Validation of short screening tests for depression and cognitive impairment in older medically ill inpatients.
Goring H, Baldwin R, Marriott A, Pratt H and Roberts C;
Commented by , 16 Jun 2004
Background
Both depression and cognitive impairment are very common in elderly medically ill inpatients. There is a need for quick, easy to use screening tools to become widely used in both hospital and community settings.
Objective
To investigate the criterion validity of the 4 item Geriatric Depression Scale (GDS4) and the 6 item Orientation-Memory-Concentration (OMC) test on medical wards.
Method
Participants were over 65s drawn from 4 acute medical wards in 1 hospital in Manchester, UK. Reasons for exclusion included inability to complete assessments due to medical instability or profound sensory loss; risk of self harm; or discharge within 3 days of admission.
Consenting patients completed the GDS4 and OMC 3-5 days after admission. Those who scored greater than the recommended cut-offs (1/2 for GDS4 and 10/11 for OMC) then completed the 30 item GDS and the Standardised Mini Mental State Examination (SMMSE), supervised by a liaison nurse. These longer instruments were administered within 48 hours of the initial screen.
Data was analysed using a receiver operating characteristic (ROC) curve, a graph which indicates the sensitivity and specificity of a test for every possible cut-off value. A Spearman correlation coefficient was calculated as a measure of association between the GDS4 and GDS30; and the OMC and the SMMSE. Cronbach’s alpha coefficient was used as a measure of internal consistency/reliability.
Results
486 patients were screened, 136 scored above cut-off on the GDS4, 183 on the OMC; 55 (11 %) scored above cut-off on both screens. Thus, a total of 265 were eligible to proceed on both screens (54.5 % of total sample), but only 153 of these (58 %) were consented to proceed to the validation stage.
GDS4 – using a cut-off of 1/2, 106/139 scored consistently on the 2 scales with 71/139 true positives and 35/139 true negatives. Sensitivity and specificity were 78 % and 75 % respectively. Using a cut-off of 0/1, sensitivity was higher, but specificity lower.
OMC – the cut-off of 10/11 gave optimal performance, with 116/136 participants in agreement with the SMMSE, and 86 % sensitivity and 87 % specificity.
Discussion
Both the OMC and GDS4 performed well as screening tools for older medical inpatients. Both have been validated in a variety of settings, though have not been adopted widely in primary care or nursing homes.
The OMC performed particularly well. As with most depression screening tools, there was a trade-off between sensitivity and specificity for different cut-offs for the GDS4. These screening tools could be used by clinical staff for routine screening for depression, dementia and delirium on medical wards, conditions that are acknowledged to be underdiagnosed and undertreated.
Screening of medically ill older people for psychiatric morbidity is a complex issue, and this paper highlights some of the difficulties. Inevitably, many of the high risk patients refuse or are unable to complete the screening rating scales, and it is not clear how the high numbers of false positives should be managed in clinical practice.
For older people with complex interacting medical, social and psychological problems, there may be no alternative to a skilled psychiatric assessment to delineate and address their difficulties.
However, if short screening tests are to be used, the GDS4 and OMC are probably the best available with current knowledge. Details of the GDS4 and OMC, and how they are scored, are given in an appendix of the article.