Monday, April 10, 2017

Pittsburgh Post-Gazette: "Medication errors in hospitals don’t disappear with new technology". Government: "It's the doctors' fault." I am cited.

The Pittsburgh Post-Gazette published an article on EHR problems yesterday entitled "Medication errors in hospitals don’t disappear with new technology."  It is based on a recent study by the Pennsylvania Patient Safety Authority, retrievable here:;14(1)/Pages/01.aspx

I am cited.  Also cited is an HHS official, Dr. Andrew Gettinger, acting deputy national coordinator for health information technology in the U.S. Dept. of Health and Human Services, who disagreed with my views.  I am familiar with Dr. Gettinger's views.  More on that later.

Medication errors in hospitals don’t disappear with new technology
Steve Twedt
Pittsburgh Post-Gazette

In the first six months of 2016, Pennsylvania hospitals reported 889 medication errors or close calls that were attributed, at least in part, to electronic health records and other technology used to monitor and record patients’ treatment.

A majority of the errors pertained to dosages — either missed dosages or an administration of the wrong dose. Of the 889 errors, nearly 70 percent reached the patient. Among those, eight patients were actually harmed, including three involving critical drugs such as insulin, anticoagulants and opioids.

The extent of the injuries was not detailed, although no deaths were recorded.  Those are the stark numbers in a new analysis by the Pennsylvania Patient Safety Authority, an independent state agency that looks at ways to reduce medical errors.

But interpretations of the report’s significance — and specifically the overall benefits and risks of information technology in a hospital setting — cross a wide spectrum.

The wide spectrum is the gap between those who believe in what might be called cybernetic supremacy (that is, the hyper-enthusiasts who ignore the real-world downsides of technology such as today's EMRs) versus those who promote what I call cybernetic sobriety (a more candid, mature attitude fostered by actual knowledge of the long history of cybernetic failures and the myriad causes of such failures).

Some view reports such as that of the Pennsylvania Patient Safety Authority (PPSA) in a reasonably patient rights-oriented manner, including the PPSA itself:

“This is the classic ‘tip of the iceberg,’” said pharmacist Matthew Grissinger, manager of medication safety analysis for the Patient Safety Authority in Harrisburg and co-author of the analysis with fellow pharmacist Staley Lawes. “We know for a ton of reasons not every error is reported.”

I've written extensively at HC Renewal on the "tip of the iceberg" issue, a phrase also used in the past by the FDA CDER (Center for Devices & Radiological Health) director Jeffrey Shuren MD JD and others.  See for example my February 28, 2010 post "FDA on Health IT Adverse Consequences: 44 Reported Injuries And 6 Deaths In Two Years, Probably Just 'Tip of Iceberg'" at as well as my January 8, 2016 post "Yet another observation that known health IT-caused injuries and deaths are 'the tip of the iceberg'" at

Of course, a PPSA disclaimer was issued, in my view perhaps to placate the health IT industry:

...Mr. Grissinger cautioned that the findings are “absolutely not” an indicator that patients are less safe, as hospitals have moved from paper to electronic records incorporating health information technology...the authors did conclude that technology meant to improve patient safety “has led to new, often unforeseen types of errors” due to system problems or user mistakes.

A more correct statement might have been that "these most current findings are yet another red flag that patients could be less safe with bad health IT, but since there are a 'ton of reasons' not every error is reported, we just don't know - and we truly need to devote a great deal of effort towards filling the gaps in our limited knowledge."

I've written on the issue of not jumping to health IT safety conclusions, one way or another, based on current data, especially when that data is admittedly limited.  For example, see my April 9, 2014 post "FDA on health IT risk:  "We don't know the magnitude of the risk, and what we do know is the tip of the iceberg, but health IT is of 'sufficiently low risk' that we don't need to regulate it" at

In that post I noted that a secret 2010 FDA internal report on health IT risk (marked "not for public use") unearthed by investigative reporter Fred Schulte stated that "...In summary, the results of this data review suggest significant clinical implications and public safety issues surrounding Health Information Technology...The absence of mandatory reporting enforcement of H-IT safety issues limits the number of relevant MDRs [device reports] and impedes a more comprehensive understanding of the actual problems and implications."

We don't know what we don't know, but to date the efforts to robustly learn the truth has been milquetoast to non-existent.  "Proof (of safety) by lack of evidence" - in an area where we admit the evidence is likely severely deficient - seems to be the default industry go-to position.  "Proof by lack of evidence", of course, is a logical fallacy.

Back to the Pittsburgh Post-Gazette:

... Frustration with the technology
In January 2015, 35 physician groups — including the American Medical Association, the American Academy of Family Physicians and the American Society of Anesthesiologists — sent a nine-page letter about electronic health records to the national coordinator for health information at the U.S. Department of Health and Human Services.

Their purpose was to convey their “growing frustration with the way EHRs are performing,” the letter stated.

“Many physicians find these systems cumbersome, do not meet their workflow needs, decrease efficiency, and have limited, if any, interoperability. Most importantly, certified EHR technology can present safety concerns for patients.”

That Jan. 2015 letter is at and speaks for itself.  Kudos to the Post-Gazette for citing it; the public is largely unaware of its existence.

I am then cited in the Gazette article:

Physician Scot Silverstein, a Philadelphia-based consultant and independent expert in electronic health records and vocal critic of such systems, calls the software “legible gibberish” better designed for handling warehouse inventory than managing and monitoring patient care in a clinical setting.

“Electronic health records are a massively complex computer application, far too complex than is needed for a clinic taking care of patients,” he said in a phone interview. “EHRs need to be toned down, be less complex, and be used less.”

Opportunities for mistakes are numerous, he said, as a physician may have to scroll through multiple screens, while each screen with a dozen or more columns plus an array of drop down menus. Some systems, he said, allow doctors to keep screens on multiple patients open simultaneously, increasing the chances of a medication mix-up.

“The software needs to be designed better.”

I am a vocal critic of bad health IT, and actually called the output of the systems to be "legible gibberish" as at my Feb. 27, 2011 post "Two weeks, two reams" at, but the quote is close enough.

Indeed, today's EHRs seem more designed for mercantile, manufacturing and management settings, and "calm, solitary office environments" (channeling Joan Ash) rather than the incredibly complex, poorly bounded and unpredictable environment of clinical medicine.

I am quoted accurately on the complexity and overuse issue, although the issue of preventing  physicians from having multiple patient screens open was actually a short term workaround known to me to have been put in effect some years back.  This was done when a major EHR was unpredictably transposing orders into wrong charts when multiple patient's screens were open (creating two potential patients at risk).

The software indeed needs to be designed better, to meet clinical needs.

Dr. Silverstein, who says his mother’s death was precipitated by a heart medication mix-up involving her electronic health record, cites federal initiatives giving hospitals financial incentive to implement electronic health systems as pushing the programs without sufficient vetting.

“The thinking was, ‘Computers plus doctors equals better medicine,’ period. But the technology was not and is still not ready for that kind of push.”

Indeed it was not ready, being experimental technology. Further, vetting in real-world settings via robust premarket surveillance, and postmarket surveillance of any rigor were, in fact, absent when massive incentives (and penalties) were announced as part of the so-called Economic Recovery Act and its "HITECH" component.

Instead, he recommends some combination of paper, with paper imaging capability so records are accessible, and electronic systems. “I don’t think paper should or ever will go away completely,” he said.

On this issue, and for a highly successful real-world example, see my August 6, 2016 post "More on uncoupling clinicians from EHR clerical oppression" at

I note with some irony about the above linked post (regarding a highly successful EMR that protected clinicians from oppressive clerical burdens) that the newly-appointed Director of the Office of the National Coordinator for Health IT (ONC), Dr. Donald Rucker (, was formerly the Chief Medical Officer of Shared Medical Systems, a hospital infrastructure IT provider.  He then became CMO of the failed Siemens Healthcare EMR effort after SMS was bought out ca. 2000.  Siemens Healthcare officials told me ca. 2007 that the real-world, highly successful invasive cardiology information system I'd developed as shown in the aformentioned Aug. 2016 post was "impractical" for commercial emulation.

Back to the Post-Gazette article.  In it, a government health IT official blames the doctors, a line I've heard dating back to the early 1990s when I was a postdoctoral informatics fellow at Yale:

A need for better training

Anesthesiologist Andrew Gettinger, acting deputy national coordinator for health information technology in the U.S. Dept. of Health and Human Services, disagrees with Dr. Silverstein.

He identified three key components to a successful electronic health record system — good design and implementation and the users’ good understanding of the system.

I have no disagreement there, only on the route to achieve those goals.

“What we find is that many clinicians who complain vociferously about the software and how many clicks it takes, and how user unfriendly it is, have not actually taken the time to understand the system,” he said.

This seems the "blame the physicians, they're just complainers and Luddites" canard I've written about for almost 20 years now.

Gettinger seems to ignore the issue of bad health IT and use error:

  • Bad Health IT is health IT that is ill-suited to purpose, hard to use, unreliable, loses data or provides incorrect data, is difficult and/or prohibitively expensive to customize to the needs of different medical specialists and subspecialists, causes cognitive overload, slows rather than facilitates users, lacks appropriate alerts, creates the need for hypervigilance (i.e., towards avoiding IT-related mishaps) that increases stress, is lacking in security, lacking in evidentiary soundness, compromises patient privacy or otherwise demonstrates suboptimal design and/or implementation.   (S. Silverstein and J. Patrick).
  • Use error (as opposed to user error) is defined by another U.S. government agency, the National Institute of Standards and Technology (NIST) as follows: "Use error" is a term used very specifically by NIST to refer to user interface designs that will engender users to make errors of commission or omission. It is true that users do make errors, but many errors are due not to user error per se but due to designs that are flawed, e.g., poorly written messaging, misuse of color-coding conventions, omission of information, etc. From "NISTIR 7804: Technical Evaluation, Testing and Validation of the Usability of Electronic Health Records." It is available at (PDF).

No amount of "training" can compensate for those issues.  Further, physicians and nurses just don't have abundant time for such training about mega-complex systems, on which they're already spending 50% or more of their time.  They especially don't have the time to learn multiple EHR's, a situation that exists for clinicians who work on more than one hospital.  I possess the physician and nurse user guides for a number of EHRs though my forensics work.  A manual for an EHR is as complex as a manual for an office suite like MS Office, or an OS such as Windows.

There's also the fact that physicians and nurses are not reimbursed for the hours they spend feeding the payers and other profit-makers the data, for free.

“Quite frankly, doctors are not always the best at signing up for training and taking the training...

Blaming the doctors again.

... , and some of the training is not always the best.”

Not that, as mentioned previously, "training" is at the root of the EHR problem.

He allowed that the usability criticism “is a very legitimate thing to look at”... 

How kind of Dr. Gettinger to acknowledge what has been known in the IT world for decades about poor usability, e.g., this mid 1980's wisdom written for the U.S. Air Force on user interfaces:

August 1986
Sidney L. Smith and Jane N. Mosier
The MITRE Corporation
Bedford, Massachusetts, USA
Prepared for Deputy Commander for Development Plans and Support Systems, Electronic Systems Division, AFSC,
United States Air Force, Hanscom Air Force Base, Massachusetts.
Approved for public release; distribution unlimited.

The design of user interface software is not only expensive and time-consuming, but it is also critical for effective system performance. To be sure, users can sometimes compensate for poor design with extra effort. Probably no single user interface design flaw, in itself, will cause system failure. But there is a limit to how well users can adapt to a poorly designed interface. As one deficiency is added to another, the cumulative negative effects may eventually result in system failure, poor performance, and/or user complaints.

Outright system failure can be seen in systems that are underused, where use is optional, or are abandoned entirely. There may be retention of (or reversion to) manual data handling procedures, with little use of automated capabilities. When a system fails in this way, the result is disrupted operation, wasted time, effort and money, and failure to achieve the potential benefits of automated information handling.

In a constrained environment, such as that of many military and commercial information systems, users may have little choice but to make do with whatever interface design is provided. There the symptoms of poor user interface design may appear in degraded performance. Frequent and/or serious errors in data handling may result from confusing user interface design [in medicine, this often translates to reduced safety and reduced care quality - ed.] Tedious user procedures may slow data processing, resulting in longer queues at the checkout counter, the teller's window, the visa office, the truck dock, [the hospital floor or doctor's office - ed.] or any other workplace where the potential benefits of computer support are outweighed by an unintended increase in human effort.

In situations where degradation in system performance is not so easily measured, symptoms of poor user interface design may appear as user complaints. The system may be described as hard to learn, or clumsy, tiring and slow to use [often heard in medicine, but too often blamed on "physician resistance" - ed.] The users' view of a system is conditioned chiefly by experience with its interface. If the user interface is unsatisfactory, the users' view of the system will be negative regardless of any niceties of internal computer processing.

Back to Dr. Gettinger for a somewhat non-sequitur 'BUT' disclaimer:

... BUT he defended the federal incentives, saying they defrayed the cost to hospitals while encouraging vendors to develop better systems.

I would say the incentives, just like the spectacularly failed subprime mortgage market a decade ago, just incented the health industry to waste hundreds of billions of dollars on half-baked, experimental technology, alienating physicians and nurses (cf.: the 2015 Medical Societies letter mentioned above).  The incented effort even put some organizations in financial jeopardy, e.g.,

 "MD Anderson to cut about 1,000 jobs due to 'financial downfall officials largely attributed to its EPIC EHR implementation project'
" at

"What is more important in healthcare, computers, or nurses and other human beings? Southcoast Health cutting dozens of jobs on heels of expensive IT upgrade" at

"Lahey Health: hospital jobs lost, but computer vendors prosper" at,

"Monetary losses and layoffs from EHR expenses and EHR mismanagement" (,

"Financial woes at Maine Medical Center: Reading this blog might have saved them millions of dollars, and prevented massive 'cost saving initiatives'" (,

and "In Fixing Those 9,553 EHR "Issues", Southern Arizona’s Largest Health Network is $28.5 Million In The Red" (

I also believe the easy money disincented the vendors from improving the techology, instead selling what they had on hand and acting to discourage innovation and competition to maximize their profits, e.g., see my April 16, 2010 post "Healthcare IT Corporate Ethics 101: 'A Strategy for Cerner Corporation to Address the HIT Stimulus Plan'" at and my August 31, 2012 post "Health IT Vendor EPIC Caught Red-Handed: Ghostwriting And Using Customers as Stealth Lobbyists - Did ONC Ignore This?" at

Finally, I regrettably note that Gettinger seems to possess a rather hard-nosed attitude about health IT harms.  I have contributed, of course, to articles about EHR's in other publications, including, among many others, Politico.  Arthur Allen at Politico wrote me this in 2015 regarding my opposition to the toothless "Health IT safety center" concept, and my promotion of a need for true HIT regulation:

On Wed, Jun 17, 2015 at 1:13 PM, Arthur Allen <> wrote:
I’m putting together a piece on the safety center with some notes from an interview I did with Andy Gettinger a few weeks ago. I asked him whether he though the RTI panel (which RTI named, apparently) would have come to the same consensus – that the safety center should be a safe harbor, not an investigatory agency – if you [i.e., me - Scot  - ed.] had been on the panel.

He said,
“he [i.e., me - Scot - ed.] may have heard what we were intending and been able to step back from specific things relative to his mother’s care and gotten to a space to see that this initiative has the potential of making real change in the EHRs used throughout the country. I would have loved to have Scot at the table.”

Any response?

In other words, if only I was able to "step back" from my mother's severe injury, year's worth of horrible suffering as a cripple before she died as a mentally-impaired vegetable, and my lovely mother being taken away from my home in a body bag as a result of a health IT mishap, I'd be able to see just how wonderful a toothless HIT safety center would be.  (Also, I was never asked to be "at the table".)

What a kind comment that was. 

In conclusion:

While I wish the Pittsburgh Post-Gazette article was longer, in its limited space its author did touch upon the major relevant issues well regarding the PA Patient Safety Authority study and its implications towards national Health IT policy.

ONC's Dr. Andrew Gettinger's responses, however, seems to reflect an unwillingness of he and the government to acknowledge Bad Health IT.  His repsonses also appear to show a lack of appreciation of the complaints about EMRs from nearly 40 medical societies.  "It's the doctors fault" for not training enough.

He does acknowledge that better IT would be a good thing, but to date the best HHS could come up with to achieve that goal is a toothless Safety Center. Healthcare IT would be the only healthcare device sector afforded that extraordinary regulatory accommodation.

The notion that all that is needed to solve EMR problems is clerical training of (resistant) physicians seems that of a computing dilettante, and/or a health IT hyperenthusiast.  Such a view ignores decades of knowledge of bad IT, and in multiple sectors.

The blaming of physicians is also decidedly unhelpful towards the reputation of the technology and its enthusiasts in government.  Bad enough that physicians are already spending 50% or more of their time at computers, distracting from patient care.  Gettinger's "solution" also fails to acknowledge that physicians often work in multiple hospitals with different EHRs. They don't have the time to become clerical experts in multiple mega-complex systems.

Claiming the national incentives promoted the vendors to make better health it is also absurd. It actually promoted them to sell the bad health IT they had on hand, and lessened any motivation to improve the technology.

What the issues really boil down to is a conflict between those who believe in cybernetic supremacy (the hyperenthusiasts who ignore the real-world downsides) versus those who promote what I call cybernetic sobriety (a more candid, mature attitude fostered by actual knowledge of the long history of cybernetic failures and the myriad causes of such failures).

-- SS


InformaticsMD said...

See also "Doctors Demand Extreme EHR Makeover...Right Now" at

Anonymous said...

With comments such as Gittinger's, the ONC ought to be shut down. Spend the savings on real science.