In PLoS medicine, the following article was recently published by researchers at the University of New South Wales in Australia:
Westbrook JI, Reckmann M, Li L, Runciman WB, Burke R, et al. (2012) Effects of Two Commercial Electronic Prescribing Systems on Prescribing Error Rates in Hospital In-Patients: A Before and After Study. PLoS Med 9(1): e1001164. doi:10.1371/journal.pmed.1001164
The section I find most interesting is this:
We conducted a before and after study involving medication chart audit of 3,291 admissions (1,923 at baseline and 1,368 post e-prescribing system) at two Australian teaching hospitals. In Hospital A, the Cerner Millennium e-prescribing system was implemented on one ward, and three wards, which did not receive the e-prescribing system, acted as controls. In Hospital B, the iSoft MedChart system was implemented on two wards and we compared before and after error rates. Procedural (e.g., unclear and incomplete prescribing orders) and clinical (e.g., wrong dose, wrong drug) errors were identified. Prescribing error rates per admission and per 100 patient days; rates of serious errors (5-point severity scale, those ≥3 were categorised as serious) by hospital and study period; and rates and categories of postintervention “system-related” errors (where system functionality or design contributed to the error) were calculated.
Here is my major issue:
Unless I am misreading, this research took place in hospitals (i.e., "wards" in hospitals) and does not seem to focus (if even refer to) discharge prescriptions.
I think it would be reasonable to say that what are referred to as "e-Prescribing" systems are systems used at discharge, or in outpatient clinic/offices to communicate with a pharmacy selling commercially and not involved in inpatient care.
From the U.S. Centers for Medicare and Medicaid Services (CMS), for example:
E-Prescribing - a prescriber's ability to electronically send an accurate, error-free and understandable prescription [theoretically, that is - ed.] directly to a pharmacy from the point-of-care
I therefore think the terminology used in the article as to the type of system studied is not well chosen. I believe it could mislead readers not experienced with the various 'species' of health IT.
This study appears to be of an inpatient Computerized Practitioner Order Entry (CPOE) system, not e-Prescribing.
Terminology matters. For example, in the U.S. the HHS term "certification" is misleading purchasers about the quality, safety and efficacy of health IT. HIT certification as it exists today (granted via ONC-Authorized Testing and Certification Bodies) is merely a features-and-functionality "certification of presence." It is not like an Underwriter Labs (UL) safety certification of an electrical appliance that the appliance will not electrocute you.
(This is not to mention the irony that one major aspect of Medical Informatics research is to remove ambiguity from medical terminology, e.g., via the decades-old Unified Medical Language System project or UMLS. However, as I've often written, the HIT domain lacks the rigor of medical science itself.)
I note that if this were a grant proposal for studying e-Prescribing, I would return it with a low ranking and a reviewer comment that the study proposed is actually of CPOE.
That said, looking at the nature of this study:
The conclusion of this paper was as follows. I am omitting some of the actual numbers such as confidence intervals for clarity; see the full article available freely at above link for that data:
Use of an e-prescribing system was associated with a statistically significant reduction in error rates in all three intervention wards. The use of the system resulted in a decline in errors at Hospital A from 6.25 per admission to 2.12 and at Hospital B from 3.62 to 1.46. This decrease was driven by a large reduction in unclear, illegal, and incomplete orders. The Hospital A control wards experienced no significant change. There was limited change in clinical error rates, but serious errors decreased by 44% across the intervention wards compared to the control wards.
Both hospitals experienced system-related errors (0.73 and 0.51 per admission), which accounted for 35% of postsystem errors in the intervention wards; each system was associated with different types of system-related errors.
I note that "system related errors" were defined as errors "where system functionality or design contributed to the error." In other words, these were unintended adverse events as a result of the technology itself.
The authors conclude:
Implementation of these commercial e-prescribing systems resulted in statistically significant reductions in prescribing error rates. Reductions in clinical errors were limited in the absence of substantial decision support, but a statistically significant decline in serious errors was observed.
The authors do acknowledge some limitations of their (CPOE) study:
Limitations included a lack of control wards at Hospital B and an inability to randomize wards to the intervention.
Thus, this was mainly a pre-post observational study, certainly not a randomized controlled clinical trial.
Not apparently accounted for, either, were potential confounding variables related to the CPOE implementation process (as in this comment thread).
In that thread I wrote to a commenter [a heckler, actually, apparently an employee of HIT company Meditech] with a stated absolute faith in pre-post studies that:
... A common scenario in HIT implementation is to first do a process improvement analysis to improve processes prior to IT implementation, on the simple calculus that "bad processes will only run faster under automation." There are many other changes that occur pre- and during implementation, such as training, raising the awareness of medical errors, hiring of new support staff, etc.
There can easily be scenarios (I've seen them) where poorly done HIT's distracting effects on clinicians is moderated to some extent by process and other improvements. Such factors need to be analyzed quite carefully, datasets and endpoints developed, and data carefully collected; the study design and preparation needs to occur before the study even begins. Larger sample sizes will not eliminate the possible confounding effects of these factors and many more not listed here.
The belief that simple A/B pre-post test that look at error rate comparisons are adequate is seductive, but it is wrong.
Stated simply, in pre-post trials the results may be affected by changes that occur other than the intervention. HIT implementation does not involve just putting computers on desks, as I point out above.
In other words, the study was essentially anecdotal.
The lack of RCT's in health IT are, in general, one violation of traditional medical research methodologies for studying medical devices. That issue is not limited to this article, of course.
Next, on ethics:
CPOE has already been demonstrated in situ to create all sorts of new potential complications, such in at Koppel et al.'s "Role of Computerized Physician Order Entry Systems in Facilitating Medication Errors", JAMA. 2005;293(10):1197-1203. doi: 10.1001/jama.293.10.1197 that concluded:
In this study, we found that a leading CPOE system often facilitated medication error risks, with many reported to occur frequently. As CPOE systems are implemented, clinicians and hospitals must attend to errors that these systems cause in addition to errors that they prevent.
CPOE technology, at best, should be considered experimental in 2012.
In regards to e-Prescribing proper, there's this: Errors Occur in 12% of Electronic Drug Prescriptions, Matching Handwritten and this: Upgrading e-prescribing system can bump up error risk to consider; in other words, the literature is conflicting, confirming the technology remains experimental.
This current study confirmed some (CPOE) errors that would not have occurred with paper did occur with cybernetics, amounting to "35% of postsystem errors in the intervention wards."
In other words, patient Jones was now subjected to a cybernetic error that would not have occurred with paper, in the hopes that patients Smith and Silverstein would be spared errors that might have occurred without cybernetic aid.
Even though the authors observe that "human research ethics approval was received from both hospitals and the University of Sydney", since patient Jones did not provide informed consent to the experimentation with what really are experimental medical devices as I've written often on this blog [see note 1], I'm not certain the full set of ethical issues have been well-addressed. It's not limited to this occasion, however. This phenomenon represents a pervasive, continual world-wide oversight with regard to clinical IT.
Furthermore, and finally: of considerable concern is another common limitation of all health IT studies, which I believe is often willful.
What really should be studied before justifications are given to spend tens of millions of dollars/Euros/whatever on CPOE or other clinical IT is this:
The impact of possible non-cybernetic interventions (e.g., additional humans and processes) to improve "medication ordering" (either CPOE, or ePrescribing) that might be FAR LESS EXPENSIVE, and that might have far less IT-caused unintended adverse consequences, than cybernetic "solutions."
Instead, pre-post studies are used to justify expenditures of millions (locally) and tens or hundreds of billions (nationally), with results sometimes like this affecting an entire country.
There is something very wrong with this, both scientifically and ethically.
 If these devices are not experimental, why are so many studying them to see if they actually work, to see if they pose unknown dangers, and to try to understand the conflicting results in the literature? More at this query link: http://hcrenewal.blogspot.com/search/label/Healthcare%20IT%20experiment
Addendum Feb. 10, 2012:
An anonymous commenter points out an interesting issue. They wrote:
The study was flawed due to its failure to consider delays in care and medication administration as an error caused by these experimental devices.
Delays are widespread with CPOE devices. One emergency room resorted to paper file cards and vacuum tubes to communicate urgency with the pharmacy. Delays were for hours.
I agree that lack of consideration of a temporal component, i.e., delays due to technology issues, is potentially significant.
I, for example, remember a more than five-minute delay in getting sublingual nitroglycerin to a relative with apparent chest pain due to IT-related causes. The problem turned out to be gastrointestinal, not cardiac; however, in another patient, the hospital might not be so lucky.
Addendum Feb. 12, 2012:
A key issue in technology evaluation studies is to separate the effects of the technology intervention from other, potentially confounding variables which always exist in a complex sociotechnical system, especially in a domain such as medicine. This seems uncommonly done in HIT evaluation studies. Not doing so will likely inflate the apparent contribution of the technology.
A "control ward" where the same education and training, process re-engineering, procedural improvements, etc. were performed as compared to the "intervention ward" (but without actual IT use) would probably be better suited to pre-post studies such as this.
A "comparison ward" where human interventions were implemented, as opposed to cybernetic, would be a mechanism to determine how efficacious and cost-effective the IT was compared to less expensive non-cybernetic alternatives.