ANNALS OF EMERGENCY MEDICINE
At our institution, ED patients are treated primarily by emergency medicine and internal medicine house officers, who are supervised by emergency medicine faculty-. EGGs are interpreted initially by the housestaff and are selectively referred, by either the housestaff or nursing staff, to emergency medicine faeuhy for concurrent review. In an attempt to ensure appropriate patient care and to comply with JCAHO dictates, we developed, in conjunction with the cardiologist who supervises the EGG reading laboratory, the following quality assurance process. First, two copies are made of each ECG performed in the ED, one of which is attached to the ED record, and the other to an ECG request form that contains space for a "preliminary reading" to be completed by the emergency physician. Second, ECGs and request forms are sent to the ECG reading laboratory- on a daily basis and are read within 24 hours by members of the cardiology faculty. Third, the reviewing cardiologists are expected to "red-flag" all EGG tracings that contain potentially critical abnormalities not documented in the emergency physician's preliminary reading; red-flagged tracings are sent immediately, by ECG laboratory personnel, to the ED, where they are given to the physicians who initially treated the patient or, in their absence, to the senior emergency medicine resident or faculty. When the cardiologists reading the ECG believe there are undetected, life-threatening abnormalities, they have been instructed to telephone the ED and speak directly to the ED faculty, in addition to flagging the tracing. Fourth, it is the reviewing emergency physician's responsibility to re-evaluate the ECG along with the cardiologist's interpretation and the patient's medical record and to arrange for appropriate follow-up. Finally, the reviewing emergency physician notes on the ECG request form what action has been taken and signs the form. A log of these activities is maintained by the ED clerical staff. Recognizing that even a normal ECG does not eliminate the possibility of cardiac disease, we chose to limit our study to an examination of the efficacy of our quality assurance system in those cases with potentially clinically significant ECG abnormalities. First, we sought to determine whether clinically significant errors in ECG interpretation were made by the treating physicians in our ED; that is, did the treating physicians fail to detect significant EGG abnormalities and, if so, did their failure to detect these abnormalities adversely affect patient care?
11/19 I
ECG REVIEW
Kuhn, Morgan & Hoffman
Next, we asked if any of the initial errors in ECG interpretation were detected by the reviewing cardiologists and whether, in such instances, this information was communicated to the emergency physicians. Finally, we sought to determine whether there were any cases in which this quality assurance system had a beneficial effect on patient care.
MATERIALS AND METHODS Four hundred consecutive ED ECGs were collected over a two-month period beginning January 1, 1988. Tracings were Figure 1.
Classification of ident~ed ECG abnormalities according to potential severity. ECGs deemed benign Normal Unchanged from previous tracings Minor ECG abnormalities Sinus tachycardia with a rate < 110 Sinus bradycardia with a rate > 55 Occasional premature atrial contractions Isolated first-degree atrioventricular block Leftward axis > - 6 0 ° Rightward axis ~ +90 ° Atrial enlargement (right or left) Intraventricular conduction delay Possible left ventricular hypertrophy by voltage criteria only Nonspecific ST-T wave changes
ECG abnormalities in this series deemed potentially clinically significant Arrhythmias that might indicate underlying disease or require evaluation and/or therapy, including: Atrial fibrillation with normal {70 to 110) ventricular response rate Sinus rates > 40, < 65, and > 110 Indications of old infarct without acute changes Left ventricular or right ventricular strain patterns Paced rhythms Right bundle branch block or left bundle branch block QRS axis < - 6 0 or > +g0 T wave abnormalities Prolonged QT interval ST changes, cannot rile out ischemia
ECG abnormalities in this series deemed potentially life threatening Indications of acute myocardial injury Potentially critical arrhythmias, including: Atrial fibrillation with rate > 200 Sinus bradycardia with rate of 35
obtained from the ECG reading laboratory and the ED clerical staff. Each of the tracings contained a cardiologist's interpretation accompanied by an ECG request form documenting the emergency physician's preliminary reading, if one was present. Evidence of the reviewing emergency physician's follow-up was s9ught on all tracings that had been red-flagged. A preliminary review of all 400 tracings was undertaken by two of the authors to determine which cases would require review of the medical records to evaluate the ECG abnormalities in light of their clinical context. The tracings were divided by these reviewers into one of three categories: 1) ECGs that were normal, unchanged from prior tracings, or contained only minor abnormalities; 2) ECGs with potentially clinically significant abnormalities; and 3) ECGs with potentially life-threatening abnormalities. The ECG abnormalities assigned to each category are outlined (Figure 1). ECGs assigned to category I were deemed sufficiently benign to eliminate the need for further review. Tracings with abnormalities assigned to categories II or III were thought to be potentially significant or life threatening, depending on the clinical context. If these abnormalities were not documented in the ECG request form's "preliminary reading" box, the medical records were obtained and reviewed. The emergency physician's notes were examined for documentation that the ECG had been correctly interpreted; if such documentation was found, the record was not examined further. If this documentation was lacking, the remainder of the medical record was reviewed by the first author. Cases that were determined to have been appropriately managed were not subjected to additional review; cases that the author believed were inappropriately managed were referred to the second author for review. As the reviewers agreed regarding the clinical mismanagement in each such case, they were therefore so designated. Charts that had been red-flagged by the cardiologists were analyzed separately to determine the efficacy of this action. The medical records for these patients were examined to determine whether the emergency physician notified had chosen to alter patient care beeause of the cardiologist's reading of the ECG. Charts of patients whose ECGs were not red-flagged by the cardiologists but were deemed by the investigators to have lifethreatening abnormalities were reviewed to determine if they had been appropriately managed and if patient care might have been altered had these ECGs been red-flagged and emergency physicians alerted one day after the ED visit. ~"
Junctional rhythm with bundle branch block versus accelerated ventricular rhythm All tracings red-flagged by reviewing cardiologist
20/12 |
ANNALSOF EMERGENCY MEDICINE
21:1
JANUARY1992
ECG R E V I E W Kuhn, Morgan & Hoffman
RESULTS Four hundred ECGs were reviewed; 289 of the ECG request forms docmnented the treating physician's initial interpretations, whereas 111 did not (Figure 2). Of the tracings with documented preliminary readings, the interpretations of the treating physician, reviewing cardiologist, and investigator agreed in 164 of 289 cases (56.7%) (Figure 2). In 89 of 289 cases (30.8%), the discrepancies in ECG interpretations were deemed minor. Twenty-five of 289 tracings (8.6%) had potentially significant ECG abnormalities for which the interpretations of the emergency physician and cardiologist differed. In seven of these cases, the emergency physician detected the ECG abnormalities, but the cardiologists did not; in the other 18 cases, the emergency physician failed to note the ECG abnormalities detected by the cardiologist. Review of the medical records for these 18 cases did not reveal a single instance in which the patient appeared to have been inappropriately managed, nor did it appear that patient care would have been altered by the additional information garnered fl'om the cardiologist's reading of the ECG. Of those ECGs without preliminary readings noted on the ECG request form, 38 of 111 (34.2%) were normal or unchanged from previous tracings, whereas 15 (13.5%) had only minor abnormalities. Fifty-one of 111 (45.9%) had abnormalities deemed to be Figure 2. Findingsfrom reviewof 400 consecutiveED ECGs.
potentially clinically significant; in 50 of these 51 cases, medical records were available for review. Thirty-eight of the medical records documented that the ECG had been appropriately interpreted in the ED and simply had not been recorded on the ECG request form. In the remaining 12 cases, there was no documentation that the ECG had been interpreted in the ED; review of the medical records indicated that all of these patients had been treated appropriately and revealed no instance in which it appeared that care would have been ahered by the information provided by the cardiologist. Thirteen cases were red-flagged by the reviewing cardiologists; ten of these cases had a preliminary reading noted on the ECG request form, whereas three did not. Each of these cases was reviewed and signed by an emergency physician; in no such case was patient care altered by the reviewing emergency physician. Care was not altered in the three red-flagged cases without documented preliminary readings because the tracings had actually been accurately interpreted by the treating physicians who had merely failed to note their readings on the ECG request form. Care was not altered for any of the ten cases in which preliminary readings failed to note findings red-flagged by the cardiologist because the emergency physicians notified of these findings thought, in each case, that the patients had already been treated appropriately. In the only instance in which the reviewing cardiologist actually phoned the ED to report a red-flagged ECG finding, the treating physicians were already aware of the ECG abnormalities and had acted on them. •
400 ECGs
111 ECGswithout preliminary readings
289 ECGswith preliminary readings 1
6
4
1
~
° cc"'a'',eto8° minor discrepancies
.
8
I'
I
%
o
2518.6%1 Potentially significant discrepancies
7
Abnormalities not detected by cardiologists
)
"h
cr'tP:'
lO~~"~|~'molities
Tracings red-flagged
0 Cases mismanaged
Tracings red-flagged Interpreted accurately
Without ED ECG interpretation
JANUARY 1992 21:1 ANNALS OF EMERGENCY MEDICINE
Case mismanaged
Not red-flagged
0
2
Cases
Cases
mismanaged
1
Cases mismanaged
716.3%) Life-threatening abnormalities
J t 51 (45.9%) Potentially significant abnormalities
Critical abnormality not detected by ED physician or cardiologist Record unavailable
18 Abnormalities not detected by ED physicians
I
38 ( Normal or unchanged and 15113.5%) minor abnormalities
mismanaged
0 Cases mismanaged
13121
I
ECG R E V I E W Kuhn, Morgan & Hoffman
There were five tracings that the reviewers thought should have been red-flagged by the cardiologists but were not. Two cases that did not have a preliminary reading documented were managed appropriately; both patients had evidence of acute isehemia and were admitted to the coronary care unit. The three remaining cases may have been mismanaged. The first was a 48-year-old hypertensive man with a history of cocaine abuse who eomplained of chest tightness. His ECG (Figure 3) was interpreted by both the emergency physician and the over-reading cardiologist as demonstrating "left ventricular hypertrophy with secondary repolarization abnormalities." His hypertension was treated, and he was discharged from the ED. The authors interpreted his ECG as indicating either anterior wall epieardial injury and lateral wall ischemia or early repotarization abnormalities.9 We believe that the tracing should have been red-flagged and the patient asked to return for re-evaluation. We also believe that the patient should have been admitted at the time of his ED visit on the basis of elinieal factors (a complaint of chest tightness in a patient with hypertension and a history of cocaine abuse) alone and that appropriate care should therefore not have depended on subsequent review of the ECG. The second ease involved an 85-year-old man whose tracing was read as showing "extreme sinus bradyeardia, rate 35; complete right bundle branch block with left anterior fascicular block and first-degree atrioventrieular b l o e k . . , represents trifascicular disease." His ECG was not red-flagged. Review of this patient's medical record indicated that the emergency physician Figure 3. ECG tracing from a 48-year-old hypertensive man with a history of cocaine abuse who complained of chest tightness. His ECG was originally interpreted as demonstrating "left ventricular hypertrophy with secondary repolarization abnormalities. "' The authors believe that his tracing was more consistent with either anterior wall epicardial injury and lateral wall ischemia or with early repolarization abnormalities.
22/ 14
interpreted the ECG accurately but failed to document his reading on the ECG request form. However because the ECG abnormalities had been present on a previous tracing, the emergency physician elected to discharge the patient. Once again, although we believed that this patient had been managed inappropriately, the decision as to how to treat him would not have been changed by red-flagging his ECG because the treating physician was already aware of the pertinent findings. The third case involved a 65-year-old woman who presented in atrial fibrillation with "occasional aberrantly conducted beats" and a heart rate of 225; her medical record revealed that she had experienced four episodes of transient atrial fibrillation in the past year. Her ventricular response rate during the visit in question was miscalculated as averaging 180. She was treated with IV digoxin, and after her ventricular rate decreased to approximately 85 she was discharged to seek follow-up at another facility. The reviewing cardiologist raised the possibility of a bypass tract; the investigators contacted the treating physicians and made them aware of the cardiologist's reading as well as the contraindications to the use of digoxin in patients with possible bypass tract disease. On review, we believed that the likelihood of a bypass tract was small; regardless of theoretical considerations, the patient did not suffer from the treatment she received, nor would next-day red-flagging have been able to prevent any. adverse effects had they occurred.
DISCUSSION
We began this study questioning whether a quality assurance system would benefit ED patients with abnormal ECGs and, if so, whether the quality assurance process in place at our institution was effective. Our study was not able to clearly answer the first question. If one defines the need for a quality assurance system by the percentage of undetected ECG abnmmalities, then the fact that 8.3% of the tracings had undetected significant or lifethreatening abnormalities would strongly support such a need. If, on the other hand, a quality assurance process is valued according to its potential impact on patient care, there were only two of 399 patients (0.5%) whose care was possibly adversely affected by the treating physician's failure to appropriately interpret the ECG, because one patient was mismanaged despite an accurate interpretation of this ECG by the emergency physician. Whether it is worthwhile to establish a quality assurance system to have an impact on so few patients depends on both the cost and effort involved and whether an effective quality assurance system can be constructed and implemented, ll~
ANNALS OF EMERGENOY MEDI01NE 21:1 JANUARY 1992
ECG R E V I E W
Kuhn, Morgan & Hoffman
Our results clearly indicate that our current quality assurance system is not effective. In the first place, not all of the life-threatening ECG abnormalities undetected in the ED were captured. The reasons for this failure are not entirely clear: in particular, two such ECGs were read appropriately but not flagged by a cardiologist who was apparently aware of the quality assurance system, as he had red-flagged other tracings in this series. Second, the process of red-flagging tracings did not result in any changes in patient care; in every such instance, the emergency physician who was notified decided that the patient had already been managed appropriately. The frequent (111 of 400, 28%) failure of the treating physicians to complete the Preliminary reading section on the ECG request form may well have contributed to the ineffectiveness of our quality assurance system. We might conjecture that this occurred because physicians in training are reluctant to commit their ECG interpretations to paper or that they failed to recognize this step as an important facet of patient care, but we have no data to support these views. An informal telephone survey of ten other academic institutions revealed that our quality assurance system is similar to that of other teaching hospitals. It is therefore tempting, given our results, to argue that the JCAHO mandate regarding the review of ED ECGs by "physicians with such privileges" should be reconsidered. There are, however, several aspects of our study design that limit our ability to make such an argument. We limited our review to those patients whose ECGs demonstrated undetected significant or life-threatening abnormalities. We categorized ECG abnormalities retrospectively, based on our review of the tracings in this series rather than according to a pre-established set of definitions. Most important, our gold standard for
CONCLUSION
We believe our findings call into question the validity of currently mandated quality assurance programs for review of ED ECGs and justify a more rigorous examination of these issues. Ideally, this would involve a combined effort by emergency physicians and cardiologists, with pre-established ECG categories and consensus processes. We are currently planning to undertake such a study. II
REFERENCES 1. Accreditation Manual for Hospitals. Ohicago,Joint Commission on Accreditationof Healthcare Organizations, 1990, p 44. 2. Lee T, Weisberg MC, Brand DA, et al: Candidatesfor thrembolysisamong emergencyroom patients with acute chest pain. Ann Intern Mad t989;110:957-962. 3. Lee TH, RouanGW, Weisberg MC, et al: Clinical characteristicsand natural history of patients with acute myocardial infarction sent homefrom the emergencyroom.Am J Cardiol 1987;60:219-224. 4. RusnakRA, Stair TO, Hansen K, et ah Litigation againstthe emergencyphysician: Common features in cases of missed myocardial infarction. Ann EmergMad 1989;18:1029-1034. 5. Heffman JR, Igarashi E: Influence of electrocardiographicfindings on admission decisions in patients with acute chesLpain. Am J Med 1985;79:699-707. 6. BeharS, Schor S, Kariv I, et ah Evaluationof electrocardiogramin emergencyroom as a decision-making tool. Chest1977;71:486-491. 7. Levy R, Goldstein,Trott A: Approachto quality assurancein an emergencydepartment:A oneyear review. Ann Emerg Mad 1984;13:166-169. 8. O'LearyME, Smith MS, O'LearyDS, et ah Application of clinical indicators in the emergency department. JAMA 1989;262:3444-3447. 0. Marriott HJL: Practical Electrocardiography, ed 7. Baltimore,Williams & Wilkins, 1984. 10. Willems JL, Arnaud P, Van BemmelJH, et ah Assessmentsof the performanceof electrocardiographiccomputerprogramswith the use of a referencedata base:Circulation 1985;71:523-534.
Address for reprints: Jerome R Hoffman, MD, FACEP, UCLA Emergency Medicine Center, 924 Westwood Boulevard, Suite 300, Los Angeles, California 90024.
assessment of ECG abnormalities as well as clinical management consisted of the judgment of the two authors who reviewed the cases; we did not develop a consensus process to resolve differenees in ECG interpretation 1° or evaluation of the appropriateness of care, nor did we obtain input from the cardiology faculty in this regard. It is worth noting, however, that although the authors who reviewed the ECGs and clinical records are full-time emergency physicians, they are both board certified in internal medicine, and the second author is also board certified in cardiology.
JANUARY 1992
2 1 : 1 ANNALS OF EMERGENCY MEDICLNE
15/23
I
ECG, quality assurance; quality assurance, ECG
Quality Assurance in the Emergency Department: Evaluation of the ECG Review Process From the Department of Medicine,
Marie Kuhn, MD, FACEP
UCLA School of Medicine; and the
Marshall T Morgan, MD
UCLA Emergency Medicine Center,
Jerome R Hoffman, MD, FACEP
Los Angeles, California. Receivedfor publication January 5, 199l. Revision received July 15, 199]. Acceptedfor publication July 19, 1991.
Study objective: TOdetermine whether the review of emergency department ECGs by cardiologists contributes to the quality of patient care. Study design: We retrospectively analyzed ED ECGsand compared interpretations of the treating emergency physicians with those of the reviewing cardiologists. We then evaluated the ED care of patients with potentially significant ECG abnormalities that were not detected by the treating emergency physicians, as well as the care of patients whose ECGswere "flagged" by the reviewing cardiologists as needing follow-up. Setting:
University hospital.
Participants:
Four hundred consecutive ECGsobtained on ED
patients. Measurements and main results: Thirty-three of the 400 tracings had undetected potentially significant or critical ECG abnormalities; this adversely affected patient care in two cases, These two tracings were not flagged by the reviewing cardiologists. Thirteen ECGswere flagged by the cardiologists; patient care was not altered in any of these cases.
Conclusion: Review of ED ECGs by cardiologists did not affect patient care at our institution. [Kuhn M, Morgan MT, Hoffman JR: Quality assurance in the emergency department: Evaluation of the ECG review process. Ann Emerg Med January 1992;21:10-15.]
18/10
I
ANNALS OF EMERGENCYMEDICINE 21:1 JANUARY1992
ECG R E V I E W
Kuhn, Morgan & Hoffman
INTRODUCTION
As part of emergency department quality assurance programs, the Joint Commission on Accreditation of Heahheare Organizations (JCAHO) requires that ECCs performed in the emergency department be reviewed by "physicians with such privileges" and that their interpretations be made available to emergency physicians. 1 Our extensive literature search (Melvyl Medline and Orion searches involving all possibly related subject headings over ten years, cheek of references in all articles identified, and survey of all books on quality assurance in a university medical library) did not, however, reveal any studies that substantiate the clinical efficacy of such a quality assurance system. Several studies concerning the diagnosis of myocardial isehemia in the ED have reported that failure to appropriately interpret the EGG was a significant factor in the treating physician's failure to diagnose myocardial ischemia. 2-4 However, others have questioned whether ECG changes do, or should, play a major role in decision making in this setting. 5-6 One published review of a teaching institution's ED quality assurance program found that during one year (1982), 74 ECGs had been misinterpreted in their ED and that 28% of those eases required immediate call-back.7 However, their study, which was designed to analyze their entire quality assurance program and did not specifically address the efficacy of their EGG review proeess, left a number of questions unanswered. The frequency with which errors in EGG interpretation occurred is unclear, as they did not include the total number of ECGs reviewed during that year. Also, the clinical significance of the ECG misinterpretations was unclear as they failed to note the criteria by which they concluded that an ECG had been misinterpreted or that urgent patient call-back was indicated. Nor did they note, beyond their decision to call the patient back, whether their system affected patient outcomes. Thus, there does not appear to be an examination of the ECG review process analogous to the study of the radiograph review process produced by O'Leary et al in which the efficacy of their quality assurance system was determined by its impact on patient outcomes. 8 We therefore decided to retrospectively review the ECGs taken in our ED to assess whether our quality assurance program succeeds in capturing all tracings with undetected, potentially significant or life-threatening abnormalities, as well as whether the identification of such tracings by the reviewing cardiologists affects patient care.
JANUARY 1992
21:1 ANNALS OF EMERGENCY MEDICINE
At our institution, ED patients are treated primarily by emergency medicine and internal medicine house officers, who are supervised by emergency medicine faculty-. EGGs are interpreted initially by the housestaff and are selectively referred, by either the housestaff or nursing staff, to emergency medicine faeuhy for concurrent review. In an attempt to ensure appropriate patient care and to comply with JCAHO dictates, we developed, in conjunction with the cardiologist who supervises the EGG reading laboratory, the following quality assurance process. First, two copies are made of each ECG performed in the ED, one of which is attached to the ED record, and the other to an ECG request form that contains space for a "preliminary reading" to be completed by the emergency physician. Second, ECGs and request forms are sent to the ECG reading laboratory- on a daily basis and are read within 24 hours by members of the cardiology faculty. Third, the reviewing cardiologists are expected to "red-flag" all EGG tracings that contain potentially critical abnormalities not documented in the emergency physician's preliminary reading; red-flagged tracings are sent immediately, by ECG laboratory personnel, to the ED, where they are given to the physicians who initially treated the patient or, in their absence, to the senior emergency medicine resident or faculty. When the cardiologists reading the ECG believe there are undetected, life-threatening abnormalities, they have been instructed to telephone the ED and speak directly to the ED faculty, in addition to flagging the tracing. Fourth, it is the reviewing emergency physician's responsibility to re-evaluate the ECG along with the cardiologist's interpretation and the patient's medical record and to arrange for appropriate follow-up. Finally, the reviewing emergency physician notes on the ECG request form what action has been taken and signs the form. A log of these activities is maintained by the ED clerical staff. Recognizing that even a normal ECG does not eliminate the possibility of cardiac disease, we chose to limit our study to an examination of the efficacy of our quality assurance system in those cases with potentially clinically significant ECG abnormalities. First, we sought to determine whether clinically significant errors in ECG interpretation were made by the treating physicians in our ED; that is, did the treating physicians fail to detect significant EGG abnormalities and, if so, did their failure to detect these abnormalities adversely affect patient care?
11/19 I
ECG REVIEW
Kuhn, Morgan & Hoffman
Next, we asked if any of the initial errors in ECG interpretation were detected by the reviewing cardiologists and whether, in such instances, this information was communicated to the emergency physicians. Finally, we sought to determine whether there were any cases in which this quality assurance system had a beneficial effect on patient care.
MATERIALS AND METHODS Four hundred consecutive ED ECGs were collected over a two-month period beginning January 1, 1988. Tracings were Figure 1.
Classification of ident~ed ECG abnormalities according to potential severity. ECGs deemed benign Normal Unchanged from previous tracings Minor ECG abnormalities Sinus tachycardia with a rate < 110 Sinus bradycardia with a rate > 55 Occasional premature atrial contractions Isolated first-degree atrioventricular block Leftward axis > - 6 0 ° Rightward axis ~ +90 ° Atrial enlargement (right or left) Intraventricular conduction delay Possible left ventricular hypertrophy by voltage criteria only Nonspecific ST-T wave changes
ECG abnormalities in this series deemed potentially clinically significant Arrhythmias that might indicate underlying disease or require evaluation and/or therapy, including: Atrial fibrillation with normal {70 to 110) ventricular response rate Sinus rates > 40, < 65, and > 110 Indications of old infarct without acute changes Left ventricular or right ventricular strain patterns Paced rhythms Right bundle branch block or left bundle branch block QRS axis < - 6 0 or > +g0 T wave abnormalities Prolonged QT interval ST changes, cannot rile out ischemia
ECG abnormalities in this series deemed potentially life threatening Indications of acute myocardial injury Potentially critical arrhythmias, including: Atrial fibrillation with rate > 200 Sinus bradycardia with rate of 35
obtained from the ECG reading laboratory and the ED clerical staff. Each of the tracings contained a cardiologist's interpretation accompanied by an ECG request form documenting the emergency physician's preliminary reading, if one was present. Evidence of the reviewing emergency physician's follow-up was s9ught on all tracings that had been red-flagged. A preliminary review of all 400 tracings was undertaken by two of the authors to determine which cases would require review of the medical records to evaluate the ECG abnormalities in light of their clinical context. The tracings were divided by these reviewers into one of three categories: 1) ECGs that were normal, unchanged from prior tracings, or contained only minor abnormalities; 2) ECGs with potentially clinically significant abnormalities; and 3) ECGs with potentially life-threatening abnormalities. The ECG abnormalities assigned to each category are outlined (Figure 1). ECGs assigned to category I were deemed sufficiently benign to eliminate the need for further review. Tracings with abnormalities assigned to categories II or III were thought to be potentially significant or life threatening, depending on the clinical context. If these abnormalities were not documented in the ECG request form's "preliminary reading" box, the medical records were obtained and reviewed. The emergency physician's notes were examined for documentation that the ECG had been correctly interpreted; if such documentation was found, the record was not examined further. If this documentation was lacking, the remainder of the medical record was reviewed by the first author. Cases that were determined to have been appropriately managed were not subjected to additional review; cases that the author believed were inappropriately managed were referred to the second author for review. As the reviewers agreed regarding the clinical mismanagement in each such case, they were therefore so designated. Charts that had been red-flagged by the cardiologists were analyzed separately to determine the efficacy of this action. The medical records for these patients were examined to determine whether the emergency physician notified had chosen to alter patient care beeause of the cardiologist's reading of the ECG. Charts of patients whose ECGs were not red-flagged by the cardiologists but were deemed by the investigators to have lifethreatening abnormalities were reviewed to determine if they had been appropriately managed and if patient care might have been altered had these ECGs been red-flagged and emergency physicians alerted one day after the ED visit. ~"
Junctional rhythm with bundle branch block versus accelerated ventricular rhythm All tracings red-flagged by reviewing cardiologist
20/12 |
ANNALSOF EMERGENCY MEDICINE
21:1
JANUARY1992
ECG R E V I E W Kuhn, Morgan & Hoffman
RESULTS Four hundred ECGs were reviewed; 289 of the ECG request forms docmnented the treating physician's initial interpretations, whereas 111 did not (Figure 2). Of the tracings with documented preliminary readings, the interpretations of the treating physician, reviewing cardiologist, and investigator agreed in 164 of 289 cases (56.7%) (Figure 2). In 89 of 289 cases (30.8%), the discrepancies in ECG interpretations were deemed minor. Twenty-five of 289 tracings (8.6%) had potentially significant ECG abnormalities for which the interpretations of the emergency physician and cardiologist differed. In seven of these cases, the emergency physician detected the ECG abnormalities, but the cardiologists did not; in the other 18 cases, the emergency physician failed to note the ECG abnormalities detected by the cardiologist. Review of the medical records for these 18 cases did not reveal a single instance in which the patient appeared to have been inappropriately managed, nor did it appear that patient care would have been altered by the additional information garnered fl'om the cardiologist's reading of the ECG. Of those ECGs without preliminary readings noted on the ECG request form, 38 of 111 (34.2%) were normal or unchanged from previous tracings, whereas 15 (13.5%) had only minor abnormalities. Fifty-one of 111 (45.9%) had abnormalities deemed to be Figure 2. Findingsfrom reviewof 400 consecutiveED ECGs.
potentially clinically significant; in 50 of these 51 cases, medical records were available for review. Thirty-eight of the medical records documented that the ECG had been appropriately interpreted in the ED and simply had not been recorded on the ECG request form. In the remaining 12 cases, there was no documentation that the ECG had been interpreted in the ED; review of the medical records indicated that all of these patients had been treated appropriately and revealed no instance in which it appeared that care would have been ahered by the information provided by the cardiologist. Thirteen cases were red-flagged by the reviewing cardiologists; ten of these cases had a preliminary reading noted on the ECG request form, whereas three did not. Each of these cases was reviewed and signed by an emergency physician; in no such case was patient care altered by the reviewing emergency physician. Care was not altered in the three red-flagged cases without documented preliminary readings because the tracings had actually been accurately interpreted by the treating physicians who had merely failed to note their readings on the ECG request form. Care was not altered for any of the ten cases in which preliminary readings failed to note findings red-flagged by the cardiologist because the emergency physicians notified of these findings thought, in each case, that the patients had already been treated appropriately. In the only instance in which the reviewing cardiologist actually phoned the ED to report a red-flagged ECG finding, the treating physicians were already aware of the ECG abnormalities and had acted on them. •
400 ECGs
111 ECGswithout preliminary readings
289 ECGswith preliminary readings 1
6
4
1
~
° cc"'a'',eto8° minor discrepancies
.
8
I'
I
%
o
2518.6%1 Potentially significant discrepancies
7
Abnormalities not detected by cardiologists
)
"h
cr'tP:'
lO~~"~|~'molities
Tracings red-flagged
0 Cases mismanaged
Tracings red-flagged Interpreted accurately
Without ED ECG interpretation
JANUARY 1992 21:1 ANNALS OF EMERGENCY MEDICINE
Case mismanaged
Not red-flagged
0
2
Cases
Cases
mismanaged
1
Cases mismanaged
716.3%) Life-threatening abnormalities
J t 51 (45.9%) Potentially significant abnormalities
Critical abnormality not detected by ED physician or cardiologist Record unavailable
18 Abnormalities not detected by ED physicians
I
38 ( Normal or unchanged and 15113.5%) minor abnormalities
mismanaged
0 Cases mismanaged
13121
I
ECG R E V I E W Kuhn, Morgan & Hoffman
There were five tracings that the reviewers thought should have been red-flagged by the cardiologists but were not. Two cases that did not have a preliminary reading documented were managed appropriately; both patients had evidence of acute isehemia and were admitted to the coronary care unit. The three remaining cases may have been mismanaged. The first was a 48-year-old hypertensive man with a history of cocaine abuse who eomplained of chest tightness. His ECG (Figure 3) was interpreted by both the emergency physician and the over-reading cardiologist as demonstrating "left ventricular hypertrophy with secondary repolarization abnormalities." His hypertension was treated, and he was discharged from the ED. The authors interpreted his ECG as indicating either anterior wall epieardial injury and lateral wall ischemia or early repotarization abnormalities.9 We believe that the tracing should have been red-flagged and the patient asked to return for re-evaluation. We also believe that the patient should have been admitted at the time of his ED visit on the basis of elinieal factors (a complaint of chest tightness in a patient with hypertension and a history of cocaine abuse) alone and that appropriate care should therefore not have depended on subsequent review of the ECG. The second ease involved an 85-year-old man whose tracing was read as showing "extreme sinus bradyeardia, rate 35; complete right bundle branch block with left anterior fascicular block and first-degree atrioventrieular b l o e k . . , represents trifascicular disease." His ECG was not red-flagged. Review of this patient's medical record indicated that the emergency physician Figure 3. ECG tracing from a 48-year-old hypertensive man with a history of cocaine abuse who complained of chest tightness. His ECG was originally interpreted as demonstrating "left ventricular hypertrophy with secondary repolarization abnormalities. "' The authors believe that his tracing was more consistent with either anterior wall epicardial injury and lateral wall ischemia or with early repolarization abnormalities.
22/ 14
interpreted the ECG accurately but failed to document his reading on the ECG request form. However because the ECG abnormalities had been present on a previous tracing, the emergency physician elected to discharge the patient. Once again, although we believed that this patient had been managed inappropriately, the decision as to how to treat him would not have been changed by red-flagging his ECG because the treating physician was already aware of the pertinent findings. The third case involved a 65-year-old woman who presented in atrial fibrillation with "occasional aberrantly conducted beats" and a heart rate of 225; her medical record revealed that she had experienced four episodes of transient atrial fibrillation in the past year. Her ventricular response rate during the visit in question was miscalculated as averaging 180. She was treated with IV digoxin, and after her ventricular rate decreased to approximately 85 she was discharged to seek follow-up at another facility. The reviewing cardiologist raised the possibility of a bypass tract; the investigators contacted the treating physicians and made them aware of the cardiologist's reading as well as the contraindications to the use of digoxin in patients with possible bypass tract disease. On review, we believed that the likelihood of a bypass tract was small; regardless of theoretical considerations, the patient did not suffer from the treatment she received, nor would next-day red-flagging have been able to prevent any. adverse effects had they occurred.
DISCUSSION
We began this study questioning whether a quality assurance system would benefit ED patients with abnormal ECGs and, if so, whether the quality assurance process in place at our institution was effective. Our study was not able to clearly answer the first question. If one defines the need for a quality assurance system by the percentage of undetected ECG abnmmalities, then the fact that 8.3% of the tracings had undetected significant or lifethreatening abnormalities would strongly support such a need. If, on the other hand, a quality assurance process is valued according to its potential impact on patient care, there were only two of 399 patients (0.5%) whose care was possibly adversely affected by the treating physician's failure to appropriately interpret the ECG, because one patient was mismanaged despite an accurate interpretation of this ECG by the emergency physician. Whether it is worthwhile to establish a quality assurance system to have an impact on so few patients depends on both the cost and effort involved and whether an effective quality assurance system can be constructed and implemented, ll~
ANNALS OF EMERGENOY MEDI01NE 21:1 JANUARY 1992
ECG R E V I E W
Kuhn, Morgan & Hoffman
Our results clearly indicate that our current quality assurance system is not effective. In the first place, not all of the life-threatening ECG abnormalities undetected in the ED were captured. The reasons for this failure are not entirely clear: in particular, two such ECGs were read appropriately but not flagged by a cardiologist who was apparently aware of the quality assurance system, as he had red-flagged other tracings in this series. Second, the process of red-flagging tracings did not result in any changes in patient care; in every such instance, the emergency physician who was notified decided that the patient had already been managed appropriately. The frequent (111 of 400, 28%) failure of the treating physicians to complete the Preliminary reading section on the ECG request form may well have contributed to the ineffectiveness of our quality assurance system. We might conjecture that this occurred because physicians in training are reluctant to commit their ECG interpretations to paper or that they failed to recognize this step as an important facet of patient care, but we have no data to support these views. An informal telephone survey of ten other academic institutions revealed that our quality assurance system is similar to that of other teaching hospitals. It is therefore tempting, given our results, to argue that the JCAHO mandate regarding the review of ED ECGs by "physicians with such privileges" should be reconsidered. There are, however, several aspects of our study design that limit our ability to make such an argument. We limited our review to those patients whose ECGs demonstrated undetected significant or life-threatening abnormalities. We categorized ECG abnormalities retrospectively, based on our review of the tracings in this series rather than according to a pre-established set of definitions. Most important, our gold standard for
CONCLUSION
We believe our findings call into question the validity of currently mandated quality assurance programs for review of ED ECGs and justify a more rigorous examination of these issues. Ideally, this would involve a combined effort by emergency physicians and cardiologists, with pre-established ECG categories and consensus processes. We are currently planning to undertake such a study. II
REFERENCES 1. Accreditation Manual for Hospitals. Ohicago,Joint Commission on Accreditationof Healthcare Organizations, 1990, p 44. 2. Lee T, Weisberg MC, Brand DA, et al: Candidatesfor thrembolysisamong emergencyroom patients with acute chest pain. Ann Intern Mad t989;110:957-962. 3. Lee TH, RouanGW, Weisberg MC, et al: Clinical characteristicsand natural history of patients with acute myocardial infarction sent homefrom the emergencyroom.Am J Cardiol 1987;60:219-224. 4. RusnakRA, Stair TO, Hansen K, et ah Litigation againstthe emergencyphysician: Common features in cases of missed myocardial infarction. Ann EmergMad 1989;18:1029-1034. 5. Heffman JR, Igarashi E: Influence of electrocardiographicfindings on admission decisions in patients with acute chesLpain. Am J Med 1985;79:699-707. 6. BeharS, Schor S, Kariv I, et ah Evaluationof electrocardiogramin emergencyroom as a decision-making tool. Chest1977;71:486-491. 7. Levy R, Goldstein,Trott A: Approachto quality assurancein an emergencydepartment:A oneyear review. Ann Emerg Mad 1984;13:166-169. 8. O'LearyME, Smith MS, O'LearyDS, et ah Application of clinical indicators in the emergency department. JAMA 1989;262:3444-3447. 0. Marriott HJL: Practical Electrocardiography, ed 7. Baltimore,Williams & Wilkins, 1984. 10. Willems JL, Arnaud P, Van BemmelJH, et ah Assessmentsof the performanceof electrocardiographiccomputerprogramswith the use of a referencedata base:Circulation 1985;71:523-534.
Address for reprints: Jerome R Hoffman, MD, FACEP, UCLA Emergency Medicine Center, 924 Westwood Boulevard, Suite 300, Los Angeles, California 90024.
assessment of ECG abnormalities as well as clinical management consisted of the judgment of the two authors who reviewed the cases; we did not develop a consensus process to resolve differenees in ECG interpretation 1° or evaluation of the appropriateness of care, nor did we obtain input from the cardiology faculty in this regard. It is worth noting, however, that although the authors who reviewed the ECGs and clinical records are full-time emergency physicians, they are both board certified in internal medicine, and the second author is also board certified in cardiology.
JANUARY 1992
2 1 : 1 ANNALS OF EMERGENCY MEDICLNE
15/23
I
