Background: The 8-item "Scale for Atypical Symptoms" (SAS) and its structured interview, the SIGH-SAD, have been developed to assess atypical symptoms of depression in winter depression. Although they are commonly used, no validation study has yet been conducted.
Methods: 270 consecutive depressed inpatients were assessed prospectively. Pearson's correlation coefficients between fulfilment of Liebowitz criteria for atypical depression and both the SAS score and the atypical balance [ratio of the AS score to the total score on the Hamilton Depression Rating Scale 29-item (HDRS-29)] were calculated. The SAS was evaluated against Liebowitz criteria using binary logistic regression. A ROC curve was performed with the atypical balance against the fulfilment of Liebowitz criteria.
Results: 18.5% of patients met the criteria for atypical depression. The presence of an atypical depression was significantly correlated with both the atypical score (r=0.42) and the atypical balance (r=0.51). The logistic regression showed that a higher score on the SAS, the absence of a somatic syndrome (ICD-10) and a lower HDRS-21 score were independent predictors of an atypical depression while age, gender and bipolarity were not. The ROC curve showed that an atypical balance of 29% was the optimal threshold for the diagnosis of atypical depression (sensitivity=0.86, specificity=0.79).
Limitation: Patients with bipolar I and II were not distinguished.
Conclusion: Atypical depression is relatively frequent in hospitalised patients. The concurrent validity of the French version of the SAS and its structured interview, the SIGH-SAD is satisfactory.