Problem-based learning (PBL) is an instructional method that uses patient problems (e.g., fever, pain) as the context of teaching medical students problem-solving skills and is also a process for acquiring basic and clinical science information through self effort and group cooperation. "Problem- based" and "case-based" are terms often used interchangeably (+1). In true problem-based learning, the problem is presented first, and the students learn the basic science and clinical concept while they solve the problem. The faculty role is to facilitate. A small- group format is thought essential to teach the skills that the student is supposed to learn through PBL (+2,+3). Thus, numerous faculty facilitators are needed.

PBL is based on learning theories, especially contextual learning, which posits that retrieval of information is dependent on the environment or context in which information is learned (for example, a student is more likely to remember the treatment of delirium tremens in the emergency room [ER] if the student learned about it in an emergency room). The main principles involved are the following:

PBL involves seven steps. Step 1 involves clarifying terms and concepts not easily understood. Steps 2 and 3 define and analyze the problem, respectively. Step 4 involves drawing a systematic inventory of problems, whereas Step 5 results in formulating learning objectives. Finally, in Steps 6 and 7, the group collects additional data and synthesizes the newly acquired information. In this process, most of the work is done by the students, and, as noted above, faculty are mere facilitators (+4).

Studies (+2,+3,+5—+10) assessing the value of PBL indicate that it has very high rates of student and faculty satisfaction. Although the differences are not statistically significant, students who regularly experience PBL also achieve higher scores on the National Board of Medical Examiners (NBME) Parts II and III (+11). Furthermore, the process of small-group PBL also promotes greater acquisition of psychosocial knowledge, fosters patient-centered attitudes, and enhances interpersonal skills (+12,+13). However, PBL is extremely labor-intensive (requiring several faculty per small group per problem) and although it is good at covering the selected problems in depth, may not always include enough breadth of knowledge. For example, in the case of a patient presenting with memory problems, students may focus on various aspects of memory and learn about memory disorders, but may not learn about the disorders in which memory complaints occur along with several other symptoms (e.g., depression and mania), and yet memory deficits in these disorders are clinically relevant. Some studies (+14,+15) disagree with this assumption; others (+16) assert that there is no assurance that a PBL session will cover various contact areas. Furthermore, when class size is more than 100, the cost of PBL can become prohibitive. Students who experience PBL also develop a thinking style termed "backward reasoning," where they think about the problem of fever and then try to investigate all the etiologies of fever. Clinically, this strategy can be extremely time-consuming, inefficient, and costly. Students also tend to become dependent on small groups, so that they tend to practice in group practice situations rather than working in a solo practice or for a large health-care organization (+6).

Specific modifications of PBL have been recommended. For example, using a more structured format taught by expert physician faculty would better ensure that adequate content has been addressed. Using expert faculty would also allow students to learn the "forward reasoning" that most experts use. For example, when approaching a problem, experts first recognize the pattern of the presentation, which reduces the differential-diagnostic possibilities and increases efficiency. Students, on the other hand, take a problem and go after endless possibilities. Retaining the advantages of PBL while correcting its limitations and practical usefulness has obvious teaching benefits. Walton et al. (+16) combined lecture format and small-group PBL and offered it to half of their second- year class. The other half received the traditional lecture format. The hybrid PBL group enjoyed the learning process, which was still faculty-intensive.

To get some of the advantages of PBL with limited faculty resources, we adopted small-group, patient-based interactive sessions used with residents and third-year clerks to a large second-year neuroscience course. Patient-based teaching is not PBL in a true sense. Patient-based teaching is not student-centered at its core. Here, students do not choose a problem and search for answers. They are not expected to collect additional outside information, but rather utilize the knowledge presented earlier.

Patient-based teaching relies on patient vignettes, in which the patient is briefly described as a physician would initially see him or her in practice. We try to keep these vignette situations as close as possible to actual situations. For example, in the case of the patient with memory deficit described earlier, the family may bring the patient because the patient had problems at work. So, during discussion we address not only differential diagnosis of the memory problem, but also rehabilitation and family education. So, unlike PBL, in this process, the faculty member will address every issue that may come up during the treatment of that patient. This method uses clinician faculty experienced with the kinds of patients to be discussed (i.e., they are clinically expert). This faculty determines the content to be discussed in each session. To ensure adequate content coverage, before actual patient-based sessions, trainees receive some didactic sessions.

Our second-year clinical neuroscience course enrolls up to 180 second-year medical students and runs two semesters, for 88 hours. It is divided into three sections: functional brain organization (e.g., frontal-lobe circuits and their syndromes; hemisphere specializations); assessments (e.g., psychopathology and the behavioral examination; cognitive testing laboratory studies); and syndromes (e.g., traditional DSM Axes I and II; stroke; brain tumors). We use the natural breaks within and at the end of each section for PBL, totaling 28 hours, or 30% of course time. The topic of each patient-based session reflects previous lecture material synthesized as a set of clinical problems generated within a patient vignette or several vignettes. Each patient-based session is used to 1) develop clinical thinking guided by principles of diagnosis and management; 2) review and integrate previously taught data bases; and 3) present some new elaborative information. Typically, over 100 students attend each patient-based session.

First, we decide which areas covered in previous lectures will be the topic for a patient-based session. Next, we decide what our main teaching points for that topic will be. For example, the topic might be depression; the main teaching points might be 1) recognition of a clinical depression under differing circumstances (e.g., following a stressor in an elderly person; associated with a general medical condition); 2) recognizing suicide risk and what to do about it; and 3) guidelines for the selection of an antidepressant. The vignette might then describe a depressed patient who reflects the above circumstances, or might be a "stem" vignette to which the different circumstances can easily be added. The original stem vignette or additions can come from a single patient or from a composite of patients. The stem vignette is distributed to the students at the beginning of the session. Any additions are presented by the faculty during the session. Appendix 1 displays a sample item vignette and several possible additions to it.

A session is structured to guide students through the stem vignette and any additional descriptions so that each teaching point is developed and discussed. We use a step-by-step outline to help keep us on track and as a prompt, if needed. The details of this outline can vary depending on the needs of the teacher. At the end of each session, students receive a handout summarizing the main points and supporting database of the session. Appendix 2 displays a generic outline of a session.

A session can be led by one or more faculty members. If one person does it, that person is an expert in most areas of the topic. If two faculty members teach in a session, their roles are clearly delineated. One strategy involves the faculty member taking responsibility for leading the students in understanding different aspects of the topic (e.g., the work-up for seizure disorder; the "do's" and "don'ts" in the outpatient treatment of a depressed patient). Faculty members alternate leading the session by their assigned topics, the alternation following the planned sequence of the main teaching points.

A second strategy involves one faculty member leading the students throughout the session, with a second faculty member acting as a resource for resolving different options or opinions raised by the students, or for restarting the process if the students reach an impasse and do not know (or remember) needed information to proceed, or cannot, as a group, make an important diagnostic or treatment decision.

After each examination, students evaluate the course. These evaluations are collected by the curriculum office, and the individual and overall evaluations are forwarded to the course director. We have been doing patient-based teaching since the 1997—98 academic year. +Figure 1 displays overall course ratings before and after patient-based teaching was introduced. Because the number of Good ratings remained the same, we compared Excellent, Acceptable, and Unacceptable ratings before and after PBL was introduced. There were no statistical differences between the number of "Good" ratings before and after patient-based seminars were introduced (chi- square_[1]=0.23; P=0.632). However, the number of Excellent ratings increased significantly (chi- square_[1]=39.9; P<0.001), whereas the number of Adequate ratings decreased (chi-square_[1]=9.4; P<0.005) and so did the number of Unacceptable ratings (chi-square_[1]=7.3; P<0.01). The response rate in 1998—99 was more than 95%. In the 1998—99 academic year, total hours of patient-based teaching was 12 (14.4%). On the basis of student feedback, we increased patient-based hours to 28 (30.4%). In 1998— 99, we asked more specific questions about patient- based teaching. +Table 1 describes student ratings for patient-based sessions. Also, we also asked students to endorse specific statements about patient-based teaching. In the winter, 93.75% of the total respondents answered this portion of the survey, and, in spring, 100% answered. Because the number of patient-based seminars in each term varied, we chose to report the data separately. We feel this separation helps us to clearly demonstrate the student preference for patient-based seminars. +Table 2 lists the results of that survey.

Our findings suggest that, as is the case in small- group, problem-based learning, the student satisfaction rate is very high for large-group, patient-based teaching. However, our data indicate that students would still like conventional lectures. The statistically significant changes in the course rating before and after patient-based teaching was introduced makes us speculate that a combination of these two methods of instruction may be preferred to either one alone. Also, this method of instruction addresses several criticisms directed toward small-group PBL (mentioned in the introduction). For example, the curriculum is faculty-directed and, therefore, more structured. We give conventional lectures just before the patient-based seminar to cover the depth of knowledge necessary to solve patient problems. Unlike the problem-based format, where students may pick a problem and work on their own, with the faculty member as facilitator, in this format, we use various patient scenarios, and the faculty member is an active leader and role model. In this way, students benefit from the faculty member's diagnostic and problem- solving capabilities. In this method of instruction, we have addressed some of the criticisms of small-group, problem-based teaching. However, as in the previous studies, we also do not have the data to demonstrate its superiority over conventional teaching. We also do not have data to demonstrate that large-group teaching retains the advantages of small-group, patient- based learning. In 1996—97, the mean psychiatry scores for our medical school (CMS) were 95% of the national mean. Although the change was not statistically significant, after patient-based teaching was introduced, mean CMS psychiatry scores on the 1997 and 1998 United States Medical Licensing Examination (USMLE) Step 2 exams went up to 97.6% of the national, and Step 2 scores for the 1998—99 year were 99.1% of the national mean. At present, we do not have Step 3 scores of those students who took USMLE Step 3 after we introduced patient-based teaching. These measures, however, may not be adequate because some of the skills learned through this method, such as efficient management of patients without compromising quality, are procedural in nature and cannot be reliably tested in a written examination.

In summary, at present, our findings confirm that patient-based teaching can be done by one or two faculty members with a large student group who will enjoy these sessions. However, these sessions complement, and not replace, conventional lectures. Better outcome measures need to be designed to evaluate the efficacy of this instructional method.