Do acute care Electronic Health Records always achieve their intended goals? The literature on “unintended consequences” has long warned of the potential downside of EHR implementations and the unique challenges associated with computerization in healthcare settings. Alert fatigue, for example, continues to frustrate clinicians overloaded with un-prioritized pop-ups that lead them to click past both inane and critical alerts.
The moral is that simply computerizing medical records and ordering electronically with CPOE does not lead effortlessly to the clinical, efficiency and organizational performance outcomes we seek. In fact, it often does harm. I was on the phone recently with a physician friend who was driving home after seeing a patient in a hospital that uses a competitive EHR. Abruptly he said, “I need to turn around and head back to the hospital – I apparently ordered a ten-fold over dose for a patient, and I need to fix it.” I asked the obvious: “Didn’t you get an alert?” “Probably”, he said, “but I flew past it just like the other zillion I get every hour.” Unfortunately, this is not an unusual story.
A major part of the challenge, as this story illustrates, is that not all EHRs are created equal. Digitizing paper charts and slapping alerts on top is not leveraging the true power of technology by any stretch of the imagination. If EHRs are to achieve their pivotal role in delivering better health outcomes at lower cost, they must incorporate the evolving power of advanced technology.
Order sets are a great example. When I say “order set” to a physician, and then ask them what they imagine, invariably it’s an 8 ½ x 11 inch sheet of paper with a check list. Most EHRs simply computerize the paper – there’s no technology in the background doing any thinking whatsoever.
By contrast, studies show advanced clinical decision support (CDS) goes beyond passive or task-oriented order sets by adjusting recommendations to reflect patient-specific characteristics, such as weight, age, medications or history. These “intelligent order sets” help clinicians make best decisions by designing in quality and safety, and designing out errors. Intelligent order sets are also locally programmable, providing recommendations for clinicians that reflect local processes and capacities, as well as advances in treatment options and evidence-based medicine (EBM). And they are readily updated for new regulatory and quality reporting requirements. Thus IT becomes a natural part of delivering care for every clinician.
As the following brief success stories highlight, front-line clinicians make better decisions and achieve improved documentation when using advanced CDS. Each of the organizations here implemented CPOE with advanced CDS powered by Sunrise from Allscripts. They each included disease and treatment-specific order sets designed to minimize errors and the need for most alerts. Order set development and the programmed logic were achieved through inter-disciplinary teams within which clinicians agreed on standards of care and workflows, while the organization’s leadership ensured cost-effectiveness. In each case, use of the order sets was voluntary, so we compared results from clinicians who utilized CDS to those who relied on their own abilities to order and document care.
Stroke Management at Summa Health System
Neurologists at Summa Health System’s Akron City (OH) Hospital, a 474-bed urban teaching facility, designed a CDS-driven order set for managing new-onset stroke care. The stroke order set, based on 13 elements of EBM, appeared within every clinician’s EHR workflow whenever the admitting diagnosis was stroke. When clinicians voluntarily chose to use the order set, it automatically launched CPOE orders, computed medication dosing and frequency reflecting patient characteristics, and notified all appropriate clinicians of timing for executing orders.
Six months of post-implementation data on stroke admissions verified the power of CDS with the EBM intelligent order set. Summa physicians who managed their patients using the stroke order set had 40.6 percent greater compliance to following and documenting the 13 EBM criteria.
But their success went beyond mere compliance. Patients managed on the CDS-driven order set also experienced:
- 9.4 percent more discharged to home versus skilled nursing facilities (SNF) or inpatient rehabilitation
- 35.7 percent fewer readmissions to the hospital within 31-days
- 7.5 percent shorter length-of-stay
- 11.4 percent lower direct cost-per-case, approximately $509 savings per case
- 12.7 percent lower indirect cost-per-case, approximately $209 savings per case
Shorter lengths of stay and more discharges to home represent improved outcomes for patients in terms of their preparedness clinically for discharge, as well as for quality of life. They also resulted in annual savings of approximately $253,454. Moreover, follow-up analyses for the patients discharged to SNF showed that those managed with the order set during their inpatient care also experienced:
- 7.7 percent fewer 31-day readmissions
- 4.1 percent shorter length-of-stay in the SNF
The magnitude of the improvements at Summa was such that, within 72 hours of the hospital publishing these results, 100 percent of their clinicians had adopted the EHR stroke order sets.
Blood Glucose Management at Springhill Medical Center
EBM suggests that patients heal faster and return to wellness quicker with optimal blood glucose (BG) management. Hospitalists and endocrinologists at Springhill Medical Center, a 252-bed private community medical center in Southwestern Alabama, implemented a CDS-supported BG order set reflecting updated EBM standards for post-operative cardiac cases.
Five months of post-implementation data for thoracic surgery patients were contrasted with five months pre-implementation data. Following statistical verification of comparability of cohorts, data showed:
- 18.2 percent improvement in mean BG day 1 post-op
- 25.8 percent reduction in per-patient variation in BG day 1
- 57.5 percent improvement in meeting BG goals Day 1, reaching 95.5 percent of cases reaching BG goals day 1
- 11.2 percent improvement in mean BG day 2 post-op
- 24.5 percent reduction in per-patient variation in BG day 2
- 136.2 percent improvement in meeting BG goals for length of stay
- 0 endocrine consults for cardiac surgery cases – all managed by cardiac surgeons
Blood sugar levels were better managed and patient outcomes significantly improved through the intelligent order sets leveraging advanced CDS. Moreover, these results were accomplished by cardiac surgeons without endocrine consults, reflecting substantial improvements in efficiency for Springhill.
Vaccination Compliance at Robert Wood Johnson University Hospital
Robert Wood Johnson University Hospital (RWJUH) in New Jersey, a leading academic medical center, achieved significant improvements in the safety and quality of its patient care through the appropriate implementation and use of its EHR with advanced CDS. RWJUH undertook a one-year program to improve compliance with Joint Commission Core Measures requiring that pneumonia and congestive heart patients be screened for pneumococcal and influenza vaccination upon discharge. The goal was to use the EHR and CDS clinical documentation system to automate the process and ensure nurses screened patients and ordered vaccines when appropriate.
RWJUH’s IT team embedded screening tools within the system’s nursing health history workflow, an example of advanced CDS that automatically enters vaccine orders on behalf of the patient’s physician whenever a pre-determined algorithm is triggered. Because the process is mandatory within the nurse documentation process, patients cannot be discharged without the screening. As a result, the hospital experienced an average 35 percent improvement in its Core Measure compliance rate, reaching 97.8 percent. When the approach was broadened to influenza vaccinations, RWJUH experienced a 40 percent increase in compliance with those Core Measures.
The RWJUH solution also provides physicians and nurses inside and outside the hospital with instant access to a list of the patient’s discharge medications from previous visits. The system enabled the hospital to document 100 percent of all medications and achieve 90 percent compliance with home medications matching the discharge instructions. As a result, preventable medication errors were avoided and patients received accurate information to help them manage their medications after discharge.
Collectively these successes are perhaps best recapped from one RWJUH physician’s testimonial: “You would have to pay me to go back.”
The moral is that we need to use computers for what they are good at – thinking and watching and recommending 24/7 while humans get work done in a distracting, busy, noisy world. As these examples substantiate, acute care EHRs with advanced CDS result in significant outcomes improvements. Because technology is working in the workflow and in the background, patients suffer fewer unintended outcomes, and experience better care and improved quality of life. Clinicians are protected from errors and better able to practice safer medicine with greater impact. In my experience, they’re also thankful at every turn for the help they receive delivering best and safest care every time. And healthcare organizations benefit from clinical and financial outcomes, organizational performance, and scores on reportable quality metrics. Collectively, everyone’s goals are better achieved with the assistance of advanced CDS.
So all EHRs are not created equal. Most do not leverage the true openness and power of computerization to help clinicians, organizations and patients. In fact, most EHRs are simply dumb-terminal technology with limited flexibility and adaptability.
Organizations that are not achieving similar improvements may want to re-evaluate their IT solutions. The overarching goal must remain to more effectively provide clinical decision-makers with active information beyond alerts alone, and beyond passive task list order sets. The bottom line is, you and your organization can very likely get a lot more benefit out of computerization than you already are.
By the way, unlike many such studies, the successes achieved in these examples came without long change cycles involving request and revision loops from the vendor followed by months of inching towards success. Because of the flexibility and programmability of Sunrise, these improvements were made quickly, with local project leadership and programming skills, assisted by Allscripts only as requested.
I would like to thank the following clients and colleagues for their inestimable help in gathering the data for these studies: Jeff St. Clair, CEO, Springhill Medical Center; Charles Ross MD, CMIO, and Pamela Banchy, System Director, Clinical Information Systems, Summa Health System; Robert Irwin, CIO, Robert Wood Johnson University Hospital; and Diane Gilbert-Bradley, MD, Chief Quality & Outcomes Officer, Allscripts.