Yves here. We’ve featured posts from the Health Care Renewal site that regularly warn about how electronic health care records are a serious hazard to patient health. Yet we’ve regularly had readers refuse to believe that, despite warnings like the ECRI Institute putting health care information technology as its top risk in its 2014 Patient Safety Concerns for Large Health Care Organizations report, or the president of the Citizen’s Council for Health Freedom warning that “EHRs are endangering your life” or press reports like this:
Arthur Allen at POLITICO Pro eHealth (http://www.politico.com/story/2014/06/health-care-electronic-records-107881.html) says government-imposed EHRs are:
- Driving doctors to distraction
- Igniting nurse protests
- Crushing hospitals under debt
“In short,” he writes, “the current generation of electronic health records has about as many fans in medicine as Barack Obama at a tea party convention.“
Some readers assume that anything must be better than hand-written and potentially difficult-to-read doctor notes. And the 50,000 foot explanation, that the systems are a huge and costly fail from a care perspective because they are designed primarily, if not entirely, for billing, seems insufficient.
This post will hopefully satisfy the skeptics by giving granular detail with real-world examples of how these electronic record systems distract doctors, regularly employ dangerous “check the box” approaches, produce voluminous and repetitive patient files that routinely go unread, give nurses contradictory instructions, and too often result in patients being given “care” that harms them.
One of my friends, the daughter of an MD who worked for the NIH and later a Big Pharma co, said she’d never go to a hospital without her own private duty nurse. That was before EHRs. Once you read this article, you’ll think twice about going to a hospital in the US without that sort of extra protection.
By Dorothy J. McNoble, MD, JD, who can be reached at Badmedicine005-at-gmail.com
In a now iconic experiment, subjects are asked to sit in bleachers watching a basketball practice and count the passes among players on one of the teams. A few minutes into the experiment a man in a gorilla costume walks across the court. Fewer than 50% of the subjects notice him.
In a variant of the experiment, a man stops a stranger on a somewhat busy street to ask directions. While they are talking, two men carrying a large piece of plywood walk between the two men and when the plywood has passed, the original questioner has been switched to a different man. Again fewer than 50% of people notice the change.
Recently, I witnessed an “invisible gorilla” episode in the hospital. I took my neighbor to the hospital after she had fainted. She had low blood pressure and a slow pulse. The nurse examined and interviewed her, but spent most of the interview facing the computer and inputting data. A few minutes later, my friend was moved two beds down and exchanged places with another patient due to some equipment problems. When the nurse returned to check on my friend, she addressed her by the incorrect name and questioned her about the symptoms of the patient who had been there earlier. I corrected her and she checked the armband to confirm.
There can be no denying that emergency rooms are busy and the staff are often overwhelmed, but I think this demonstrates that the new “three way” which dominates patient interactions – the patient, the computer and the nurse or doctor, risks turning patients from the central focus of all interactions into the invisible gorilla.
Anyone who has tried to wade through their own hospital records or watched as a primary care physician tried to decipher the “data dump” which is supposed to summarize the events of a recent hospitalization, will recognize that the promise of the efficient, orderly modern electronic record is far from being realized. In theory, the computer based electronic record should be perfectly suited to its task. In recent decades health care, especially inpatient hospitalization, has become increasingly complex. There are many more participants, doctors, nurses, dieticians, consultants, occupational therapists, respiratory therapists, social workers and the interventions and therapies and medications administered during a hospitalization have also increased dramatically.
The electronic record, with its ability to prompt clinicians with reminders, organize large amounts of data and allow access from any point in the hospital and even remote locations, seems the perfect tool to create an organized, complete, flexible document free of errors and redundancy. The EHR as a working document during the hospitalization should be able to immediately reflect changes in the patient’s condition, accommodate instantaneous changes in medication and therapy, allow input from a host of clinicians and remain clear and comprehensible. After the patient discharge, the EHR should be an easy to understand narrative of the event of the patient’s hospitalization with the patient as the obvious central figure.
However, instead, the EHR has become an unreadable, unholy mess in which the patient is increasingly eclipsed. How did this happen? Was it due to limitations of software capacity? Insufficient funds devoted to the development of the EHR?
All of these problems undoubtedly contribute to the difficulty of developing the optimal EHR system. However, I believe that the main impediment to the creation a good EHR is not technical limitations or financial constraint. Rather it is due to the decision to utilize the EHR as a billing document. Many of the decisions about how to organize the medical record, how to format the document, and what data to include or exclude arise from the need to use the record as the support for and documentation of “billable events” during the hospital stay.
As noted above, there are many more parties participating in a patient’s care than there were two decades ago, and each of them has more documentation required and more interventions to conduct. The checklist format is appropriately suited to this task as it helps the clinician to do a complete exam, appropriate for his or her special area.
However, reliance on checklists often diverts attention away from the patient as the clinician taps away on the computer, entering various “required fields” and ignores the patient. Moreover, while clinicians each have their tailored checklists, there is a mind-numbing amount of repetition and overlap. This wastes time and results in a bloated, redundant record.
For example, while assisting my friend in the emergency room, I noticed that no fewer than five doctors, nurses and consultants confirmed that she had obtained her flu vaccine and that had had a screening colonoscopy within the last year. In fact, during her ER stay and her hospitalization, there were dozens of these repetitive, checklist inquiries, always obtained while the clinician was facing a computer and not the patient. Added up, all of this box-checking seriously detracts from the time which could be spent actually examining or talking to the patient.
The checklist, originally intended as a way to assist clinicians to combat fatigue or forgetfulness, has now become a housekeeping chore and a billing tool.
Moreover, the checklist has been used as a mechanism for outright billing fraud, specifically Medicare fraud. Medicare has historically designated five levels of increasing patient complexity and severity of illness and allowed physicians to select the appropriate level, and bill for care delivered to that patient. Obviously, the highest code was to represent the sickest patients and the reimbursement would be accordingly increased. Quite clearly, Medicare auditors could examine only a limited number of charts (especially in the days prior to electronic records), to determine if the selected billing code conformed to the patient’s clinical status. Auditors developed a shorthand system to decide if the code matched the patient’s clinical status and this was a checklist system. In its simplest form, the more elements of the patient’s history and physical exam the clinician documented, the higher the billing code would be allowed.
Clinicians very quickly figured out a way to create expanded checklists for most of their patients and bill accordingly, irrespective of the patient’s actual severity of illness. Eventually, Medicare recognized the widespread fraud and began auditing, hired “bounty hunters” and engaged in “clawbacks” of fraudulent billing. Many physicians and hospitals have been convicted of Medicare fraud on this basis.
Checklists should be tailored to the patient’s need, should not be redundant or repetitive and should not be used in lieu of appropriate examination and evaluation, and most certainly should not be used to “upcode” and charge more than is warranted by the clinical situation. Judged by these criteria, the modern EHR is a failure and helps neither the clinician nor the patient.
The most dangerous of the checklists are the order sets. These are a list of self-contained orders for all aspects of patient’s care from diet and activity to medications and invasive procedures. These expanded order sets can be initiated by the physician with a single click of the mouse.
Initially designed, as were all of these checklists, as an aid to preventing errors and omission due to fatigue or forgetfulness, they now serve less salutary goals.
First, the order set absolves the physician of any responsibility for reviewing the extensive medications and therapies to which he is subjecting the patient. These order sets routinely contain 20-40 separate orders and the physicians rarely edit them for an individual patient.
Second, the order set often contains directives to the nurse for managing a myriad of events or problems which might arise, events such as a drop in blood pressure, low urine output, fever, fast heart rate, even change in blood count. The nurse is given a therapeutic intervention to initiate with no directive to call the physician. The physician relinquishes his or her responsibility to carefully follow the patient’s clinical course and intervene as needed.
Third, they are a means of insuring that various billable events end up in the chart.
Finally, a hospital inpatient typically has multiple physicians caring for him, all of whom rely on their own order sets. The result is another data dump of overlapping, redundant and often dangerously inconsistent and conflicting orders. For example, a patient who is admitted from the emergency room and goes to surgery, then to the recovery room and the ward may have as many as 150 orders entered in the first few hours in the hospital. The surgeon, the internist, various consultants and the pain management physician are all likely to be using a “single click” order set. There are few safeguards to identify overlapping or inconsistent and incompatible orders and this poses a real danger for the patient. These are not benign or purely theoretical charting inconsistencies.
One example which occurred frequently in my own practice was the problem of bowel and dietary management of patients who had had often complex bowel resections. These patients had delicate areas of bowel suturing which required decompression and protection from any stretching or tension which would result if bowel contents were allowed to flow through them. The therapy was a nasogastric tube which pulled all the bowel contents out of the stomach to protect the area of bowel suturing. I always ordered that the tube be kept on suction and no liquid or contents be placed in it.
Unfortunately, the order sets of the other physicians almost always contained a “laxative cascade” which provided for laxatives for patients with no bowel function. In addition, internists frequently ordered that oral medicines be placed down the tube. The nurse thus had conflicting orders about how to handle the nasogastric tube and when to begin trying to use the patient’s GI tract. Many times, the nurses simply followed the order of the two or three internists and, contrary to my order, “challenged” the patient’s bowel function in a dangerous way.
This type of conflict arose in many areas such as patient activity, when I knew that the patient’s complex abdominal closure demanded bedrest, but the internist, adhering to a non-indicated safety protocol, would order early ambulation.
There was the constant battle over anti-coagulation. Another reflex safety protocol demanded blood thinners be given to post-op patients to prevent blood clots to the lung. However, in many instances blood thinners posed a danger to a patient with a hemorrhagic dissection at surgery. At times, I did not catch the internists order for blood thinners and the patient would be endangered. This order chaos is a seemingly immutable feature of the modern EHR and the patients are not served by this system.
The Auto–Fill and the Carry Forward
Physicians, nurses and all clinicians caring for patients in the hospital are under increasing pressure to deliver more care in less time, and to document that care more extensively so that it can be billed for. As a result of this pressure, two features of the modern EHR have been implemented: the auto-fill and the carry forward.
The auto-fill utilizes a widely used feature of modern software. When the physician sits at the computer to write a daily note about the patient’s condition and the plan for continued care, he can click one button and, as determined by an internal program in the software, the computer searches various data bases and enters a myriad of data in the note. Specifically, the daily physician note has historically documented the objective findings of the previous 24 hours: vital signs, medications, dietary intake, urine output, labs, x-ray results. This is the data on which the clinician must rely to proceed to the next part of the note, the assessment of condition and the plan for care. Again, historically, the clinician had to find the data either in the electronic record or, in the written chart, and decide which data to include and then transcribe the information.
However, it is trivially easy for the computer to search the various data bases and enter all of this information WITHOUT THE PHYSICIAN EVEN BEING REQUIRED TO LOOK AT IT. This unquestionably saves time and is more efficient than having the clinician manually type in the information. However, physicians routinely fail to examine all or part of the data and simply include it in the record without scrutiny. Hospital administrators, physicians and quality assurance personnel will deny vociferously that this happens, but I encountered this very problem constantly during my own clinical practice.
One particularly memorable case occurred during my last few years in practice. I had performed a complex bowel resection on a patient with two separate tumors. She seemed to be doing well post-operatively, but one day when I went to examine her, I noticed that the internist had entered a five page note in the computer. This was my first experience with auto-fill, and I assumed that a note so extensive indicated that the patient had had a problem the night before and I had not been notified. Without reading the data carefully, I rushed into the patient’s room to check on her. She seemed fine and to be recovering on schedule. I went back to the computer to examine the note and realized that all of the patient’s previous CT scans had been included as well as a four page pathology report. When I asked the internist about the note, he said that he had not read the CT scans or pathology report himself, but had just included them in the data portion of his note “automatically.” More data equals more billing, pure and simple. As a corollary, less physician time spent with patients or completing records equals more work the physician can do and more billing for the hospital and the physician.
So ubiquitous and meaningless has this auto-fill portion of the note become that in one hospital, this key data, on which all decisions should be made, has been relegated to the end of the daily patient note, like a tangential footnote instead of critical information.
The carry forward is a trivially technologically easy version of “cut and paste,” and it is used by clinicians for the same reason as auto-fill – easy documentation, less time spent, more billing available. The carry forward is used in the assessment and plan portion of the physician record. Quite simply, the physician incorporates his previous listing of patient problems and his plan for solving them. I have admitted patients with a perforated bowel, performed surgery and been a week into an uneventful recovery and noticed that the internist note continued to list “perforated bowel” under the patient problem list and included a repeating reference to the findings on the admission CT scan. These constant carry forward notes are redundant, confusing and useless for patient care. They simply provide padding to justify billing and serve as a “time saver” for the physician.
These documentation shortcuts are not new in medicine. In the era of the paper chart, physicians used the shorthand “WNL” to describe various aspects of the physical exam. WNL was shorthand for “within normal limits,” but everyone knew that it meant “we never looked.”
We are now in an era of “WNL on steroids. That is, with the convergence of checklist medicine, the auto-fill and carry-forward features, and the piling on of tangential safety and bureaucratic entries in the chart, the physicians and nurses seem to have relinquished their obligation to carefully monitor the patient’s clinical condition and assume responsibility for each of the therapeutic interventions taken.
Feeding at the Trough
Healthcare is the largest single industry in the country and the source of the greatest job growth. However, the growth in clinical care positions are not responsible for most of this increase. Medical and nursing schools have at most a negligible increase in graduates, and ancillary clinical training, such as occupational therapists is also growing slowly.
A large part of the increase in participants in the healthcare industry is due to the dramatic increase in federal, state, and local health care bureaucrats as well as the increases in hospital administrators, auditors, plan administrators and other non-clinical participants. These government and administrative bureaucracies are theoretically in place to insure efficiency in the delivery of care and to monitor and insure patient safety. The security of their roles as integral to patient care is assured if they can require that their particular area of concern is a mandated part of the medical record. For example, there must now be documentation in the EHR about smoking cessation, potential for elder abuse, vaccination status, use of seat belts use of child safety seats and a variety of other issues. Although the inclusion of these global safety and care concerns is laudable, the medical record has become bloated with repetitive, inappropriately placed mandatory documentation of these often peripheral and distracting subjects.
For example, when I reviewed the emergency room record of the patient I had brought in for fainting and low blood pressure, I saw that the first two pages had no reference to the reason for her admission, but had entries about vaccination status, the name and phone number of her outpatient pharmacy, the use of seat belts, the query about abuse in the home, a smoking cessation notation, the placement of the IV, including a standard description of the sterile technique used, a description of her living arrangements and the names and phone numbers of her emergency contact and her physicians. Only at the bottom of the second page was there a brief discussion of her main symptom, and there was not elaboration on the event itself – had she just taken a new medication? Had she been dehydrated, not able to eat or drink for some reason? Had this happened before? The chief complaint and history of present illness, which should be documented immediately, was never even placed in the emergency room record. Instead, somehow the monitors of the bloated crushing, non-clinical superstructure are able to justify their existence by having their marginally relevant concerns dominate the record.
Surely, the powerful EHR software could allow these myriad, disparate notations to be made, but categorized in an appropriate area of the EHR, not on the first two pages of the ER record. In particular, the detailed description of placement of a standard IV with standard IV fluid does not need to occupy three lines on the first page of the record. This is done only because of the necessity to “capture” this billing event. In fact, the emergency room record is more a narrative of billable events than medically proper narrative of the patient’s emergency room course.
Focus Fatigue and Limited Bandwidth
As discussed above, the structure of the EHR is designed to serve its purpose as a billing document, but makes it very challenging as a dynamic health care management tool during hospitalization or a concise, complete, well organized, non- redundant narrative after discharge. These structural features are a real impediment to clinicians trying to care for patients or understand what happened to patients after the fact.
There is an ample body of literature discussing the inefficiency and inaccuracy which results from multi-tasking. When doctors and nurses attempt to examine and obtain histories from patients while scrolling between various lists on the computer screen to be certain that all the billable bases are covered, it is clear that their ability to attend to the patient is compromised.
In addition, when the record itself is lengthy, repetitive and contains large amounts of prominently placed, but extraneous information, the clinician is likely to lack the mental stamina to wade through the document, switching between screens to find the relevant information. Moreover, the much touted “safety” features of the EHR are themselves often so ubiquitous and distracting that they lose their efficacy.
For example, recently a hospitalized patient was prescribed an antibiotic by a physician and a pharmacy generated warning appeared on the screen about a possible allergy. The physician believed that this might be an “auto-allergy” alert, that is, the patient denied the allergy, but the computer continued to generate the warning because it was pre-programmed to do so. The patient was given the antibiotic, and eventually manifested an allergic reaction. When the record was reviewed, no fewer than 17 physicians and nurses had hit “override” when seeing the warning. The records are now replete with these safety warnings, as well as with many checklist programs which demand that the data be entered in a particular order.
What may have begun as an attempt to amplify the safety of the “checklist manifesto” has become so arcane and distracting, that clinicians cannot attend to its voracious appetite and still provide efficient, expeditious patient care. Our cognitive limitations mean that we have only so much bandwidth with which to work, and if the record itself, with its demand to navigate computer screens and satisfy often marginal or useless documentation requirement consumes this precious cognitive resource, the patients themselves risk becoming the “invisible gorilla.”
The Electronic Tower of Babel
Finally, the reliance upon a market based competition model to determine how an EHR system is selected has resulted in an electronic Tower of Babel. Hospitals and physician practices select among and contract with competing EHR companies, and the patient finds himself in the center of system of non-communicating silos.
Patients often have multiple providers and, in an urban area, may have care at different hospitals as well. The EHRs are seldom compatible and there are always delays as the various offices and hospitals try to obtain printed records from another facility. While the theory is that the best EHR systems would emerge, the patient’s interest may be better served by a single system which can be understood and utilized by all health care entities.
Certainly, Medicare standardizes their billing submission requirements and many physicians think that Medicare should have insisted upon a single medical record system to facilitate communication about patients, especially when that communication is time-sensitive, as in the case of an emergency.
Thus, the usurpation of the patient electronic health record by its need to serve as a billing document, and the utilization of documentation shortcuts, such as auto fill and carry forward, coupled with the very human limitation to focus on t a limited number of matters simultaneously has turned the electronic record into a monstrous, unwieldy document which serves neither the patient nor the physicians well.