Several authors have cited curriculum approaches to teaching and applying evidence-based medicine (1–5). Education applied in the realm of direct patient care engages busy clinicians as effective evidence users (3). Building on traditional approaches of formal education and knowledge gained from journal clubs, we sought to teach clinicians how to perform their own practice-generated research as the next step to learning evidence-based medicine.

Subjects

We selected 18 clinicians from a combination of clinicians who had attended the previous year’s Basic Research Skills class, expressed a desire to learn content, potential to develop research projects, and an interest in sharing expertise. Four clinicians dropped out within 2 weeks and were not included in the total. The resulting class consisted of 14 clinicians, including physicians, pharmacists, nurses, and respiratory therapists. Eight out of the 14 clinicians registered had taken the previous seminar, designed to teach how to ask a researchable question. The other six clinicians were selected based on prior research involvement and interest in research. Participants’ research and education background ranged from the baccalaureate to the doctoral level.

Design

Ten 1-hour classes were offered in weekly succession at the end of the workday. Class objectives and content are outlined in Appendix 1. The primary instructors were two experienced and respected individuals in the hospital: a biostatistician and a clinical nurse specialist. Their varying viewpoints initially created tension in the classroom. A sense of commitment to colleagues who had registered for the course, along with a willingness to share expertise, motivated the primary instructors to develop as a team. To this end, they initiated pre-class “live” prep sessions, where the biostatistician lectured and interacted with the clinical nurse specialist, posing as a student. During this time, the biostatistician’s content was placed in a clinical context by the nurse educator and the nurse educator’s understanding of statistics increased. As a result, the instructors complemented each other and allowed for a comprehensive approach to presenting clinical and statistical topics.

Each clinician was provided a copy of a standard textbook on designing clinical trials (7). To engage clinicians in thinking about doing their own research projects, they were asked to submit a published research paper from their area of interest before the first class. Two articles were selected and utilized as tools to teach elements of study design as well as statistics. Weekly homework assignments consisted of answering questions at the end of assigned chapters in the text or from their research projects. Clinicians were asked to submit an outline of their proposed study individually or in combination with collaborators.

Measurements

Statistical knowledge and understanding of study design were assessed with a test consisting of 27 questions. Six questions related to study design, and 21 questions focused on statistical and epidemiological definitions and interpretations of results. The question sequence varied between the pre- and posttests to minimize question order memory and increase the assessment of knowledge.

Data Analysis

We used exploratory data analysis (e.g., skewness, stem-leaf plots, normality tests) to assess the distribution of change scores (pre- minus posttest) prior to actually testing the null hypothesis of no difference in test scores. After concluding that the distribution of change scores did not significantly deviate from normality, a one-sample (two-sided) t test was applied to assess the change in test scores. A two-sample t test was utilized to compare participants who completed both pre- and posttests to the subgroup that did not take the posttest. An F test was conducted to compare variances between completers and noncompleters, determining whether to use the standard or an adjusted (e.g., Satterthwaite) estimator for the two-sample t test.

Four participants dropped out within the first 2 weeks because of conflicting schedules or health concerns and were not included in the measurements. Of the 14 who remained for the entire course, an attendance rate of 79.5% (95% CI=70%–89%) was achieved.

Interest in participation was measured by weekly voluntary attendance and preparation for class. Of the 14 who attended each class, 12 (87%) completed all homework assignments as evidenced by references made to the text and contributions to the progress of projects. The remaining two clinicians were not consistent in their level of engagement. The last five classes were especially rich in terms of class discussion and participation. In retrospect, it was during these classes that clinicians and instructors merged into one learning team. In addition, all clinicians collaborated on six research projects.

Knowledge gain related to statistics and study design was assessed by comparing pre- and posttests. All 14 participants completed the pretest, and eight clinicians (57%) also completed the posttest. A one-sample two-sided t test revealed a significant 24% improvement in test scores (pre- [51%] versus posttest [75%], df=7, p=0.002). A two-sample t test showed that the eight participants who completed both pre- and posttests did not significantly differ from the six who did not complete the posttest with regard to age (45 versus 46) and pretest scores (51% versus 51%), indicating that the analysis was performed on a representative sample of the 14 participants who had completed the pretest.

Six local research projects progressed as a result of this second research class. Two oncology studies resulted in acceptances for conference poster presentations. Two local pilot projects were designed to increase patient involvement with health-related decisions in the areas of women’s health and oncology. The fifth project was designed to study the correlation between fall risks and medications in the inpatient geriatric veteran patient. And the sixth project involved evaluating data collected from an inpatient surgical service at the Veterans Hospital for the National Surgical Quality Insurance Program.

Ultimately, clinicians interested in research learned about statistics and study design as they developed their own research activities. As an unexpected outcome, a biostatistician and a clinician successfully adapted their varying perspectives about research design to teach the shared importance of both statistical and clinical significance. A subsequent third course would focus on how to prepare a research proposal for Institutional Review Board (IRB) approval.