Saturday, December 14, 2013

Yet Another "Anecdote" - Inpatient Results of Electronic Prescribing "Disappointing"

Many of those in the Medical Informatics community, especially the academics in the upper echelons of the American Medical Informatics Association, are not of a risk recognition / risk management mindset.  Typical of academics, they are often also hostile towards dissent from the party line, as you can read about at my post "The Dangers of Critical Thinking in A Politicized, Irrational Culture" at

The academics, with a few exceptions, have repeatedly conflated scientific anecdotes of supposed positive results from health IT with risk management-relevant incident reports of bad outcomes and 'near misses', as an Australian colleague wrote about, via me, on August 17, 2011 at "Anecdotes and Medicine, We are Actually Talking About Two Different Things" at

They dismiss the latter incident reports, even when from well-qualified observers, while giving great attention and credence to the former [a few years as a safety manager in a large urban transit authority disabused me of that type of behavior - ed.], when the former conveniently fit their most cherished beliefs about the beneficence and efficacy of today's health IT.  Further, quality or lack thereof of the former type of evidence is often not considered.  See for instance my post of March 9, 2011 "ONC: The Benefits Of Health Information Technology: A Review Of The Recent Literature Shows Predominantly Positive Results" at for a stunning example of this phenomenon directly from the national leadership of health IT.

In my view, this phenomenon has led to a substantial loss of focus on health IT realities needed in order to remediate the industry and realize the true benefits of which the technology is capable.

Now there's yet another "anecdote" at Med Page for the experts to chomp on:

E-Prescribing: Inpatient Results Disappointing
Published: Dec 12, 2013
By Sarah Wickline , Contributing Writer, MedPage Today

ORLANDO -- Electronic prescription order entry and medication reconciliation reduced some errors in hospital settings but increased others, and did not meet overall expectations, researchers reported here.

After implementation of a computerized prescriber order entry (CPOE) system, one hospital experienced a 29.2% increase in medication dispensation errors (P less than 0.05), Ramadas Balasubramanian, PharmD, PhD, of the Carolinas Medical Center-Pineville in Charlotte, N.C., and colleagues reported at the midyear meeting of the American Society of Health-System Pharmacists.

It is likely these are qualified observers who in fact might be expected to be biased towards showing good results of this technology.

In a second study, another hospital experienced a 12% improvement in accuracy (P less than 0.001) after implementation of electronic medication reconciliation charts, according to Jill Covyeou, PharmD, of Ferris State University in Big Rapids, Mich., and colleagues.

For the tens of millions of dollars likely spent to achieve a mere 12% improvement, one wonders if a far less expensive investment in experienced human resources might not have accomplished the same results or even far better.

For their study, Balasubramanian and colleagues looked at the impact of CPOE on medication errors in a community hospital setting.

At the end of 2011, when the Carolinas Medical Center-Pineville hospital had only 119 beds, officials there implemented the CPOE-CANOPY system. In early 2012, the hospital nearly doubled in capacity to 210 beds and opened the pharmacy to 24-hour operation. The researchers looked at medical errors from October 2008 through October 2012.

The categories of medication errors included: drug omission, administration at the wrong time, unauthorized drug, wrong dose, and wrong form of dose.

Any of which, of course, can harm or kill, I note.

There was a 57% increase in medication doses from prior to the CPOE system to after implementation, but even after volume adjustments, the number of errors per 1,000 dispensed medications still increased by 29.2% (P less than 0.05).

Bad health IT such as CPOE systems designed for (per Joan Ash) "calm and solitary office environments" would be expected to perform more poorly as caseloads increase and clinicians have less time for computer fritter.

Unauthorized drug dispensation and improper dose of medication decreased post-CPOE, but drug omission and administration at the wrong time were responsible for the increase.

Those seem to match the expressed concerns of another large group of "anecdote-profferers", e.g., the nurses at my Nov. 17, 2013 post "Another 'Survey' on EHRs - Affinity Medical Center (Ohio) Nurses Warn That Serious Patient Complications 'Only a Matter of Time' in Open Letter" at

From those nurses' Open Letter to management on health IT risks:

... Some of the concerns that nurses have brought to the attention of management include:
  • Medication errors/scanning issues - perhaps the biggest concern of all RNs
  • RNs unable to access patient records for hours at  a time
  • Incorrect descriptors and inaccurate drop-down menus
  • Incorrect calculations in the I&O and MAP [mean arterial pressure - ed.] portions of the chart 
  • Inaccurate medication times and the inability of RNs to ensure medications are scheduled correctly
  • Endless loops of computer prompts that are unable to be dismissed by RNs in an emergency

Back to the current Med Page article:

The use of CPOE software created a time cut-off issue that explained the wrong time of administration increase, the study authors told MedPage Today. If drugs were ordered 5 minutes after the cut-off for the morning medication administration, they would not make it to the patient until the evening rounds unless a special alert was sent to the nurse staff.

That is, a time-eating and fragile (and thus potentially dangerous} workaround.  I note that one does not have to work around something that is not standing in their way.

Balasubramanian and colleagues suggested that the lack of flexibility in the CPOE software was responsible for the drastic increase in medication errors, despite the fact that they thought it would improve error rates across the board.

In other words, the software is not truly fit for purpose.  See definition of "bad health IT" below.

... The authors suggested that electronic prescribing would be better with electronic medical records. "[E]lectronic prescribing alone may fail to increase medication list accuracy to the extent we would like," Covyeou wrote.

Replace "may fail" with "in our case, did fail."

The point about adding an EHR is certainly in the category of "wishful thinking."

If the EHRs are bad health IT, I opine the situation would likely get even worse (cf. Affinity Health, above).

From my site at 

Bad Health IT ("BHIT") 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, compromises patient privacy or otherwise demonstrates suboptimal design and/or implementation.

In summary, the technology is not a panacea, and can in fact be worse than paper.  Those who blindly ignore that reality and push for mass rollout of this still-experimental technology "no matter what" do not share, in my opinion, the ethics I was taught in Medical School.

-- SS


Anonymous said...

Wow, that is shocking. Who would have thunk that $multi-million systems of devices known as CPOE would increase the error rate by nearly a third?

Welllll, al they had to do was read the Kopel paper in JAMA 2005 showing about that many new errors.

Soooo, 7 years later, they are reaffirming what was known in 2005, demonstrating that the absence of regulation did not result in innovation by the vendors of EHR devices.

Anonymous said...

EHR and CPOE, are, wishful thinking. Great concept and would be fabulous if it worked well. But, face it. It is too complicated and the risks are high. People will die and or be harmed.

Wake up. It is only a dream.

Steve Lucas said...

I was struck by the use of the word academic and remembered a couple of instructors I had as an undergraduate, but also the problem instructors I had in my graduate program.

In graduate school instructors are much freer to impose their viewpoint on the students and force the students to at least pretend to accept this point of view. A friend’s son is having problems in his first semester with a PhD instructor who has been criticized by a number of students, will impact his grade point, but is being allowed to finish their tenure and retire. This is something I have seen on more than one occasion.

The real world result is that often we see people with academic credentials espousing ideas that have no merit, but due to their position they are accepted. Then following the pharma model they become KOL’s.

Technology has been the big beneficiary of this attitude as progressive thinking people all agree all we need is more technology. After all, down at the university we have small projects just waiting to be expanded.

Like many progressive ideas the reality is very different from the academic model.

Steve Lucas