DIAGNOSTIC ACCURACY OF THE SLU AMSAD SCALE FOR DEPRESSION IN NON-DEMENTED ELDERLY

Abstract

Introduction Despite its significant impact, geriatric depression remains both underdiagnosed and undertreated ( 1 ). In the absence of specific diagnostic criteria, the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria currently represent the gold standard diagnostic tool for late-life depression. Given the fact that most elderly patients receive mental health care in primary care settings ( 2 ), there is an increasing need to develop practical diagnostic tools to be used by primary care practitioners. The Saint Louis University (SLU) AMSAD tool is a brief (5-item) questionnaire that was recently developed to screen for late-life depression. The 5 items reference appetite, mood, sleep, activity and death ideation using simple language and scaling. Previous research has supported its validity and reliability in cognitively intact older adults, in relation to the Geriatric Depression Scale (GDS)-15 and the Montgomery-Asberg Depression Rating Scale (MADRS) ( 3 ). However, few studies have examined the diagnostic accuracy of depression screening instruments in older adults. The objective of this study was to evaluate the accuracy and reliability of the SLU AMSAD regarding diagnosis of major depressive disorder (MDD) per DSM-5 criteria in a sample of older adults without major neurocognitive disorder. Methods A convenience sample of 50 patients, ≥ 65 years of age was enrolled through our specialized geriatric psychiatry outpatient clinic. Patients with a clinical suspicion/diagnosis of major neurocognitive disorder as reflected by a Saint Louis University Mental Status (SLUMS) score of Results Descriptive Analysis The sample had a mean (SD) age of 73.5 (7.1) years; 58% (n = 28) were women, 88% (n = 44) were Caucasian, and 86% (n = 43) had at least a high school level of education. The mean SLUMS score was 25.9 (2.9). For the depression measures, mean scores were 6.5 (4.5) for the GDS-15; 15.6 (11.4) for the MADRS; and 4.9 (3.2) for the SLU AMSAD. The SLU AMSAD evidenced adequate internal consistency reliability (alpha = .77). 30% (n = 15) of patients met DSM-5 criteria for MDD diagnosis. Diagnostic Accuracy and Correlations In the initial ROC analysis, total scores (continuous variables) for the GDS-15, MADRS, and SLU AMSAD were evaluated in relation to the DSM-5 designation of No MDD vs. MDD. AUC values were uniformly high (≥ .93), with correspondingly high levels of sensitivity (.93) and specificity (≥ .80). Optimal cut-offs were 9+ for GDS-15, 18+ for MADRS, and 7+ for SLU AMSAD. The SLU AMSAD was equivalent to GDS-15 and superior to MADRS in these analyses. ROC analyses using established categories for the depression measures were then executed. When score results were categorized as No vs. Any (Mild, Moderate, or Severe) Depression, AUC values ranged from .69-.79, with perfect sensitivity (1.0) but modest specificity (range = .37-.57). When score results were categorized as No-Mild vs. Moderate-Severe Depression, AUC values were again high (range = .82-.89), with adequate levels of sensitivity (.87-.93) and specificity (.71-.86). Moreover, total score correlations were nearly uniform across the depression measures (.69-.70). Scores categorized as No vs. Any (Mild, Moderate, or Severe) Depression yielded correlations with DSM-5 MDD that ranged from .39-.53; scores categorized as No-Mild vs. Moderate-Severe Depression yielded correlations ranging from .59-.75. When scores were categorized according to the optimal cut-off value, SLU AMSAD and GDS-15 had correlations of .75 with DSM-5 MDD, while the MADRS correlation was .68. Conclusions The results indicated strong diagnostic accuracy for all three depression scales in relation to the DSM-5, by ROC and correlational analyses. The SLU AMSAD performed at least as well as the GDS-15 (three times the length of the AMSAD) and slightly better than the much longer and complex MADRS. However, the clinical superiority of the SLU AMSAD for depression screening in the elderly, in relation to the other measures, is supported by the fact that it encompasses only 5 simply-worded, simply-scaled items. Thus, the SLU AMSAD emerges as a potentially strong candidate for depression screening in busy clinical settings, including primary care but also in the emergency department. This research was funded by No funding was received for this research.