It’s Stupid to Take a History

Absolutely idiotic. And the sooner we realize that it’s incredibly inefficient and ineffective for physicians to question patients about the nature of their medical problems, the faster we’ll be able to increase the quality and decrease the cost of healthcare. I’m dead serious in making this point, albeit a wee bit overly dramatic. I figured that’s the best way to get your attention focused on one of our most powerful opportunities for improving both the healthcare and health status for all the inhabitants of this planet.

Think about it. Clinicians today, in the Age of Managed Care, are pressured to spend minimal time in each patient encounter. And a significant proportion of that time is taken up by asking the patient a substantial number of questions, the vast majority of which end up being answered in the negative. Waste of (rather expensive) doctor time. And there’s no way that physicians could ever find the time to elicit all the information relevant to determining the cause of a patient’s problem. Even if they did, it wouldn’t matter. Because it’s highly doubtful – surely for primary care providers, but also for specialists – that they could possibly have learned all that they need to know about all the patient problems they might confront, have remembered all that they have learned, and can effectively match that knowledge to the symptoms they hear…all within the constraints of a distressingly brief encounter. In fact, the limited time available takes this from “highly doubtful” to “virtually impossible”.

I can predict the reaction of most physicians to this assertion. I predict ridicule. I sense hatred. I hear passionate claims that, “By watching and listening to the patient’s response to my questions, I can skillfully hone in on the cause of the problem. I know the questions to ask, and after just a few minutes of interaction, the Art of Medicine enables me to understand the patient’s condition and to know what action to take.” Baloney. With all due respect, Doctor, you are playing a game of probabilities. The chief complaint indicates a few questions that will determine which common cause could likely be the etiology. You will generally go with that one. Your thinking is that if you’re wrong, the patient will come back when the treatment doesn’t provide a remedy. A workable strategy in many cases. But what about the serious, rapidly progressing disorders?!

This approach was fine in the Age of Hippocrates – and even in the Age of Dr. Marcus Welby. But not in the Age of the Computer. And it’s outrageous in the Age of the Internet.

“Practicing the Art of Medicine.” There’s something terribly wrong with this phrase. (Read it slowly.) But it’s also disturbingly accurate. Methinks that – given the importance of our healthcare and health – we patients deserve far more than what practicing an art can provide. But how do we begin to change the process? How do we apply the Science of Medicine to every clinical encounter? Because then we can let physicians practice their art from a higher platform, starting with all the useful information and guidance the science can provide. And once we’ve done that – here’s a revolutionary concept?! – let’s analyze what we learn from correlating the inputs, process, and outcome of those gazillion clinical episodes, and advance the Science of Medicine. Most physicians are highly dedicated, and will embrace the opportunity to provide better medical care. They are handicapped today not just by managed-care constraints, but even more so by the lack of the right tools and the right model for acquiring, documenting, and processing clinical data.

Yeah, well just how do we get things to where they need to be? While it’s stupid to take a history, it is surely not stupid to acquire a history – to get the needed data documented efficiently. First we have to recognize that the computer has the potential to do a much better job of acquiring a patient history than a physician can. And then we have to realize (this is really going to get me into trouble) that the computer has the potential to do a much better job of processing the patient history than a physician can. That’s how this article got its title. Let’s assume that you, the reader, have bought into these assertions at least enough to be willing to see how they could be implemented. Here’s a summary of one possible process….

We will start at the time the patient determines that s/he might have a (likely non-emergent) medical problem. That patient should get on the Internet, or go to a kiosk in a shopping mall or post office. After entering the nature of the problem (headache, stomach pain, rash, etc.), the computer software – which we will call CGS, for “Clinical Guidance System” – asks the patient a series of branching questions, seeking all the data that the current Science of Medicine believes will help determine the cause of that particular type of problem. An automated triage is then performed. The patient is advised to either: immediately go to an emergency room, schedule an appointment with a primary care clinician (within 48 hours, 2 weeks, or 4-6 weeks), bypass primary care and schedule with a specialist, or undertake specified self care.

When the patient arrives at the doctor’s office, the summarized history is already available in the computer-based patient record system (remember, we are idealizing here), and the CGS has even suggested some potential causes for the problem, providing probabilities and supportive knowledge as appropriate. Additionally, the CGS specifies what additional information – physical examination findings, laboratory tests, medical imaging – should be acquired to most cost effectively arrive at a definitive and accurate diagnosis. The physician can document an elaboration of the patient’s history as well as the physical exam, via structured/codified input and/or free text, using either digital dictation or speech recognition to generate the free text.

Once the diagnosis is reached, the CGS recommends a management plan, which is optimized for the patient’s unique set of presenting findings and test results. The patient is followed up, and the outcome is entered into the record. Of course, if the outcome is anything other than a satisfactory remedy, then it serves as new input to another assessment by the CGS.

Now here comes the fun part. All those data – patient history, physical exam, test results, treatments, and outcome – are shipped off (with relevant encryption of identifiers) to a centralized Clinical Analysis System (CAS). Imagine: comprehensive, codified data on every clinical episode throughout the nation or across the globe, available for analysis by salivating biostatisticians, epidemiologists, and other medical researchers. Although we clearly don’t have a controlled environment here, like with double-blind studies, powerful statistical techniques can correct for most of the potentially confounding effects of tertiary variables. And, if there is still some uncertainty about the conclusions, we can use the results to guide (the invariably expensive and narrowly focused) controlled studies.

One major goal of the CAS is to identify the initial signs of disease, so that we can improve our skills at triage and early diagnosis. Another is to determine which are the best treatments for patient problems, taking into consideration each patient’s specific findings, not just the general diagnosis. A third important goal is to improve the health status of every person in the world. In order to accomplish this particular mission, of course, we’ll need to gather up data on lifestyle, baseline medical history, and family history.

The final thrust is to insert all this new knowledge back into the CGS – implementing a feedback loop, continuous quality improvement, learning from experience, or whatever you want to call it. We will then not only have advanced the Science of Medicine, we will also have brought all our learning to where it belongs: the point of care.

This brief article certainly oversimplifies the challenge of healthcare documentation, and also overly summarizes the recommended clinician-data-input model. And there is little doubt that most physicians will resist such a major shift in the way they provide and document care. As Larry Weed has said, “If physicians were in charge of airports, there would be no radar…just intensive care units all around the periphery.”

Dr. Weed has put his reengineering efforts where his mouth is, developing innovative, knowledge-based tools^1,2 that address issues raised in this paper. We should not allow more time to pass before exploring the power of Weed’s groundbreaking work and other relevant research in the area. Because we need to start making dramatic improvements in healthcare now! And perhaps the very first step on the path to a healthy tomorrow is recognizing that it’s stupid to take a history.

References

1. Weed LL. New connections between medical knowledge and patient care. BMJ. 1997; 315:231-235.

2. Burger, C. The use of problem knowledge couplers in a primary care practice. Healthcare Information Management. 1997; 11:4:13-26.

Joe Weber is CEO of Narratek, Inc., provider of dictation, transcription, and speech recognition software – and clinical-documentation consulting. He has over 35 years’ experience in medical research, healthcare administration, marketing, and consulting. Joe has a B.A. in Biology from Brandeis, an M.S. in Biostatistics from Columbia, and an M.S. in Management from M.I.T. joeweber@alum.mit.edu.

This paper was originally published in Advance for Health Information Professionals, May 10, 1999