Friday, March 12, 2010

I love Electronic Health Records but.....

The EHR freight train has left the station and is barrelling down the tracks. There is a lot of excitement and enthusiasm for EHRs and their potential role in saving money and improving quality. The government is giving out grants to encourage adoption of EHRs (when used meaningfully) and for training professionals to support needs of health information technology.
There are several studies showing how EHRs can improve quality of care and decrease cost of care. EHRs have a huge potential to increase efficiency in our health care system. Unfortunately there is not much data that shows an EHR created by a vendor and implemented in a large setting has changed outcomes for a large group of patients in a meaningful manner. Thus studies may show that there is an increase in compliance with guidelines for ordering a HbA1c (Average sugar test) but we don't know if that lead to actual improvement in diabetes outcomes like fewer patients going on to dialysis or blindness. Granted one reason we don't have such data is that such studies would take a long time to do and EHRs are relatively new.
But in the absence of such studies we have taken a leap of faith and are proceeding apace with implementing the EHRs as universally as possible. There has been a lot of focus on making electronic data interfaces, setting standards for sharing data, and certifying EHRs based on features like electronic prescribing, clinical decision support, etc. These are potentially huge factors in making EHRs actually deliver on their potential.
One factor that has been somewhat neglected is the impact of EHRs on a physician's day-to-day clinical experience. There are several potential problems with EHRs:
  1. The User Interface, design of the functionality- the iPhone has been described as feeling like an extension of their body. A device that seamlessly integrates into one's daily life, that seems to know what you might want to do on each screen and presents those options. Think of how you want to grab a cup of coffee, your right hand goes to the handle and grasps it without any specific conscious instructions. Most EHRs are quite the opposite. They make simple tasks more difficult to do. Some examples-
    • The screens are cluttered - one very popular EHR has 2 horizontal toolbars at the top, 2 vertical navigation bars at the left leaving about 1/3 of the screen workspace available at the bottom right.
    • Simple tasks require you to constantly take your hand from the keyboard to the mouse and back.
    • For tasks that you know that in most cases there is only one option, the program will lead you through multiple screens where you have to accept the defaults to proceed to the end. A good example is the request for a medication refill where the necessary information (name, strength, directions, refill# etc) are all presented on one screen. It is very possible you remember the patient and want to just approve this medication. One EHR software requires 5 clicks of the mouse on 3 separate screens to approve this medication. It has a button to "Approve" and then another to "Sign" instead of having a button "Approve and sign".
    • You get a lab result showing abnormal kidney function. You would want to go back and see the last several results for this same test to see if this is a change. If it is worse than the last time it was checked 6 months back, you would want to select that 6 month time frame and look for any new medications prescribed during that time, any tests done with iodine contrast or any serious illness. While all this data is there in the EHR it would take 10-15 minutes to review this information with many clicks and screen loads.
  2. The alerts for the clinical decision support system - The CDS have the potential to make EHRs the tool for improving quality and decrease cost. The problem is that they are very poorly implemented at most places. Some examples -
    • Your patient has a reported allergy (a rash) with Sulfa in the EHR. You are about to refill a medication for this patient that has a remote chance of causing the same allergic reaction in patient with Sulfa allergy. The patient has been on this medication for several years and has not had any allergic reaction. The CDS unless it is well designed will pop up an alert when do submit the this refill request. The physicians come to expect that an alert will come up and they hit "Enter" without even reading the useless alert. This has led to even possibly useful alerts being ignored and we are losing a very valuable tool from our EHR.
    • Alerts can help to prevent duplicate tests. But in one EHR product the alert is unable to distinguish between tests ordered for the future. Thus when I see a patient with diabetes and order a lab test for 4 months in the future, I get an alert telling me the patient had this same test done today, am I sure I want to order it again!
  3. Potential barrier to patient physician communication in the exam room- The presence of a computer in the exam room can completely change the dynamics of the encounter. Theoretically the access to the huge amount of patient and reference information at the fingertips is a huge boon. Physicians can adapt to the presence of the computer and even leverage it to improve the dynamics and provide better quality care. But many patients and physicians struggle with this. The non-intuitive and poorly designed interfaces require physicians to focus more on the monitor. The need to use both keyboard and mouse to navigate the screen requires the physician to sit more squarely facing the monitor. This leads to poor eye contact, missing non verbal cues and even changes the driver of the interview from the patient to the computer prompts. The short visit times combined with more information to review and the pressure for timely completion of documentation adds a very tangible stressor in the equation. In the past, physicians would sit squarely facing the patient, taking notes in the paper chart in a very natural and unobtrusive manner. History taking can correctly diagnose over 50% of complaints in the ambulatory setting. Without a good history, physicians are more likely to take a shotgun approach and order more tests.
  4. The quality of the documentation - EHRs allow several mechanisms for increasing more rapid documentation. There are smart phrases, templates, and ability to copy and paste. In the past when one had to hand write ones notes, we were parsimonious and focused. Every word and every phrase had a meaningful thought behind it. Now we have reports of voluminous notes where it is difficult to find the real information. Physician payment for an encounter is associated with the detail of the documentation (which is not necessarily the same as the thoroughness of the clinical encounter). The more thorough a clinical encounter, the less time a physician has to document all facets of the encounter. Templates can help to solve this problem. There is a financial incentive for more detailed documentation. The temptation to create and use templates that support a higher level of coding is huge. . The balance between appropriately detailed, cognitively rich, non-templated notes and a quick templated, copy pasted note may be difficult to achieve for many providers.
  5. The quality of the information in the EHR - There is a risk of over reliance on the EHR for information that should be obtained from the patients. There are several examples of how the EHR information may be inaccurate:
    • Entering information using drop downs and selections can cause problems. Thus instead of open-texting something like "Osteoarthritis of the left knee medial compartment" the physician might choose something like "Osteoarthritis generalized" Critical information is lost as data entry is difficult or time consuming
    • Instead of asking a patient whether she smokes, a physician may just look this up in the EHR social history tab noting that this was updated at the visit with a colleague. He may not realize that the colleague made the same assumption and the information was not truly verified for a long time.
    • A patient being admitted to a hospital from the Emergency room usually has the entire history and physical documented in the EHR by the ER physician. While waiting for the patient to reach the hospital floor, a physician may review this information and then not go through this process with the patient again. Often the patients will promote this, being tired and ill and just wanting to get some rest, they will say something like, "It is all in that computer, why are you asking me this again". The admitting physician writes a note based on the ER physician note and if there are any inaccuracies in there, they get cemented into the patients records.
    • A patient when he first walks into the office, gets his Blood pressure checked by a nurse who enters it in a discrete field in the EHR. Often the physician may check this again and enter a different value into their notes. This information is not easily retrieved and the only information available as subsequent encounters is the one recorded by the nurse. This can lead to wrong decisions e.g. a medication being started if the BP appears to be poorly controlled.

These are some potentially major problems with the use of EHRs. Some of them fall into the no man's land and no one may take the responsibility of fixing them. Others have solutions that can be built into the program but will need to be implemented appropriately at each institution. These problems if left uncorrected have the potential of defeating the entire potential of EHRs.

I am taking the liberty to embed a recent Dilbert strip here