Abstract: Background: Novel statistical methods are constantly being developed within the context of biomedical research; however, the rate of diffusion of this knowledge into the field of general / internal medicine is unclear. This study highlights the statistical journal articles, the statistical journals, and the statistical methods that appear to be having the most direct impact on research in the field of general / internal medicine. Methods: Descriptive techniques, including analyses of articles’ keywords and controlled vocabulary terms, were used to characterize the articles published in statistics and probability journals that were subsequently referenced within general / internal medicine journal articles during a recent 10-year period (2000-2009). Results: From the 45 statistics and probability journals of interest, a total of 597 unique articles were identified as being cited by 900 (out of a total of about 10,501) unique general / internal medicine journal articles. The most frequently cited statistical topics included general/other statistical methods, followed by epidemiologic methods, randomized trials, generalized linear models, meta-analysis, and missing data. Conclusion: As statisticians continue to develop and refine techniques, the promotion and adoption of these methods should also be addressed so that their efforts spent in developing the methods are not done in vain.
Keywords: phase I clinical trials, standard algorithm, likelihood method, evidential paradigm. Abstract: In phase I clinical trials, the standard ‘3+3’ design has passed the test of time and survived various sample size adjustments, or other dose-escalation dynamics. The objective of this study is to provide a probabilistic support for analyzing the heuristic performance of the ‘3+3’ design. Our likelihood method is based on the evidential paradigm that uses the likelihood ratio to measure the strength of statistical evidence for one simple hypothesis over the other. We compute the operating characteristics and compare the behavior of the standard algorithm under different hypotheses, levels of evidence, and true (or best guessed) toxicity rates. Given observed toxicities per dose level, the likelihood-ratio is evaluated according to a certain k threshold (level of evidence). Under an assumed true toxicity scenario the following statistical characteristics are computed and compared: i) probability of weak evidence, ii) probability of favoring under (analogous to 1-α), iii) probability of favoring under (analogous to 1-β). This likelihood method allows consistent inferences to be made and evidence to be quantified regardless of cohort size. Moreover, this approach can be extended and used in phase I designs for identifying the highest acceptably safe dose and is akin to the sequential probability ratio test.