If we broke off the rear-view mirror, took a meaningful look at what data and functions physicians really need from an EMR to provide actionable patient information for swift, comprehensive, interactive diagnosis and treatment monitoring—what kind of system would we create? For starters, “the EHR needs to be a robot that understands the context of the encounter,” recommends W. Ed Hammond, MD, professor emeritus of biomedical engineering at Duke University and an original developer of the school’s “The Medical Record” that dates to 1970. We’d likely add some artificial intelligence, standardized ontology, decision-support tools and make it collective so multiple caregivers can associate notes and orders with specific problems, too.

With more than $19 billon in federal money now set aside for health information technology, the Congressional Budget Office (CBO) estimates that 90 percent of physicians and 70 percent of hospitals nationwide will have electronic medical records (EMRs), computerized physician order entry (CPOE), clinical decision support (CDS) and other clinical IT by 2019—about 25 percentage points higher than it would be without the stimulus.

Even without meaningful healthcare payment reform, the investment will save taxpayers a net $12 billion in Medicare, Medicaid and government employee health expenditures over the next decade, CBO says, by preventing medical errors, reducing inappropriate and duplicative procedures and by cutting administrative costs.

In reality, the savings could be much more. The adoption rates reflect those who would qualify for Medicare or Medicaid bonus payments or avoid penalties under the stimulus law, and does not account for future advances in technology. Late last year, the CBO estimated that by 2019, 40 percent of physicians would have EMRs or EHRs that meet interoperability standards for that year, when criteria likely will be far more stringent than they are now.

“At the moment, EMRs are not very smart,” says Steve Margolis, MD, CMIO at 1,780-bed Orlando Health in central Florida.

He says many EMRs replicate paper charts with electronic “tabs” for orders, results, nursing notes and ancillary notes. “I personally think that’s a flawed metaphor for workflow in the electronic world,” Margolis says.

He notes that informatics professionals often equate the physician view of an EMR with the instrument panel in a fighter jet, but, unlike a doctor, a flier never has to look down to see the display. “I have to go from tab, to tab, to tab,” Margolis says. “Imagine if the fighter pilot had to go to a different tab while trying to shoot down another aircraft.”

Hammond has often seen physicians at Duke flip through the electronic chart, then copy down information on paper because the record simply duplicates paper processes. “I don’t think we’ve ever really challenged what an electronic health record does,” he says.

“The EHR should be an interactive exchange with the doctor,” Hammond says. “It shouldn’t be rote. It shouldn’t be [trying to recall] everything you learned in med school.”

“What we need to do is say, ‘I need a new tool.’ Rethink from the very beginning what the system needs to show me when the patient walks through the door,” according to Hammond.

In the future, Margolis envisions an EMR that simplifies workflow by pulling all pertinent information a physician needs to document against into a single view, such as a problem list, medication list and only the test results that are directly relevant to each clinician, so the user doesn’t have to gather all this information that is already in the system. “When it comes to writing a discharge note, he can see seamlessly what the patient was on at home,” Margolis says.

He says present-day clinical decision support mostly is very “rigid” if-then logic. He believes that future iterations will have more intelligence behind it and suggest possible diagnoses and treatments.

John Halamka, MD, CIO at CareGroup Healthcare System in Boston, says that CDS should work with laboratory, medication and imaging ordering, and alerts should incorporate “event-driven medicine,” which Halamka defines as “the transformation of data into information, knowledge and wisdom based on decision support, business rules and timely notification of clinicians.”

“The problem today is that we have too much data,” Halamka says. “We need to find a way to distill it to what the doctor needs to know, and provide actionable information.”

The EMR also should have an easy-to-read clinical summary of active problems, medications, doctor visits and test results, in a format that’s readily exportable to a personal health record. The lists should follow a standardized ontology such as Snomed-CT and should be collective so that multiple caregivers could associate notes and orders with specific problems, Halamka says.

Before the patient even arrives, the EMR should have the intelligence to present a snapshot of what appears to be wrong, according to Hammond.

Margolis would like useful auxiliary information on the same screen as the patient dashboard. “I’m looking for the latest information from journals to be presented to me peripherally in my vision.” It also might be helpful when making a diagnosis to know, for example, that the patient recently visited Africa or is of a specific genotype. The latter, of course, will become more prominent as the field of genomics grows and works its way into mainstream medicine.

Margolis compares this kind of support to GPS. “If you go off course, it gives you corrective information,” he says. If the car is low on gas, where’s the nearest station? If it’s lunchtime, where are some convenient places to eat? “Based on my current patient’s information, what’s the likelihood of various situations?”

While there are some data aggregators out there already—Microsoft Amalga is a well-known one—they tend to be limited to presenting information and don’t have inputs for order entry or documentation. Optimistically, Margolis believes the more advanced technology is two years away, but realistically it will take five to 10 years for it to be in widespread use.

Inside certification

The Certification Commission for Healthcare Information Technology (CCHIT), a federally sanctioned private organization, has toughened its criteria each year since beginning certification of ambulatory EHR software in 2006, inpatient systems in 2007 and, in 2008, for products related to health information exchange. In late March, CCHIT announced that it would start certification for CDS, interoperability, quality reporting and interoperability in 2010, a year earlier than originally planned.

“Had certification not existed, we would not be where we are now,” says David Brailer, MD, PhD, the former national coordinator for health IT in the Department of Health and Human Services (HHS). Where we are now is with health IT on the national agenda, a majority of U.S. hospitals at least with plans to install EMRs and a big pile of money available in the not-too-distant future. “CCHIT has solidified itself as a durable, valuable institution in health IT,” Brailer says.

The stimulus law calls for the Office of the National Coordinator for Health Information Technology (ONC) to have a means of certifying health IT. Whether the certification process includes CCHIT going forward or not—and that will be up to the Obama administration’s new team of HHS Secretary Katherine Sebelius and National Coordinator David Blumenthal, MD—EMRs should and will get better.

Indeed, HHS still needs to work out many details before distributing the stimulus money (see story, page 10). Although it is early in the process, H. Stephen Lieber, chief executive of the Healthcare Information and Management Systems Society (HIMSS) has been hearing that physician documentation will be required, along with CDS and CPOE, equivalent to Stage 4 on the HIMSS Analytics EMR Adoption Model (see chart, page 5), an 8-stage scale toward a paperless environment. “It’s only at this level of functionality that you can truly realize the savings,” including greater efficiency and reduced errors, Lieber says.

Less than 6 percent of all non-federal U.S. hospitals had achieved this stage by the end of 2008, according to HIMSS Analytics, a research subsidiary of HIMSS. But the fact that a few have reached the higher echelons indicates that the next generation of healthcare IT already is here in a few select institutions.

NorthShore University HealthSystem in Evanston, Ill., formerly known as Evanston Northwestern Healthcare, has three of the 15 hospitals set to reach Stage 7 at the top of the HIMSS Analytics scale for 2008; the other 12 are all Kaiser Permanente facilities in California.

Stage 7 covers more than just an EMR. NorthShore started in 1996-97 with a new lab system and a radiology information system at its flagship Evanston Hospital, then added a picture archiving and communication system (PACS) in 1999 and 2000. The decision to purchase an Epic Systems EMR came in 2001, and the technology started going live in early 2003, a phase mostly completed that year for Evanston, Glenbrook and Highland Park Hospitals.

In 2004, NorthShore began installing the EMR in the 65 offices of employed physicians. The network now includes 50 independent physicians who contribute to the same database.

CIO Thomas W. Smith continued to expand the installation, beginning with oncology in 2004 and adding labor and delivery last year. This year, the health system will add a module for operating suites as part of its ongoing quality-improvement efforts. “It’s not like we did it and stood still,” Smith says.

He is devoting much of his time now to automating Skokie Hospital in Illinois, which North Shore acquired at the beginning of 2009, and also connecting ambulatory clinics to the Surescripts (formerly SureScripts-RxHub) e-prescribing network to deliver patient-specific formularies and medication histories to doctors at the point of care. “It gives back to us every medication that was paid for by that insurance plan,” Smith explains.

“It’s an expectation for patients to be able to order the one with the lowest cost,” he notes. With real-time formulary checking, the doctor can prescribe the appropriate medication with the lowest out-of-pocket cost to the patient. There is more value to the patient here than in automating consent forms, which remain on paper for now.

“Is it worth it to spend the time to drive out every piece of paper?” Smith asks. He says it is worth it to scan many documents received from outside the organization just so they can be viewed electronically, but scans are nothing more than images, not computable data.

“There’s a general problem with EMRs in that it’s still difficult to get some information into discrete forms,” Smith says. That means data in a structured format rather than as free text. Templating can be clunky and uncomfortable for physicians used to dictating or even typing notes.

“It’s not really a shortcoming of products,” Smith says. “We still allow dictation, and people tell a story. It’s tough to take that story and put it into discrete data.”

And that comes back to workflow. “The art of medicine has to change. It’s wasteful if it doesn’t,” says Hammond.

“Healthcare in the future is not about physicians. It’s about me, the patient,” he says. It doesn’t matter if it takes the doctor an extra minute to get a piece of information if the information is pertinent to the case at hand. “The whole purpose of this is improving health and healthcare. If we’re not doing that, we’re doing things wrong.”