Conventional wisdom holds that electronic medical records will reduce medical errors in the United States, but there’s surprisingly little evidence to back that up
When doctors found a nodule on her thyroid gland in 2001, Massachusetts resident Jerilyn Heinold was relieved that the growth was noncancerous. About three years later, her primary care physician suggested a visit to a specialist to make sure the tumor was not growing.
The endocrinologist determined that the tumor had not grown but, as a precaution, referred Heinold for an ultrasound test. The ultrasound report came back with the disturbing finding that the tumor had grown. The endocrinologist then told Heinold she needed a biopsy.
Heinold knew better. She dug into her medical records, available to her online as a patient of one of the forward-looking hospitals in the Boston area, and looked up the size of the original tumor. In fact, the tumor had not grown. The radiologists somehow had received incorrect information about the growth’s size in 2001.
Such anecdotes often help make the case that widespread use of electronic medical records would reduce the incidence of medical errors. That assertion is made every day, appearing on the Web sites of the Office of the National Coordinator for Health Information Technology and many others.
However, scientific evidence about the efficacy of EMRs in preventing errors is rather slight. In a report published last year in the Annals of Internal Medicine, researchers affiliated with the Southern California Evidence-Based Practice Center said they found few rigorous and generalizable studies of the effects of health IT.
“Although we did a comprehensive search,” the research team that Dr. Basit Chaudhry led reported, “we identified only a limited set of articles with quantitative data. In many important domains, we found few studies.”
One of the most frequently cited studies of EMRs and medical errors took place in the late 1990s at Brigham and Women’s Hospital in Boston. Researchers found a 55 percent decrease in serious medication errors when it introduced an e-prescribing system, known in the health IT world as a computerized provider order entry system. When the hospital enhanced the CPOE system with better decision support features, the reduction in serious medication errors reached 86 percent.
Medication errors top the list That reduction is significant because it shows IT can have a positive effect in preventing the largest category of medical errors. An Institute of Medicine report in 2006 states that drug-related errors harm at least 1.5 million people every year, and the extra medical costs of treating drug-related injuries that occur in hospitals conservatively amount to $3.5 billion each year.
The National Committee for Vital and Health Statistics has reported that with handwritten prescriptions, more than three-quarters of doctors’ signatures are illegible, and a quarter of prescriptions are incomplete. The Food and Drug Administration, which regulates prescription drugs, has expressed concerns about the use of completely different drugs whose manufacturers give them similar names by accident.
The proliferation of drug products — more than 11,000 of them are available — makes it difficult for doctors to stay abreast of developments in pharmacology. “Physicians absolutely cannot keep in their heads all the current information,” Heinold said. “It’s just not possible.”
Meanwhile, many CPOE systems do more than merely convey prescriptions to a pharmacy. They also warn doctors of potentially harmful interactions with other drugs a patient is taking, validate dosages and sometimes suggest drugs suitable for a patient with a given diagnosis.
For example, the system at Harvard-affiliated Brigham and Women’s Hospital displays a menu of the doctor’s most-prescribed medications along with standard doses and other instructions. Most of the time, doctors tend to prescribe only 30 to 50 different medications, said Dr. John Halamka, chief information officer of Harvard Medical School and a health IT leader.
“There may be 27 different pain medications,” Halamka said, “but I usually stick with two or three.” Giving doctors one-click ordering for their customary prescriptions increases their willingness to use the system, he said.
Bar coding is better If a CPOE system can help prevent medication errors, a bar code-based system to monitor drug handling can perform even better. The Department of Veterans Affairs first implemented its homegrown Bar Code Medication Administration system in 1999 and has been enhancing it ever since.
The BCMA system relies on bar codes imprinted on the medications and on patients’ wristbands. A nurse scans both bar codes at the bedside, and the system verifies that patients receive the right dose of the right medicine at the right time. It also adds to a patient’s EMR a record of the dosage.
Nurses can generate a list of doses that they need to administer during a shift or other time period, and they can see on a laptop PC screen a patient’s allergies to any medications. The VA administers half a million doses of medication through the system every day.
Several VA medical centers have studied BCMA’s effect on patient safety, and the results are striking. At the Martinsburg, W.Va., medical center, for example, five years after installing BCMA, medication errors decreased to one-third of their pre-BCMA level.
The Defense Department has its Pharmacy Data Transaction Service, a system that links CPOE capabilities to the Armed Forces Health Longitudinal Technology Application, DOD’s EMR system. PDTS holds medication histories for 9.1 million military personnel and their family members. Records come from DOD hospitals, other health facilities and commercial pharmacies where those enrolled in the Tricare health plan may get prescriptions filled.
The PDTS system identifies potentially harmful drug interactions, duplicate prescriptions and allergies at the pharmacy. Military Health System officials say it screens the request and responds to a pharmacist’s computer within 2.8 seconds in the continental United States and 3.2 seconds elsewhere.
In the past six years, the system has averted more than 200,000 potentially life-threatening drug interactions, said Col. Thomas Beach, a physician in the Air Force Medical Corps and AHLTA’s project officer.
The VA and DOD systems are among the most advanced in the world, experts say.
Beyond CPOE Medication errors represent the largest category of medical errors, but there are other types, including:
Making an incorrect diagnosis.
Performing surgery or otherwise treating the wrong site or organ of a patient.
Mixing up hospital patients and giving one a treatment or surgery meant for another.
Failing to provide proper treatment after an ailment is diagnosed.
Although using a Magic Marker may be the best way to avoid surgical mishaps, IT systems help prevent most of those errors. Clinical decision support systems, for example, can review a patient’s symptoms and suggest not only a diagnosis but also a plan of treatment. Those systems generate reminders of tests and treatments that the patient should receive on a regular schedule, such as colonoscopies for those older than 60. Bar coded patient wristbands are a solution to mix-ups about patients’ identities.
However, IT systems are not a cure-all. Some software for CPOE and other functions causes errors. For example, a Journal of the American Medical Association study conducted in 2005 found that a single CPOE system in use at an unnamed big city teaching hospital increased the likelihood of medication errors.
The system crashed on occasion, losing some medication orders, and dosage menus were based on what the pharmacy had in stock, rather than on clinical dosing guidelines. Twenty other aspects of the system had the potential to induce mistakes.
“Bad software, implemented badly, causes problems,” Halamka said, agreeing that some systems are not helpful to doctors. He said systems should be developed collaboratively with users and improved after their initial deployment — a textbook approach to building successful systems.
Health IT hospital leaders That approach to building systems was used at four hospital systems where Chaudhry’s team found the effects of systems on medical errors had been verified, at least to some extent.
The hospitals were LDS Hospital and the affiliated Intermountain Healthcare organization in Utah; the VA hospital system; the Regenstrief Institute in Indianapolis and affiliated health care institutions; and Brigham and Women’s Hospital and its affiliate, Partners Health Care, in the Boston area.
Those systems, all connected with medical schools, are widely regarded as the country’s leaders in implementing health IT. None is a typical American health delivery system. Regenstrief is a wealthy foundation that has invested millions in health IT. The Boston system is affiliated with Harvard. LDS is considered one of the best hospitals, and the VA is a one-of-a-kind system.
The typical hospital, in contrast, is likely to buy commercial software packages that are not modified for local environments. That hospital would not be able to take years to gradually implement the system, and it will not have the time or money to do in-depth studies of the system’s effects on the institution.
Chaudhry’s team concluded, as a result, that too little is really known about the potential benefits of health IT and recommended more studies. Some are under way now with support from the Agency for Healthcare Research and Quality.
Halamka, meanwhile, wrote an editorial for the Annals of Internal Medicine that took issue with the Chaudhry team’s conclusions. “I believe that proof of effectiveness in some settings is enough to proceed with widespread implementation” of EHRs, he wrote.
Common-sense approach In an interview with Government Health IT, Halamka was even more emphatic. “How could it be bad to have an accurate prescription of the right med to the right patient?” he asked. “The evidence is limited to academic health centers right now, but it just seems pretty obvious that you don’t need a controlled trial on the efficacy of parachutes against gravity.”
Partly to prove that CPOE’s effectiveness in reducing errors is not limited to a small group of elite hospitals, Massachusetts is undertaking measures to provide CPOE to most hospitals in that state.
A survey found that 14 already had the systems, leaving 60 to be automated. The project is beginning with a group of a dozen hospitals that scored high on a readiness test and met other criteria.
One of the project’s goals is to make a dent in the startling statistics about medical errors. Although the landmark Institute of Medicine study in 1999 indicated that as many as 98,000 people die in hospitals each year because of medical errors, subsequent studies suggest that estimate may have been low. For example, federal statistics indicate that 90,000 people die each year from hospital-acquired infections. Most of those infections are preventable.
The fear of lawsuits inhibits discussions about medical errors. However, the latest thinking is that lawsuits can be less of a problem if doctors are willing to admit errors to their patients and explain how they will prevent similar mistakes in the future.
The payment system is another disincentive for change because it rewards mistakes by paying more. If patients are injured in the hospital, their stays are extended and more treatments are administered, which in turn brings more revenue into the hospital. A few payers recently have begun to refuse reimbursement for avoidable errors — known in the medical profession as adverse events — and that trend is likely to grow.
Despite the lack of comprehensive evidence, there’s little question that health IT will prove to be an effective tool against medical errors. Acting National Health IT Coordinator Robert Kolodner said about one-fifth of the errors are because of a lack of immediate access to information about patients, and that’s a lack that only information systems can remedy.
“Safer is just the tip of the iceberg” when the benefits of health IT are added up, Kolodner said.
From the battlefield to the home front: Managing medical data
Government Health IT presents Col. Claude Hines Jr., program manager for the Defense Health Information Management System, in this recent InSight eSeminar. Col. Hines discusses the health information technology and tactical challenges faced by the military medical community in Iraq, Afghanistan and other areas of conflict. In doing so, he describes the current information technology solutions for transferring clinical data between battlefield care givers to health care personnel at military treatment facilities worldwide.
