Teaching Stata and statistics in contexts of evidence-based medicine and clinical trials

1. Teaching Stata and statistics in contexts of evidence-based medicine and clinical trials Glenn W Jones BSc MD FRCPC MSc Credit Valley Hospital 905-813-2200 (Canada)

2. Outline

3. Background Training • Applied math & stats in BSc • Clinical Epidemiology & Biostatistics MSc Learning • Full-time practice in Radiation Oncology • Mycosis Fungoides • Director of a Data Management Centre (2002-present) • Reviewer & Associate editorpositions (e.g. Evidence-Based Med.) • Ontario Cancer Research Ethics Board Teaching • Critical appraisal • Evidence-Based Medicine • Health Economics • Trial technology • Biostatistics Venues • Courses, rounds for fellows, residents, students • Seminars for senior and junior investigators • Staff mentoring Funding • Sanofi-Aventis Educational Grants (2007-2010) • International Atomic Energy Agency Contracts (2002-present) • Ontario Ministry of Health salary Affiliation • Dept. Of Medicine, McMaster University Perspective • Physician scholar • Clinical Epidemiology

4. Five principles

5. Managing emotions Emotions are essential to motivation, coping and learning How to minimize anxiety about “statistics” • Use minimal expressions, definitions, symbols and formulae • Introduce any theory & software (Stata) as tasks unfold • Learning such “structure” is intentionally relegated to being incidental to the doing of tasks (research, decisions, communicating ) How to engage “positive” emotions • Active discovery learning, doing the tasks • Students construct strategies, identify patterns, and assign meaning

6. Social Networking Meet social needs of students in the class environment • To work closely with several people Experience a shared practiceof statistics & research • Work in small teams to solve problems and answer posed questions Transition from a teacher-driven process to peer collaboration • Peer-to-peer dialogue and tutoring can clear information and creates shared knowledge

7. Goal is autonomy Do practical tasks that mirror later work that the staff will do • To answer genuine clinical questions (phenomena, events, observation, research question of interest); increases transfer of learning • Tool is a combination of the data-sets and software (Stata) • Pick up practical skills about data management, using software (Stata), workflow, collaborating • “Hot cognition” arises from synergy of kinetic, affective and mental engagements; increases later recall Use a graduated approach • Students may not have any prior knowledge, or be very mistaken about it • Use guided inquiry-discovery, through a well thought out and logical sequence of smaller tasks matched with the appropriate types and levels of tools Migrate to independence • shifting to student problems and projects (i.e. creative production)

8. Understanding research & EBM Picking up the methods of research • Technologies of research (e.g. randomized trials; design & measurement; causality) • Records of research (forms, data-sets; becoming facts when validated) • Results represented in transformations (e.g. recoding, tabulations), graphs and p-values Making knowledge claims • Knowledge, interpretation, explanation (Internal Validity) • Generalizability (External Validity) Making value claims • Criteria for quality of “evidence” (rules) • Change in clinical practice, budgets

9. Clarity of methods & concepts Use of data to answer interesting statistical questions • Develop strategies, concepts and statistical reasoning (while engaging with data) • Relate those to the event, object or research question that has been posed (for that data-set) Over sequential questions and data-sets, reflection facilitates growth in knowledge • Establish elaborating patterns to work and methods of analysis • Accommodation or re-organization of information (interpretation, mulling over) Establish the meaning and value of statistics • To assign an appropriate role to statistics in work and life • To construct a more complete understanding of the world (humankind and nature)

10. Tailored Curriculum Research Strategies & EBM Data Management & Statistics Data-set plan Data management, Records Data Capture, Study monitoring Workflow for data/analyses Graphing Descriptive statistics Associations & Modeling Predictions & Diagnostics Survival-type methods Longitudinal-panel methods Multivariate methods Research Question Eligibility, Sampling Design, Architecture, Bias Chance, Power, Randomization Measurement, Test statistics Sample size & Equivalence Protocol, Ethics, Grants Good Practice, Operations Communication, Publication Note synergistic interplay of core statistics parts with the adjacent contexts (see Gowan’s Vee mapping)

11. Example:“A classical medical study” Students in pairs/threes & Laptops with software(Stata) & data (.dta) & Problem-based learning module: Introduced as a cross-sectional case-series Expanded into a cohort study Summarized as a classical Clinical Experience paper Extended into a randomized clinical trial Further elaboration

12. Learning moduleMini-protocol to set the stage Background (literature, importance) Research question (who, what, so what) Eligibility/Population Design of study Measures Sample size Description of ethics, operational problems Description of the .dta (wide-long) Posing the statistical question “Problem-based learning, therefore addenda/replacements provided for each sequential step and extension”

13. Strategy Matrix Integrated EBM, data management, and statistical tasks

14. Strategic extensions • Add more cases • Append cases having different treatment(s) • Add the second arm of the randomized trial • Test study-arm for p-value, assess biases, calculate power • Add two more arms to create a 2x2 randomized design • Test for interaction and interpret • Add more time-dependent events • Merge dates and codes for distant recurrence & profiles of toxicity scores • reshape & append to get data into correct formats for tests/graphs • Lab test (e.g. spline or fractional regression due to curvilinear changes) • Add other outcome measures • Fractures (for neg. binomial regression) • Economic data (counts, units of resources) • Sub-study validation of another QOL measure (construct and convergence-divergence, item-response theory, canon) Eventually, transfer learning/skills: Students use their own research study or existing data-base working with peers and a mentor

15. Summary • Assumption • “students” have only rudimentary acquaintance with statistics and research methods; little or no formal training in Evidence-Based Medicine though a strong cultural influence; practice is in a narrow area (e.g. health-care); statistical anxiety; highly practical focus & impatient! • Method • Small group, problem-based learning (similar to McMaster Medical School) but a more guided-discovery method (initially not self-directed) • Mentor model with research project is possible (even as a second course); could be a self-study program-text (vs. typical texts) • Strategy • Focus on health-care questions and related literature (publication, sub-cultural formats); clear learning curriculum can be mapped (in statistics, EBM, research methods, work processes and skills) • A step-wise approach is used to build self-efficacy and confidence; an expandable data-base is used to introduce new aspects logically • Skills • Developed and repeatedly reinforced in: reasoning, using Stata, collaborating, workflow, writing (protocols, manuscripts)

16. References Pendlebury S. Chapter 2: Luck, responsibility, and excellence in teaching. In: The Educational Conversation, Closing the Gap. Eds: Garrision JW, Rud Jr AG. SUNY Press, 1995. Pressley M, Woloshyn V. Cognitive strategy instruction that really improves children’s academic performance. 2nd Ed. Brookline books, 1995. Waite-Stupiansky S. Building Understanding Together. A constructivist approach to early childhood education. Delmar Publishers, 1997. Zieffler A, Garfield J, Alt S, Dupuis D, Holleque K, Chang B. What does research suggest about the teaching and learning of introductory statistics at the college level? A review of the Literature. J Stat Educ 16(2), 2008 • Cherney ID. The effects of active learning on students’ memories for course content. Active learning in higher education 9(1):152-171, 2008. • Daedalus. Education yesterday, Education tomorrow. J Amer Acad Arts Sciences 127(4), 1998. • Lane DM, Peres SC. Interactive simulations in the teaching of statistics: Promise and pitfalls. ICOTS-7, 2006. • Merrill PF, Hammons K, Vincent BR, Reynolds PL, Christensen L, Tolman MN. Computers in education. 3rd Ed. Allyn & Bacon, 1996. • Novak JD, Gowin DB. Learning How to Learn. Cambridge U Press, 1984. • Ormrod JE. Educational psychology. 2nd Ed. Merrill Prentice Hall, 1998.