|
|
 |
GHS 201C
Note: This course is in the process of finalization and topics listed below may be rearranged depending on textbooks chosen and course timeline.
Global health policy and development depends on the nature and validity of evidence, which in turn depends on appropriate research methods and analysis. This course will cover instruction in research study design, measures of disease occurrence and disease association, the different sources of error in observational research, and a conceptual approach to multivariable analysis. The course will cover types of data, their summarization, exploration, and explanation with special emphasis on means, proportions, regression coefficients and contingency tables. Throughout the course, the software program STATA will be used. The course will integrate basic descriptive and analytic statistics with research methods, allowing the students to conceive and develop their own research project within the framework of practical, ethical and logistical constraints.
Teaching format: Lectures, seminars, independent study, assigned projects, and practice with statistical software (STATA)
Course credits:
4 units over one quarter
- 2 hours of lecture per week
- 4 hours of seminar plus 2 hours of independent study per week
Competencies:
By the end of the course the student should be able to:
- Describe the uses of epidemiology and the 5 types of epidemiologic studies (descriptive, ecological, case control, cohort and randomized clinical trials)
- Define the measures of disease frequency (prevalence and incidence) and give examples
- Define variable types (e.g. categorical, continuous), distributions, and measures of central tendency and variance
- Demonstrate an understanding of standardization (direct and indirect) and give examples of comparative statistics of two populations (e.g. standardized incidence ratio, age standardized rate ratios)
- Know the major measures of effect (excess risk) such as rate, risk ratios, odds ratio, and confidence intervals.
- Describe measures of public health impact such as population attributable risk
- Be able to conduct descriptive and unadjusted bivariate analyses
- Describe the relative merits of different study designs and the main analytic methods available to estimate disease association (outcome) with predictor variables
- Define the sources of error in epidemiologic research (chance, bias, confounding), cite examples of each, and enumerate methodological or statistical strategies to deal with them in both the design and analysis phases
- Know Hill’s postulates of causation and be able to use these to critically analyze literature and studies to estimate causality.
- Critically analyze a published epidemiologic paper, citing strengths, weaknesses, and validity of the inferences made
- Become familiar with survival analysis (Kaplan-Meier) and measures associated with this analytic technique
- Understand principles of regression methods (linear regression; logistic regression)
- Be able to navigate the STATA statistical package and conduct univariate and bivariate data analyses
- Plan and design an epidemiological or clinical study taking into consideration study designs and methods of analysis. Write a proposal to include: study question/hypothesis, outcome and predictor variables to be measured, target study population, sampling, sample size, analysis plan, and timeline.
Course content (1a – 5b basic biostatistics; 6a-10b epidemiologic methods):
Biostatistics (weeks 1-5, 2 sessions per week)
- Lecture 1a: Variables- nominal, ordinal, categorical, continuous. Descriptive statistics: distributions, means, standard deviation, standard error, and variance; predictors, outcomes, exposure variables and their measurement
- Seminar 1a: Introduction to data and data collection; Epi-info or other “data screen” for data to be collected by students (= class data set).
- Lecture 1b: Measurement of disease in populations: prevalence, incidence, rates
- Seminar 1b: Introduction to STATA. Descriptive statistics (means, medians, frequencies)
- Lecture 2a: Measurement of disease association I: the 2x2 table; prevalence ratio; odds ratio, attributable risk; confidence intervals
- Seminar 2a: STATA examples of disease measurements and rates; standardization; review a journal article that illustrates learning concepts
- Lecture2b: Measurement of disease association II: Confounding; chi square and Mantel-Haenszel tests; stratification
- Seminar 2b: STATA exercise: descriptive statistics of class data set
- Lecture 3a: Cohort studies: incidence, risk ratios, rate ratios, PYAR
- Seminar 3a: STATA: Bivariate analysis of class data set; review a journal article with longitudinal data.
- Lecture 3b: Confounding
- Seminar 3b: STATA Confounding: how to prevent in study design; how to detect and adjust in analysis, using class data set
- Lecture 4a: Introduction to multiviariate methods (linear and logistic regression)
- Seminar 4a: STATA exercises in multiple regression analysis using class data set
- Lecture 4b: Non-parametric statistics: t test; Mann-Whitney; Wilcoxon; Interaction
- Seminar 4b: STATA workshop: parametric vs. non-parametric statistics.
- Lecture 5a: Review of biostatistics (TBA)
- Seminar 5a: Biostatistics final exam: Analysis of new data set with which students are not familiar using STATA
Epidemiology (weeks 6-10, 2 sessions per week)
- Lecture 5b: Observational study designs: descriptive, ecological, case control, cohort
- Seminar 5b: Journal article review: ecological and descriptive case studies
- Lecture 6a: Interventional studies: the randomized clinical trial (RCT); stratification and randomization methods
- Seminar 6a: Journal article review: RCT case studies; clinical and community examples
- Lecture 6b: Designing Epi/Clinical research project and essential components: study question/hypothesis, independent and dependent variables, sampling, sample size calculation, analysis plan. Everything that has to be included
- Seminar 6b: Review design of proposed study
- Lecture 7a: validity and precision and error. Sources of error in observational studies: chance, bias and confounding. Bias in all its guises.
- Seminar 7b: Review potential biases of proposed study
- Lecture 7b: Methods to minimize error. Dealing with bias: design, measurement, etc. Detecting and avoiding bias in case studies; misclassification,
- Seminar 7b: Review a paper, examine bias
- Lecture 8a: Sample size; power calculations; matching; restricting.
- Seminar 8a: Perform sample size calculation for your study
- Lecture 8b: Designing clinical research (DCR) review:
- Seminar 8b: Work on DCR project
- Lecture 9a: Ethics of epidemiologic studies and clinical trials. Consent, assent and ethics of cross-cultural investigation
- Seminar 9a: Informed consent case studies
- Lecture 9b: Meta-analysis
- Seminar 9b: Critical analysis of a paper or review
- Lecture/seminar 10a,b: presentation of student projects in small groups
|
|
|
