On the record

Brian Dickson was installing shelves and moving boxes around his garage over President’s Day weekend in San Jose when the familiar seasick feeling washed over him. Shaky. Sweaty. Fight or flight. He dropped what he was doing and went into the house. Sat down in front of a fan, drank a glass of water and tested his blood sugar.

The glucometer confirmed his suspicion: His level had dropped below 70 milligrams per deciliter, an undesirable reading for Dickson, 52, who has type 2 diabetes. He took a glucose tablet and ate a cheddar and raspberry jam sandwich, then spent the rest of the weekend trying to get back on track.

At 1:15 p.m. on the following Tuesday, an email notification floated across Dickson’s phone as he worked from his home office.

“Hi Brian,” wrote Nancy Shin, PharmD, a clinical pharmacist at Stanford Primary Care in Santa Clara, in the message. “I got a few alerts over the weekend that your blood sugar was in the 60s. Wanted to see if you were doing OK.”

She reminded him of how to take his three medications around meals and asked if anything unusual had caused his blood sugar to drop. She offered to talk.

Dickson felt reassured. Shin was there for him. Like a coach on the field. “It was as if she said, ‘You fumbled the ball, so now you need to come up with a new play for how to prevent that from happening,’” he said.

Flowing securely into his electronic health record, Dickson’s readings had triggered a pre-programmed alert for Shin. Such remote monitoring of chronic conditions is just one way computerized systems are moving beyond replicating paper medical files to improve how health providers do their jobs.

Gone are the days of the late-night pilgrimage to track down a chart for an early morning procedure, or managing a complex case in the emergency room while the patient’s medical file languishes on an unknown cart.

Also in the past are large-scale transitions to massive documentation systems and classroom tutorials on basic keystrokes and menus. Tech-savvy doctors who stepped in to help with IT troubleshooting are now seasoned leaders, many of them board-certified, in the emerging field of clinical informatics. Each year, new trainees emerge who have never seen a paper chart.

Moving past early frustrations

In this moment — deemed “the second inning of a nine-inning game” by one medical informaticist — doctors are working through early frustrations, communicating with each other more seamlessly about individual patients, and identifying potential new uses for their digital collections of medical numbers and notes.

Health leaders are exploring remedies for procedural inefficiencies. Researchers are building records-based tools to aid clinical care at the bedside. And patients are collaborating electronically with providers about their health in real time.

At Stanford Medicine, physicians are moving beyond workflow challenges to design innovative ways for electronic health record systems to enhance patient care.

Lloyd Minor, MD, dean of the School of Medicine, believes the time is right to capitalize on the potential of digital documentation.

“We can dramatically improve the care we deliver to patients,” he said at the recent Electronic Health Records National Symposium at Stanford, “if we’re better able to derive information from the vast amounts of data that oftentimes sit trapped inside electronic health records.”

Pioneering providers began using electronic systems for medical documents as far back as the 1960s, attracted by the prospect of more efficiently storing patient information. As technology evolved, other potential benefits emerged: more sophisticated tracking of individual and population health, easier communication with patients, improved safety and cost savings.

For years, though, the cost of installing the systems, along with privacy concerns and substantial technical and logistical challenges, remained largely insurmountable.

Incentives changed on a grand scale in 2009 with the passage of the national stimulus bill. The American Recovery and Reinvestment Act provided $27 billion for an incentive program to help doctors and hospitals adopt and use electronic health records systems, and billions more was later set aside for training and assistance.

Tech-savvy doctors who stepped in to help with IT troubleshooting are now seasoned leaders, many of them board-certified, in the emerging field of clinical informatics.

By 2016, the government reported that more than 97 percent of hospitals and 70 percent of office-based physicians participating in the incentive program were using electronic health record systems. Both Stanford Health Care and Stanford Children’s Health had transitioned by the mid-2000s, adding functions over subsequent years.

Christopher Sharp, MD, chief medical information officer for Stanford Health Care, said a crucial part of the process was convincing doctors to embrace the far-reaching change. “We called it driving adoption,” he told the audience at the EHR symposium.

Though few wish to return to paper files, the medical world’s transition to electronic records generally has been rocky. Some of the struggle comes simply from learning to navigate a complicated new software system and related programs. But other challenges arise from increased — and changing — documentation requirements from payers and the government, along with decreased opportunities to delegate.

Adding to the frustration, doctors report spending more time navigating digital files than interacting directly with their patients. In a Stanford Medicine survey, conducted with The Harris Poll in March 2018, primary care physicians across the nation said that of 31 minutes they devoted to a patient on average, they spent 19 minutes — or 61 percent of the overall time — interacting with electronic health records during or outside a visit.

In a separate needs assessment conducted in late 2015, more than a year after Stanford Children’s Health installed a new system that brought ambulatory care clinics on board, physicians there said they spent five hours a week, on average, using electronic records systems after clinic hours.

The results were eye-opening. “They were trying to be as efficient as they could, and still it wasn’t good enough,” said Tzielan Lee, MD, the associate chief medical information officer for Stanford Children’s Health and a pediatric rheumatologist and clinical associate professor.

A team led by Lindsay Stevens, MD, used the assessment, along with user data from the system and direct observations, to create personalized learning plans. They rolled out the plans in one-on-one sessions over nine months in 2016, naming the training Home 4 Dinner. Stanford Health Care has a similar program called Epic Concierge.

Tandy Aye, MD, was one of more than 370 health care providers to complete the Home 4 Dinner training. In years of practicing medicine, Aye, a pediatric endocrinologist and associate professor, was accustomed to structuring her notes a certain way. She found the pre-formatted templates in the new electronic records system frustrating and was pleased to learn how to create customized templates for her notes — for example, having a different plan and counseling advice for an 11-year-old versus a 16-year-old.

“Home 4 Dinner made it a lot more efficient,” Aye said.

Others who completed the program reported similar results. A follow-up survey showed a significant improvement in providers’ comfort with the system’s tools and in their feelings about control of their workload, said Stevens, a clinical assistant professor and medical director for training and communications at Stanford Children’s Health. “One of the biggest things that this program did was make people feel heard,” she said.

Now that most providers have completed the training, Stevens has turned her attention to developing a version for new hires.

Still, Aye and others are not completely pleased with electronic records. For example, the system automatically pulls in pronouns — or even patient names — that are incorrect for young transgender patients she treats. Aye must override the automated information to correct the errors, which takes time.

Lee said requests for such sophisticated changes in software are more common now that clinicians are more adept with the system.

“I definitely think there is a learning curve,” Lee said. “Now, instead of questions like ‘Where is that button?’ we hear ‘How can I use the system to improve my whole job?’”

Improving care through innovation

Teams of doctors and nurses at Stanford Health Care and Stanford Children’s Health devote hours outside their clinics to answer precisely that question. They work with Stanford IT staff, as well as vendor representatives dedicated to supporting Stanford, to design and implement solutions for requested improvements, such as displaying upcoming operating room staffing assignments on mobile phones or integrating drug infusion pumps with the electronic health record system so dosages in patient files are automatically updated.

“We have all these ways of continuously checking how the system is working. Where are the issues coming up? What can we do to make the system better?” said Natalie Pageler, MD, chief medical information officer for Stanford Children’s Health.

Some researchers are pursuing more far-reaching advancements, such as using speech recognition and machine learning technology to automatically generate a progress note, or designing a way for physicians with difficult cases to query electronic health records data for information on the care of similar patients.

Each year, two new scholars take on a challenge related to electronic records, or health information systems generally, through Stanford’s two-year clinical informatics fellowship program, which in 2014 became the first in the country to be accredited by the Accreditation Council for Graduate Medical Education.

Jonathan Palma, MD, the program director, sees innovation in clinical informatics as a way to extend the functionality of electronic records. “I think of it in three buckets: to improve the patient experience, to improve provider workflow or satisfaction, and then the holy grail, which is to improve care with the data you’re generating and analyzing,” he said.

That kind of breakthrough is the goal for an EHR tool developed at Stanford for the neonatal intensive care unit. Premie BiliRecs helps clinicians decide when preterm babies should receive phototherapy for jaundice. The calculus about whether to take the measure is not easy.

The yellow tint of a jaundiced baby’s skin and eyes is the result of a buildup of bilirubin, a substance formed when red blood cells break down. Too much bilirubin in the blood, a condition called hyperbilirubinemia, can cause brain damage when the substance circulates in a baby’s tissue. Light absorbed through the baby’s skin helps eliminate excess bilirubin. Phototherapy is a well-proven treatment, but too much too early carries its own risks, particularly for preterm babies.

“In the well-baby nursery, we rely on bilirubin levels as a function of postnatal age to identify those that needed treatment and closer scrutiny,” said Vinod Bhutani, MD, a Stanford neonatology professor. “But a number of factors make premature babies more vulnerable. A bilirubin value that is not toxic to a term baby becomes extremely toxic to a baby at 28 weeks.”

“I think of it in three buckets: to improve the patient experience, to improve provider workflow or satisfaction, and then the holy grail, which is to improve care with the data you’re generating and analyzing.”

Jonathan Palma, MD

A tool had already been created to guide clinicians through treatment for babies born after 35 or more weeks of gestation, an effort led by Christopher Longhurst, MD, a pediatrician and former chief medical information officer at Stanford Children’s Health. Bhutani, who is renowned for his research and clinical expertise in treating hyperbilirubinemia in babies, wanted a similar application for infants who arrive early.

Premie BiliRecs would be more complicated because data are limited by preterm infant fragility and the wide variability of their health concerns. Bhutani, Palma and Yassar Arain, MD, who was then a neonatology scholar at Stanford, used linear algebra to flesh out recommendations Bhutani had co-authored, matching a baby’s age — adding gestational age and days since birth — with a threshold bilirubin level for beginning phototherapy or, in serious cases, exchange transfusion, in which an infant’s damaged blood is replaced. They then worked with Joshua Faulkenberry, a senior web applications engineer with Stanford Children’s Health, to build the tool.

The recommendation and related data from Premie BiliRecs deploy automatically in the electronic record of preterm babies. Previously, nurse practitioners and residents turned to a table in the 450-page neonatal intensive care unit guide with less precise values. Arain used a screenshot of the page bookmarked to his phone.

“Clinicians are making tens if not hundreds of decisions a day, and that’s just one less decision,” Arain said. “With the click of a button, they can move on with their day.”

As Bhutani sees it, the benefits of Premie BiliRecs are twofold. First, because it was designed to collect data, it can eventually be used to inform evidence-based standards after gathering sufficient information from enough users. Secondly, with a more detailed source at their fingertips, clinicians are more likely to follow the established guidelines with ease and trust.

Early results in a study by Palma and Arain have borne this out. Previously, doctors and nurses in the NICU were more cautious about treatment, initiating phototherapy earlier than the guidelines in more than 50 percent of cases. With Premie BiliRecs, that number dropped to under 30 percent with no adverse effects.

“I was always concerned whether people would conform to recommendations from a ‘black box’ because we all trained as capable individuals and are keen to rely on our individual intellect,” Bhutani said. “We respect that individual judgment, and you don’t want to supplant their decision-making process.”

Lance Downing, MD, an assistant professor in biomedical informatics research, has looked at clinicians’ reactions to some EHR-based decision support tools and witnessed pushback against what has been termed “cookbook medicine.”

“The flip side,” he said, “is that it has been shown to really improve care on the whole.”

Engaging patients more

But digital reminders are not confined to physicians in the new era of electronic health data. Through Bluetooth-compatible devices and special apps, patients’ vital measurements can transfer directly into their medical files — available for view and comment from their health care providers in near real time.

Shin, the clinical pharmacist at Stanford’s primary care clinic in Santa Clara, manages patients’ diabetes and hypertension drug therapies, and is involved in a clinical trial that examines how patients’ hypertension management is affected when they wear a blood pressure cuff that automatically transmits data into their electronic record.

Regular reports on blood sugar readings for diabetic patients, like Dickson, and on blood pressure for hypertensive patients have been helpful, she said.

“This is information that we want to know anyway,” Shin said. “Chronic diseases occur outside the clinic.”

Downing, the principal investigator for the virtual hypertension management study, said the hope is to use devices in a way that’s more effective for patients and more efficient for them and providers. “Those are two important themes,” he said, “using electronic health records and IT infrastructure to connect data generated by patients, and moving away from a hospital-centric model into these other models of remote monitoring and care.”

One concern is ensuring that providers have sufficient time to track the data flowing in from the field. For Shin, scanning a handful of reports a day hasn’t been burdensome. What’s more, she’s found that periodic review of health data on chronically ill patients saves time during clinic visits — when she’d normally be pulling the readings — and it allows for a quicker assessment of the effects of medication changes and other adjustments.

The increased interaction outside of the clinic also fosters trust with patients. For some, like Dickson, the system flags readings that are particularly high or low, and Shin responds by reaching out with emails for routine situations or check-in phone calls for more serious concerns.

Dickson finds the communication reassuring: “It makes me a little bit more aware,” he said. Knowing that Shin sees his blood sugar measurements between visits also makes him more accountable. And he’s had fewer readings outside the acceptable range.

“It’s improved my health,” said Dickson, a network and software engineer. “I’m aware of what’s going on and I can do a better job of planning my lifestyle, in terms of food and exercise and everything else.”