Using clinical decision support to get the right diagnosis the first time

Joseph Britto is co-CEO of Isabel Healthcare, a clinical software vendor that helps clinicians with diagnosis. He practiced medicine in the UK before joining with co-CEO Joseph Maude to start Isabel, named after Joseph’s daughter who was wrongly diagnosed with Chicken Pox and nearly died as a result. Joseph has a personal connection as he was the physician in charge of Isabel’s recovery.

Remember President Bush’s goal, first stated in the 2004 State of the Union message, of giving “every American” his own EMR by 2014?

That goal seems as elusive as ever, especially in light of a recently released study by the The Center for Studying Health System Change which found a discouragingly low rate of EMR adoption among physicians. The new study, released last month, reported that only 29 percent of the hospitals surveyed were actively supporting physician acquisition of EMRs through financial or technical support. This number was disappointing in light of the current government initiative that has relaxed federal rules on physician self-referral and made available hundreds of millions of dollars in various subsidies for EMR adoption by physicians.

Many health policy experts believed that “if you subsidize it, they will come.” While that approach has worked in persuading people to take mass transit, it hasn’t lured many physicians into using EMRs.

Why the reluctance? One reason is cost. On September 25, 2008, the Certification Commission for Healthcare Information Technology (CCHIT) issued a report that reviewed 90 EMR incentive programs (state, federal, private) with a total funding of $700 million available.

Health Industry Insights,
a private research firm, reviewing the CCHIT report, estimated that the
cost to implement an EMR is roughly $25,000 per physician and that the
$700 million currently available would represent only 9% of the
expected total national cost  to furnish all physicians in private
practice with an EMR.

As we endure financially difficult times, we believe it is time for IT leaders at hospitals and medical groups to take second look at a different set of technologies, one which, like EMRs, can align physicians and hospitals in the shared goals of improving patient care and reducing clinical risk.

Clinical decision support (CDS) technology is not new, it has been available in various forms since 1986, but as computer hardware has become vastly more powerful, the newer versions of the systems have become faster and more practical for physicians to use. One particular kind of CDS technology, diagnosis decision support (DDS), has been adopted in many hospitals in the past two years, as medical executives realize its value in attracting leading physicians who understand and value medical knowledge tools. 

DDS systems, as defined by a leading medical textbook, “link health observations with health knowledge to influence health choices by clinicians for improved health care.”

DDS systems include two key components: a dynamic medical knowledge data base and an inferencing or logic engine to sort and select decision options for clinicians.

Medical leaders have been looking at computers as potential tools to improve clinical decision-making since the early mainframe systems were first installed in hospitals in the 1960s.  After many years of development, one of the first  DDS systems, Dxplain, was installed at Massachusetts General Hospital in 1986, with a data base including 2,000 diseases.

These early versions of DDS technology were frustratingly slow. At the heart of these early systems was a crude form of artificial intelligence (AI). The software required the input of multiple experts to provide semi probabilistic relationships between thousands of clinical features and hundreds of diseases.

These early systems required physicians to spend a considerable amount of time interacting with them, answering a hierarchy of questions. Published trials reported that it took physicians 20 or 30 minutes to enter the data and arrive at a final set of decision options.

Dr. Robert Wachter is Associate Chairman of the Department of Medicine at the University of California, San Francisco and author of two books, Internal Bleeding: The Truth Behind America’s Terrifying Epidemic of Medical Mistakes and Understanding Patient Safety. He has oft written of the frustrations and “overhyping” of the early diagnosis decision support programs.

According to Dr. Wachter, “the disappointment over the ineffectiveness of the early programs led to widespread skepticism that any DDS could help physicians be better diagnosticians. This skepticism is getting in the way of today’s markedly improved systems, such as the Isabel System, from gaining the traction they deserve.”

Today’s computer systems are thousands of times more powerful than those of the 1980s. This vastly improved performance has enabled a variety of different clinical decision support systems to be adopted in hospitals, large and small, across the country. CDS systems today provide clinicians with prescribing decision support, image recognition and interpretation, therapy planning and patient alerts.

My company, Isabel Healthcare, makes a diagnosis decision support that uses  advanced pattern recognition software from Autonomy Inc. The use of pattern-matching software, rather than keyword searching or Boolean query techniques, enables searches to be made against a complex cluster of signs and symptoms entered in free text and the results to be returned as a list of likely diagnoses . Diagnoses are then linked to knowledge effectively allowing the DDS to mobilize knowledge from disparate sources around specific diagnoses right at the point of care.

The Isabel System contains a knowledge library of more than 100,000 documents made up of medical journal articles and textbooks and it is continually updated. Physicians using laptop computers can enter queries in free text and receive a list of the most likely diagnoses in seconds thereby helping them reduce excessive bed days due to delays in diagnosis, improve the appropriateness of testing and reduce clinical risk.

While hospitals have found value in many different kinds of clinical decision support systems, we believe diagnosis support is particularly important. As Dr. Wachter has noted, until we resolve the issue of diagnostic errors, we face a fundamental problem in patient safety:that hospitals can be rewarded for high quality in performance even if every diagnosis was wrong.

Categories: Uncategorized

Tagged as: , ,

8 replies »

  1. I do consider all the ideas you’ve offered in your post.
    They are really convincing and can definitely work.
    Nonetheless, the posts are very short for newbies. May just you please prolong them a little
    from next time? Thanks for the post.

  2. I don’t even know the way I stopped up here, but I thought this post used to be good. I don’t realize who you might be but definitely you are going to a famous blogger for those who aren’t already. Cheers!

  3. Who says doctors feel threatened? I don’t feel threatened at all, but I still hate our EMR. I hate the fact that I have to spend hours every day watching that damned hourglass. Click a check box, watch the hourglass, click another check box, watch the hour glass. I’d rather be spending my time doing something else. My job has suddenly stretched to fill most of my waking hours. I’m miserable.
    Switching to an EMR involves so much more than just installing a new word processor. It involves changing the work flows for a large team of people. It’s actually a very big change. Add the fact that the EMR, at least the “award winning” one that we just bought, is very poorly implemented; slow, crappy user interface and unreliable, and it turns into a giant fiasco. So no, it’s not that I feel “threatened”, it’s that I feel overworked and frustrated.

  4. Why do doctors feel threatened when technology can help them to do a better job ? Isabel is like a senior consultant who can offer friendly reminders, to ensure that the doctor has considered all possible diagnoses. What’s wrong with this ?
    Dr Aniruddha Malpani, MD
    Medical Director
    HELP – Health Education Library for People
    Excelsior Business Center,
    National Insurance Building,
    Ground Floor, Near Excelsior Cinema,
    206, Dr.D.N Road, Mumbai 400001
    Tel. No.:65952393/65952394

  5. to call me a proto-luddite is cute, but, unfortunatly, no cigar. Think of me more as someone who avoids the first generation iphone, knowing that before knocking back a lot of cash, something better will come along. The evidence NOT to use something is no reason to throw money after it…if I could use my rolodex better than an Apple Newton (how many Luddites remember that?), then why would I buy one and switch? I also fully agree J Bean; EBM is great, but I do I really need a computer program to help me do more than remind me to give flu shots (this luddite DOES do that).
    I don’t buy into structured systems…all successful works for the user and not to thwart the user.

  6. The continued hostility of many physicians to evidence-based medicine is remarkable. Any particular manifestation of it can be flawed or misguided, but setting the goal of making more medical decisions on evidence-based models is as close to a no-brainer as it gets.
    It’s not clear to me why you are conflating EBM and CDS systems. EBM can be performed quite adequately in an all paper and pen setting. You can’t conflate EMR/EHR and CDS either. Many of the EMR tools currently available provide no CDS or CDS that is so flawed and basic that it is unusable. I’m no expert and have no knowledge of the Isabel system, but what I have seen is definitely not ready for prime time yet.

  7. Ira, while your response isn’t Luddism, it is defensive and offers no solution to improve the way we do medicine. I hope you are not saying that current DDS solutions aren’t perfect, so therefore they aren’t good enough and shouldn’t be used.
    Is there any evidence at all that it is better not to use these systems, than to use them where they are applicable with the understanding that they can be over-ridden in unusual circumstances?
    The continued hostility of many physicians to evidence-based medicine is remarkable. Any particular manifestation of it can be flawed or misguided, but setting the goal of making more medical decisions on evidence-based models is as close to a no-brainer as it gets.
    As for having to slow down when using EHR systems, this cuts both ways. There is actually a point to entering more detailed data into an EHR system, as tedious as it may be. It provides structured data that can be used so much more powerfully than unstructured data (particularly unstructured data in disjointed paper silos scattered among providers who don’t communicate or coordinate well). Structured data isn’t just useful for point of care DDS systems, but also for a slew of analytical and reporting tools at the individual and population levels. It is also useful for payment systems that don’t just reimburse the treatment, but pay based on outcomes.

  8. The real reason that internists and other PCPs have been slow to adopt the EHR systems (EMR is sooo 20th century)is that they are not ready for prime time. Data entry is slow to the point of ridiculous, and while “decision support” may work in a hospital where lots of available data can be condensed into an algorithm, its still not real useful for the question of whether to give antibiotics in the case of sinusitis. Yes, I know there is an algorithm for that, but the reality is I don’t always follow it and neither do the guys who wrote it (if they actually take care of patients).
    An accurate history is the key to any diagnosis. While EHRs can poke, prod and help, there is a reason that we get the medium sized bucks. A computer can, with some simple inputs, draw up a perfectly good limited liability partnership. But many of us still prefer a lawyer: their expertise is something we don’t expect a computer can match.
    Good luck with your system. But remember that the most important decisions occur with extremely limited information. If you think your programs language capability is good enough to understand the nuance of an intern’s decent history I will be impressed and amazed.