The materials on this course website are only for the use of students enrolled in this course for purposes associated with this course and may not be retained or further disseminated. The materials on this course website may be protected by copyright; any further use of this material may be in violation of federal copyright law.
This course is
designed to provide a working knowledge of statistical methods suitable for
longitudinal data analysis. Topics
covered include: time-to-event outcomes, Kaplan-Meier methods, Cox regression,
repeated measures, and mixed models. Emphasis will be on medical and epidemiologic
examples and practical applications using SAS.
Professors |
Teaching Assistant |
|
Raymond R. Balise Balise at Stanford |
Brit B. Turnbull Brit decimal Turnbull at
Stanford |
Jim Murphy Jdmurphy at Stanford
|
HRP 259 and HRP 261, plus HRP 223 recommended, or two previous courses in statistics, including ANOVA and linear regression..
Monday 3:15-4:45 in LKS 101 for the rest of the quarter.
Wednesday 3:15-4:45 in M206.
Balise by appointment in Redwood Building T213D. Directions can be found here: www.stanford.edu/~balise/FindBalise.htm
Turnbull by appointment in Redwood Building T260.
We will be using a Google group groups.google.com/group/hrp262-2011 . After you fill out the survey shown under Lecture 1 you will be given access.
Regression
methods in biostatistics by Vittinghoff -- Ray’s EG Code for Vittinghoff’s
Regression Methods in Biostatistics book is here.
Survival
Analysis using SAS: A Practical Guide by Allison – Ray’s EG code for
Allison’s Survival Analysis using SAS is here
and the link to the data is here. If you unpack the zip archive into a folder
called C:\Projects\books\AllisonSurvival
the EG project should behave well.
Applied Longitudinal Analysis by Fitzmaurice, Laird and Ware – Ray’s SAS/EG Code for Chapers 1-10 in Fitzmaurice Laird and Ware’s Applied Longitudinal Analysis is here.
The Little SAS Book by Slaughter and Delwiche
The Little SAS book for Enterprise Guide 4.2 by Slaughter and Delwiche
Participation: 20%
Midterm Project (Survival Analysis): 40%
Final Project (Longitudinal Data Analysis): 40%
You many work on projects in groups (maximum 3 per group) and hand in a single report. For both projects, you will be given a dataset and will analyze it using SAS. Your report will include tables, figures, a results section, and SAS code. For hints on writing results sections of manuscripts, see: http://www.stanford.edu/~kcobb/writing/lecture6.ppt
Both exams will be submitted via email to the TA.
Each of the assignments will be due at the beginning of class on the day specified.
The programs that you turn in must run on Windows SAS 9.2 and/or Enterprise Guide 4.3. We can provide good support for Windows or a Mac running parallels (http://www.parallels.com/).
If you would like to purchase a SAS license for your
personal computer see itservices.stanford.edu/service/softwarelic/sas
Pictures
showing my SAS 9.2 TS2M3 and EG 4.3 install are here.
Notes
on configuring SAS Enterprise Guide 4.3 are here.
Email
Dr. Balise if you do not have permission to access the slides.
Some
of the slides are done. Known broken hyperlinks are in red.
Reading: Vittinghoff Section 3.5.1, 7.1.1, 7.1.2, Allison Ch. 1
Logistics
Introduction to Survival Analysis
Functions in Survival Analysis
Kaplan-Meier Estimate of the Survival Function
The PowerPoint slides are here.
Tell us who you are with the survey here.
Tell us what you know with the survey here.
Reading: Allison Ch. 1-3
Configuring SAS EG
Importing data
Creating a well-organized analysis project
Basic Descriptive Statistics
Creating formats to label formats
Doing Basic KM estimation
Making Survival Curves
Loading Keyboard Macros
Smoothed Hazard with confidence limits
Survival Curves with number remain at risk
The PowerPoint slides are here.
An Enterprise Guide project covering the first few labs is here.
Reading: Vittinghoff Section 7.1, Allison Ch. 2-3
Kaplan-Meier Detailed Example
Kaplan-Meier Confidence Limits
Comparing Groups
Competing Risks
The PowerPoint slides are here.
Reading: Allison Ch. 2-3
Using formats to make data meaningful
Survival plots with confidence bounds and confidence limits
Comparing two groups
Making publication ready KM plots
Adding recoded variables
Grouping continuous predictors into categories
Comparing multiple groups
An Excel 2010 file showing the manual calculation of the survival function is here.
The PowerPoint slides are here.
Reading: Vittinghoff Ch. 4 (review), Sections 7.2 and 7.4, Allison Ch. 5.
Hazard Ratio and Proportional Hazards
Review of Linear Regression and Model Selection
Modeling Survival Data
Introduction to the Cox Proportional Hazards Model
The PowerPoint slides are here.
Reading: Allison Ch. 4
Hazard functions
Peering down SAS output
Parametric hazard estimation
Building Hazard Curves
With code and GUI
The planned PowerPoint slides are here.
The PowerPoint slides I covered are here. (Updated fixing DSD bug 4/17/201)
An Enterprise Guide project covering the first few labs
(with imbedded datasets) is here. (Updated
fixing DSD bug 4/15/2011)
A problem set allowing you to practice KM and Cox Regression
is here.
Time varying covariates
Accounting for tied event times
Diagnostics for proportional hazards
Hypothesis testing
The PowerPoint slides are here.
Reading: Allison Ch. 5
Hazard functions
Basics of Cox Regression
Trying to analyze categorical variables with Cox in EG
Basic checks on proportional hazards
Interactions
Coding Interactions with time to check proportional hazards
The PowerPoint slides are here. (Slightly updated 4/20/2011)
An Enterprise Guide project for labs 1-4 is here. (Slightly updated 4/20/2011)
Survival analysis completion
Review
Allison Ch. 5–6
Building datasets for survival analysis
Competing risks
The PowerPoint slides are here.
The Enterprise Guide project for lab 5 is here
Reading: Fitzmaurice, Laid and Ware Ch. 1-3 & Vittinghoff Ch. 8.
Paired t-test vs. single sample t-test
Correlation and Covariance
ANOVA
ANCOVA
A couple PowerPoint slides showing modeling are here.
The PowerPoint slides are here.
Reading: Fitzmaurice, Laid and Ware Ch. 1-3
Loading data from text files
Labeling and formatting values
Doing paired t-tests
Calculating two point change score
Comparing two point change score between two groups with unpaired t-tests
Converting wide/broad data sets to long form
Calculating summary statistics on long form repeated data
Plots of data where subjects are not measured consistently
Sampling a few patients for plotting
The PowerPoint slides are here.
The Enterprise Guide project for lab 5 is here
Repeated Measures ANOVA
Fixed Random and Mixed Effects models
The PowerPoint slides are here.
Reading: Fitzmaurice, Laid and Ware Ch. 5
Converting long form data to wide/broad format
Wilcoxon-Mann-Whitney tests on changes
Univariate repeated measures ANOVA
Multivariate repeated measures ANOVA
Repeated measures mixed effects models
Checking subject specific patterns
The PowerPoint slides are here.
An Enterprise Guide project is here.
Modeling
Mixed Effects
Covariance structures
The PowerPoint slides are here.
History of repeated measures
Univariate repeated measures ANOVA
Multivariate repeated measures ANOVA
Repeated measures mixed effects models
Wide to Narrow datasets
Checking subject specific patterns
Overall mean plot
Fixing degrees of freedom ddfm = kr;
Looking at correlation structure
Correlation by time distance plot
Comparing models with -2 log likelihood, AIC, AICc, BIC
Using maximum likelihood to compare fixed effects in nested models
Influence Diagnostics
Means across time
Custom Contrasts
The PowerPoint slides are here.
An Enterprise Guide project is here.
Review
A solution to the midterm problem set
KM methods
Cox methods
Dealing with interactions in survival analysis (take 2)
The PowerPoint slides are here.
A sample Enterprise Guide solution to the midterm is here.
Repeated measures on outcomes that are or are not normally distributed
GEE
Generalized Linear Mixed Effects Models
Cookies!!!!
The PowerPoint slides are here.
An Enterprise Guide project is here.
Keyboard
macros for SAS 9.2 and EG 4.3 are here.