The included studies were carried out among a wide spectrum of patients mostly in general practice settings, with the exception of 1 in a nursing home and 1 receiving home care. Two studies specifically included patients with dementia. Nine different instruments were used; most had 20 or fewer questions and were relatively easy to administer.
Overall test performance in detecting major depression was similarly favorable among the instruments, with sensitivities ranging from 67% to 100% and specificities ranging from 53% to 98%. All but 2 studies5,6 reported sensitivity and specificity based on optimal cutpoints determined by post-hoc receiving-operating characteristic (ROC) curve analyses, possibly exaggerating test performance in comparison with the studies testing predetermined cutpoints.
Five studies6,11,19,22,23 explicitly stated that interviewers performing the criterion standard exam were blinded to the results of the screening test; the remainder did not report on blinding, although in most cases blinding was implied by the use of a second “independent” rater.
Geriatric Depression Scale. The GDS, the Center for Epidemiologic Studies Depression scale (CES-D), and the SelfCARE(D) were the most-evaluated screening instruments. The GDS has both a 30- and 15-item version and was designed in a yes/no format for self- or caregiver administration, making it easy to use. It minimizes questions about somatic and vegetative symptoms, which can overlap with symptoms of concurrent medical illness.
The GDS has been validated repeatedly in psychiatric settings.23-27 Nine studies5-10,12 evaluated its use in primary care elderly, most using the 15-item version and a cutpoint of 3 to 5. Sensitivity and specificity ranged from 79%–100% and 67%–80%, respectively.
Center for Epidemiological Studies Depression Scale. The CES-D can be self-administered. It lists 20 statements addressing depressive symptoms over the last week, asking the participant to rank the frequency of these feelings from “rarely” to “most of the time.” Its psychometric properties have been consistently strong in younger adults in the community.
In the 5 studies13-16 that evaluated this instrument, cutpoints varied from 9 to 21. The resultant sensitivities were 75%–93%, with specificities ranging from 73%–87%. One study16 also specifically evaluated the performance of the CES-D in mildly demented subjects with an average Mini-Mental State Examination (MMSE) of 19, and showed similar test characteristics to the patients without dementia. This instrument was perceived as generally easy to administer, except in a nursing-home population where the questions had to be repeated multiple times.
Papassotiropoulos et al17 used the CES-D and the General Health Questionnaire (GHQ) to identify subthreshold depression in a community sample in Greece. They defined subthreshold depression as fewer than 5 depressive symptoms in a 2-week period; brief, monthly depressive symptoms not occurring for a 2-week duration; and, any significant single depressive symptom not specified by duration or frequency. Accuracy was poor for delineating these syndromes, with sensitivities below 50% and specificities of 75% and 72%, respectively.
Lyness and colleagues15 used the CES-D, as well as the GDS-15, to identify minor depression in their cohort. They defined minor depression as having sad mood or loss of interest and at least 2, but fewer than 5, additional depressive symptoms within a 2-week period. The CES-D revealed a sensitivity of 40% and specificity of 82% for detecting minor depression, while the GDS-15 had a sensitivity and specificity of 70% and 80%, respectively.
SelfCARE(D). The SelfCARE(D) is a self-administered instrument that requests responses to 12 items on a Likert scale, reflecting depressive symptoms over the last month. It was derived from a larger, previously validated instrument used in England.18
In 1 of 3 included studies, Bird and colleagues18 reported the original results in a 1987 outpatient sample, showing a sensitivity of 77% and specificity of 98%, with a cutpoint of 5. Since then it has been validated again in general practice and in home care.19,20 Both studies revealed sensitivities in the 90% range, but the specificity in home care was 53% vs 86% in general practice.
Caribbean Culture–Specific Screen. In an effort to address the potential cultural limitations of common instruments, Rait and colleagues11 tested the Caribbean Culture–Specific Screen (CCSS) in the growing contingent of Caribbeans of African descent in the United Kingdom. They found that it performed well, but not better than the Brief Assessment Schedule Depression Cards or the GDS-15. Each had a sensitivity of 92%, with specificities ranging from 71%–84%.
Similarly, Abas et al12 tested the CCSS and the GDS-15 in an African-Caribbean population, reporting sensitivities of 82% for both instruments, and specificities of 68% for the CCSS and 82% for the GDS-15.
Cornell Scale for Depression in Dementia. Dementia poses barriers to effective screening for depression given the obvious limitations in self report due to cognitive impairment. The Cornell Scale for Depression in Dementia (CSDD) was specifically designed for this population and calls for the clinician to use both patient and caregiver information to complete the screen.