Less favourable trajectories of depressive mood across adolescence into early adulthood are associated with current and later psychopathology, impaired educational attainment and social dysfunction, yet the genetic and environmental risk factors associated with these trajectories are not fully established. Examining what risk factors are associated with different trajectories of depressive mood could help identify the nature of depression symptoms and improve preventative interventions for those at most risk.

To examine how genetic and environmental risk factors differentially associate with trajectories of depression symptoms between ages 10 to 24 years.

In a longitudinal cohort study established in 1990 and currently ongoing (Avon Longitudinal Study of Parents and Children), we used growth mixture modelling to identify trajectories of depression symptoms in 9,394 individuals. We then examined associations between different risk factors and membership in these trajectories. Analysis was conducted between August 2018 and January 2019.

Trajectories were composed from depression symptoms measured using the short mood and feelings questionnaire (SMFQ) across nine occasions between the ages 10 and 24 years. Risk factors included sex, a polygenic risk score (PRS) taken from a recent genome wide association study of depression symptoms, maternal postnatal depression, partner cruelty to the offspring’s mother between ages 2-4 years, childhood anxiety at age 8, being bullied at age 10.

Data on all risk factors, confounders and the outcome were available for 3,525 individuals, including 1,771 (50.2%) females. Trajectories were assessed between 10.65 years and 23.8 years. We identified 5 distinct trajectories of depression symptoms: ‘stable low’ (71.1%), ‘adolescent limited’ (9.2%), ‘childhood limited’ (5.8%), ‘early-adult onset’ (11.1%) and ‘childhood persistent’ (2.8%). Of all the associations between risk factors and trajectories, sex (OR, 6.45; 95% CI, 2.89-14.38), the PRS for depression symptoms (OR, 1.47; 95% CI, 1.1-1.96) and childhood anxiety (OR, 1.3; 95% CI, 1.16-1.45) showed strongest associations with the childhood persistent trajectory, compared to the stable low trajectory. Maternal postnatal depression (OR, 2.39; 95% CI, 1.41-4.07) had the strongest association with the early-adult onset trajectory, whilst partner cruelty to mother (OR, 2.3; 95% CI, 1.36-3.9) had the strongest association with the adolescent limited trajectory. Bullying (OR, 8.08; 95% CI, 4.92-13.26) showed the strongest association with the childhood limited trajectory.

The least favourable trajectories of depression symptoms (childhood persistent and early adult onset) were associated with both genetic and environmental risk factors. However, the two trajectories of limited duration that had resolved by early adulthood (childhood limited and adolescent limited) were not associated with the PRS or maternal postnatal depression (two more ‘genetic’ risk factors). Instead, bullying (a more ‘environmental’ exposure) was strongly associated with both the childhood onset and childhood limited trajectories, suggesting that this risk factor may have a time-specific impact. Our findings suggest that examining genetic and multiple time-specific environmental antecedents could help identify trajectories of varying onset and chronicity. Understanding whether there are different risk profiles (including timing/nature of exposure and prior vulnerabilities) may offer more opportunities to target interventions at certain stages of development.
Original languageEnglish
Article numbere196587
Number of pages14
JournalJAMA Network Open
Issue number6
Early online date28 Jun 2019
Publication statusPublished - 28 Jun 2019

Structured keywords

  • SoE Centre for Multilevel Modelling


  • depression symptoms
  • trajectories
  • growth mixture modelling
  • gene-environment


Dive into the research topics of 'Genetic and Environmental Risk Factors Associated with Trajectories of Depression Symptoms from Adolescence to Young Adulthood'. Together they form a unique fingerprint.

Cite this