Biostatistics Support for Medical Student Research (MSR) Projects Allen Kunselman Division of Biostatistics and Bioinformatics Department of Public Health Sciences Penn State College of Medicine
MSR Process for Biostatistics Support l Not all MSR projects require a biostatistics consultation, but support is available l Limited free biostatistics support l Student is expected to do their own data collection and analysis with advice from statistician l Request biostatistics support well in advance of any deadline l In general, don’t procrastinate
Medical Student Research Process for Biostatistics Support l Complete the Department of Public Health Sciences (PHS) consultation request form at: l A meeting will be arranged and the student will receive statistical advice from the assigned statistical consultant l If at all possible, have your advisor attend the meeting l After the statistical consultant agrees with the proposed design and analysis, the student or statistical consultant will place the student's MSR Approval Form in Allen Kunselman's mailbox for a signature and the form will be returned.
4 Reasons Why Investigators Seek A Statistician’s Collaboration l Advisor/mentor told them to do so l Their study did not show significance l Abstract or manuscript was rejected n They want the statistician to “work some magic” and resurrect their study l Reviewer of manuscript or grant told them to seek statistical advice l The investigator has planned and designed the study, conducted the experiment and collected the data, but now has no clue how to analyze it.
5 Reasons Why Investigators Seek A Statistician’s Collaboration l Statistician viewed as a “necessary evil” needed in order to get sample size and analysis plan to appease the Internal Review Board (IRB) or potential grant/manuscript reviewers l Good experience collaborating with statisticians in the past n The statistician helped get a grant funded, manuscript published, results interpreted, etc. l Ideally, while the study is still in the planning and design phase, they seek a statistician because they really value the statistical advice and collaboration.
6
Statistician: What My Collaborators Think I Do
8 What can a Statistician Provide l Clarification of study objective n With respect to clinical trials: u Superiority: efficacy of treatment E is greater than that of treatment C –C can be a placebo or an active control u Non-inferiority: within a certain margin, efficacy of treatment E is at least as good as that of active control treatment C u Equivalence: within a certain margin, efficacy of treatment E is the same as that of active control treatment C
9 What can a Statistician Provide l Experimental Design n Completely Randomized n Randomized Complete Block n Latin Square n Incomplete Block n Split-Plot n Factorial n Crossover n Bayesian n Adaptive
10 What can a Statistician Provide l Sample size and power n Scientifically Meaningful Effect u Study must be “big enough” to statistically detect a scientifically (clinically) meaningful effect. u Study must not be “too big” where an effect of little scientific importance is nevertheless statistically detectable. n Money, Money, Money u Undersized study wastes resources and money by not having the capacity to produce useful (definitive) results. u Oversized study uses more resources and money than necessary.
11 What can a Statistician Provide l Sample size and power (continued) n Ethical Issues u Undersized study exposes subjects to potentially harmful treatments without advancing knowledge. u Oversized study exposes more people than necessary to a potentially harmful treatment or denies a potentially beneficial treatment. n Grant reviewers (and IRBs) are looking for sample size and power considerations, especially for prospective studies!
12 What can a Statistician Provide l Explicit definition of primary and secondary outcomes l Identify potential sources of random error and bias l Randomization l Blinding methods l Interim analysis and stopping rules l Methods for handling missing data
13 What can a Statistician Provide l Consideration for potential confounders and effect modifiers l Statistical data analysis l Interpretation l Reporting and graphics l Reproducible analysis l Data collection, management, monitoring, and archiving
14 Keys to Successful Collaboration: A Two-Way Street l Involve statistician at beginning of project (planning/design phase) l Specific objectives l Communication n avoid jargon n willingness to explain details
15 Keys to Successful Collaboration: A Two-Way Street l Respect n Knowledge n Skills n Experience n Time l Embrace statistician as a member of the research team l Fund statistician on grant application for best collaboration n Most statisticians are supported by grants, not by Institutional funds
16 Symbiotic Relationship for Investigator and Statistician l Investigator: scientific knowledge of the disease, outcomes, device, etc. l Statistician: technical skills to incorporate investigator’s scientific knowledge into an appropriate study design and analysis plan l In short, statisticians are not the enemy! Statisticians are there to contribute and support investigator research in order to obtain valid results.