HLTH 447

Concepts

Epidemiologic research questions can be translated into statistical questions that can be addressed using appropriate data and quantitative data analysis methods. Data analysis should be informed by an understanding of the epidemiologic study design and data collection methods that gave rise to the data, and limitations of the data such as missingness, information bias, and selection bias. Many epidemiologic research questions involve comparisons across groups, where associations between exposures and outcomes are likely confounded by other variables due to a lack of randomization of exposures in the study design. Standardization and regression are two methods of reducing confounding when estimating associations. Biostatistical procedures rely on concepts of probability and probability distributions. Using these concepts, samples are used to estimate unknown population parameters, and precision of estimates is shown with confidence intervals.

Procedures

Summarize nominal, ordinal, and numerical data in tables and graphs Calculate sensitivity, specificity, positive and negative predictive values Select a probability sample using random numbers Standardize means, proportions, and rates across groups for comparison Calculate point estimates, confidence intervals, and P values Compare means using t tests and linear regression Compare proportions using chi-square tests, logistic regression, and binary regression Compare rates using Poisson regression Calculate life tables and Kaplan-Meier curves for survival data Compare survival times using log rank tests and Cox proportional hazards regression Combine results from multiple studies using meta-analysis procedures

Modes of Communication

Communicate research findings to a scientific audience in writing (primary focus) Communicate research findings to a scientific audience orally (secondary focus)

Concepts

Epidemiologic research questions can be translated into statistical questions that can be addressed using appropriate data and quantitative data analysis methods. Data analysis should be informed by an understanding of the epidemiologic study design and data collection methods that gave rise to the data, and limitations of the data such as missingness, information bias, and selection bias. Many epidemiologic research questions involve comparisons across groups, where associations between exposures and outcomes are likely confounded by other variables due to a lack of randomization of exposures in the study design. Standardization and regression are two methods of reducing confounding when estimating associations. Biostatistical procedures rely on concepts of probability and probability distributions. Using these concepts, samples are used to estimate unknown population parameters, and precision of estimates is shown with confidence intervals.

Procedures

Summarize nominal, ordinal, and numerical data in tables and graphs Calculate sensitivity, specificity, positive and negative predictive values Select a probability sample using random numbers Standardize means, proportions, and rates across groups for comparison Calculate point estimates, confidence intervals, and P values Compare means using t tests and linear regression Compare proportions using chi-square tests, logistic regression, and binary regression Compare rates using Poisson regression Calculate life tables and Kaplan-Meier curves for survival data Compare survival times using log rank tests and Cox proportional hazards regression Combine results from multiple studies using meta-analysis procedures

Modes of Communication

Communicate research findings to a scientific audience in writing (primary focus) Communicate research findings to a scientific audience orally (secondary focus)