Home Evidence-Based Medicine Why evidence-based medicine isn’t always a black-and-white matter

Why evidence-based medicine isn’t always a black-and-white matter

September 2004

Published in the September 2004 issue of Today’s Hospitalist

If you could search the literature before treating patients, would you find that your decision-making follows good clinical evidence, or would you change some of the treatments you prescribe for patients? A recent study found that in many cases, the answers to both questions might be yes.

Researchers at Chicago’s Cook County Hospital examined whether physicians were following good clinical evidence when treating inpatients. They examined treatment decisions made for 130 patients cared for by 33 attending physicians.

The process began when researchers talked to attendings about the patients who had been randomly selected to be included in the study. They would ask about the primary treatment issue that warranted hospitalization for the patient, perform a literature search and then produce a research packet for the physician.

Most treatment decisions (86 percent) were based on solid evidence, a figure that echoes other recent research. What was more interesting, however, was the fact that a significant number of physicians who were following evidence indicated that they would change their patients’ treatment anyway.

Changed treatment

According to Brian P. Lucas, MD, the study’s lead author and head of the section of inpatient medicine in the division of general internal medicine at Cook County Hospital and Rush University in Chicago, researchers identified potential alternative treatment options for 35 percent of all patients. When presented with those options, physicians said they would change treatment in 23 (18 percent) of 130 patients because of information produced by the literature searches.

That’s not to say that physicians who changed treatments weren’t following good clinical evidence. In only one case did a physician’s treatment plan change from non-evidence-based to evidence-based. Dr. Lucas says that the study’s results, which were published in the May 2004 Journal of General Internal Medicine, point out the variety of factors that go into medical decision-making.

In one case, for example, researchers gave the attending physician information about the value of giving trimethoprim/sulfamethoxazole (Bactrim) to prevent pneumocystis pneumonia in a patient with metastatic brain cancer. The attending physician wasn’t aware of the information until he received the search packet, and he said he would add the drug to his treatment plan.

(Dr. Lucas notes that researchers didn’t count the example as a change from non-evidence-based medicine to evidence-based medicine because the other aspects of the treatment plan followed good clinical evidence.)

The single example of a treatment that went from nonevidence-based to evidence-based was fairly clear cut. In that case, Dr. Lucas explains, a patient with liver cancer was scheduled to be treated with chemoembolization. The search packet, however, pointed out that the clinical evidence indicated that the benefit of that treatment was fairly unimpressive. As a result, the attending decided the treatment was not warranted.

“The most common reason that attendings changed from one evidence-based therapy to another was that we found there was better evidence for a different therapy than the one they were prescribing,” Dr. Lucas says, “even though the initial therapy was often evidence- based. There are different strengths of evidence, so while two therapies may both be evidence-based, one may be based on better evidence than the other.”

Countering EBM orthodoxy

One of the interesting ideas raised by the study is the fact that information was given to a passive physician audience and not sought out by doctors. The attending physician who was treating the patient with brain cancer was unaware of the evidence about using Bactrim, Dr. Lucas explains, so he probably wouldn’t have found information about it if he had done his own literature search. “It speaks to the idea that physicians don’t know what they don’t know,” he adds.

In fact, when researchers talked to attendings about patients who had been selected for a literature search, the physicians would often respond by saying that the patient wasn’t very interesting and probably didn’t merit a look at the evidence.

Dr. Lucas says the results are interesting because the strategy used by the researchers defies some of the tenets of evidence-based medicine.

“The evidence-based medicine orthodoxy has said that you identify the patient and generate a patient-centered question, then you approach the literature based on that,” he explains. “Our search method was very different in that it was standardized and generic. You could apply this generic search to a random sample of patients, as we did, and just searching on the treatment for their particular problem, you could find that 20 percent of those attendings would change treatment.”

Defining evidence-based treatments

The study also raises questions about what exactly constitutes evidence-based treatment.

“Patients often receive multiple treatments,” Dr. Lucas says. “If a patient is receiving four different medications for high blood pressure, one of those medicines may be evidence-based, but is the combination of those four medications evidence-based, and do you refer to that patient as receiving evidence-based treatment or not?” The fact that so many physicians who were already judged to be practicing evidence-based medicine in the study changed their treatment course anyway “shows that there are other factors that play into the complicated decision- making process of deciding on a therapy for patients,” he explains. That includes interactions with physicians like consultants, who in the study would occasionally nix new evidence presented to the attending physicians.

While about half of the physicians said the search packets were helpful or somewhat helpful, they didn’t always change their treatments based on the search packets. In 23 cases, in fact, attendings decided that the evidence presented did not warrant a change in therapy. They either believed the evidence did not apply to their individual patient, was not convincing, or it was not the preferred route identified by a consultant.

In some cases, patients were the ones who decided against evidence uncovered by researchers. Dr. Lucas gives the example of a patient who was taking doxazosin for high blood pressure. When the attending suggested a different drug, noting that doxazosin has been shown to increase hospitalization for congestive heart failure, the patient responded that doxazosin was the only drug that controlled his prostatic symptoms. “Even though there might be evidence to take the patient off doxazosin,” Dr. Lucas says, “the patient prefers to stay on it for other reasons.”

Complex interplay

Researchers who reviewed the evidence presented to attendings didn’t always agree about whether changing therapies would help or hurt patients. Reviewers said that in most of the 23 patients (78 percent) where treatment was changed, the postsearch treatment was better or equal to the presearch treatment. In about one-quarter (22 percent), however, reviewers thought the attending physicians’ presearch treatment was actually better.

That interplay between evidence and other factors that the literature can’t take into account, says Dr. Lucas, is the real significance of the study’s results. “What’s news is the fact that whether patients were receiving evidence-based treatment wasn’t the sole determinant in whether or not physicians decided to change their treatment,” he explains. “It speaks to the complexity of physicians having to make treatment decisions. It’s not simply about trying to pick the one treatment that’s evidence-based, because there are other things involved.”