Algorithms Assisted By Informatics

April 15, 2004

Several international algorithm projects are underway, some of which have been in use for years. What are the roadblocks for their successful implementation?

The science of informatics is now being utilized by algorithm and guideline projects to address problems with keeping recommendations updated and diffusing innovations.

David Penniman, M.D., dean of the University at Buffalo School of Informatics, is experienced in the use of informatics in assisting psychopharmacology algorithm development. He was a speaker at the initial meeting of the International Psychopharmacology Algorithm Project (IPAP) and the Chinese Psychopharmacology Algorithm Project in conjunction with the Ministry of Health of the People's Republic of China in Beijing in 2000. He is the consultant for the IPAP Schizophrenia Algorithm Project, endorsed by Collegium Internationale Neuro-Psychopharmacologicum (CINP). Representatives of CINP and IPAP are preparing for April talks in Geneva with Benedetto Saraceno, M.D., director of the Department of Mental Health and Substance Dependence for the World Health Organization (WHO), to explore ways in which psychopharmacology algorithms and guidelines crafted by an expert international faculty might be useful in WHO's efforts to promote mental health.

Asked about the rationale for informatics' contribution to treatment algorithm issues Penniman said, "While the application of information technology to medical issues (referred to as medical informatics) has been an area of interest for many years, a more general interest in successfully merging people, information and technology in a wide variety of settings is evolving. The School of Informatics at the University at Buffalo is one of the first of such schools in the United States concerned with the more general study of informatics."

Informatics research includes the study of the diffusion of information, information acquisition and screening methods, the effective presentation of information to specific audiences, and the social effects of information and information technology.

The expertise represented in this emerging area can greatly benefit those researchers and practitioners concerned with the development and dissemination of new treatment practices in a variety of medical fields. One pressing issue involves the development and use of treatment algorithms for mental illnesses including schizophrenia and depression. While the development of such algorithms is a well-accomplished art (some might say science), the successful diffusion of such algorithms is not well-understood. Furthermore, the challenge of maintaining up-to-date algorithms is a daunting task given that much of what goes into such algorithms is buried in clinical data or, even more challenging, in tacit knowledge held in the heads of researchers and clinicians.

For instance, IPAP is at the stage in its development of algorithms for the treatment of schizophrenia where it must address the challenges of a) ensuring that the treatment algorithms under development will be accepted and embraced by health care providers for use on patients with schizophrenia and b) assuring that the algorithms are kept up to date as new knowledge is acquired.

The collaboration among experts in the treatment of such mental illnesses and experts in the areas of research known as informatics holds promise in addressing the ultimate challenge--assuring that the best of what we know is used to benefit those who need it the most.

The use of algorithms and guidelines in residency education is in its early stages. Dan Stein, M.D., professor of psychiatry at the University of Stellenbosch, said, "Guidelines from Mental Health Information Center of South Africa are used in training residents and family medicine practitioners and are distributed to a fair number of primary care practitioners annually."

David Osser, M.D., and Robert D. Patterson, M.D., both of Harvard Medical School, have maintained and revised a teaching algorithm on the Internet since 1996. Since 1999, the Web site has been used extensively in the psychopharmacology course for residents in the department of psychiatry's Harvard South Shore Program. This two-year course emphasizes evidence-based medicine, algorithms and practice guidelines as important decision-making tools. Two surveys show that residents who have completed the course find the Web site valuable and use it frequently. The American Society of Clinical Psychopharmacology Model Curriculum for Psychopharmacology, 3rd edition (in press; American Society of Clinical Psychopharmacology, Inc.) includes a description of the Web site and flowcharts of its algorithms. Earlier editions of this curriculum were purchased by half of the U.S. residency programs.

Miles K. Crowder, M.D., director of psychiatric residency education at Emory University School of Medicine, department of psychiatry and behavioral science, says that instruction in using guidelines is not a formal part of the lecture program.

At Vanderbilt University School of Medicine, algorithms and guidelines are not currently routinely used in training residents in psychiatry, according to Oakley Ray, Ph.D. The literature has very little data regarding the level of awareness, much less the use, of algorithms and guidelines in psychiatry.

Dieter Naber, M.D., director of the Clinic for Psychiatry and Psychotherapy of the Psychiatric University Hospital in Hamburg, Germany, opines that there is much resistance to the use of guidelines among German psychiatrists.

However, in China, university and public officials are incorporating psychopharmacology algorithms in their curricula. Shu Laing, M.D., Peking University professor, has studied algorithm-based care in multiple hospitals. Yu Xin, M.D., also of Peking University, has proposed the use of psychopharmacology algorithms in residency training and for clinicians. The Harvard South Shore Psychopharmacology Algorithm Project is getting widespread notice and increased use. Its Socratic, consultative format recommends it. Osser reports that since the project went online, "There have been millions of hits. Individuals from 66 countries have identified themselves prior to downloading the algorithms for off-line use, and many more have viewed the site and used the algorithm programs without downloading. Translations of the algorithms into Chinese and Russian and placement on local Web sites in those countries have extended utilization ... national algorithm projects in Argentina, China, and Japan have studied the Web site and its updates.

"Individuals from around the world have spontaneously e-mailed comments about the site, 98% of which are complimentary. A recently added feature of the algorithm software encourages users to make a one-click response to rate the usefulness of the specific information they obtained. While not many users have given feedback, their ratings are overwhelmingly 4 to 5 on a 5-point scale."

Osser also noted, "Non-physician use also appears to be significant. I received an Exemplary Psychiatrist Award in 2000 from the National Alliance for the Mentally Ill (NAMI) because of nomination by members of the Florida branch that use the Web site.

"The [NAMI] Chicago branch has distributed materials from the [NAMI] Web site (Strategies for patients who are non-compliant and Levels of recovery from psychotic disorders) to 500 members," he added. "Instructors at the University of Minnesota division of health sciences teach from the Web site. Many Web sites have requested permission to create a link to the algorithm Web site. Despite the above, the potential for use would be far greater, based on experience in local performance improvement field trials and the informatics literature, if this kind of decision support were integrated into the workflow through computerized medical records. Answers to questions must be accessible extremely quickly and at the moment the question occurs to the clinician or the information will rarely be used. This is a major challenge."

Patterson and Osser also believe that clinicians would be more apt to use algorithms if they covered a greater number of specific clinical situations, especially those involving comorbid psychiatric and physical illnesses. Another challenge to the widespread use of algorithms is their need for desktop computers, which are difficult to use in typical clinical encounters. To help solve this problem, they noted that handheld versions are planned.

Madhukar Trivedi, M.D., professor at the University of Texas Southwestern Medical Center, department of psychiatry, said the algorithm products associated with the Texas Medication Algorithm Project are in use in 13 states. Trivedi has followed the work of Clement J. McDonald, M.D., who recommends that the algorithm be placed at the site of clinical decision for ease of reference.

Trivedi and colleagues noted (Methods Inf Med 2002;41:435-442):
Research indicates that computerized decision support systems (CDSSs) can improve clinical performance and patient outcomes, and yet CDSSs are not in widespread use. Physician guidelines, in general, face barriers in implementation. Guidelines in a computerized format can have novel aspects that have to be, considered, aspects such as technical problems/support and user interface issues that can act as barriers. Though the literature points out that human, organizational, and technical issues can act as barriers in the implementation of CDSSs, studies clearly indicate that there are methods that can overcome these barriers and improve CDSS acceptance and use. These methods come from lessons learned from a variety of CDSS implementation ventures. Notably, most of the methods that improve acceptance and use of a CDSS require feedback and involvement of end-users. Measuring and, addressing physician or user attitudes toward the computerized support system has been shown to be important in the successful implementation of a CDSS.

Making the algorithm available at the point of the clinical treatment decision, whether by PDA or other CDSS, can assist in utilizing the algorithm. A second way is utilizing the informatics field of innovation diffusion. Arun Vishwanath, Ph.D., M.B.A., professor at the University at Buffalo School of Informatics, says that an algorithm, like a product or an idea, can have its diffusion "exponentially increased by converting the opinion leaders to its use."

Stanford University department of psychiatry's Psychotic Depression Algorithm has had a suboptimal hit rate, according to Charles DeBattista, M.D., despite having many useful features and a large reference base. Thus, to paraphrase the father of information science Claude Shannon: Technical accuracy and semantic precision do not equate with effectiveness of diffusion.

The current project between CINP and IPAP to produce algorithms for the treatment of schizophrenia has to deal with the disparate formularies in different countries. A second issue is cost, which prohibits the use of a number of newer medications to some degree in all countries, but particularly in developing countries. The algorithms proposed for the treatment of schizophrenia can look very different, depending on the regulatory and financially practical formularies.

I am aware of early discussions between the developers of guidelines algorithms and officials involved in Medicaid and private managed care with a purpose of simultaneously focusing on quality of care and cost containment. Vishwanath pointed out that innovation diffusion is very different in closed or hierarchical systems than in the open population of potential users.

Perhaps the early courtship between guideline and algorithm projects and health economists has begun. The issue of how cost is factored into algorithms and guidelines will surely become more salient and transparent, one hopes, for the benefit of our patients.