The Biostatistics Component will support currently funded projects. In order to support the aims of a broad range of projects, the Biostatistics core provides a wide array of statistical and data management services. These services include: database development, data processing, data summarization in tabular and graphical formats, and analysis. Statistical support can be provided to studies of various designs, such as case-control, clinical trials, cross-sectional and laboratory experiments.
The study design phase includes specification of hypotheses, outcomes, sample size estimation, eligibility criteria, stratification factors, randomization and blinded designs. The analysis phase may consist of comparison of outcomes between groups, estimation of the association between variables, or one variable measured at different time points. The staff of the Biostatistics core has experience with both univariate and multivariate statistical methods. A partial list of statistical techniques include: analysis of variance, simple and multiple regression, Chi-square, logistic regression, Poisson regression, survival techniques such as life table estimation and Cox proportional hazards regression, non-parametric methods and longitudinal analysis.
For investigators choosing to use data provided by the Human Subjects core, the Biostatistics component assists by creating data files of the selected variables in various formats (Excel, text, Access). Researchers can request the data in raw format (individual data points for each case), or is a summary format, which can include data analysis for specific hypotheses.
For example, analysis was done to support an expanded aim of a current clinical trial. Investigators wanted to identify possible risk factors for early otitis media (OM ). Data from a previous epidemiologic study of the incidence of OM in children from birth to 2 years of age was used to identify potential risk factors. Chi-square analysis and multivariate logistic regression was used to develop prediction models. This allowed the current study to focus attention on factors that are most apt to lead to early otitis media.
The above services will be provided on a fee for service basis. For additional information, contact Bruce Lindgren: bruce@biostat.umn.edu