Case-finding algorithm for post-stroke depression in the veterans health administration.

OBJECTIVES

Post-stroke depression (PSD) is prevalent, often undiagnosed, and undertreated. The accuracy of detecting patients with post-stroke depression in administrative databases has not been examined.The objective was to validate a case-finding algorithm for post-stroke depression (PSD) among veteran stroke survivors.

METHODS

We conducted a retrospective cohort study of veterans admitted to two local VHA facilities for an inpatient episode of care for acute ischemic stroke. Our cohort included all patients from two medical centers who were identified in the fiscal year (FY) 2001 VHA inpatient database using high specificity stroke ICD-9 codes. FY 2002 VHA and Medicare inpatient, outpatient, and pharmacy data were used to examine the patients' 12-month PSD status by using ICD-9 depression codes and antidepressant use. We assessed our accuracy about patients' PSD from the administrative databases through standardized chart reviews.

RESULTS

Of our 185 subject cohort, 50 (27%) were identified as having PSD. The most sensitive case-finding algorithm for PSD included having an ICD-9 code diagnosis for depression or receiving a prescription for an approved-dosage of antidepressant medication. However, the algorithm of receiving an ICD-9 code for primary or secondary diagnoses of depression revealed the largest positive predictive value.

CONCLUSIONS

A case-finding algorithm using outpatient ICD-9 codes or medication was the most sensitive in identifying cases of PSD. The use of ICD-9 codes alone may be adequate for characterizing a cohort with PSD. Intention for use should be considered when choosing an algorithm to detect PSD.