Reducing unnecessary testing

Efforts to test more judiciously represent a dramatic shift in the way physicians use tests to aid diagnosis and guide treatment.

The advent of a simple blood test to measure brain natriuretic peptide (BNP) almost 10 years ago was heralded as one of the most significant advances in heart failure diagnosis in decades. Today, it's considered an essential tool in the emergency department, and it has become a staple of inpatient management for heart failure, as well. In an era of health reform, however, the latter practice has raised questions.

“BNP can be a useful diagnostic tool. For example, BNP tests can differentiate heart failure from pulmonary disease for a patient who comes into the emergency department with shortness of breath and lowered oxygen saturation,” said Jasminka M. Criley, FACP, associate residency program director for internal medicine and director of inpatient services for graduate medical education at St. Mary Medical Center in Long Beach, Calif. “But serial measurements of BNP are not as helpful when managing the care of a patient, such as monitoring the effectiveness of therapy, or deciding treatment options.”

Photo by Thinkstock
Photo by Thinkstock.

After a pilot study found that physicians at St. Mary were ordering four BNP tests per patient stay, on average, Dr. Criley launched an initiative to educate physicians on the latest research and evidence-based guidelines on measuring BNP. (The American Heart Association/American College of Cardiology heart failure guidelines state that “the value of serial measurements of BNP to guide therapy for patients with heart failure is not well established.”) The effort paid off: Over about two and a half years, the average number of BNP tests ordered per patient stay dropped from four to two, saving the hospital approximately $500,000 without any negative effect on clinical outcomes.

“Before our initiative, the common practice was to follow an initial BNP test for an admitted patient with automatic, daily measurements,” said Dr. Criley. To reduce the number of unnecessary BNP tests, “we asked physicians to base their decisions on documented clinical evidence with less reliance on repetitive BNP testing.”

Efforts like this one represent a dramatic shift in the way physicians use tests to aid diagnosis and guide treatment. The traditional approach is to be as thorough as possible, using tests to consider every possible diagnosis and back up every treatment decision. But physicians like Dr. Criley are increasingly advocating laboratory appropriateness—ordering the right tests at the right time based on evidence and effectiveness.

“With health care reform, there is the expectation of lowering costs while improving quality and the patient experience,” said Todd A. May, MD, chief of the medical staff at San Francisco General Hospital and clinical professor in family and community medicine at the University of California, San Francisco. “[Preventing unnecessary testing] has always been a good thing to do, the right thing to do, but now pressure is on us to reduce costs, and this is a big area where we can do that.”

A cultural shift

A recent study on why physicians order unnecessary preoperative tests identified several themes that might apply to the ordering of any test. Physicians and nurses surveyed for the study said that many decisions to order tests are driven by habit (it's what's always been done) or what they believe other physicians want done (primary care physicians, for example, said they ordered tests based on surgeons' requests). Many seemed unaware of evidence-based guidelines or hospital policy on preoperative testing that may have guided their decision-making, leading the authors of the May 2011 Family Medicine study to conclude that “substantial barriers” exist to lowering inappropriate use of preoperative testing.

“Those themes are not unique to preoperative testing but pretty universal to test ordering in general,” said Dr. May. “The authors looked at one area of lab ordering which should be really straightforward—people should know what's necessary and what's not—but we failed miserably there.”

The study underlines the significant challenges to permanently changing physician behavior, Dr. May added. As a case in point, he referenced his own study that employed a simple intervention: implementing an electronic order function to prevent doctors from ordering recurring phlebotomy-lab tests during inpatient stays. The intervention was a short-term success, resulting in 12% fewer inpatient tests over one year, according to the results published in the August 2006 American Journal of Clinical Pathology. But Dr. May noticed the physicians eventually gravitated back to old habits.

“The intervention resulted in a dramatic response, saving money that could be redeployed to improve our phlebotomy service,” said Dr. May. “But fast forward to today and I bet we are pretty close to where we started again. You still can't order recurring labs, but physicians adapted and started ordering labs every day.”

It's not surprising that physicians err on the side of over-testing in a medical culture that “doesn't tolerate any misses,” said David B. Feinbloom, ACP Member, a hospitalist at Beth Israel Deaconess Medical Center and assistant professor of medicine at Harvard Medical School. “You might be 90% sure it's one thing, but if there's a small chance that it's something else, you get the test.”

Show them the evidence

At Danbury Hospital in Danbury, Conn., Damanjeet Chaubey, FACP, chief of hospital medicine, has had some success convincing residents to resist ordering unnecessary tests. As a key member of the hospital's Diagnosis-Related Groups (DRG) cost-efficiency council, Dr. Chaubey compared the charges associated with a group of common DRGs, such as pneumonia and stroke, with corresponding evidence-based recommendations.

The exercise revealed that many physicians were unnecessarily repeating lab tests every day during a patient's stay and that there was a tendency to order a whole battery of tests at admission to the ED instead of selecting those appropriate to the patient's symptoms. For example, physicians typically ordered electrolyte panels and complete blood counts (CBC) every day, even though the results changed very little day-to-day and usually did not affect how the patient was managed.

Dr. Chaubey knows that lecturing residents won't have much impact, so she engages them in an active learning process. At the bedside, she asks them to explain their rationale for ordering tests and involves them in researching evidence-based practices.

“It puts the housestaff into a self-questioning mode,” said Dr. Chaubey. “Whenever they order a test, I ask them to pause and think, ‘Why am I ordering this test? How will it change the management of the patient?’ Then, if it's still needed, order it.”

She also encourages residents and hospitalists to defer non-urgent tests to outpatient facilities to reduce costs. For example, when a hospitalist identifies lung nodules in a chest X-ray ordered to diagnose pneumonia, the radiologist will suggest a CAT scan of the chest that is performed in the hospital. But ordering the CAT scan on an outpatient basis saves money without compromising the management of the patient's pneumonia.

In one year, the efforts led to $2 million in cost savings on inpatient testing associated with the targeted group of diagnoses. The DRG council also discovered a “halo effect” on other diagnoses, resulting in a total decrease in inpatient testing charges of $6 million, or a 10% reduction in unnecessary inpatient tests.

The focus on reducing testing costs has spread to analysis of other costs, as well. Earlier this year, the council turned its attention to costs associated with prescription medications. For example, the council discovered that Lovenox (enoxaparin sodium injection) once a day was being used 80% of the time for deep vein thrombosis (DVT) prophylaxis even though the evidence suggested that generic heparin (three times per day) was just as effective at a much lower cost.

“The difference was in the hundreds of thousands of dollars,” said Dr. Chaubey. “As a result, we removed Lovenox from our order set as a first choice for DVT prophylaxis for the majority of patients and we started looking at pharmacy costs across the board.”

Computer alerts must be on target

While pop-up messages built into electronic health records can alert physicians to research or guidelines that could influence whether or not they order a test, physicians tend to override or ignore alerts when they start occurring too often. Yet the strategy can be effective if narrowly targeted to specific situations, according to a study at Kaiser Permanente Colorado.

The study, published in the November 2010 American Journal of Managed Care, had physicians in a group of Kaiser Permanente Colorado ambulatory clinics receive the following alert whenever they placed a computerized D-dimer test order for patients age 65 or older: “D-dimer testing is not recommended for patients 65 years of age and over because the accuracy is only 35% in this age group. Instead, order diagnostic imaging as appropriate.”

After 20 months, the number of D-dimer tests ordered per 1,000 visits for patients over age 65 decreased from 5.02 to 1.52 compared with a decrease from 3.14 to 2.11 in control clinics, where physicians received training and education about D-dimer testing but did not see the alerts. After the alert was activated in control clinics, the rate decreased further to 0.81 per 1,000 visits.

At the same time, the number of imaging tests increased. While these cost more than D-dimer tests, the latter often result in false positives in patients over age 65, which leads to ordering follow-up radiology tests, the authors noted.

The results point to the need for alert systems to become more sophisticated as electronic health systems mature, said Ted E. Palen, FACP, a clinician researcher at Kaiser Permanente's Institute for Health Research in Denver who led the D-dimer study. “If alerts aren't targeted, important, and add benefit for decision support, people might ignore them, just like pop-up ads on the Internet,” said Dr. Palen. “But as they become more targeted and specific, they really can aid in our decision-making.”

Physicians may have been less likely to ignore this alert because they were given an alternative strategy—imaging—as opposed to just being told not to order a D-dimer test, the authors wrote.

While the D-dimer study was performed in outpatient clinics, another recent study suggests its findings may apply to hospitals. The article, published online April 18, 2011 in Pediatrics, found that computerized decision alerts based on current evidence decreased the use of red blood cell transfusions without increasing mortality. The authors noted that the results could have wide applicability.

“We speculate that this strategy can be effectively applied to a wide array of therapies in which nonevidence-based practice and variability persist despite widely accepted standards,” the authors concluded.

Knowledge is key

While evidence-based educational initiatives and alerts can help reduce unnecessary testing, disagreement among clinicians about what is “necessary” is likely to persist. There is wide variation in the amount of uncertainty physicians are willing to tolerate that relates to where they trained, how long they have been practicing, and the medical-legal environment in which they work, Dr. Feinbloom said.

“When time is short and patient volume is high, it is always tempting to get one more study or test to ‘make sure’ you have all the bases covered,” he said.

The key for doctors is to understand the pre-test probability of the disease in question and the operating characteristic of the test they are ordering, and to be clear on what they intend to do with the information. Doctors also need to limit the number of tests that get repeated for no other reason than that a physician does not know the tests have already been done, Dr. Feinbloom said.

“Interoperable electronic health systems would be the single most helpful tool to minimize costs in an aggregate sense,” he said.

As it stands now, test results that reside outside of a hospital's proprietary system cannot be accessed. In fact, it is often easier to repeat a CT scan that has already been done than to wait for medical records at the other hospital to create a CD-ROM and ship it over, then have it uploaded into the radiology system and re-read by a radiologist at one's own hospital—a process that can take longer than the hospitalization itself.

Danbury Hospital has implemented interoperability on a regional scale with some success. The hospital's information technology department helped develop a regional health information exchange that links primary care physicians, specialists and hospitalists across the community with the goal of improving information exchange during transitions of care.

“If I admit a patient, I can view which prior as well as outpatient testing has been done,” Dr. Chaubey said. “If this patient just had an echocardiogram or CAT scan, even if it wasn't at our hospital, that information will help in reducing unnecessary testing.”

As hospitals face greater scrutiny under health care reform to improve cost-efficiency, it's no longer an option to leave cost out of the decision-making process, said Molly Cooke, FACP, an internist and professor of medicine at the University of California, San Francisco who penned an editorial on teaching cost-consciousness in patient care published in the March 31, 2010 New England Journal of Medicine.

“I was taught and people are still taught that when I get into the exam room or bedside my job is to advocate for anything that is of any conceivable benefit to that patient regardless of cost or the small likelihood of benefit,” wrote Dr. Cooke. “Given that the problems associated with costs of healthcare are well recognized, I don't think we can have people who say it's not my job to think about cost. It needs to be everybody's job.”