The Test-Retest Reliability and Familial Concordance of Individual Symptoms of Major Depressive Disorder
thesisposted on 01.05.2020, 00:00 by Ariela J Kaiser
Major Depressive Disorder (MDD) is a heterogenous syndrome, which likely contributes to the diagnosis’s questionable test-test reliability (Regier et al., 2013). MDD consists of nine core symptoms, most of which are unaggregated (e.g., sleep disturbance includes hypersomnia or insomnia). Individual depression symptoms may therefore exhibit different psychometric properties. This study examined two psychometric properties of the individual symptoms of MDD: (1) test-retest reliability and (2) familialness (i.e., whether they run in families). These two psychometric properties were examined for (1) the categorical diagnosis of MDD, (2) aggregated symptoms, and (3) unaggregated symptoms. Lifetime depression symptoms were measured in 504 young adults (237 sibling pairs) using an adapted version of the Structured Clinical Interview for DSM-5 (SCID; Shankman et al., 2018). Fifty-one people completed a second SCID within three weeks of their first SCID (M = 8.5 days, SD = 4.31). The test-retest reliability and familialness of each lifetime MDD symptom was evaluated using Cohen’s Kappa and established conventions for agreement (Cohen, 1960). The lifetime diagnosis of MDD had substantial test-retest reliability (k = .68). The test-retest reliabilities for aggregated and unaggregated symptoms fell into the moderate to substantial range (k’s ranged .52-.72). The lifetime diagnosis of MDD had fair familial concordance (k = .21). The familial concordance for the aggregated symptoms were highest for anhedonia (k = .28) and depressed mood (k = .21). At the unaggregated symptom level, the symptoms that comprise anhedonia—loss of interest (k = .21) and loss of pleasure (k = .25)—had the highest familial concordance and were in the fair agreement range. Given the increasing focus on the differential validity of individual MDD symptoms, our findings help illuminate whether interview-based assessments of individual symptoms reach an adequate level of reliability and validity.