Influence of Highly Correlated Cross-Basis Functions in the Distributed Lag Nonlinear Model

• Activity Number: 534 - Contributed Poster Presentations: Section on Statistics in Epidemiology
• Type: Contributed
• Date/Time: Wednesday, August 1, 2018 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Lung-Chang Chien* and Yunqi Vicky Liao and Michael Swartz and Kristina Whitworth
• Companies: University of Nevada, Las Vegas and University of Texas Health Science Center (UTHealth) School of Public Health at Houston and University of Texas Health Science Center (UTHealth) School of Public Health at Houston and University of Texas Health Science Center (UTHealth) School of Public Health at Houston
• Keywords: Distributed lag non-linear model; cross-basis function; lag; RV-coefficient

Abstract

The distributed lag non-linear model (DLNM) is frequently used to explore delayed effects of environmental exposures on human health. The DLNM represents delayed non-linear exposure-outcome relationships through cross-basis functions. Although investigators have used multiple cross-basis functions in a single DLNM, the correlation between those cross-basis functions has not been evaluated, and the impact of using multiple cross-basis functions in a DLNM is unknown. A simulation analysis of 500 data sets, each with 1,000 observations of X1, X2, and Y, was conducted in this study. RV-coefficients between the cross-basis matrices of X1 and X2 were calculated to quantify the correlation between the cross-basis functions of X1 and X2. Six quasi-Poisson DLNMs with a cross-basis function of X1 using lag = 10 and a cross-basis function of X2 using lag = (10, 11, 12, 13, 14, 15) were fitted for each set. The largest averaged mean square errors and the largest averaged standard errors of X1 estimates both occurred in the model where the lag periods for X1 and X2 were the same. To sum up, using two highly correlated cross-basis functions in a single DLNM results in more biases in estimates.