Statistical Thinking in Biomedical Research
Division of Biostatistics & Epidemiology
Department of Health Evaluation Sciences
University of Virginia School of Medicine
Rob Abbott, Viktor Bovbjerg, Mark Conaway, Frank Harrell
hesweb1.med.virginia.edu/biostat/
teaching/clinicians

Office of Continuing Medical Education
Fall 2002

Slide: 1


Section 1: Introduction to Statistical Concepts ¾ Frank Harrell


Slide: 2


What Do Biostatistics & Epidemiology Offer?


Slide: 3


Example


Slide: 4


Statistical Inference ¾ Examples


Slide: 5


Infinite Data Case


Slide: 6


Finite Dataset


Slide: 7


Steps Involved in Statistical Inference


Slide: 8


Study Design Issues


Slide: 9


Response Variables


Slide: 10


Types of Studies/Believability of Results


Slide: 11


Randomized Experiments


Slide: 12


External Validity of Study Findings


Slide: 13


Pitfalls in Analysis & Interpretation


Slide: 14


Descriptive Statistics


Slide: 15


Analysis of Paired Observations


Slide: 16


What's Wrong with Percent Change?


Slide: 17


Objective Method for Choosing Effect Measure


Slide: 18


Biostat / Epi Resources at UVa


Slide: 19


How to Collaborate with Statisticians & Epidemiologists?


Slide: 20


Collaboration Issues


Slide: 21


Education Opportunities at UVa

References

[1]
D. G. Altman and J. M. Bland. Absence of evidence is not evidence of absence. British Medical Journal, 311:485, 1995.

[2]
J. C. Bailar III and F. Mosteller. Medical Uses of Statistics. NEJM Books, Boston, second edition, 1995.

[3]
C. Begg, M. Cho, S. Eastwook, R. Horton, D. Moher, I. Olkin, and et al. Improving the quality of reporting of randomized controlled trials. The Consort statement. Journal of the American Medical Association, 276:63-39, 1996.

[4]
T. C. Chalmers, H. Smith, B. Blackburn, B. Silverman, B. Schroeder, D. Reitman, and A. Ambroz. A method for assessing the quality of a randomized control trial. Controlled Clinical Trials, 2:3-9, 1981.

[5]
T. J. Cole. Sympercents: symmetric percentage differences on the 100 loge scale simplify the presentation of log transformed data. Statistics in Medicine, 19:310-125, 2000.

[6]
CPMP Working Party. Biostatistical methodology in clinical trials in applications for marketing authorizations for medicinal products. Statistics in Medicine, 14:165-682, 1995.

[7]
P. C. Gøtzsche. Blinding during data analysis and writing of manuscripts. Controlled Clinical Trials, 17:28-93, 1996.

[8]
L. Kaiser. Adjusting for baseline: Change or percentage change? Statistics in Medicine, 8:118-190, 1989.

[9]
R. A. Kronmal. Spurious correlation and the fallacy of the ratio standard revisited. Journal of the Royal Statistical Society A, 156:37-92, 1993.

[10]
T. A. Lang and M. Secic. How to Report Statistics in Medicine: Annotated Guidelines for Authors, Editors, and Reviewers. American College of Physicians, Philadelphia, 1997.

[11]
J. S. Maritz. Models and the use of signed rank tests. Statistics in Medicine, 4:14-53, 1985.

[12]
L. Törnqvist, P. Vartia, and Y. O. Vartia. How should relative changes be measured? American Statistician, 39:4-6, 1985.

1
RCTs include crossover studies, which can be of excellent quality when there are no carryover effects or when carryover effects are understood well enough to be ``subtracted out''.
2
Because absolute risks of events vary with disease severity, dictating that risk differences must vary.
3
Because of regression to the mean, it may be impossible to make the measure of change truly independent of the initial value. A high initial value may be that way because of measurement error. The high value will cause the change to be less than it would have been had the initial value been measured without error. Plotting differences against averages rather than against initial values will help reduce the effect of regression to the mean.