An Assessment of the Screening Performance of a Single-Item Measure of Depression From the Edmonton Symptom Assessment Scale Among Chronically Ill Hospitalized Patients

作者：Pantilat Steven Z^*; O'Riordan David L; Dibble Suzanne L; Landefeld C Seth

来源：Journal of Pain and Symptom Management, 2012, 43(5): 866-873.

DOI：10.1016/j.jpainsymman.2011.05.022

• Depression
• validity
• Geriatric Depression Scale
• Edmonton Symptom Assessment System
• chronically ill hospitalized elderly patients

Context. Few studies have examined the validity of using a single item from the Edmonton Symptom Assessment Scale (ESAS) for screening for depression.
Objectives. To examine the screening performance of the single-item depression question from the ESAS in chronically ill hospitalized patients.
Methods. A total of 162 chronically ill inpatients aged 65 and older completed a survey after admission that included the well-validated, 15-item Geriatric Depression Scale (GDS-15) and four single-item screening questions for depression based on the ESAS question, using two different time frames ("now" and "in the past 24 hours") and two response categories (a 0-10 numeric rating scale [NRS] and a categorical scale: none, mild, moderate, and severe).
Results. The GDS-15 categorized 20% (n=33) of participants as possibly being depressed with a score >= 6. The NRS for depression "now" achieved the highest level of sensitivity at a cutoff >= 1 (68.8%), and an acceptable level of specificity was obtained at a cutoff of >= 5 (82.2%). For depression "in the past 24 hours," a cutoff of >= 1 achieved a sensitivity of 68.8% and a cutoff of >= 7 a specificity of 80.3%. For the categorical scale, a cutoff of "none" provided the best level of sensitivity for depression "now" (65.6%) and "in the past 24 hours" (81.3%), with an acceptable level of specificity being obtained at >= "mild" (68.8%) and >= "moderate" (68.8%), respectively.
Conclusion. These single-item measures were not effective in screening for probable depression in chronically ill patients regardless of the time frame or the response format used, but a cutoff of >= 5 or "mild" or greater did achieve sufficient specificity to raise clinical suspicion. J Pain Symptom Manage 2012; 43: 866-873.

• 出版日期2012-5