Shared and Study-Specific Dietary Patterns: a Novel Approach to Replicability and Validity

• Activity Number: 435 - SPEED: Sports to Fire: Fascinating Applications of Statistics
• Type: Contributed
• Date/Time: Tuesday, July 31, 2018 : 3:05 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Roberta De Vito* and Carlo La Vecchia and Giovanni Parmigiani and Valeria Edefonti
• Companies: and Università degli Studi di Milano and Harvard T.H. Chan School of Public Health / Dana-Farber Cancer Institute and Università degli Studi di Milano
• Keywords: dietary patterns; external reproducibility of dietary patterns; generalizability of dietary patterns; laryngeal cancer; multi-study factor analysis; oral cavity and pharyngeal cancer

Abstract

We adopt the "multi-study factor analysis" approach in order to identify dietary patterns (DPs) across different studies. The goal is 1) to distinctly estimate shared and study-specific DPs across multiple populations, and 2) to validate the DPs by examining their association with cancers of the oral cavity and pharynx (OCP) and larynx. To assess these issues we use individual-level pooled data from 7 case-control studies (3,844 cases and 6,824 controls from Europe and the United States) participating in the International Head and Neck Cancer Epidemiology consortium. We identified 3 DPs that were shared among all studies: the Antioxidant vitamins and fiber DP was inversely associated with risk of OCP and laryngeal cancers, the Fats DP was inversely associated with risk of OCP cancer, whereas the Animal products and cereals and the Fats DPs were positively associated with risk of laryngeal cancer. Each study from the US expressed an additional study-specific DP. We provided a valuable tool to fully understand multi-study replicability in a DP analysis and insight into DP validity.