This classic EKG is an example of why doctors get paid to think. The computer interpreted this patient as having the following abnormalities: sinus tachycardia, paired ventricular premature complexes, aberrant complex, possibly supraventricular, right atrial abnormality, nonspecific intraventricular conduction delay, probable left ventricular hypertrophy and consider anterior infarct.
As a hospitalist, I have an obligation to review electrocardiogram tracings on my patients and understand what I'm seeing. If I don't understand the consequences of the data in front of me, I have an obligation to determine whether it has significant or insignificant relevance to the patient's care plan. For example, patients nearing end of life with an abnormal EKG do not need cardiologist opinions on what to do next. All data should be interpreted with the patient's clinical situation in mind.
In addition, we have an obligation to review data that computers may not have the sophistication to decipher. In this example EKG below, the computer interpreted numerous abnormalities, some that could require urgent bedside evaluation. As a hospitalist, this abnormal EKG is usually handled by phone. I'll get a page from the nurse telling me the EKG computer has a reading suggesting heart attack. If the EKG was obtained on my patient I have been rounding on, I have an good understanding of the clinical scenario for which it was obtained.
However, if the EKG was performed on a cross covering patient for whom limited knowledge of the nuances of situation are unknown, I am left with a higher level of uncertainty when abnormal computer EKG readings are obtained. Many times a computer will read a chronic finding as a critical SST elevation myocardial infarction instead of the chronic changes seen on all 92 EKGs they have had done in the last six months.
What about the EKG pictured below? The computer has seven individual diagnoses, including sinus tachycardia, paired ventricular premature complexes (PVCs), aberrant complexes possible supraventricular complexes, right atrial abnormality, nonspecific intraventricular conduction delay, probably left ventricular hypertrophy, and consider anterior infarct. However, that's not how I would interpreted the ECG. This is a poorly obtained EKG with a wandering baseline. I would recommend getting another EKG right after administering Ativan to both the patient and the ECG technologist who allowed this EKG to become a permanent part of the patient chart.
How reliable are EKG computer interpretations? The computer algorithm can only be as successufl as the input data. That's why it's important to give the computer a fighting chance in making a correct diagnosis. That's why great electrode to skin contact and correct lead placement is imperative. The skin should be cleaned and shaved well for placement of the electrode pads. I found a great resource detailing the sensitivity and specificity of EKG computer interpretations. In the referenced 1991 study, computers made the correct read 69.7% of the time compared to 76.3% for cardiologists. This linked resource goes on to describe how good computer algorithms are with ST segment changes, a finding that would define the need for thrombolytic therapy and they describe the recommended prudence of computer reads from the consortium statement by the AHA/ACC and HRS:
Make sure to see all my posts about EKGs.
As a hospitalist, I have an obligation to review electrocardiogram tracings on my patients and understand what I'm seeing. If I don't understand the consequences of the data in front of me, I have an obligation to determine whether it has significant or insignificant relevance to the patient's care plan. For example, patients nearing end of life with an abnormal EKG do not need cardiologist opinions on what to do next. All data should be interpreted with the patient's clinical situation in mind.
In addition, we have an obligation to review data that computers may not have the sophistication to decipher. In this example EKG below, the computer interpreted numerous abnormalities, some that could require urgent bedside evaluation. As a hospitalist, this abnormal EKG is usually handled by phone. I'll get a page from the nurse telling me the EKG computer has a reading suggesting heart attack. If the EKG was obtained on my patient I have been rounding on, I have an good understanding of the clinical scenario for which it was obtained.
However, if the EKG was performed on a cross covering patient for whom limited knowledge of the nuances of situation are unknown, I am left with a higher level of uncertainty when abnormal computer EKG readings are obtained. Many times a computer will read a chronic finding as a critical SST elevation myocardial infarction instead of the chronic changes seen on all 92 EKGs they have had done in the last six months.
What about the EKG pictured below? The computer has seven individual diagnoses, including sinus tachycardia, paired ventricular premature complexes (PVCs), aberrant complexes possible supraventricular complexes, right atrial abnormality, nonspecific intraventricular conduction delay, probably left ventricular hypertrophy, and consider anterior infarct. However, that's not how I would interpreted the ECG. This is a poorly obtained EKG with a wandering baseline. I would recommend getting another EKG right after administering Ativan to both the patient and the ECG technologist who allowed this EKG to become a permanent part of the patient chart.
How reliable are EKG computer interpretations? The computer algorithm can only be as successufl as the input data. That's why it's important to give the computer a fighting chance in making a correct diagnosis. That's why great electrode to skin contact and correct lead placement is imperative. The skin should be cleaned and shaved well for placement of the electrode pads. I found a great resource detailing the sensitivity and specificity of EKG computer interpretations. In the referenced 1991 study, computers made the correct read 69.7% of the time compared to 76.3% for cardiologists. This linked resource goes on to describe how good computer algorithms are with ST segment changes, a finding that would define the need for thrombolytic therapy and they describe the recommended prudence of computer reads from the consortium statement by the AHA/ACC and HRS:
"Computer-based interpretation of the ECG is an adjunct to the electrocardiographer, and all computer-based reports require physician overreading. Accurate individual templates should be formed in each lead before final feature extraction and measurement used for diagnostic interpretation."Given the world of computer generated data and template care, it's easy to see how physician failure to correct bad computer data will get easier and easier as more information becomes algorithmic in nature.
Make sure to see all my posts about EKGs.
Happy,
ReplyDeleteWhy can't you bill for the EKG interpretation?
So, find out who is not reading the ekgs, but billing for them. Then turn them in for fraud to the Medicare police. After all, they are committing fraud by billing for a service they never provided. But, that might cause some political problems at your hospital I bet. Might threaten your employment. By definition, that makes you complicit as well.... You are, in fact, protecting your own income stream by overlooking obviously illegal activities at your place of business.
ReplyDeleteYou can get an echo at 3 am?!
ReplyDeleteAs for all the other stuff you said, right on, man.