Let’s look at two very different positions from a single article on Saturday (10/11/14) morning. It’s just too bad that this blog has to be in the wake of personal tragedies around ebola.
From a provider and electronic health record (EHR)/health information technology (HIT) user/thought leader:
“Over the years, we’ve seen problems with overhype and overenthusiasm of these systems, leading to design and implementation failures, and a total lack of regulation,” said Dr. Scott Silverstein of Drexel University. He is a physician who has helped build clinical record systems and advocates improving the systems. – From The Dallas News, Online
And now Epic, a multi-billion dollar EHR vendor:
The company that built the system for Presbyterian is Epic, a Wisconsin-based software technology giant that is estimated to control 20 percent of the U.S. market in electronic hospital records. It declined to discuss the company’s work but said in a brief statement that “there was no flaw” in its technology at the hospital. – From the same Dallas News article [Bold is my emphasis.]
The first statement is one that is undeniably true, and any frontline healthcare provider or hospital/office IT implementation staff will enthusiastically affirm the truth of “overhype and overenthusiam,” (EHR companies sold the Feds a “bait and switch” when they positioned their systems as the panacea of safety and quality) and yet the systems (all of them) are rife with design issues that create frontline clinical workflow disruption, inhibit the flow of information due to poorly designed/executed or non-existent interfaces, and implementation failures due to a lack of understanding of the organizational (and, in fact, unit-level) culture, workflow and other processes, and on-the-ground technology. The simple act of forcing folks out of established clinical workflows creates risk, and all one has to do to find a safety risk in the post-EHR implementation organization is look for where the frontline has found or is developing (forced) work-arounds. Workarounds reflect a significant disconnect between the planned and actual use of the system in a specific environment or context.
Broken “One-size-fits-all” Approaches EHR Implementation
Even the basic aspect of EHR implementation is wrongheaded, not taking into account unit-by-unit cultural and job-specific workflow differences, differing unit-level readiness to adopt technology, and differing unit-specific frontline clinical goals and needs. What seems incomprehensible in the implementations is that the vendors almost never have anyone as part of the on-the-ground implementation team that has a frontline clinical background — someone who can actually listen, contextualize, and advocate for the the frontline user’s concerns to the vendor, and, God forbid, actually proactively “fix” broken workflow challenges or operational safety risks upfront before go live. This was true in every EHR implementation I was part of as a physician, and remains true in organizations I consult with who are in the throes of implementation. Worse, is that feedback mechanisms for the frontline folks to voice ongoing or evolving safety concerns, whether regarding usability or usefulness, with the system’s functionality or output are remarkably ill-defined, poorly functioning, or, in many cases, nonexistent.
Epic is known for its “big bang” approach to implementation – a one-size-fits-all-implementation-plan — an approach that was demonstrated years ago to be a colossally bad idea by Rob Kling, a Father of Social Informatics, and senior author of Understanding and Communicating SOCIAL INFORMATICS: A Framework for Studying and Teaching the Human Contexts of Information and Communication Technologies. As Kling and co-authors note in Chapter One, “This body of knowledge comes from more than thirty years of systematic, empirically anchored investigation, extensive analysis, and careful theorizing.” Yet this body of work seems largely unknown to (or ignored by) EHR vendors. More specifically, the current state of affairs in HIT was predicted by Kling, and available to anyone interested in avoiding known pitfalls in social informatics — the very space in which HIT/EHRs function:
“The unambiguous conclusion of the highest quality empirical Social Informatics research is that technology-centered organizational interventions often fail. When they fail, it is rarely a technological failure (though that can happen, such as when prototype systems don’t scale up well). Some failures may be in project management. More often, failures seem to be sociotechnical – workplace requirements are poorly understood by information systems designers, information systems are not well integrated into preexisting workflows, information systems are underused because they don’t resolve the issues of professionals who are supposed to use them (perhaps they were “best practice” for a different kind of organization) or system use conflicts with organizational incentive systems (a major issue with knowledge management, but even with older concepts, such as expert configuration systems for large computer system sales support). Well-integrated sociotechnical interventions seem to be most workable, though even they are not foolproof.” [Bold and underline is my emphasis.]
While EHR challenges (and failures) are clearly sociotechnical in nature, (culture, process and technology) – there are major issues with interoperability between systems, as well as more proximate and critical issues with getting systems within one organization to talk to each other through functioning, reliable interfaces, including providing semantic interoperability so there is transfer of consistent and reliable meaning. In other words, there are technological failures involved, despite Epic’s assertion that “there are no flaws.” Many, if not most of these failures, as Kling notes, could have been mitigated by a thoughtful sociotechnical development approach. And frankly, gleaning such information from the frontline is not difficult. In my consulting career, I have never been within a hospital where it takes longer than a few minutes of questioning to identify major frontline EHR/HIT functional issues — including issues that are inherently unsafe related directly to the EHR itself. And mind you, I’m not asserting that this stuff is easy… EHRs are incredibly complex systems working within complex cultures and processes — but don’t try to tell me, or the public, “there are no flaws.” And for Heaven’s sake, can we please stop throwing technology on top of broken technology, back up, and remember we need to have culture and processes ready to meet the technology, too — remembering what Sydney Dekker tells us:
Safety in complex systems is created by people through practice — at all levels of the organization. It’s only people who can hold together the patchwork of technologies and tools and do real work in environments where multiple, irreconcilable goals compete for their attention (efficiency, safety, throughput, comfort, financial bottom line).
EHRs were positioned, by the vendors, and heartily supported by the Feds (due to highly biased, partial information provided by large, powerful lobbying groups of the vendors), as “the solution” to the patient safety challenges faced by the U.S. healthcare system. But as Kling tried to tell us, “technology-centered organizational interventions often fail,” as has EHRs in this endeavor. Why? Because healthcare is a complex, adaptive, sociotechnical industry, and the frontline caregivers have been marginalized in the rush to get HIT into place. In that rush, workflows were trampled, workarounds abound, and the “information systems are underused because they don’t resolve the issues of professionals who are supposed to use them.”
Bigger issues reside with the incredibly common fact that frontline usably has largely been an afterthought, forcing frontline caregivers out of important and effective patient care workflows developed and made efficient over time. How many times have we heard a frontline provider say regarding the EHR they are now forced to use, “What used to take me 3 minutes now takes me 20.” Or 30. Or…” Many RNs I interact with in consulting to their organization tell me that anywhere from one- to two-hours per day is “non-contact time with the computer.” In other words, time is taken away from actual patient care to deal with EHRs that have poor usability, which to date, seems endemic to all EHRs systems. I can’t imagine anyone making the assertion that this is a good thing, or have the hubris to claim, “But the care is safer.” With frontline clinical folks having so much on their plate, EHRs need to “be the plate,” not just another item added that they can’t take a bite of, much less eat…
The tragic situation of the ebola patient at Presbyterian Hospital Dallas highlights many of these issues, and there is plenty of room for detailed sociotechnical analysis – culture (were the clinicians communicating clearly? staffing issues; burnout, and many others), process (standardized verbal communication practices and handoffs; workflow; etc.), and technology (Epic functionality; interfaces; alerts; etc.). But, of course, in this situation, Epic had nothing to do with it…”there were no flaws.” It’s a position that is laughable, untrue, and frankly offensive to the frontline providers. I wonder if Epic would allow a third party IT consultancy into Presbyterian to confirm the “flawless performance” of their system? (Will you, Epic?) If the government wants more control over healthcare quality, I have a suggestion: monitor the vendors, EHR functionality, interfaces, usability, usefulness (all determined by frontline feedback and all harm measures) and ROI and ROS (see below).
But Thank Goodness! EHRs make us more productive, right? NOT!
Not only have EHRs failed to produce large-scale, sustainable safety improvements, they have failed to enhance productivity — in fact, in many (most?) cases, EHRs have negatively affected productivity. Economists have named this dynamic – that is, the underlying assumption that IT improves productivity – the Productivity Paradox. This name resulted from Nobel laureate Robert Solow’s (1987), an economist, observation, “You can see the computer age everywhere but in the productivity statistics.” It’s “a sentiment that is ignored by the technophilic press,” as Kling noted. And when will we stop talking aboutthe financial return on investment (ROI) and start talking about the ROS — return on safety for HIT vendors, based on all forms of harm captured by a robust system?
“Surely EHRs have contributed to healthcare somehow?” you ask. Yes. Certainly they have pushed the conversation about universal availability of patient information to the forefront, even if not solving it. Unfortunately, perhaps the most consistent contribution that EHRs have made to healthcare is to increase costs and create a new cadre of million- and billionaires. Real advances in HIT-driven safety seem to come from small innovators capitalizing on the disarray and siloed data sets trapped in the big vendors systems, like TheraDoc, now part of Premier, that truly created a ROS from electronic information. And for goodness sake, can we quit talking about “big data,” and start talking about “useful data?” I suspect, like anything else, 80% of safety contributed by HIT will come from 20% of the data. We just need to get it right in small bits of data, instead of throwing it all on the wall to see what sticks.
So where are we with EHRs?
We aren’t delivering generally safer care. We aren’t effectively communicating with each other (in part because we’re staring at a screen, hoping what we need to know is there, and often it isn’t or we can’t find it). And we aren’t more productive at a time when staff retention is critical, yet workforce shortages and provider burnout are approaching crescendo levels, and to which the EHRs are actually contributing.
Isn’t it time for healthcare organizations to embrace a sociotechnical approach to understanding and implementing HIT, before having a EHR vendor come on the ground—with a focus on meeting the needs of the two most important constituents: patients and their frontline caregivers. Isn’t it time to demand EHR vendors listen to the client’s needs and adapt to them, as opposed to foisting a one-size-fits-all approach onto the organization? (In other words, let’s learn from that which is known – Kling and colleagues!) And isn’t it time success metrics for EHR vendors be, simply: 1. Safer care for our patients — ROS — based on an all harm measure (documented consistently from vendor to vendor); 2. Enhanced communication between providers and providers and patients, both verbally and electronically, and; 3. Improved efficiency, measured in direct time for providers to use the system (not dollars).
Any organization at any stage of EHR/HIT implementation or evaluation would be well served to slow down and consider using a thoughtful sociotechnical evaluation before continuing down the wrongheaded path (as the huge experiential and research-based body of knowledge has shown us) of a “technology-centered organizational intervention.”