This proposal concerns the transfer and implementation of statistical methodology applicable to the study of serial measurements of intermediate markers of disease status. We plan to synthesize several new statistical methodologies; to produce a manual showing illustrative and comparative examples and to produce associated software which would be made available free of charge to applied researchers using disks, listings or via the electronic archive STATLIB. Information concerning the manuals and programs will be posted on electronic bulletin boards such as STAT-L, S- NEWS etc. The applicability and validity of the methodology will be demonstrated with the use of data from two large cohort studies, which investigate the effectiveness of serial prostate specific antigen (PSA) measurements for the diagnosis of prostate cancer. The benefits of the successful conclusion of this project could include: More effective diagnostic measures for early detection of cancer in a screening program; Earlier detection of disease recurrence while monitoring patients in a cancer treatment or prevention trial; Improvement of the efficiency of designs of epidemiologic and clinical studies, by using the methodology to aid the identification of high risk patients for inclusion in the study population.
| Slate, E H; Turnbull, B W (2000) Statistical models for longitudinal biomarkers of disease onset. Stat Med 19:617-37 |
| Strawderman, R L; Parzen, M I; Wells, M T (1997) Accurate confidence limits for quantiles under random censoring. Biometrics 53:1399-415 |
| Natarajan, R; Turnbull, B W; Slate, E H et al. (1994) A computer program for the statistical analysis of repeated event data using a mixed effects regression model. Comput Methods Programs Biomed 42:283-94 |
