Date of Award


Document Type

Open Access Thesis

Degree Name

Medical Doctor (MD)

First Advisor

Thomas Gill, MD


Antidepressant medications are commonly prescribed for older adults with depressive symptoms who may not have a major depressive disorder. Yet, the effect of antidepressants on depressive symptoms over time in this population is largely unknown. We sought to determine whether the use of antidepressant medications is associated with a reduction in the severity of depressive symptoms over time. Participants included 754 community-dwelling adults, aged 70+ years, who were followed at 18-month intervals for 90 months. Depressive symptoms were assessed using the 11-item CESD scale, with a higher score indicating worse depressive symptoms. A linear mixed effects model, adjusted for demographic features, number of chronic conditions, cognitive status, and physical frailty, was used to evaluate the effect of antidepressant use on change in depressive symptoms score over time. In addition, among persons with clinically significant depressive symptoms (i.e., CESD score >=20), we evaluated whether antidepressant use was associated with a transition to a non-depressed state (CESD score <20) using a Generalized Estimating Equations (GEE) model. At baseline, participants taking an antidepressant (n=75) had higher mean CESD scores than those not taking an antidepressant (15.1+ 9.2 vs. 8.5+8.3; p<0.001) and were more likely to be female (p<0.001). Average unadjusted CESD change scores ranged from -3.4 to 1.7 and 0.4 to 1.5 among those taking, and not taking, an antidepressant, respectively (for the different 18-month intervals). Adjusted CESD scores worsened, on average, for participants taking an antidepressant as compared with those not taking an antidepressant. These differences were statistically significant between baseline to 18 months (p=0.03), 36 to 54 months (p=0.02) and 72 to 90 months (p=0.01). The longitudinal findings indicated that CESD scores worsened by 2.2 points, on average, among participants taking an antidepressant as compared with those not taking an antidepressant, although this difference was not statistically significant (p=0.14). Among participants with clinically significant depressive symptoms, use of antidepressants was not associated with transitioning to a non-depressed state (OR=0.85, 95% CI 0.5-1.4). Our findings raise concerns about the effectiveness of antidepressant medications, as prescribed in clinical practice. Additional research is needed to better understand the real world use and benefit of antidepressants among older adults.