Depict the Trajectory of Depressive Symptoms using Quantile Regression

• In Event: Poster Session 05
  In Poster Session: PS 05 Topic - Health, Growth, Injury

Abstract

A better understanding of the developmental trajectory of depressive symptoms over time can provide information for early detection and intervention. Several studies have depicted the trajectory of depressive symptoms using data with various age groups (Contoyannis & Li, 2017; Musliner, Munk-Olsen, Eaton, & Zandi, 2016; Shore, Toumbourou, Lewis, & Kremer, 2018). These results generally showed that the mean level of depressive symptoms increased from late childhood to adolescence and gradually tapered off around late adolescence and young adulthood. Nevertheless, significant heterogeneity, including the peak of depressive symptoms and the timing of having the greatest rate of change, has also been identified. As suggested by Kwong et al. (2019), the lack of consensus in the shape of trajectory can be due to having short-term follow-up information and wider time gaps between measurements. In addition to the study design, the analysis methods that focusing on the mean trajectory cannot appropriately capture the diversity of developmental patterns related to different severity of depressive symptoms. By estimating the trajectories at various quantiles, rather than the mean, the quantile regression has the strength to address diversity. This method also allows to explore specific factors that have strong effects on the developmental trajectory for each subgroup. The goal of the present study is to use quantile regression modeling to depict the trajectory of depressive symptoms over time. The results can bridge the research gap by addressing the diversity of the development of depressive symptoms over time.
Data from 1,885 individuals participated in the National Longitudinal Survey of Youth: Children and Young Adults study form 1994-2016 was employed in the preset study. The present study focused on individuals who were White (51% male) with follow-up data between ages 15 and 30. Depressive symptoms were measured by seven-item CES-D scale. Background information, including gender, birth cohort, education, and mother’s characteristics, was included to address the heterogeneity of the developmental pattern. Results from multilevel modeling (focusing on the mean) and quantile regression (.25, .50, and .75 quantiles) analyses were presented.
From age 15 to 30, the depicted trajectories among male demonstrated a significant inverted-U shape, whereas those for females were flat over time. There were also significant cohort differences among male. Factors that significantly influence the development of depressive symptoms varied across mean and quantile trajectories. The mean trajectory for female showed that mother’s early depressive symptoms significantly increased individuals’ depressive symptoms at age 15. In contrast, results from the quantile regression showed that the impact of mother’s depressive symptoms was not only significant but also greater in magnitude for the .75 quantile trajectory, compared to those for the .25 and .50 quantiles. Furthermore, whereas the mean trajectory for female found that receiving an associate or a college degree reduced depressive symptoms, this effect was only significant in the .50 quantile trajectory. Similar differences can also be found in the trajectories for male. The results suggest the importance to further study the diversity in depressive symptom trajectories and identify significant factors among subgroups using quantile regression.
 

Authors

• Yi-Ju Eva Chen, University of Northern Iowa
  Presenting Author

• Eli Yi-Liang Tung, National University of Singapore
  Non-Presenting Author