The Emergency Department, by its nature, is interruption driven given it is “uncontrolled and unpredictable and punctuated by intermittent time-critical activities (Chisholm 2000). On average, each Emergency Physician is interrupted 6.6 times per hour while at work; 11 percent of all tasks were interrupted, 3.3 percent of them more than once (Westbrook 2010). One study (Westbrook 2010) calculated that physicians were multitasking 12.8 percent of the time, while not returning to the same task 18.5 percent of the time. Anecdotally, one of the major factors in interruptions in the ED can be associated with interpretation of EKGs. EKGs are typically initiated in triage through nursing protocols to avoid delays in care. In accordance with American Heart Association guidelines, EKGs performed on patients with suspected acute coronary syndrome (ACS) must be interpreted by an Emergency Physician within 10 minutes of patient presentation (Antman 2004). This study looks evaluates the accuracy of an EKG machine interpretation as “normal” in an attempt to reduce interruptions.
Do triage EKGs interpreted as “normal” by computer analysis have any immediate clinical significance or can emergent human review be bypassed?
Patients >18 years old that had EKGs obtained at triage at an academic Emergency Department over 16 weeks. EKGs were performed by triage nurse or EKG technician within 10 minutes of arrival and immediately given to the attending physician for review for chief complaints of: chest pain, chest pressure, chest tightness, weakness, fatigue, palpitations, syncope, dyspnea, any atypical symptoms such as nausea and vomiting or pain in the jaw, upper back, or upper abdomen.
Single center prospective cohort study
EKGs were obtained using MAC 5500 machines and interpreted using Marquette 12SL software. All EKGs were uploaded to a hospital server where board certified cardiologists entered the final EKG interpretation into the EMR. Each EKG interpreted by the computer as normal was compared to the cardiologist’s final interpretation. If both interpreted the EKG as normal, it was considered an accurate computer interpretation. If they differed, the EKG was presented to 2 board certified Emergency physicians blinded to patient presentation, patient care and goals of the study and asked to evaluate for clinical significance defined as a something that would alter standard triage care. If both EPs agreed that the EKG would not alter care, then the EKG was interpreted as normal.
- Of the 855 EKGs obtained, 222 (26%) were reported as “normal” by the computer software.
- Of these, 13 (5.8% of normal EKGs) were flagged as being abnormal by a cardiologist, all found to have mild, non-specific abnormalities.
- One of these 13 EKGs (0.4% of normal EKGs) was interpreted by one of two EPs as having an abnormality that would alter triage care to “[get a] bed immediately”. This patient was ultimately discharged from the department with a normal stress test the next day.
- Overall, the negative predictive value for a triage EKGs interpreted by the computer as “normal” was calculated to be 99% (95% CI = 97% to 99%).
- Cardiologist and interpreting EPs were blinded to patient presentation. Often only the chief complaint is known when EKG is handed to the provider
- Uses EP interpretation with regards to actionable items versus just their clinical interpretation of the EKG, which is more consistent with real-life practice
- Single center study
- Small sample size. Study intended to capture 379 “normal “ EKGs but fell short of this goal
- Incidence of STEMI was low in this cohort (0.57%)
- Only evaluates one EKG software
- Cardiologists not blinded to EKG software read
“Our data suggest that it may not be necessary for the EP to immediately review computer normal EKGs. No delay in patient care or poor outcome was associated with computer-interpreted normal electrocardiograms. This strategy has the potential to reduce interruptions in direct patient care provided by EP physicians.”
While unlikely to fully replace physician interpretation of EKGs, it is possible for the computer to designate those EKGs which are clinically insignificant (those interpreted by the computer as “normal EKG”) in order to prevent physician interruption through immediate interpretation.
Potential Impact To Current Practice
This may allow for less interruption by directly filing the EKG into the patient’s chart (or waiting until the EP is changing tasks) to hand them a normal EKG for evaluation.
EKGs that the computer interpreted as “normal” led to no delay in patient care or adverse outcome. It is possible for the computer to identify clinically insignificant EKGs in order to help prevent physician interruption.
Antman EM et al. American College of Cardiology/American Heart Association Task Force on Practice Guidelines, ACC/AHA guidelines for the management of patients with ST-elevation myocardial infarction. Circulation 2004;110:588–636. PMID:15289388
Berg LM et al. Interruptions in emergency department work: an observational and interview study. BMJ Qual Saf 2013;22:656-663. PMID: 23584208
Chisholm CD et al. Emergency department workplace interruptions: are emergency physicians “interrupt-driven” and “multitasking”? Acad Emerg Med 2000;7:1239–43. PMID: 11073472
Jeanmonod R et al.The nature of Emergency Department interruptions and their impact on patient satisfaction. Emerg Med J 2010 27: 376-379. PMID: 20442168
Westbrook JI et al. The impact of interruptions on clinical task completion. Quality and Safety in Health Care 2010;19:284-289. PMID: 20463369