From time to time, we’ve featured post from the informative Health Care Renewal blog on the appalling state of electronic medical records. Many readers have doubted its message, since the assumption that anything computerized has to be better than doctors maintaining handwritten records, often in famously illegible handwriting. However, as HCR stresses, underlying, well-established procedures and practices were revised to conform to the dictates of computer systems, with result being crapification of the activity. Second, the design priority was the money, meaning billing and doctor control in mind, with patient outcomes taking a back seat. The result has been in some cases to worsen medical outcomes. Indeed, HCR has noted that electronic medical records have been found in hospital systems to be a top cause of patient risk.
Confirmation of HCR’s dim view comes via a new article in ComputerWorld (hat tip Chuck L) on how the screw-ups resulting from lousy electronic medical records are large and frequent enough to have caught the attention of attorneys. The interesting twist is that it is the software companies that are the litigation targets.
From the ComputerWorld story:
As electronic medical records (EMRs) proliferate under federal regulations, kludgey workflow processes and patient data entry quality can be problematic…
Keith Klein, a medical doctor and professor of medicine at the David Geffen School of Medicine at UCLA, described four such cases where judgments reached more than $7.5 million because the data contained in an EMR couldn’t be trusted in court…
EMRs require physicians to perform their own data entry, stealing precious face time with patients. What had been a note jotted into a paper record, now involves a dozen or more mouse clicks to navigate a complex EMR workflow…
Data administrators may copy and paste patient information from an older record to a newer one, supposing that the data would remain the same. And the sheer complexity of EMRs pose issues with accuracy, as being able to track who has entered what data, and when, over time can become confusing…
One recent lawsuit involved a patient who suffered permanent kidney damage when he was given an antibiotic to treat what was thought to be an infection resulting in elevated creatinine levels. The patient was also suffering a uric kidney stone, which precludes the use of the antibiotic. Because of the complexity of the EHR, none of the attending physicians noticed the kidney stone.
Detracting from the EMR’s validity was the fact that a date related to a previous intravenous drip was repeated over and over on all 3,000 pages of the record.
While his physicians claimed they’d documented his care properly, the EMR was so complex and filled with repetitive data, the judge found it in inadmissible…
n another case, the physician was accused of plagiarizing data entered from another healthcare provider because he copied and pasted basic patient information.
Rita Bowen, senior vice president of health information management for Healthport in Atlanta, a records audit management and tracking technology firm, said she’s seen duplicate data, erroneous data and copied data in EMRs.
“I’ve seen records where someone has copied and pasted from older records, ‘The IV will be removed today,’ over and over again. Well, was it removed?” Bowen said, illustrating how admins may copy and paste older information into newer records…
But the problem isn’t solely human error. The way EMRs and electronic health records (EHRs) are designed can prompt error-prone entries. For example, drop down menus for diagnoses can automatically enter data if a mouse is hovered over them too long.
“We’ve seen 92-year-old women getting diagnosed as crack addicts because of drop down menus,” she said.
It’s instructive to contrast this data negligence with the eagerness of health insurers to mine medical data when they can make, or more accurately, save a buck though patient monitoring. Consider this story that shows how insurers are starting to reward customers for engaging in prisoner-level all-the-time surveillance. From Fusion (hat tip Dr. Kevin):
The New York Times reports that John Hancock will be the first life insurance company that will offer discounts to Americans who agree to wear an activity tracker:
People who sign up will receive a free Fitbit monitor, which can be set to automatically upload activity levels to the insurer. The most active customers may earn a discount of up to 15 percent on their premiums, in addition to Amazon gift cards, half-price stays at Hyatt hotels and other perks.
Traditionally, life insurance companies have set their rates based on a customer’s medical history and age, among other factors. But with the advent of fitness trackers, they can set rates in real-time, drawing on data about their customers’ lifestyles. Andrew Thomas, a 51-year-old South African who has been sharing his activity with his insurer for years now, including his cholesterol level and exercise habits, tells the Times he gets “points” for good behavior, which translates into money back on his premiums. “Every Saturday morning, just for playing golf, I get points,” he told the Times. (He would probably get more points for a sport that requires serious exercise. For the sake of his premiums, let’s hope he walks the course instead of driving a cart.)
The author points out that these monitoring systems will allow the insurers to know when and where you are having sex, which would also give them an informed guess as to whether you were faithful (sex at a local hotel, for instance). But the more general issue is that insurers seem perfectly capable of doing a good job with information technology when they see a benefit, but too bad of anyone that might suffer adverse consequences.
30 comments
Comments are closed.
Re: fitness trackers. A couple of years ago a research project into obesity and exercise on children used trackers to measure overweight kids activities. It found that many of the fatter kids were much more active than the slim, fit kids. When they investigated more closely they found the ‘active’ kids had just strapped the tracker to their pet dogs while they played video games.
Not that I’m suggesting anyone should do this to get health insurance discounts….
Patient caseloads for hospitalists I know have quadrupled in the past 20 years. Cut and paste seems like a good option when you have four times as many patients to see and they are sicker than they ever were when they could stay in the hospital to get stable. Common sense has left the arena and you can lose a lawsuit over a duplication error and a typo.
In the end with EHR’s perhaps less (data entry) is sometimes better.
That used to be a pharmacy perogative: data entry at point of DUR.
All of this was a no brainer, and shows that they have forgotten the lessons we learn back in the 70s on how to build computer systems. Mainly, that you have to build them for what the user actually does, and not for what some outside party thinks they do, Doing it backwards leads to instant crapification on day one, and does so in a manner that the system probably cannot be reverse engineered into correctness.
This of course was inevitable as the major corporate tasks all started getting turned over to the bean counters who treat staff like equal interchangeable little blots, each “work unit” being reassignable to the next available (often outsourced) blot of “equal” rating. I’ve noticed this especially as the newer operating systems come out with increasing amounts of small malfunctions that show that no one is left in change of overall quality and integration of the design phase of these efforts, and I wonder as we’ve laid off more experienced (and expensive) professional whether o not we even have the ability anymore to correct this “crapification”..
[Aside: And just wait until they put in the ICD 10 codes !!!]
I totally agree, but as I’m sure we’re all aware, the systems weren’t built to improve patient health. They were designed to reign in physicians and maximize profit. For insurers and software billing companies first and foremost of course.
Computerization came to make the billing efficient . All else came secondary. Very few inputs were solicited from physicians/Nurses re clinical care, quality issues. Patient care is at the bottom of this totem pole. As retired Radiologist, NONE of this surprising!
ICD 9 (current) has 13000 codes
ICD will be ( Oct ’15) 60,000 codes!
More the merrier. I guess!
The problem with software is universal … it costs 3x as much to write good software, and takes 3x as long vs the bad kind. In free market capitalism this means that “save development money and time” equals crappy software. Cost/benefit tradeoff doesn’t work if the tradeoff accrues to two different organizations/individuals. So with software, it is always a race to the bottom.
Of course there were errors in paper records too, but that was harder to find during the “discovery” phase of litigation. The powers that be have the same solution here, as with other tort law … simply don’t allow lawsuits from ordinary people. If it were true that 90,000 patients die early/unnecessarily in American hospitals each year … the resulting lawsuits (if every family filed one) would immediately bankrupt the medical system.
15% of Medicare patients hospitalized for surgery die of preventable medical errors. Don’t know how many Medicare patients are admitted for surgery each year, but 600,000 seems low.
In response to “no one” who posted on links of 4/14 that MediCal support line was answered by Xerox employees, I wrote in part:
Theoretically, IF implemented correctly, EHR’s CAN not only lower costs but improve care by saving practitioners time, improving accuracy of documentation, and allowing better tracking of processes for risk potential and inefficiencies, improving care broadly, and already have in many ways. For example, analytics might note those units in a hospital having an unusually high rate of medication errors had recently implemented a new unit dosing system, easily escaping simple human observation. EHR’s can also flag fields which still require entry for healthcare practitioners, thus reducing chance that care given is undocumented (and perhaps repeated, or the institution being held liable for negligence……. a well-used axiom holds “if it wasn’t charted, it wasn’t done”), or alert to care still ungiven. Entries can’t be deleted or backdated. Drug allergies trip automatic warnings, supplies reordered automatically when stock is low. Reading pages of illegible notes are history, as physician progress notes were not usually transcribed, and other providers had no access to transcription services. Thus docs NEVER read them prior to EHR’s even if they could locate them, deemed too time-consuming for their little value. (Nurses had nothing to say of importance, and if they did, were expected to provide phone or face-to-face notification……. because nurses had ample spare time to offer personal babysitting services. /s)
I designed some clinical software for behavioral health institutions in the mid-90’s and found it really exciting. However, it was a small company that had folks with clinical backgrounds making the important design decisions, and user feedback was continually sought. Integration? A moot point as there was little to integrate. However, things have changed and EHR has become complex software with lots of players on the field. Clinical input to design has taken a back seat to those with business and programming backgrounds. Like most software, the philosophy is “ship today, fix later” and I won’t even get into problems associated with offshoring of coding to centers in India (company above has facilities in Bangalore). Just as network security that leaves a system vulnerable to attacks doesn’t translate to wanting to abandon efforts to nail down a system, EHR’s can be implemented poorly. A drop-down menu that saves time for most users can lead to serious mistakes by rushed or sloppy users. Traditional dictation of notes by physicians came with its own sets of problems and required docs to proofread notes once typed, often not done. Books got written containing entertaining entries found in charts due to transcription errors, and words and phrases would be left to (maybe) be filled in later if not heard by the transcriptionist.
Some bugs remain to be worked out. The biggest is the lack of compatibility between clinical systems. There should have been an entity designated to develop common standards, as was done with html for web pages BEFORE mandating the transition to EHR’s. This was the same people that launched Obamacare, perhaps the most ambitious software project ever undertaken, without allowing at least several months for testing. The dream team called in after the disastrous launch deserves major kudos for pulling a rabbit out of the hat. Obama, and his chosen advisors (Valerie Jarrett), know jack shit about computing. But ten years from now paper charts will be as unthinkable as typewriter editing and payment processing.
*Having trained providers, physicians are notoriously the most problematic group of users. Not knowing a patient has a kidney stone because the record is too complex is a bs excuse that deserves no sympathy. Would it fly if Exxon’s accountants’ claimed the financial records were too complex to find tax subsidies? One local hospital around 2000-1 had to recruit somebody solely to train their physicians as their existing staff refused. Resistance to change was common, esp. among older workers, but apparently their physicians were particularly hostile, feeling they should be excepted, and their trainers insufficiently thick-skinned and/or unable to insist ALL students vent their anger elsewhere (which is NOT passive-aggressive).
Lucy Lu – cut the boo hooing about those “problematic” physicians who didn’t get all giddy about jumping on your new tech hobby horse.
The reason docs are so problematic and don’t swallow tech training hook, line and sinker comes from years of training in the art of medicine. Medicine is every bit as much an art (meaning decision trees are equally apt to bog down the process as to help along the process) as a science. A doctor who is blind to this reality is a doctor with limited capabilities in solving complex problems.
What utter and complete BS.
The reduction of medical diagnosis and treatment to a “drop down menu,” the equivalent of a multiple choice test (choose the MOST CORRECT or LEAST INCORRECT “answer”), should be an unforgivable offense to every “physician” who ever borrowed $200,000 to get his/her degree.
Effective diagnosis and treatment is the result of HUMAN education, observation, judgement, experience and even intuition which is UNIQUE to each individual case and simply cannot be “standardized,” an imperative required for computerization.
I can understand EMR where prescriptions/bad handwriting are concerned. I can also understand it where “portability” or “transferability” is concerned. The fact that “communication” between systems appears to be a low to nonexistent priority speaks volumes.
But “drop down menus”????
Note to the “lawyers”: try defending a misdiagnosis by pointing out that the correct “answer” didn’t appear in the “drop down menu.”
It was the software’s fault that the patient died.
The programs limit free text entries forcing the user to check a box in the drop down menu. There are so many options it can take forever to find the desired option.
Another frightening thing for those of us that like our privacy, researchers can search for all patients, on “x” medication, with “x” diagnosis.
MD offices, pharmacies, hospitals are all starting to share the same programs. At my facility (this was a few years ago) I believe the program was unable to search patient’s files free text notations, perhaps that feature cost more.
It is all about billing and research.
“….limit free text entries….”
Sounds like a job for Twitter. 140 characters. Hieroglyphics 2.0.
“There are so many options it can take forever to find the desired option.”
It’s my recurring nightmare–a multiple choice test with unlimited choices and no obviously wrong ones. I wake up in a cold sweat every time.
You do realize that in most EMRs, *there is no option* aside from the drop-down menu. Every diagnosis must be selected from the ICD-9 codes because everyone from hospitals to insurance companies want to data mine your notes for their own purposes and rather than devise a system that could interpret physician notes, they want to go the easy route and make us physicians choose numbers, so that their computers are easier to program.
There is no room for free-text entry of what you think is going on. These ICD-9 codes are detested by most physicians because they don’t really model the way physicians make diagnoses. Yet we’re forced to use them. But no worry, they’re soon to be replaced by ICD-10 which has 10x the codes while somehow being even less true to the way we decide upon diagnoses.
ICD-10 is so ridiculous that entire websites have sprung up dedicated to finding the most outlandish and stupid codes. I’ll give you a few examples:
Y92146, or, “swimming pool of prison as the place of occurrence of the external cause.” (because to make a medical diagnosis of drowning, it depends whether the swimming pool was in a prison or not)
Burned when water skis caught on fire: Y91.07XA
W56.22xA, “struck by orca, initial encounter.”
And yet basic diagnostic entities are frequently missing, or buried within literally thousands of codes. And you have to select the proper codes for every single note you write, within the 5 minutes you have to write the entire note for all 40-50 patients you might see in a day.
NB: I’m a physician and I’m also highly technically skilled (have been running servers in my basement since I was a high school student and have had an email address since the 1980s). And I detest EMRs and would resist using them if I could. The main reason is that EMRs do not appreciate the fact that medical communication, even if written freehand on notes, is a highly structured form of communication that has evolved over literally thousands of years, through numerous languages, marked changes in our body of knowledge, etc., to model complex disease processes and our thinking of them in a way that captures both what’s known and what’s unkown, in order to accurately communicate this to another physician who may never see the patient, and indeed, may not even speak your language. A significant portion of medical education is spent learning this language and structure until it becomes second nature both to the way you communicate and the way you *think*.
Every criticism of hand-written physician notes begins and ends with doctor’s poor handwriting. Fine. Make us type (or dictate) the notes. But don’t replace the incredibly sophisticated structure of our notes because some computer programmer doesn’t understand the reason why a SOAP note has been around longer than most countries in the world, and doesn’t appreciate that, thanks to that structure, I can read case reports from hundreds of years ago and understand what the doctor is thinking about his patient.
Even doctor’s orders, even when written freehand, have a very precise structure to them. And if most of the errors are due to illegibility, again, make us type the order. But replacing the precise structure of our orders with a cumbersome system of menu options, checkboxes, etc. that doesn’t model the way we think through our orders just means you’ve replaced illegibility (something that can be dealt with by typing) with errors that are harder to detect (until the patient is injured).
If you think this is a bad way to utilize a highly trained person, you will find near universal agreement among physicians, who are now being turned into data entry clerks, and poor ones at that, by EMR systems that don’t really care about what happens to a patient as long as the billing goes through without being rejected.
P.S. Personally, I’m finding that thanks to EMRs, physician notes have become largely billing and legal instruments, and that actual physician communication is now using a different technology: text messages. Imagine that. A freehand, legible, easy to use, unobtrusive system that is interoperable across systems, vendors, hospital systems, across the world! And it’s even mobile and can be carried around in your lab coat. And guess what, those crusty old luddite physicians who hate technology are flocking to it in droves. So tell me again why we’re paying billions of dollars for those other systems?
You DO realize that I agree with you, right?
I did after I re-read your post. Sorry for the confusion :-)
No disagreement here either. As noted in my post above, there’s been a failure to include the needs of providers in software design, i.e. there has been a failure to include those with clinical backgrounds during development. Check boxes and drop-down menus can be convenient and time-saving options as well as reducing errors but shouldn’t be provided at the expense of free text entries. Handwritten notes WERE highly problematic and in my own experience, physicians were focused on resisting ANY unwanted changes, e.g. using computers period, rather than providing input into ensuring changes preserved the integrity of documentation (most marked among older physicians, who also protested the loudest). Unlike physicians, other providers were accustomed to having change forced upon them. Whether this would have made a difference in the end, I can’t say.
I have to disagree somewhat. I’m glad you agree that providers need to be included in software design, but that’s only part of the issue. There are several reasons why many physicians (not just older or technically illiterate ones) were / are against current EMRs.
1) As mentioned before, EMRs are designed for billing and legal purposes. This is true *even* for physician-designed EMRs. That is, while we physicians bellyache about hospital systems that are designed to optimize billing, we don’t admit that EMRs for small offices, which usually do cater to physicians since physicians make the purchasing decisions, also are designed to optimize billing because that’s what *we* select for. When an EMR vendor comes to a doctor’s office to sell their wares, their sales pitch is based on three things: 1) increasing revenue by better billing, 2) decreasing time spent on documentation (using templates, etc.) 3) decreasing staff costs by automating processes (e.g. scheduling, etc.). Nowhere in there is a focus on improving quality of care.
2) In defense of my colleagues, we’re not interested in using EMRs to improve quality of care because there’s precious little evidence that it does so. Forgetting about theoretical benefits, if you asked EMR vendors what quality benefit does their software provide *today* to patients, they will universally bring up allergies and drug interactions. That’s usually it. (Occasionally they’ll bring up schedules for vaccinations and other preventive care).
If we’re spending billions of dollars to essentially replace the bright red sticker we used to place on paper charts with a patient’s allergies with a bright red icon on an EMR, I submit the quality improvements of an EMR are minimal.
Furthermore, whatever quality improvements come from current EMRs, I assert, come from improving the operational reliability of the rest of the hospital processes. That is, once an order has been entered, it’s far better to have it in a computer where it can be reliably transmitted to the pharmacist, down to automated medication dispensers (Pyxis machines), down to bar code scanners that nurses use to ensure the right medication is given to the right patient at the right time. But that doesn’t mean that physician processes should be automated as well, especially with the same system. That’s akin to saying that since a car is built by CAD files that get translated to machine instructions that power the robots on the line and allow the line workers to know exactly where and how each part fits together, that the original designers of the car shouldn’t be allowed to work on pencil and paper or at least a computerized freehand drawing program.
Indeed, with paper systems, orders were still entered into a computer system by the pharmacist at which point we gained the operational reliability described above without needing the billions of dollars to automate (poorly) the physician portion of it.
Am I saying that physician processes can’t be improved by computerization? No. I’m saying *current* EMRs don’t, and software companies have enough history of vaporware that I’m wary of promises of future benefits.
3) Final reason (there are more, but I’ll stop here :-). While I agree that physician input is needed, it needs to be way before any computer code is ever written. It sounds like in your company you had that, which is good. But to most EMR companies, physician input into design means asking a focus group of doctors whether the buttons should be bigger, or whether this window should be on the right or left. In other companies it’s worse, and amounts to “how can we get you guys to use our system by pretending to listen to your concerns?”
How many EMR designers actually spent a month in a hospital, watching every person and creating a detailed description of each of their jobs? How many interviewed doctors, nurses, therapists, etc. to understand how they approach their tasks, before sitting down to design an EMR to help them with those tasks?
McDonald’s has an entire warehouse in Chicago complete with mockups of their kitchens dedicated to testing how even the smallest change in a menu (from new products to changes in ingredients or cooking methods) can affect the operation of its restaurants, whether it’s time, cost, quality, etc. They then complement this with detailed video analysis of the changes’ effects in beta testing in actual restaurants. They do all of this before releasing any change to its franchisees’. All in the name of a better burger. How many EMR companies do even a fraction of that analysis before unleashing changes that carry the risk of injuring patients?
So in the end, EMRs, as they stand today, are a waste of time and money, and whatever quality improvements they provide could have been gained with much less expense and hassle. But why is that a problem when you can successfully lobby the government to force people to buy your product, with the govt footing the bill to boot?
Gotta love it!
wait a minute……..text message??? What about HIPAA? Are you telling me you are not encrypting the messages? Oh my God…….someone might realize you just operated on an 80 year old’s fractured hip who now has pneumonia……..How could you do such a thing? That is almost as bad as having the paging operator ask if Ms. Smith’s family could come up to the OR waiting area.
DAVID GEFFEN SCHOOL OF MEDICINE?
I guess if you have money they’ll put your name on anything
Had to look the guy up. Get this, albeit from Wikipedia:
“After a brief appearance as an extra in the 1961 film The Explosive Generation, Geffen began his entertainment career in the mailroom at the William Morris Agency (WMA), where he quickly became a talent agent. In order to obtain the WMA job, he had to prove he was a college graduate. As he later reported in an interview, he claimed in his job application at WMA that he had graduated from the University of California at Los Angeles (UCLA). Because he worked in the mailroom, Geffen was able to intercept a letter from UCLA to WMA which stated that he had not graduated from UCLA. He modified the letter to show that he had attended and graduated, then submitted it to WMA.[7]
“Philanthropy ” as absolution. I swear, this sh*t never ends.
Even in the early 90s, paper filing systems were becoming a lost art. It’s gotten much worse, where “moving a document through a workflow” has taken the place of properly cataloging and indexing. It’s more like moving a blob of data around and hoping to extract information from it at some point in the future. It’s the “programmer who spends two hours automating a one-off data entry task that would have taken ten minutes” syndrome writ large. And potentially fatally.
Dr. Richard Kimble: It wasn’t me, it was the one-armed drop-down menu!
Recently started seeing a new endocrinologist. He uses a laptop with voice recognition to enter his notes during my visit, which is highly entertaining to watch as it happens, AND he has a worker in the room with another laptop keying data during the visit as well. Just try to tell me this is more efficient.
That sounds better than my experiences over the last decade or so, where my physicians spend 99% of the visit glued to a monitor. Turning customers into cashiers is one thing, but turning doctors into scribes seems dangerous in my opinion.
There’s a trade-off between adaptability and efficiency. When the integrity of my mortal coil is at stake, I’ll take adaptability every time.
Why does he have the second person?
The financialization of health care records has had the same predictable course as had the financialization of every other damn thing. I am a physician, as well, and every day thank the Lord that I have not yet been reamed by these digital devils.
“underlying, well-established procedures and practices were revised to conform to the dictates of computer systems, with result being crapification of the activity.”
Sorry, but I am enough of a computer programmer to know better. The computer is a Good Soldier Schweik, it does what it is told to do. There is no “computer system” to which procedures and practices were forced to conform. The computers could have been programmed for carrying out the well established procedures and practices. If they were not, you can be sure that there were people who did not want them to be, and who got their way.
Wow. Am I the only one who notes that EMRS, with the letters switched around, spells MERS?
Apparently letters of the alphabet are not the only things they haven in common…
My Mother and Sister are both MDs, another Sister is an RN. And I am a lawyer, although I don’t do medical malpractice. I haven’t seen mentioned in the comments about the costs of EMRs the additional time it takes to see patients and the change of focus from doctor-patient personal interaction to doctor/nurse-computer interaction. Mom says the time per patient visit has at least doubled but the interaction between the MD and the patient has dropped. Less than ideal.