Integrating social science and behavioral genetics: testing the origin of socioeconomic disparities in depression using a genetically informed design

Abstract

Objectives: We tested 3 hypotheses-social causation, social drift, and common cause-regarding the origin of socioeconomic disparities in major depression and determined whether the relationship between socioeconomic status (SES) and major depression varied by genetic liability for major depression.

Methods: Data were from a sample of female twins in the baseline Virginia Adult Twin Study of Psychiatric and Substance Use Disorders interviewed between 1987 and 1989 (n = 2153). We used logistic regression and structural equation twin models to evaluate these 3 hypotheses.

Results: Consistent with the social causation hypothesis, education (odds ratio [OR] = 0.78; 95% confidence interval [CI] = 0.66, 0.93; P < .01) and income (OR = 0.93; 95% CI = 0.89, 0.98; P < .01) were significantly related to past-year major depression. Upward social mobility was associated with lower risk of depression. There was no evidence that childhood SES was related to development of major depression (OR = 0.98; 95% CI = 0.89, 1.09; P > .1). Consistent with a common genetic cause, there was a negative correlation between the genetic components of major depression and education (r(2) = -0.22). Co-twin control analyses indicated a protective effect of education and income on major depression even after accounting for genetic liability.

Conclusions: This study utilized a genetically informed design to address how social position relates to major depression. Results generally supported the social causation model.

FIGURE 1—

Predicted probability of past year major depression by difference between participant’s own education and education of main support in childhood: Baseline Virginia Adult Twin Study of Psychiatric and Substance Use Disorders, Virginia, 1987–1989. Note. DZ = dizygotic; MZ = monozygotic. Data were adjusted for age and zygosity, stratified by major depression (MD) status of co-twin. Whiskers indicate 95% confidence intervals.

FIGURE 2—

Bivariate twin model of past year major depression and educational attainment in adulthood: Baseline Virginia Adult Twin Study of Psychiatric and Substance Use Disorders, Virginia, 1987–1989. Note. A = Additive genetic effects; AIC = Akaike Information Criteria; C = Common environmental effects shared by twins that make them more similar; E = Unique environmental effects that make twins different; MD = major depression. AIC_{BEST MODEL} = −1759.4 (df = 4296) versus AIC_{FULL MODEL} = −1756.7 (df = 4293). Values are standardized path coefficients (95% confidence interval).

FIGURE 3—

Predicted probability of past year major depression by (a) education of main support in childhood and (b) adulthood educational attainment: Virginia, 1987–1989. Note. DZ = dizygotic; MZ = monozygotic. Data were adjusted for age and zygosity, stratified by major depression (MD) status of co-twin. For 3b, predicted probability and SE were combined. Whiskers indicate 95% confidence intervals.