Age as Substitute of Age at Onset in Mood Disorders Studies

Age at onset is an important variable in the study of mood disorders. Different ages at onset may identify distinct subtypes [1]. However, recall bias can reduce its value. This is not a problem with age, per se. The aim of the present study was to determine whether patient's present age could be used instead of age at onset to distinguish mood disorder subtypes.

Three hundred and ninety-nine consecutive DSM-IV bipolar II and unipolar major depressive episode (MDE) outpatients were interviewed by the author with the Structured Clinical Interview for DSM-IV, the Montgomery Asberg Depression Rating Scale (MADRS), and the Global Assessment of Funtioning Scale (GAF). The variables studied, derived from studies comparing bipolar with unipolar disorder [2–7], were age, age at onset of the first MDE, gender, number of previous MDEs, chronicity MDE/MDE without full inter-episode recovery lasting more than 2 years), atypical and psychotic features, axis I comorbidity and MDE severity. Age and age at onset were correlated with linear regression and non-parametric Spearman's test. Differences between age and age at onset were tested with t-test and non-parametric Mann–Whitney test. Linear regressions between both age and age at onset, and the study variables were calculated (stata 5 statistical software; stata College Station, TX, USA). All p-values were two-tailed, and the significance level was 0.05.

Correlation between present age and age at onset in the whole sample was r = 0.65, p = 0.0000; rho = 0.59, p = 0.0000. Correlation between age and age at onset in bipolar II patients (n = 199) was r = 0.62, p = 0.0000; rho = 0.54, p = 0.0000. Correlation between age and age at onset in unipolar patients (n = 200) was r = 0.64, p = 0.0000; rho = 0.61, p = 0.0000. The age difference [mean 39.7 (SD = 13.3) vs 46.5 (SD = 16.9) years, median 39 vs 44 years] between bipolar II and unipolar patients was significant (t = 4.4, p = 0.0000; z = 3.8, p = 0.0001). The age at onset difference [mean 25.5 (SD = 11.4) vs 31.1 (SD = 15.2) years, median 23 vs 26.5 years] between bipolar II and unipolar patients was significant (t = 4.1, p = 0.0000; z = 3.6, p = 0.0003). Linear regression between both age and age at onset, and the study variables gender, duration of illness, recurrences, psychotic features, chronicity, MADRS, GAF, bipolar II, unipolar, comorbidity, atypical features, gave the following results respectively: r = 0.03, p = 0.4890, and r = −0.02, p = 0.6837; r = 0.53, p = 0.0000, and r = −0.28, p = 0.0000; r = 0.19, p = 0.0001, and r = − 0.25, p = 0.0000; r = 0.09, p = 0.0575, and r = 0.07, p = 0.1277; r = 0.16, p = 0.0009, and r = −0.05, p = 0.2569; r = 0.08, p = 0.0744, and r = 0.03, p = 0.4753; r = 0.00, p = 0.9803, and r = 0.00, p = 0.9599; r = −0.21, p = 0.0000, and r = −0.20, p = 0.0000; r = 0.21, p = 0.0000, and r = 0.20, p = 0.0000; r = −0.33, p = 0.0000, and r = −0.20, p = 0.0000; r = −0.20, p = 0.0000, and r = −0.19, p = 0.0001.

Conclusions drawn from this study are that the strong correlations between age and age at onset, the significant difference in age corresponding to a significant difference in age at onset, the same significant and non-significant correlations (often in the same direction) between both age and age at onset, and many of the study variables, suggest that present age could substitute for age at onset in some comparisons between bipolar II and unipolar disorder. Age does not have the problem of recall bias of age at onset.