David Gerstman has an interesting piece up at Legal Insurrection about the IT panacea for Obamacare. He notes an op-ed by Thomas Friedman that received an endorsement by Health and Human Service Secretary Kathleen Sebelius that paints a glowing picture IT investments made under the act will have at providing better and cheaper medical care. Gerstman then follows up Friedman’s breathless piece with another that asks a simple question, if the impact of IT on health care is so wonderful Why Is Your Doctor Typing? Forbes’s Steve Denning writes about his experience at his doctor’s office where he watches his doctor typing on a computer during his exam.
Surely, I said, computerized medical records generate benefits. They are easily retrievable. They can be transferred from one practice to another and accessible to the many different service providers—hospitals, laboratories, specialists, radiology and so on—that might be involved in any one patient.
“In theory, perhaps,” he replied. “But in practice, it’s a horrible and costly bureaucracy that is being imposed on doctors. I spend less time with patients, and more time filling out multiple boxes on forms that don’t fit the way I work. Often I am filling out the same information over and over again. A lot of it is checking boxes, rather than understanding what this patient really needs.”
What about retrieving information? Isn’t that easier?
“Again, in theory, retrieval should be easy and quick,” he said, “But you can’t flip through these records the way you do with a paper file and easily find what you want.
I mentioned the articles to Dr. Wife and she said, “The only thing EHRs have done is make it easier to read a doctor’s handwriting.” Since the US is projected to spend $6 billion on EHRs by 2015, that’s a lot of money spent trying to make up for the failure of primary education to teach penmanship.
Being married to a doctor and an IT professional specializing in “big data” in the financial industry, I have watched the Wife’s experience with various EHRs with levels of amazement and dismay. It’s as if the lessons learned by the financial industry in the 1990s, such as poorly designed software that is incompatible with other software will cost more money to replace than it did to implement in the first place, have been completely lost by the lemming-like rush towards electronic health record (EHR, also known as electronic medical records EMR) systems.
The basic problem is that EHRs are not designed to suit the ways doctors practice. This is complicated by the fact that the way doctors practice varies between specialties, an orthopedic surgeon doesn’t practice medicine the way a primary care physician does, and by the additional complication that how doctors practice varies within the same specialty, often the same office. Even the same doctor will treat patients differently depending on what he feels works best for each patient. Yet these variances between specialties are only rudimentarily addressed within EHRs, and handle variance within specialties one of two ways, providing either a set workflow that dictates to the doctor the way she should practice, or one that provides so much flexibility that she is lost trying to get basic tasks.
The key decision in any software development is to address who the software is for and the key needs it is meant to address. Judging by the current EHR systems available none were designed for doctors. Instead they were designed for the employers of doctors such as large health systems, insurance companies and the federal government who are interested in aggregated data in order to answer questions such as “How many patients are uncontrolled diabetics?” or “How much is being spent on obesity-related illness?” These are questions which might be of interest to a doctor in general, but they are not what he’s thinking about when he’s facing his patient, say a morbidly obese, uncontrolled diabetic medicaid patient. Instead he is interested only in that particular patient’s problems. Is her agoraphobia contributing to her obesity, or is it the result of it? How can he wean her off HFCS soda and begin to move and diet when getting her into his office requires so much effort? Most of all, how can he encourage her to take an active role in her own medical care and help him treat her?
Current EHR systems will be very good at picking up his patient as an uncontrolled diabetic, and the data can be used by medicaid to threaten to cut his reimbursement for her treatment as is under discussion to control health care costs. But his patient’s needs and his attempts to deal with them will be lost in the sea of data the EHR generates because current systems are modeled on existing software developed in the financial industry which was the first to successfully integrate the technology with its existing business. Even that integration wasn’t painless, occurring over decades after many fits and starts, adoption of dead-end technologies and gargantuan piles of wasted money.
A key difference between the medical and financial industries is in the nature of the data itself. Financial data is transactional, meaning that money is traded for a good. Transactional systems are repetitive. For example, a store will sell a loaf of bread for $2.59 to every person who comes into the door and asks for it, but a doctor seeing a sore throat today knows 99% of the time her patient likely only has a viral condition, and that remaining 1% can present with a sore throat but have much more serious, perhaps even fatal, underlying conditions. Doctors are taught in medical schools to “think horses when you hear hoof beats, not zebras,” but the problem is that in reality zebras are not limited to the Serengeti Plains: they are mixed in with the horse. So while a doctor should think horses when he sees an 8 year old with a high fever and sore throat, he always must rule out he’s hearing a zebra. This is why when you see your doctor complaining of head and neck pain she makes you touch your chin to your chest: doing so rules out meningitis, a rare but very serious infection, a zebra running with horses.
The equivalent of this repetition and poor data is handling would be going to the store and buying a loaf of bread with your debit card. This bread would be tailored to your specific needs on site. Prefer no crusts? The crusts would be removed. Like thicker slices? The store would slice the bread to your exact specifications. The cash register would report the sale to your bank via fax. A person at the bank would read the fax transaction and key it into the bank’s debit card system which would then debit your account for the payment to the store. Since the store’s financial records are kept at another bank, your bank would then email the credit to the store’s systems, and someone at that bank would open the attachment, read it and add the amount to the store’s bank account. Such a transactional system would be costly to run, inefficient with the same task performed multiple times, and time consuming. A similar system already exists today with check processing, but that is limited to a handful of data elements such as the bank, amount paid and the account number of the person writing the check, and the name, the account number and bank of the payee depositing the check. That’s six pieces of data that costs banks billions to process every year. Banks hate checks which is why they have backed the current system of debit cards working to replace them.
From a physician’s perspective, what should an electronic medical records system do? It should provide her with the treatment plan from the previous encounter. Most systems hide this information from a doctor, making her search for the notes from the last visit. The system should provide lab work and test results directly from the laboratory providing the test results. Currently labs do not have set data standards, and electronic medical records systems do not have the capability to receive these records directly. Instead the records are either faxed or sent via email where they are “attached” to a patient record. This is akin to attaching a picture to an email, meaning that the contents of the picture remain completely unreadable by the system. The email system doesn’t know if the picture is a snap from your trip to the beach, whether its of a sunset or a personal portrait. Data in a picture or as commonly sent PDF format cannot be read and translated into a data record directly. Instead either the doctor, mid-level or medical tech must look at the results in the attachment and physically key them into the system.
Dr. Wife tells me her current system, one of the top used in the US, can only report weight and BMI results from last visit. Lab values and other pertinent information is hidden in attachments or non-indexed patient notes. Prior to the EHR she would open a patient’s chart and look at the lab result for a patient’s hemoglobin a1c result. Since the labs were in a separate section of the paper chart she could open it up then flip backwards through the stack to immediately find the results of previous tests. Similarly she could open the chart and see the notes from the patient’s last visit to see what recommendations she had then, or flip back further to see how the patient’s condition had changed with time. To do this in her current EHR is much more difficult than flipping through pieces of paper. Instead she has to search for and find lab result attachment which may not only be located in the lab result folder, but which may have been filed mistakenly by a medical tech into the fax folder because the lab result may have arrived via fax, and was scanned and added as a patient communication. Since the information is not indexed, there is no way for the physician to type in a search box “hemoglobin a1c” and have all documents that contain the phrase pop up. Instead she has to open each attachment to determine what it is and whether it’s the lab result she is looking for. Since EHRs are rarely known as fast and responsive, opening each attachment takes 5-15 seconds depending on size and EHR file complexity, making a search which would have taken three or four seconds flipping through a paper charts several minutes to complete. When a doctor is allotted 15 minutes per patient, anything that makes a doctor’s job harder for no benefit to him or his patient whatsoever will not be appreciated. Yet hospital administrators and software companies wonder why medical practitioners loathe electronic medical records systems?
Here’s what Dr. Wife described as her dream medical records system. First, the entire encounter would be recorded to protect her from future litigation or in case anyone needed to review or document anything from the patient encounter later. Next she would be able to choose from a set of predefined dropdowns or checkboxes the treatment plan for the patient. Lab values would be available on the right side of the screen, and she would be able to click on any one of them to see details or trends. These would be automatically populated by the labs themselves without any input from the doctor or practice staff, and could be signed off by the doctor simply by clicking the value. It would be a simple app that would run on an iPad. Suri would be used to transcribe a brief note after the visit, which would allow Dr. Wife to spend more time with her patients and doing what she is paid to do, diagnose illnesses and develop treatment programs, instead of typing, filing and other busy work skills that is so devalued in today’s workplace that much of it is offshored.
Another alternative would be to hire scribes, medical technicians who are trained to enter data into the EHRs. Many optometrists who must use their hands and eyes in concert use scribes already to notate lens dimensions and other key patient facts, so their presence in the exam room wouldn’t be completely new. Such positions would pay $12-15/hour with benefits, about what medical technicians commonly earn today, and would offer advancement thanks to the coding skills and familiarity with the software developed with experience. Of course adding a scribe for each physician would increase personnel costs, but ask yourself, does it make sense to pay someone $75 an hour to do a job that can be done by someone making $15 an hour? And from the patient’s perspective, would they rather pay an extra $4 a visit to have the undivided attention of their doctor for 15 minutes instead of watching him divide his attention between them and his computer?