270001 Using Orthogonal Coefficients to Test for Trends in Health Survey Data

Monday, October 29, 2012 : 3:15 PM - 3:30 PM

Lei Zhang, PhD MBA , Office of Health Data and Research, Mississippi State Department of Health, Jackson, MS
William Johnson, PhD , Pennington Biomedical Research Center, Louisiana State University System, Baton Rouge, LA
Jerome Kolbo, PhD , College of Health, University of Southern Mississippi, Hattiesburg, MS
Background and objective: Trend analysis is often used for public health surveillance and monitoring, for forecasting and program evaluation. Many health surveys were conducted within the equal time intervals or the same elapsed time. The main goal of trend analysis for public health surveillance is to discern whether the measures of the indicators have increased or decreased over time

Methods: This study used the 2005, 2007, 2009, and 2011 Child and Youth Prevalence of Obesity Survey (CAYPOS) as an example. CAYPOS measured height and weight of grades K-12 students in Mississippi. The orthogonal polynomial contrast coefficients were constructed using SAS IML procedures. The logistic regression used orthogonal linear and quadratic coefficients to model longitudinal trends in overweight and obesity while controlling for students' gender, race, and grade level. SUDAAN 10.0 was used for all data analysis due to complex design of the surveys.

Results: The linear coefficients (-3, -1, 1, 3) and quadratic coefficients (1, -1, -1, 1) were generated and assigned to the years 2005, 2007, 2009, and 2011, respectively. Neither linear (p = 0.0845) nor quadratic (p = 0.7090) trends were observed in prevalence of overweight and obesity.

Conclusion: The orthogonal coefficients can be easily constructed and assigned to the datasets. In addition, they can be used in the regression models. The model outputs can explain why the prevalence of overweight and obesity for all students in grades K-12 no longer appears to be increasing in Mississippi.

Learning Areas:
Biostatistics, economics

Learning Objectives:
Participants can learn how to construct, assign, and explain linear and quadratic coefficients in trend analysis using complex health survey data.

Keywords: Obesity, Biostatistics

Presenting author's disclosure statement:

Qualified on the content I am responsible for because: I have been PIs for several population-based surveys (such as YRBS and PRAMS) in the state level. I also helped the CAYPOS data analysis.
Any relevant financial relationships? No

I agree to comply with the American Public Health Association Conflict of Interest and Commercial Support Guidelines, and to disclose to the participants any off-label or experimental uses of a commercial product or service discussed in my presentation.