An Informaticist takes stock of electronic health records in EDs.
Part 1 of 2
I saw a Facebook post the other day – an EM doc lamented, “we go live on our new EHR next week – so there are four retirement dinners this week.”
My heart sank. It’s 2019, and Emergency Medicine careers are still being cut short because of our EHRs. Of course, this isn’t anything new, but it’s discouraging to still hear it happening.
It seems like a good time to reflect. It’s been 10 years since the HITECH Act (and its Meaningful Use incentive program) started. Ten years since electronic health records in EDs went from an eclectic mix (if present at all), to widespread adoption of a few enterprise vendors like Epic and Cerner.
Usability and Burnout
There was sort of a chicken-and-egg theory 10 years ago: “Are customers reluctant to invest in EHR systems, because they’re not easy to use? Or do EHRs have usability problems, because they don’t yet have enough paying customers for vendors to invest in usability?” The hope was that once everyone was using EHR, the influx of cash and interest would drive improvements in usability. Incentives to adopt (and penalties, for holding out) would get hospitals over the hump, and we’d soon have systems as quick and intuitive to use as our laptops or phones.
That really hasn’t happened. Meaningful Use did spur widespread adoption – hospitals with a fully functional, enterprise-wide EHR went from under 10% in 2009, to above 90% today. A similar transformation occurred in ambulatory settings.
The MU incentive program didn’t just want to give hospitals cash in exchange for installing an EHR – hospitals had to demonstrate ‘meaningful use’ of the EHR. To gain a few percentage points of extra reimbursement on CMS patients, the hospital’s EHR had to support problem lists, medication lists, drug interaction warnings and other basic features we take for granted now. Quality measures had to be reported, via EHR. Computerized order-entry had to occur on a small fraction of hospitalized patients, and a fraction of patients had to have been logged into the patient portal.
Program implementation occurred in stages – a hospital that missed stage one could still catch up and gain incentives for meeting stage two requirements. But the bar was set higher for stage two – more EHR features would be required, and higher participation rates for order entry and portal logins would be mandated. And failing to participate didn’t just mean lost incentives – penalties were imposed as well.
From the perspective of increased EHR adoption in the US, the MU program worked. Yet, at a hearing looking at the MU program aftermath, Sen. Lamar Alexander remarked, “The hope was that the program would improve care, coordination, and reduce costs. The evidence suggests these goals have not been reached… A survey found nearly 70 percent of physicians say their electronic health records systems have not been worth it.”
Atul Gawande summed up his EHR experience in the New Yorker, writing, “I’ve come to feel that a system that promised to increase my mastery over my work has, instead, increased my work’s mastery over me.”
We hear more these days about physician burnout, and the potential responsibility of EHRs. AMA President Barbara McAneny recently stated, “Doctors are spending excessive time on data entry, contributing to physician burnout, with implications for quality of care.”
Emergency Physician Ted Melnick led a recent study to determine the relationship between EHR usability and burnout. His paper showed “EHR usability scores were strongly and independently associated with physician burnout in a dose-response relationship…The usability of current EHRs as assessed by US physicians using a standardized metric of technology usability is markedly lower than for most other technologies, and falls into the grade category of F.”
Many of the frustrations from using EHR 10 years ago remain today. Some aspects have improved, but not as much as one would hope. There’s a lot of blame to go around – and a lot of excuses. Maybe the problem is that doctors and nurses, while the prime “users” of EHRs, aren’t really the “customers” – that would be the hospital system leadership, who care more about charge capture and pop-up compliance alerts, than user experience and provider satisfaction. Maybe the problem is CMS and the federal government, with their 90s era charting requirements, as well as their shifting goals for measuring quality and performance? Maybe the Meaningful Use requirements themselves were wrongheaded – which never explicitly prioritized interoperability or usability?
Essentially, the problem is the modern enterprise EHR system is still, at its core, a billing and coding engine. Over the years, additional features were bolted on, like limbs onto Frankenstein’s monster: results review, order entry, note-writing, messaging, quality dashboards, a patient portal. These areas of the EHR all have different user interfaces and moving between them in the course of daily work is jarring for clinicians. The problem is compounded as the user base has expanded beyond doctors and nurses to include social work, respiratory therapists, administrators and scribes. A hospital running an enterprise EHR is like an “enterprise restaurant” where all the staff used just one huge app for booking reservations, taking menu orders, cooking food and washing dishes.
EHRs are capable, but extremely complex. Meanwhile, the patients are getting older and sicker, ED volume keeps ticking up even as more EDs are closing, and new regulations and mandates put more stress on existing processes. The ED doc is in the middle – forced, each shift, to review charts, take action, and document care decisions on dozens of new patients.
Physicians report that, for each hour of face-to-face contact with patients, there’s one- to two- hours spent on the EHR, or other clerical tasks. They’re staying later after work to finish charts than ever before, even with the rise of scribes and advanced dictation. Discussion of burnout and the burden of EHR is rampant, even among policymakers. And when new systems are chosen for implementation, too many EPs opt for retirement.
What went right, what went wrong?
When I started as an ED attending, we were on a “best of breed” EHR — it was pretty fast, tailored to our particular ED workflows, but also somewhat isolated. We couldn’t see the clinic’s notes, when they’d send a patient to the ED. Inpatient docs couldn’t see our notes, when we admitted a patient. We couldn’t see a patient’s hospital discharge summaries, unless we took the effort to log into another system. So ED patients truly merited the textbook EM approach – thorough histories and exams, broad differentials.
A switch to an enterprise EHR vendor was thrust upon us, and the rest of the medical center, to take advantage of the Meaningful Use incentives. We lost some of the intuitive, EM-centric design of our old system – it’s safe to say that Epic and Cerner don’t make sales on the strength of their ED modules. But when we switched, we did gain new, quick access to patient information from clinic notes, inpatient discharge summaries, and a unified view of old labs, EKGs and radiology reports. Also, around this time, we got single-sign on access to our city’s burgeoning Health Information Exchange (HIE) which opened up a lot of records and results. Plus, we had a relatively big group of IT analysts who helped us customize our EHR’s ED module as best we could, to streamline our workflow and make it feel similar to our old system.
As far as these transitions go, ours went well. Certainly there were frustrations, a temporary dip in speed and productivity, and a few early retirements. But charge capture went up, and with time we adjusted well enough to the new way of doing things. Also, while I grasped that it would be helpful to have clinic notes and old DC summaries, I didn’t realize how often it would help. Almost every shift, when a patient begins to talk about how their symptoms remind them of a prior hospitalization, I can fill in the details from my “chart biopsy” and potentially avoid a redundant workup or medication trial. Sometimes it feels like cheating.
Similarly, we probably see as many drug seekers today as we used to — but they don’t get as far. Between detailed visit summaries from our clinics and urgent cares and associated EDs, and the EHR’s network, and our local HIE, and our state’s PDMP, dealing with drug-seekers doesn’t feel like as much of a time-suck or soul-suck anymore.
I once worked in an ED where, right after the patient was discharged, the chart went into a huge pile in a back room… and by evening, all the charts were carted to another part of the hospital for billing. If the patient happened to come back — for a prescription problem or question or anything — if the original EP was gone, the patient would realistically have to start over, or wait an unreasonable amount of time for a clerk to bring back the paper chart. It was ridiculous and embarrassing to have to practice like that, and we probably don’t give EHRs enough credit for obviating these kinds of scenarios. I actually wonder how our specialty would have evolved, with the kind of access to past visits that we have today.
Generously, you could say another thing that’s improved is standardization of practice. For well-defined clinical entities, we’re often using the same order sets, and patients are getting the same work ups, with uniform meds and doses. Every stroke patient is getting a swallow study, because it’s a pre-checked box in the stroke order set. The sepsis patients are getting their early antibiotics, and second lactates. Hospital administrators seem happy with compliance audits. And DKA, post-intubation sedation and other scenarios involving drips are less prone to improvisation; getting the right doses and titration parameters is faster an easier than before.
Even an EHR-skeptic like Rick Bukata has said, in these pages, “What some call ‘the art of medicine’ I call an unacceptable level of physician practice variability. Electronic medical records now allow us to compare apples to apples, and begin bringing over- and under-utilizers in line with the evidence.”
I happen to agree. And yet, it’s hard to know if patients are really doing better, or whether they’re even safer than they used to be. If you were expecting clear, affirmative answers to the impact of EHRs on patient outcomes and safety, you’d be disappointed. This is, in part, because determining outcomes and errors from the paper-chart era isn’t easy. But there’s also no question that, even as EHRs have made some physician errors far less likely, they’ve introduced the potential for new kinds of mistakes.
Errors and Outcomes
Bob Wachter’s book, The Digital Doctor, is chock-full of examples for data display and order-entry screens that present information to doctors in unfamiliar, unintuitive ways. I still long for the familiar layout of electrolytes and CBC I learned in med school. Computerized order entry still makes it too easy to order the code dose of epi, in non-code situations.
Even if a more intuitive EHR redesign were possible, upgrading would require expensive, disruptive retraining. When an error attributable to the EHR does happen, it’s addressed locally, behind closed doors. There’s a Root Cause Analysis, and a Corrective Action Plan. Often, that plan involves more pop-ups and alerts, which seem reasonable to administrators, but just contribute to alert fatigue. These alerts shift the risk and burden of bad design further toward the doctor. Even if a Corrective Action Plan involves a clever EHR fix that reduces the chance for future errors, the idea is unlikely to be shared with other institutions.
Over the years I’ve heard proposals and recommendations for confidential reporting systems, where doctors could annotate a screenshot of, say, a misleading data display, or a confusing order, and submit it to some specialty body or federal agency for review. Vendors could be evaluated on their responsiveness to safety concerns. But despite these reasonable, feasible ideas, nothing’s changed. Hospital safety committees are usually stuck reinventing the wheel, and tackle concerns case-by-case, site-by-site.
The biggest meta-analyses of EHR implementations thus far have shown modest improvements in guideline adherence and fewer adverse drug-events. There’s been no detectable impact on mortality. Some people hailed these studies as a validation of the huge cost and effort to introduce EHRs.
Others said, “That’s it?” After all the hype from EHR vendors, all the incentives and expense, and all our personal frustrations with these systems, it’s reasonable to expect more from Electronic Health Records.
In Part II I’ll talk about some lessons from specific shortcomings of EHR, whether more technology can help, and what will be ultimately be necessary to improve the EHR experience for doctors.