In the past few
months there have been multiple studies demonstrating that we still have a long
way to go in bridging the human-computer interface, at least as far as electronic
medical records go. The full potential of electronic health records and
clinical decision support tools has yet to be realized.
First was a study of
children’s hospitals using the pediatric version of the Leapfrog Group’s CPOE
evaluation tool (Chaparro
2016). We like the Leapfrog tool as it
applies to adult hospitals (see our previous columns for July 27, 2010 “EMR’s
Still Have a Long Way to Go”, June
CPOE Simulation: Improvement But Still Shortfalls”, March 2015 “CPOE
Fails to Catch Prescribing Errors” and May 3, 2016 “”).
The adult and pediatric tools assess the ability of a hospital’s CPOE system to
identify potentially problematic orders. It was used to evaluate 41 pediatric hospitals over a 2-year period, assessing
overall performance as well as decision support categories (eg,
drug-drug, dosing limits). CPOE systems at these hospitals overall were able to
identify 62% of potential medication errors in the test scenarios, but ranged
widely from 23-91% at individual institutions. The highest scoring categories
included drug-allergy interactions, dosing limits, and inappropriate routes of
administration. Looking at hospital performance longitudinally over time, they
found that hospitals with longer periods since their CPOE implementation did
not have better scores upon initial testing, but after initial testing there
was a consistent improvement in testing scores of 4 percentage points per year.
The latter exemplifies the ability of hospitals to learn from using the
Leapfrog tool and improve their systems.
In our May 3, 2016 Patient
Safety Tip of the Week “”
we discussed the results of The Leapfrog Group’s most recent report on
how adult hospitals perform on their CPOE evaluation tool (Leapfrog
2016). To fully meet Leapfrog’s standard, hospitals must:
In 2015 nearly two-thirds of hospitals (64%) fully met the
standard, showing a considerable improvement compared to 14% in 2010. The
hospitals also demonstrated improved performance in medication reconciliation.
However, on the 2015 Leapfrog Hospital Survey, hospitals’ CPOE systems failed
to flag 39% of all potentially harmful drug orders, or nearly two out of every
five orders. The systems also missed 13% of potentially fatal orders.
Another recent, FDA-sponsored project analyzed medication
errors potentially related to computerized prescriber order entry (CPOE). Amato
and colleagues (Amato
2016) reviewed all patient safety medication reports that occurred in the
medication ordering phase from 6 participating sites. They found that 51.9% of
the medication error reports were related to CPOE. Of these, CPOE facilitated
the error in 13.1% and potentially could have prevented the error in 86.9%. The
most frequent CPOE issues involved transmission errors (eg.
orders not being received in the appropriate place), erroneous dosing, and
duplicate orders. These resulted in delays in medication reaching the patient, patients
receiving duplicate drugs, or receiving a higher dose than indicated.
Also from the
pediatric literature was a column on malpractice related to EHR’s (Oken
2016). Oken cited a medical liability
insurer’s recent review of EHR-related malpractice claims showed that 42% were
derived from system errors and 64% were from user factors. System errors include lack of interoperability of various
systems and lack of clinical decision support in some systems. He notes that
not all electronic prescribing programs in pediatric offices are able to
validate the pediatric drug dosage, check drug interactions or include
prescriptions from other physicians. Lack of pediatric functionalities in many
EHRs, such as weight-based dosing, pose significant safety risks to children.
On the user side, copying and pasting clinical information from previous
medical encounters (sometimes known as “cloning”) can be a major risk. And
somewhere in between system and user error is the problem of failure to
recognize and appropriately follow up on reports that come back to the EHR.
A recent systematic review on the effects of healthcare IT
on patient outcomes was informative (Brenner 2016).
Using quite rigorous criteria for inclusion of studies, the authors found that
such studies in the outpatient or long-term care settings were scant. The
majority of studies demonstrating a positive impact of healthcare IT on patient
outcomes came from inpatient settings and were likely single-center studies.
And only 36% (25 of the 69 studies meeting criteria for inclusion) actually
showed a positive effect on patient outcomes for the primary outcome measured.
The rest showed mixed effects or no effect on patient outcomes. And one study
showed a negative effect. Quality of the included studies was variable and most
studies were observational or cohort studies rather than randomized controlled
trials. The review really emphasizes the need for high quality studies with
better design and larger populations, particularly on the outpatient side where
the majority of care is rendered.
A recent AHRQ Web M&M case (McGreevey 2016)
illustrated a CPOE-related problem that appeared after the transition from
paper order sets to computer-based order sets. In the old paper-based world
order sets for electrolyte replacement or correction included segments for
hypokalemia and hypomagnesemia in close approximation (hypomagnesemia often
occurs in patients who are hypokalemic). However, when the order sets were
converted to the electronic format, attempts were made to simplify the
information provided on individual computer screens. As a result, order options
for magnesium no longer appeared on the same screen as those for potassium. In
the case presented there was failure to address the hypomagnesemia when the
patient’s hypokalemia was addressed. Loss of the visual cue to prompt assessment
of the magnesium levels was considered contributory to the adverse outcome in
That case highlights
a problem we’ve previously discussed. The vulnerability of CPOE systems when
changes are made to various components or files was also noted in our May
3, 2016 Patient Safety Tip of the Week “”.
There we discussed a study from Brigham and Women’s Hospital (BWH) in Boston,
which probably has the most robust CDSS of any healthcare organization anywhere
2016) and found several errors that were often very difficult to detect.
They identified several contributing factors:
The studies collectively demonstrate that it is never enough
to simply implement a CPOE system or e-prescribing system with clinical
decision support systems and assume your patients will be safe from medication
errors. Clearly, ongoing evaluation and assessment using validated tools are
important to identify vulnerabilities that may be unexpected. We, of course,
should expect better design and function from our IT vendors. The Wright study
clearly shows that problems may arise even when the initial design and
implementation were good yet changes to systems or files result in gaps that may
go unidentified for long periods.
reinforce some other common errors related to computer order entry. Another
recent AHRQ Web M&M case (Wears 2016) illustrated the “picklist” error (also known variously as the mouseclick error, drop-down list error, cursor error,
stylus error, or juxtaposition error depending upon the setting and device
being used) that we so often continue to see. A physician intending to
order a CT scan of the abdomen and pelvis with oral contrast but without IV
contrast mistakenly clicked on the CPOE order for a CT scan that included IV
contrast. The patient subsequently developed a rise in serum creatinine,
consistent with contrast nephropathy. Also, truncation of some items in
picklists may lead to errors when physicians fail to scroll fully side-to-side
Timely in this
regard is a new report prepared for the Office of the National Coordinator for
Health Information Technology on the safe use of picklists in ambulatory care
2016). The report really focuses on two key issues:
While many wrong-patient errors occur from having multiple
patient charts open simultaneously, picklist errors remain an important
contributing factor to wrong-patient errors. The authors note that simply
having adjacent items in a list can predispose to picklist errors but other
design factors, such as pick list length and organization of the items in the
list, may also contribute. They note that personalizing
the list of patients to those that the provider has seen and allowing for sorting, filtering, and/or grouping the
patients into categories can result in a shorter list of patients from which
the provider can select, thus increasing the likelihood of selecting the
We’ve done multiple columns on patient identification errors.
You’ll find links to most of them and links to the literature in our January 19, 2016 Patient Safety Tip of the
Week “”. That includes discussions on the “ID-verify alert”, the “ID-reentry
function”, and use of patient photographs
that are also discussed in the new ONC report as ways to force active
verification of patient identity.
List length is
one factor contributing to wrong-medication errors. Obviously, the similarity
of drug names is another. We’ve done many columns on LASA (look-alike, sound-alike) drug errors and the use of Tall Man lettering to help minimize
these (see, for example, our December
1, 2015 Patient Safety Tip of the Week “” and our July 2016 What's New in the Patient Safety World column “”).
Allowing ordering providers to customize lists of medications they most commonly use is another
way to reduce wrong-medication errors. But keep in mind that unanticipated changes to lists can also
be a source of errors. We’ve seen cases where a physician is used to choosing
the first item from a medication category and then a vendor updates the
software and a different medication is now first on the list. If the physician
is not paying careful attention, he/she may just click on that first item.
Requiring input of the indication
for each medication ordered is something we’ve long advocated. That helps,
for example, when you accidentally click “digoxin” rather than “Dilantin”. Your
clinical decision support tool would alert you that digoxin does not meet the
indication of “anticonvulsant” that you entered. But one problem, particularly
in hospitalized patients, is that you may not know the indication for which a
patient had been started on previously and many medications (eg. beta blockers) have multiple different indications. But
we still feel requirement of an indication is a useful safeguard in many cases.
of the orders entered in a session is another way to have the ordering
provider review the orders and potentially catch errors.
The ONC report has several appendices that contain the
recommendations and best practices to reduce the likelihood of picklist errors.
Use them! They provide details for each of the following recommendations:
There is also a self-assessment form you can utilize to help
identify your vulnerabilities. The report is replete with examples of various
types of picklist errors and has excellent references.
The government incentives and “meaningful use” requirements
have achieved the enviable goal of extending electronic health records to the
vast majority of medical practices and hospitals. But the proliferation of
systems has led to widespread variability of performance and multiple flaws
that have impacted patient outcomes. Fortunately the ONC (Office of the National Coordinator for Health
Information Technology) has just published a federal rule that will
provide the ONC more authority to regulate and oversee the EHR tools that get
certified via the Certified EHR Technology (CEHRT) program (DHHS 2016).
The new rule basically focuses on catching flaws in EHRs already being used by
providers and will address some of the usability, patient safety, and workflow
design flaws found in the current generation of health IT tools. That is a
welcome addition that should benefit patients, physicians, hospitals, and a
whole host of stakeholders.
It’s time for us to
begin realizing the full potential benefits that the world of healthcare IT
See some of our other
Patient Safety Tip of the Week columns dealing with unintended consequences of
technology and other healthcare IT issues:
Chaparro JD, Classen
DC, Danforth M, et al. National trends in safety performance of electronic
health record systems in children’s hospitals. J Am Med Inform Assoc 2016; Epub ahead of print
2016 Sep 16
The Leapfrog Group. Hospitals’ Computerized Systems Proven
to Prevent Medication Errors, but More is Needed to
Protect Patients from Harm or Death. The Leapfrog Group 2016; April 7, 2016
Amato MG, Salazar A, Hickman TT, et al. Computerized
prescriber order entry–related patient safety reports: analysis of 2522
medication errors. JAMIA 2016; First published online:
27 September 2016
Oken RL. Lessons learned from
EHR-related medical malpractice cases. AAP News 2016; August 8, 2016
Brenner SK, Kaushal R, Grinspan Z,
et al. Effects of health information technology on patient outcomes: a
systematic review. J Am Med Inform Assoc 2016; 23:
McGreevey JD. Unexpected Drawbacks
of Electronic Order Sets. AHRQ Web M&M. Cases & Commentaries. Published
Wright A, Hickman T-T T, McEvoy D, et al. Analysis of
clinical decision support system malfunctions: a case series and survey. JAMIA
2016; First published online: 28 March 2016
Wears RL. Unintended Consequences of CPOE. AHRQ Web M&M.
Cases & Commentaries. Published October 2016
Rizk S, Oguntebi
G, Graber ML, Johnston D. Report on the Safe Use of Pick Lists in Ambulatory
Care Settings. Issues and Recommended Solutions for Improved Usability in
Patient Selection and Medication Ordering. (prepared
for the Office of the National Coordinator for Health Information Technology).
DHHS (Department Of Health And Human Services). 45 CFR Part
170. RIN 0955–AA00. ONC Health IT Certification Program: Enhanced Oversight and
Accountability. Federal Register 2016; 81(202): 72404-72471. Wednesday, October