Abstracts

Rationale: Depression is one of the most commonly reported psychiatric disorders in persons with epilepsy (PWE). Numerous methods to identify depression in PWE exist, including self-report, short questionnaires and clinical interviews. The most appropriate scale for determining the presence of depression in PWE has yet to be definitively established in a well-characterized large consecutive cohort of PWE using an adequate gold standard for depression diagnosis.Methods: 300 consecutive PWE were recruited from the only epilepsy programme in a large tertiary care center (serving over 1 million persons). Participants completed a questionnaire (demographics, the Patient Health Questionnaire (PHQ)-9 & -2 and the Hospital Anxiety and Depression Scale (HADS)). Within two weeks of their appointment, participants underwent a Structured Clinical Interview for DSM-IV (SCID) diagnostic interview. The PHQ-9 was scored according to the suggested algorithm, which attempts to reflect DSM-IV criteria, and cut-point (10/27) methods and the PHQ-2 (3/6) and HADS (8/21) were scored using cut-point methods. Using the extensively validated SCID as the gold-standard, the diagnostic properties of each scale were examined. Areas under the curve (AUC) were compared for all questionnaires. Results: Of the 300 people presented with the full study consent, 268 agreed to participate (89.3%). 182 PWE completed the SCID, along with the PHQ-9 and HADS (67.9%). 52.2 % of participants were female, with a mean age of 40.5 years. The prevalence of current and past depression according to the SCID was 16.5% and 30.8% respectively. 28.0% of respondents reported clinically significant symptoms of anxiety according to the HADS. The PHQ-9 cut-point method resulted in a 24.2% prevalence of depression with a sensitivity (Se) compared to the SCID of 70.0%, specificity (Sp) 84.9%, positive predictive value (PPV) 47.7% and negative predictive value (NPV) 93.5%. The PHQ-9 algorithm scoring method resulted in a prevalence of depression of 3.85%, a Se of 20.0%, Sp of 99.3%, PPV of 85.7%, and NPV of 86.3%. Using the PHQ-2, the prevalence of depression was 17.6%, with a Se of 40.0%, Sp of 86.8%, PPV of 37.5% and NPV of 88.0%. The HADS resulted in a 17.0% prevalence of depression, a Se of 60.0%, Sp of 91.4%, PPV of 58.1% and NPV of 92.1%. A cut-point of 7 or 9 did not significantly affect the diagnostic accuracy of the HADS. The PHQ-2 had a significantly smaller AUC (63.4%) than both the PHQ-9 cut-point (77.4%) and the HADS (75.7%), p = .005 and .032, respectively. The PHQ-9 algorithm had a significantly smaller AUC (59.7%) than the PHQ-9 cut-point and the HADS, p = .0003 and .0023, respectively. The AUC did not differ between the PHQ-9 cut-point and HADS or between the PHQ-2 and the PHQ-9 algorithm. Conclusions: Four scales were compared to a gold-standard to determine their diagnostic accuracy in measuring depression in PWE. The PHQ-9, using the cut-point scoring method, produced the best sensitivity and specificity, as evidenced by the largest AUC. The PHQ-9 appears to be the ideal choice amongst the examined questionnaires.