EHR-Driven Medical Error: The Unknown and the Unknowable

Politico’s Arthur Allen has written a useful report on recent findings about EHR-related errors. We must keep in mind, however, that almost all EHR-related errors are unknown, and often unknowable. Why?

  1. The most common errors involved with EHRs are medication prescribing errors. But we seldom find those errors because those type of errors seldom manifest themselves because so many hospitalized patients are old and sick, have several co-morbidities and are taking many other medications. Key organs, like the liver, kidney and heart, are compromised. Bad things can happen to these patients even when we do everything right; conversely, good things can happen even when we do much wrong. We usually miss the results of, say, a wrongly prescribed medication. (Note: these types of ‘missed’ medication errors contrast to leaving a pair of hemostats in the gut or to wrong-site surgery—where most errors soon become obvious).
  2. As the experts referenced (Dr. Bob Wachter, Dean Sittig and Hardeep Singh) noted, very, very few cases make it to litigation, further reducing the numbers examined in the study discussed.
  3. Perhaps worse, few clinicians want to report problems even if they know about them. This is a litigious society and few medical professionals want to spend time in court. Also, as the authors Allen interviewed (all of them my friends and respected colleagues): some of the errors that were known did not result in harm and many were caught by others or by the professional involved in the error before they harmed.

In one of my papers we studied all medication orders stopped by the prescriber within 45 minutes. It emerged that 2/3rds of those orders were either wrong or at least suboptimal. To my knowledge, however, not a single one of those errors was reported as an error (or near miss). More, if the patient already received the medication, I also know of no case where the actual error was reported.

  1. The non-disclosure clause in EHR vendors’ contracts prevents clinicians from publicly reporting EHR problems to the public. Yes, they can report errors to the vendor, to the FDA, and to a patient safety officer. But if they took a screen shot of a dangerous or deceptive EHR screen and published that to their colleagues or to the web, they could be sued for millions. Moreover, the “hold harmless” clause in those contracts means that vendors are not responsible for errors—even if 2000 doctors have reported them. Of course vendors don’t wish to harm patients and make corrections, but they have not removed those two clause in the 9 years since I exposed them in an article in JAMA.
  2. EHRs are very difficult to use, and not all of the problems are due to lousy programming by the vendors. The systems must interact with hundreds of other IT systems that are constantly changing. Also there are always new environments, e.g., patients with differing diseases, new clinicians with different training, new equipment, and new requirements. New medications and new disease protocols require constant updating. The quality of the implementation teams differ, as do the institutions’ experience with technology.

All of that said, EHRs suffer from significant usability challenges, e.g.,

  • Confusing presentations of patients’ data and general poor usability
  • Drop-down lists that continue to several screens (with the existence of the extended often hidden from the clinician)
  • Pop-ups that hide medication or problem lists
  • Medication lists and problem lists that can’t be seen when ordering medications
  • Lab reports presented in erratic or absurd formats and sequences
  • Herds of decision support alerts that obscure the screen
  • Data that should be contiguous separated by three screens and multiple clicks
  • Critical information on the patient is lost because of proprietary EHR software, idiosyncratic device data formats, and refusal to accept data standards, and
  • Lack of true interoperability.
  1. As with any complex software system, one is often not sure if the user is the source of the problem. Did I fail to read a recent update memo? Did the hospital IT department make a change? Did the vendor make a change? Did another colleague alter the patient’s record in some way? Did a diagnosis change and thus the patient is re-classified in some way? Is there a bug? Did I put in the right combinations of passwords? Am I authorized to use this computer on this floor for these types of patients? Like all of us with very complex software, even physicians can feel like sinners in the hands of a capricious god. Few rush to report a “problem” that may be uncertain, due to something or someone else, and may well not even be a problem.
  2. As the piece by Allen also notes, the ONCs effort to construct a patient safety reporting system has been consistently refused by congress and the administration. Even if a clinician knows about a problem, there is no single place that systematically collects them.

In sum, we know just a tiny fraction of errors associated with EHRs. This report, while most welcome, does not reflect the tip of the iceberg. More, it’s a scratch on the tip of the iceberg.

NOTE: parts of this text are taken from some of my previous publications.



Senior Fellow, Wharton’s Leonard Davis Institute of Healthcare Economics;

Senior Fellow, Center for Public Health Initiatives, Perelman Sch of Medicine;

Adjunct Professor (full) Sociology Department;

Principal Investigator and Affil Prof of Medicine, Perelman School of Medicine;

Senior Investigator, School of Engineering and Applied Sciences

Also:  Chair, Clinical Information systems, AMIA; & Research Prof. Biomedical Informatics, SUNY@Buffalo



Categories: Uncategorized

5 replies »

  1. Rocky said “EHR is meant as a solution to the fragmented healthcare delivery system”

    Actually, EHR is most probably meant as a way to upcode as many visits as possible and to make most notes read the same, so it will matter less who is doing the documenting.

  2. I have been practicing ID for 38 years. Most of my referrals take the form “what should I give this patient”, which of course is a surrogate for “what’s wrong with this patient?”

    We are changing EHRs on December 1. The organization has spent two years and tens of millions on the transition, and they have done as well as anyone could do.

    But the new EHR demands, before you start to enter words or orders, that you specify the diagnosis. This now leads, immediately, to helpful hints, clinical decision making guidance, guidelines, etc, etc.

    But the enormous amount of decision support the new system will be providing begs the question: What if the diagnosis is wrong? That is, after all, one of the main reasons patients get referred to specialists, and even the idea of “wrong diagnosis” is probably improper while a patient is going through the process of establishing the final, correct diagnosis.

    My chief at Sloan Kettering, W. P. Laird Myers, used to say that 20% of the patients referred to Memorial Hospital did not have cancer at all, and the next 20% DID have cancer, but not the one they were referred for.

    What is going to happen now when clinicians trained from medical school to use systems like this start to receive text pages from the decision support specialists? “This is how you treat pneumonia”. That’s great – if you have pneumonia.

  3. One thing that stood out to me from Koppel’s blogpost is that the software is not tailored for its users and does not have venues to accept feedback. It seems like clinicians are not being heard by the government and the EHR vendors about their needs for the EHR. The U.S. government pushed for the adoption of EHR, creating laws that fine people for non-compliance. There are also many commercial companies who develop EHR systems that are not completely compatible with one another. EHR is meant as a solution to the fragmented healthcare delivery system but it brings additional separation among providers, government, and the vendors.
    A study done by Milstein et al evaluated hospital participation in EHR. 80.5% hospitals adopted EHR system at least on the most basic level. The researchers categorized EHR usage into two categories: basic and advanced. Basic level involvement with EHR includes 10 electronic capacities such as patient demographics, physician notes, medication orders, and laboratory reports. Advanced involvement includes additional features that are tied to performance evaluation and patient engagement like patients’ ability to view, download, and share their medical information online. Milstein found that among hospitals with advanced usage of the EHR are providers with more resources. Because patient participation in their medical decisions is critical to their decision making, the difference in EHR functions in hospitals may result in a more differential delivery of healthcare. Patients attending hospitals with higher IT and administrative capacities may already be characteristically different from those going to a safety-net provider. EHR, intended to improve health outcome for all, becomes another manifestation of health disparities due to SES in this country.
    Koppel is not the only one who is complaining about the EHR. Another major concern is privacy and information security. Would the ease of personal health information exchange compromise patient privacy? There have been breaches of HIPAA regarding the inappropriate use of patient information. But in my opinion, most security breaches result from inadequate privacy training of physicians and EHR users. Some healthcare providers download unencrypted confidential information on their portable personal devices which are prone to theft. I believe that most clinicians have good intentions to improve healthcare in any possible way, but they do need more training to ensure that the fear of losing data stored in the EHR does not turn patients away.
    How can the EHR be improved? Even though there are mixed reviews of the system, no discussion of complete withdrawal of the program has been initiated. With proper structure built in the EHR systems, patients can benefit from coordination of health professionals, reduced medical costs incurred for duplicate tests and excessive paperwork, and improved quality control of medical care. First, medical errors should be made more transparent. Hospital administrators should provide a platform that allows for report of minor medical errors. Some threshold of the severity of problems can be established so not everything will go public. Physicians should feel comfortable reporting issues and feedback to EHR vendors. EHRs should be optimized based on user needs. Healthcare workers who have proficient coding knowledge can also participate in the development process. The interoperability between EHR systems should also be enhanced so providers using different EHRs can “talk” to each other. In addition, the government should adjust the “Meaningful Use” incentives to encourage the rollout of advanced EHR uses in small or rural hospitals for underserved populations. Perhaps we can pair low- and high-resourced hospitals so that hospital groups have greater negotiating power with the vendors. Lower-resourced hospitals could also learn from hospitals who have the advanced features to reduce the costs of initiation. To address the privacy concerns, healthcare workers should receive advanced training about treatment of sensitive data. EHRs could be designed to prohibit downloads to personal computers. Encrypted data can be stored in clouds on VPN if remote access is necessary.
    We cannot expect EHR to be perfect in one step, but we can improve the system little by little, adding features that are deemed beneficial for health care delivery. For example, the National Institute for Occupational Safety and Health recommends EHR to include patient work information to enable physicians to identify disease risks. Working information allows for the surveillance and epidemiological studies for job-related conditions. We are pretty much at the point of no return and it is a group effort to make it work. It is for the good of all.

  4. In November 2015, I suggested that one solution to reduce EHR errors was to mandate that the source code of all EHRs be “public information.” Unfortunately, I was unsuccessful at promoting the idea among the medical community, the IT community, the professional organizations which represent physicians or the academic community.

    I still believe it is a solution which has no down-side and (probably) a serious upside. The recent ECW code-cheating scandal, where the Federal government fined ECW for “cheating” on their MU Certification exam, only reaffirms my belief that had my proposal been enacted, the ECW code-cheating scandal would have been unearthed much sooner.

    Here is the link to that proposal…
    A Proposal to Increase the Transparency and Quality of Electronic Health Records

    Hayward Zwerling

  5. Among the many med errors within a hospital as compared to a person’s person’s Comprehensive Healthcare Plan (CHP), most occur during a transfer between levels of care: especially out of an intensive care unit. Of course, the other is a person’s hospital discharge. For a root cause, we have no widely agreed-upon means to support the responsibility of and economic support (aka usable EHR) for a Primary Physician’s active commitment to maintaining a written CHP for each of their patient’s. Furthermore, we also do not have a nationally sanctioned and community driven process to assure that Primary Healthcare is equitably available to and ecologically accessible by each of its community’s citizens. This isolated paradigm represents the true “elephant in the closet” for solving the following riddle: if you keep doing what you have always done, you will receive what you have always received. The unacceptable cost and quality problems of our nation’s healthcare will not be solved without a dedicated strategy to build TRUST, COOPERATION and RECIPROCITY as the basis for the renewal of Primary Healthcare, neighborhood by neighborhood and community by community. It is likely that such an effort could be managed successfully with a fixed budget of $1.00 per citizen annually and managed by a new, semi-autonomous institution Chartered and funded by Congress.
    see http://www.nationalhealthusa.net/home/rationale/