Skip to main content
ARS Home » Research » Publications at this Location » Publication #213483
Title: Growing Growth curves using PROC MIXED and PROC NLMIXED

Author
item GOSSETT, JEFF - ACHRI-DAC
item SIMPSON, PIPPA - ACHRI-DAC
item PATRICK, CASEY - ACHRI-DAC
item WHITESIDE-MANSELL, LEANNE - UAMS
item BRADLEY, ROBERT - UALR
item JO, CHAN HEE - ACHRI-DAC
item Bogle, Margaret

Submitted to: Meeting Proceedings
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/26/2007
Publication Date: 6/3/2007
Citation: Gossett, J.M., Simpson, P.M., Patrick, C.H., Whiteside-Mansell, L., Bradley, R., Jo, C., Bogle, M.L. 2007. Growing growth curves using PROC MIXED and PROC NLMIXED: SAS Survey Prodedures and NHANES. SAS Global Forum 2007, April 16-19, 2007, Orlando, Florida. Paper PR03-2007. Available: http://www.lexjansen.com/pharmasug/2007/pr/pr03.pdf.

Interpretive Summary: Understanding how children grow and the impact of growth on obesity is a major challenge. For example, some hypothesize that rapid early growth is associated with obesity later in life. The challenge is how to reduce a set of measurements on a child and summarize it in a useful manner. Intervention data also tends to have measurements at multiple times to evaluate how end points change over the life of the intervention and to evaluate whether the intervention is working. We look at several statistical methods that can be used to summarize this growth, and to determine factors that affect growth.

Technical Abstract: Being able to describe growth appropriately and succinctly is important in many contexts, including biology, epidemiology, and statistics. Various approaches are used varying from differential equations, deterministic modeling, and statistical approaches like regression. Often, with epidemiologic data we want to model growth in the context of demographic variables and other potential mediators. However, growth is non-linear, so the addition of covariates may be problematic. Additionally, measurements may be unevenly spaced, and there may even be missing data, so some form of modeling that will deal with this is needed. We investigate growth models implemented in SAS/STAT using anthropometric data (height, weight, and body mass index,) from the Infant Health and Development Program (IHDP). This program was an eight-site randomized controlled trial testing the efficacy of early intervention to enhance the cognitive, behavioral, and health status of low birth weight, premature infants. These infants are not expected to grow in the same way as normal birth weight infants. Thus, we characterize their growth by gender subgroups in a similar way to the CDC growth curves. We use the control group (n=608) to model the growth pattern. This group received pediatric follow up and referral services only.