Clinical Computing: Treatment for All: The Computer as a Patient Assistant

In the last Clinical Computing column (1), I provided an overview of computer interview assessments for screening, diagnosis, and monitoring. Computer assessments based on structured interviews can be made widely available as low-cost screening tools to identify persons with psychiatric disorders and to monitor changes in the severity of their illness during treatment. They provide accurate and reliable information on which many clinical decisions can be based and with which they can be improved.

In controlled research, computer assessments reduce biases inherent in human ratings, thereby increasing the likelihood of separating an effective treatment from placebo. Use of these standardized assessments in clinical practice will provide critical data that are so costly to obtain by clinician interview that they are essentially unavailable.

Although clinicians can do screening, structured diagnostic interviews, and repetitive monitoring, these tasks are time consuming. Performing them is not the best use of clinicians' abilities to synthesize knowledge and experience into the clinical wisdom and decisions that have always been hallmarks of the best practitioners. Clinicians seldom collect assessment data reliably, and they also have difficulty processing and scoring algorithms accurately.

Aided by computer assessments, the average clinician improves performance, the better clinician approaches brilliance, and all benefit from knowing how treatment is proceeding. On the clinical stage, computer assessments are playing supporting roles of increasing importance and to growing applause. Waiting in the wings are computer programs that advance into a heretofore-exclusive clinician domain—treatment. This column presents the rationale for the development and use of computer treatment programs and reviews experience with early examples of this genre.

An ideal world

In an ideal medical world, every patient would have an hour with William Osler and another with Sigmund or Anna Freud, with additional visits as necessary. In reality, more than 40 million Americans have no health insurance, and many with insurance are granted only ten (not even 15) minutes of Warholian fame during visits with their primary care physician-gatekeeper. And those ten minutes are divided among face-to-face contact between the patient and the doctor, interactions with support staff, and record keeping. Little time remains to explain to patients the findings and diagnoses and what can be done about them, and even less time for is left for psychotherapy.

Talk-time has always been an expensive commodity in medicine. We're often dismayed by how little patients retain from their hurried visits and annoyed when they are "noncompliant" with our instructions. In 11 studies the proportion of information given by the physician and forgotten by the patient ranged from 28 to 71 percent (2). Given the crush of pressures on clinicians, it's remarkable that treatment goes as well as it does much of the time. The saying "So many books, so little time" is amusing; "So many patients, so little time" is serious. How might computers help?

More than 30 years ago, Warner Slack, then at the University of Wisconsin and soon to move to Harvard, conducted the first patient-computer interview (3). He saw the potential educational and treatment benefits of computer interviews then and has continued innovative work on computer therapy for emotional problems (4). Warner's new book, Cybermedicine, is a must-read for anyone interested in clinical computing (5).

What the pioneers in these applications recognized, and current practitioners are acting on, was that some aspects of psychiatric practice are so well codified that they can be written out. If they can be written out, they can be given to patients to use for self-help. Once a treatment is so well established that it can be committed to paper, it can be moved to a computer program. And once it has become a computer program, it can be accessed by patients in a variety of ways, each offering interactivity akin to that of a patient-clinician interaction.

For the computer treatment programs now in use, every step along the way had to be tested, and its efficacy demonstrated. It was shown that the codified treatment did indeed help patients, and that patients could benefit from a written self-help program. It was then shown that patients did not object to using a computer to work on the same self-help program, and that in some instances they preferred giving information to a computer.

In the remainder of this column, I describe specific work that has been done to treat some disorders with the help of a computer. I hope it will be clear that the techniques can be applied to treating other disorders.

Computers for patient education

First, a word about using computers to educate patients about disorders and treatment, which is the intermediate phase between assessment and therapy. Many clinicians provide patients with booklets as part of the education process. However, booklets are a passive medium that depend on functional literacy and a reading habit. It is sobering to remember that 21 to 23 percent of adults in the United States—40 to 44 million people—are functionally illiterate (6), and that only 56 percent of U.S. adults read books (7). We have learned that other media may convey our information more effectively to some individuals. Studies of self-help programs show that when two media are employed, such as an audiotape or a videotape plus a booklet, treatment effect size more than doubles (8).

Multimedia computer programs have now been developed to help patients learn important information about specific disorders and available treatments (9). These programs improve patient understanding and satisfaction with treatment, presumably because the patients become enlightened participants in decisions about their health care. In psychiatry, Madoff and colleagues (10) found that a desktop computer program was as helpful as nurses in providing information that inpatients retained about their medications for schizophrenia.

For patients, support persons, and clinicians, CATERS, an educational program that describes schizophrenia and schizoaffective disorder, medications useful in their treatment, and factors affecting quality of life has been developed. It was pilot-tested in a community mental health center and a Veterans Affairs hospital, where it was well received and was revised based on feedback. It is now being readied for widespread use. Because it is available by interactive voice response from any touch-tone telephone via a toll-free number, CATERS has the potential to be accessible at all hours, even when staff are not.

The World Wide Web is expanding rapidly, and more than 2,000 health sites of widely varied quality are currently available. More than 30 million Americans already have the capability to connect to the Internet, and the number is growing rapidly. For about $15 a month, those with a personal computer, a modem, and browser software can explore this intriguing world of interlinked pages and self-help newsgroups, listserves, mailing lists, and chat rooms. Web jargon is fast becoming part of our idiom. The Internet is a source of information—and misinformation—that is already performing an educational function, and it has the potential for interactivity between patients and mental health professions, probably soon rather than later.

Computers providing treatment

Some computer programs go beyond education to provide psychotherapy. Careful studies are being conducted to answer the many questions that have been posed about this approach to treatment. Does it undermine the most interpersonal and intrinsically humane of therapies? How many patients are being denied treatment because there are not enough trained therapists? How much effort and expense would be involved in training enough therapists? Once they were trained, when and where would they be available, and at what cost? Because not all therapists are created equal, how can we identify those who excel and ensure minimum levels of initial competence and continued acceptable performance by all therapists?

If the cost and effort of providing enough human therapists proves too great, and if there are concerns about quality with increased numbers, would a computer program that provides treatment compare favorably in efficacy, availability, and cost? Use of computer assessments makes comparison with therapists along these dimensions feasible for the first time. My hypothesis is that the best therapists are better than today's computer therapy programs in terms of efficacy, but they may not be better on the other dimensions of access, convenience, and cost. Computer programs may have greater efficacy than less skilled therapists, and their low cost, convenience, and capability for widespread dissemination may make them accessible to many who would otherwise not receive help.

Current computer therapy programs are focused on disorders for which self-help is integral to treatment success. They guide patients in designing and doing treatment, and they also monitor patients' status as treatment progresses. Although most programs are intended for prescription by clinicians who provide other aspects of treatment, such as medications, and oversee patient care, it is possible that some of these programs could be offered directly to patients by the manufacturers in the same way that patients try self-help books without the supervision of a clinician. An effective self-help program guides users through an assessment of whether they should seek professional help rather than continue on their own.

Because development and study of computer therapies is still in the early stages, far more questions than answers exist about their best uses and limitations. Will better results be obtained when the programs supplement what a clinician provides? What training would clinicians need to make best use of computer programs? How would the programs perform without clinician supervision? What range of severities of a disorder would respond to computer treatment?

Some of the early work on computer treatment programs has been directed toward depression, phobias, and obsessive-compulsive disorder. In one study, depressed patients treated with a desktop-computer program of cognitive-behavioral therapy improved as much as patients treated by a cognitive-behavioral therapist (11). At the end of six weeks of treatment and at a two-month follow-up, both groups had improved more than a treatment-on-request control group (11). In another study, phobic patients improved as much when they used a desktop computer without therapist assistance as patients who were treated for three hours by therapists and those who read a chapter from the self-help book Living With Fear (12) and were also treated for three hours by therapists (13).

The health care delivery trend I find most exciting involves the use of the interactive voice response (IVR) technology I described in my previous column on assessment (1). One program, called BT STEPS, treats patients with obsessive-compulsive disorder (14,15). Patients access a programmed text using a toll-free number and a touch-tone telephone, which allows them to interact with the education and treatment modules according to their specific needs. They learn about the disorder and behavioral and medication treatments, describe their rituals, estimate the monetary and time costs of having the disorder, and identify triggers of their obsessions, discomfort, and rituals.

Using the program, they design very specific exposure goals: "I will touch the mossy brass doorknob at work, making sure I get some of the contaminated moss under my wedding band. I will then go directly home without washing and will 'contaminate' the countertops in the kitchen and bath linen and bed sheets and then neither wash nor check for three hours afterward." They then record their personalized exposure goal in their own voice so it can be played for them before each exposure session and refined and rerecorded based on their experience. The program allows them to fine-tune their exposure technique and carry out repeated exposure and ritual-prevention sessions for their triggers until habituation reduces their discomfort to a level of 0 or 1 on an 8-point scale. The BT STEPS program concludes with a relapse prevention phase.

Patients who have used the program averaged seven hours and 58 minutes of telephone contact time over 12 weeks of treatment, and they spent many more hours with the exposure and ritual-prevention homework they designed in concert with the program. Seventy-two percent of calls to BT STEPS were placed outside the usual office hours of Monday to Friday, 9 a.m. to 5 p.m. Phase 1 and 2 trials have demonstrated efficacy and a high level of patient acceptance (14,15), and further evaluations are under way.

COPE, another self-help IVR treatment program, has been tested as a treatment for mild-to-moderate depression (16). It provides IVR modules and booklets on constructive thinking, pleasant activities, and assertive communication, as well as a booklet on grief. During preliminary evaluation studies, patients recorded their major problem in their own voices. Several commented that hearing their distress unmitigated by modulating interactions with a therapist increased their motivation for treatment. The program worked well in studies of its efficacy, was enhanced in response to those studies, and is now available.

Conclusions

Computer-administered treatments have several advantages. They can be made widely available on desktop or laptop computers and will be nearly ubiquitous via interactive voice response on touch-tone telephones. Costs are far less than for comparable contact time with therapists—cents rather than dollars per minute. The computer's faithful execution of treatment algorithms ensures customization of each person's interactions with the program. Regular reports on patient status can be generated and made available for a clinician's use. The high standard of consistency that is possible in computer therapy ensures each user the best that the program has to offer. Modular design permits the addition of new components and systematic dismantling to assess contributions of each component.

In addition, the IVR delivery system offers other benefits. In the study of the COPE program, 68 percent of the IVR calls were made outside 9-to-5 weekday office hours, a convenience patients commented on. In the study of the BT STEPS program, patients in England who did not own a touch-tone telephone were loaned one, and some who did not have home telephone service used a public telephone. The ramifications in terms of accessibility are part of the excitement about IVR. The telephone link to a central computer can be worldwide—the patients in England used a toll-free number to call a computer in Madison, Wisconsin. Use of the programs does not require the individual clinician to buy more computer hardware or software or to hire additional nonmedical staff.

As with assessments, we simply cannot disseminate and systematically study effective, standardized, authoritative, interactive, low-cost treatments without computers. Using computer programs, we can complement, supplement, extend, and enhance the best clinician treatments of today and improve them for tomorrow.

Dr. Greist, who is editor of this column, is distinguished senior scientist at the Dean Foundation for Health, Research, and Education, 2711 Allen Boulevard, Middleton, Wisconsin 53562, clinical professor of psychiatry at the University of Wisconsin Medical School, and chief executive officer of Healthcare Technology Systems, both in Madison (e-mail, [email protected]).