Evidence-Based Medicine Aims to Improve Patient Care

The everyday decision of which medication will work well for which patient ideally requires a tremendous amount of data and time-consuming analysis by the prescribing psychiatrist. Recently, researchers and educators have been trying to make these decisions faster, easier, and more reliable through standardized guidelines and medication algorithms.

But the most difficult question about this may be whether physicians will use the guidelines and algorithms once they have been established.

“The gap between best practice in psychopharmacology and typical practice is uncharacteristically wide,” said David N. Osser, M.D., an associate professor of psychiatry at Harvard Medical School and general editor of the department’s Psychopharmacology Algorithm Project. Guidelines using evidence-based medicine for choosing the most appropriate psychotropic medications do exist, Osser said during an APA annual meeting presentation in New Orleans last month, but “those guidelines are in fierce competition with industry-supported educational activities.”

In addition, Osser said, there are significant barriers to using evidence-based medicine (EBM) to make medication choices for patients. The decision requires “laborious thinking,” he said, based on the clinician making a complete and thorough DSM-IV-TR diagnosis.

EBM guidelines further require extensive research and development before they can be applied in clinical practice.

“Most importantly, though,” Oesterheld concluded, “we need to find out if EBM really works. ”

Rather than use EBM guidelines, physicians make more reflexive decisions. “They will make bias-driven judgments about medications that they categorize as part of the ‘art’ of medicine,” Osser said. Many physicians lack experience in using recommended treatments and in general, he said, have an overall sense of rejection toward what many call “cookbook medicine.”

The problem becomes one of an unfinished picture, said Osser. Physicians rely far too heavily on the pharmaceutical industry to tell them how, when, and why to prescribe particular medications.

Ira D. Glick, M.D., a professor of psychiatry at Stanford University Medical Center, agreed with Osser. “You’ve got the fox teaching how the chicken tastes! It’s nuts,” Glick emphasized.

Physicians must learn to include industry-supported data in their decisions in an educated manner, knowing that there is a potential for some bias included in the presentation of the data, the two speakers said.

Osser has spent the last several years at Harvard developing the Psychopharmacology Algorithm Project along with Harvard colleague Robert D. Patterson, M.D. The project includes algorithms for the pharmacotherapy of schizophrenia, depression, and anxiety in patients with a history of substance abuse.

The Algorithm Project team said that the challenge continues to lie in the rapidly advancing and increasingly complex knowledge base upon which the principles of psychopharmacology rest. It is nearly impossible for clinicians to keep up with developments, Osser said, and especially difficult for clinicians to keep those principles in mind during actual clinical decision making about medications.

The Harvard project’s team members aimed to improve the quality of psychopharmacological thinking and care by promoting EBM practice through the use of computerized algorithms. The evidence cited in the algorithms includes high-quality empirical research studies, but also some uncontrolled studies and compilations of expert opinion and practice guidelines derived from clinical experience.

In addition to the actual algorithms, the project’s Web site at www.mhc.com/algorithms includes extensively referenced information on dosing strategies and side effects.

The algorithms each provide a sort of “virtual consultation,” which asks the clinician about the patient, Osser noted. The user then gets a set of fully referenced recommendations that include both the merits and the problems associated with alternatives to the first-line therapy.

Each recommendation comes with a confidence rating that indicates the strength of the scientific support for the recommendation.

In addition to working through the algorithms, psychiatry residents also progress through 17 learning models presented over their third and fourth postgraduate years. Each module focuses on one of 11 psychiatric disorders and contains clinical case conferences, literature reviews, and medication algorithms based on the best literature specific to treating that disorder.

“One major point here,” Osser said, “is that residents become experienced and educated consumers of evidence-based medicine.” He added that industry-supported data are included in the project’s literature reviews; however, students are taught to look for problems in design, methodology, and conclusions so they can “take some results seriously and others with a grain of salt.”

At Stanford, Glick has been working on revising the model psychopharmacology curriculum distributed by the American Society of Clinical Psychopharmacologists (ASCP), a project he has been involved with since the early 1980s. Unlike Harvard’s Algorithm Project, the ASCP curriculum does not include medication guidelines, but relies on traditional teaching methods such as lecture and discussion outlines, a teacher’s guide, and extensively referenced appendices.

“The curriculum,” said Glick, “addresses the long-standing problems of how to teach a dynamic and complex subject effectively and with getting the right people who have the knowledge to teach that knowledge well.” Glick and his colleagues review the development and content of the model curriculum in the summer issue of Academic Psychiatry.

Jessica Oesterheld, M.D., now semiretired and in private practice in Maine, helped review the ASCP curriculum for accuracy and said she used both the ASCP model and Harvard Algorithm Project extensively as a professor of psychiatry at the University of South Dakota.

“Both the algorithms and the ASCP method,” Oesterheld said during the annual meeting presentation, “respect the snowball effect of learning as being cumulative. They simply give different approaches to achieving the same goal.”

An advantage of the ASCP model curriculum, said Oesterheld, is that it allows instructors—regardless of their proficiency with the subject matter—to start with an expert lecture and then adapt to their own needs by changing either the text or corresponding slides.

The real advantage of the Algorithm Project is its focus on teaching residents to “think like an evidence-based psychopharmacologist,” she said, which encourages residents to reflect on how a pharmacological decision is made. Unfortunately, Oesterheld said, the method requires teachers with substantial skills. “They have raised the bar, perhaps too high, and many of the teachers just can’t do it.”

Ideally, Oesterheld said, the two curricula should be the ultimate complement to each other, with individual adjustments appropriate to the needs of specific training programs.

The Harvard Psychopharmacology Algorithm Project is posted on the Web at www.mhc.com/algorithms. The article “How Should We Teach Psychopharmacology to Residents: Results of the Initial Experience With the ASCP Model Curriculum” is posted at http://ap.psychiatryonline.org/. More information on ASCP’s teaching modules is available by calling (212) 268-4260. ▪