Modeling Risk Factor Trajectories When Measurement Tools Change Sequentially During Follow-up in Cohort Studies: Application to Dietary Habits in Prodromal Dementia.

4 years ago

Modeling Risk Factor Trajectories When Measurement Tools Change Sequentially During Follow-up in Cohort Studies: Application to Dietary Habits in Prodromal Dementia.

Samieri, Wagner, Dartigues, Proust-Lima

Modeling risk factor trajectories is critical to understand the natural history of diseases; yet, the measurement tools used to assess risk factors often evolve during follow-up in cohorts and such change prevents longitudinal analyses using standard models. The authors propose to address this issue with a Latent Process Model. Trajectories of average intakes of five food families (Fish, Meat, Fruits, Vegetables, Carbohydrate-rich foods) were described in prodromal dementia during the 10 years prior to diagnosis of cases and compared to those of controls, using a case-control sample nested within the Three-City Study, Bordeaux, France (1999-2012). Food intakes were measured by two to three different subquestionnaires across five repeated food frequency questionnaires. The sample comprised 205 incident cases and 410 controls matched for age, gender, education and number of repeated food frequency questionnaires. Intakes of fish, fruits and vegetables decreased at the approach of diagnosis among cases, suggesting reverse causation. This study demonstrates that Latent Process Model approach constitutes a powerful framework for modeling risk factor trajectories, even when measurement tools sequentially change over time. Coupled with a case-control approach to contrast trajectories in prodromal disease versus healthy status, it helps understand the dynamic, causal relationships between risk factors and diseases.

Publisher URL: http://doi.org/10.1093/aje/kwx293

DOI: 10.1093/aje/kwx293