Check out this short but impactful essay entitled, “In the US, patient data privacy is an illusion” by my friend and mentor Dr. Harlan Krumholz, in the British Medical Journal. The piece describes just how little control patients in the United States have over what happens to their healthcare data stored on modern computers, networks, and “in the cloud.”
With electronic medical records being increasingly integrated, both good and bad things can occur. For me, as an emergency physician, if I have a sick patient who is unable to tell me their medical history, I might be able to glean key information quickly—when time is of the essence—both from prior records found in our own hospital’s network and in some cases by accessing patients’ medical records from outside institutions, using a feature in Epic (our electronic medical record) called Care Everywhere. Care Everywhere allows healthcare providers to see at least some patient health information from other healthcare organizations that also use Epic. That feature has saved me and my colleagues countless hours of having to call (or fax!) other places to find out simple things like “What did this patient’s most recent cardiac ultrasound show?” Information like that can be life-saving. Care Everywhere in Epic can help patients. In a system lacking one central national database, it’s better than nothing.
But there are potential problems. As Harlan points out, too often, "healthcare providers find out more information than they strictly need to know. Just this evening, Harlan and I were talking about an example of a test that gets overused and therefore stands to harm patients (by jeopardizing privacy) more than it stands to help them: drug screening tests. These are low-yield tests (especially in ERs) that almost never change the treatment plan for a patient. So why do we do them? In some cases, we have no choice—a psychiatric patient simply won’t be accepted into an appropriate facility without that drug test.
But why? If a patient took a drug a week or two ago, their test could still come back positive. What information would that add? A positive drug test does not mean the patient is intoxicated and therefore has not been “medically cleared” and therefore ready for psychiatric care only. The question is not whether a patient has a “clean” drug test, but whether they are clinically intoxicated at that moment. A drug test does not tell us that information. Talking to the patient and examining them tells us that. All the drug testing does is create a paper trail. That paper trail does nothing to help a patient. Drug tests in the ER provide useless information most of the time, leaving side how inaccurate they can be. Adding these results to the “permanent record” of a patient’s electronic chart is unhelpful—and in some cases unethical.
Then there’s the fact that patient data too often gets mined for profit. Even quasi- or completely-anonymized data ends up being sold to outside companies, and usually without the direct knowledge or approval of patients. Patients should have a say in this, Harlan argues. It’s a principled argument, I must say.
Still, when patient data are truly anonymized, the insights we gain (from “big data”) can be remarkable. In the digital age, though, consent is crucial.
In enigmatic cases where the diagnosis is unclear, obtaining an accurate history often leads to the correct diagnosis. Patients, including those with disorders of brain function, are often unable to give an accurate history, and collateral sources, including electronic medical records, can be the key to arriving at the correct diagnosis. While issues of privacy are a concern, often obtaining vital information is delayed because of fear of the consequences of releasing medical information. When physicians must make critical life-saving or life-altering decisions, this can be a barrier to optimal care.
Although I understand your reasoning that drug screening is often superfluous, there are situations where it is useful.