| Abstract: | Background/Aims Over 50% of depressed patients fail to remit after an adequate antidepressant (AD) treatment course, and 35% remain symptomatic after two adequate treatment courses. Patients with treatment-resistant depression (TRD) have higher risks of morbidity and mortality, and substantially higher healthcare expenditures. This study aims to develop and validate algorithms to identify patients with TRD in claims databases. Methods We first identified Harvard Pilgrim Health Care members aged 18 years or older who had a diagnosis of depression and new use of selective serotonin reuptake inhibitors or serotonin- norepinephrine reuptake inhibitors (after at least 365 days of no AD use) in 2000-2009. Among these patients, we identified those who received adequate treatment, defined as treatment initiated at or greater than the recommended starting dose based on practice guidelines and taken for at least 8 weeks. We will further identify patients with TRD, i.e., patients who are treated adequately but fail to remit. Although multiple definitions for TRD exist in the guidelines and literature, we will use the emerging consensus of failure to remit after two adequate treatment courses as our definition of TRD. We will consider various markers of treatment resistance such as switching ADs (particularly to those reserved for second-line), adding atypical antipsychotics or other non-AD medications commonly used for depression, and invasive nonpharmacologic intervention. We will validate our algorithms via chart review. Results In preliminary results, 114,002 patients meeting inclusion criteria initiated an AD and 63,882 (56.0%) completed an adequate treatment course. Among these patients, 35,547 (55.6%) continued that treatment, 10,255 (16.1%) stopped treatment, and 5,753 (9.0%) switched to another AD. Among switchers, 3,625 (63.0%) achieved adequate treatment with the second AD. After examining treatment patterns and markers of treatment resistance, we will select the most promising algorithm and validate 300 randomly selected potential cases identified by the algorithm beginning in December. Discussion Claims databases have potential to identify TRD, but algorithms to identify such patients must be developed. Once such an algorithm is validated, these databases can be assessed to answer important questions about the safety and effectiveness of treatments for TRD patients. Primary: Mental Health. |
