Measuring antidepressant prescribing practice in a health care system using administrative data: implications for quality measurement and improvement

BACKGROUND: Up to one in eight Americans experiences an episode of depression that requires treatment in his or her lifetime. The direct and indirect costs associated with major depression are high but may be reduced with appropriate treatment. To decrease the probability of relapse, guidelines specify that treatment with antidepressant medications should continue for at least 4 months after symptom remission and that adequate doses of antidepressants be used. A study was conducted in 1997-1999 to examine how different specifications in the construction of quality of care measures for depression treatment influence conclusions about the adequacy of antidepressant prescribing practices. METHODS: Subjects were all adult members of two United Healthcare plans who each had at least one outpatient or inpatient claim with a diagnosis of depression during the years 1993-1995 and were continuously enrolled for 12 months. Pharmacy claims data were used to construct measures of duration of treatment, dose, and type of antidepressant. The effects of two different definitions of a new episode (4-month versus 9-month clean period) and two different ways of identifying an episode of depression (one visit versus two visits with a code for depression) were examined on conclusions about adequacy of antidepressant prescribing practices (dose and duration). Whether antidepressant type was related to the likelihood that antidepressants were prescribed at therapeutic doses was also examined. RESULTS: Patients with two or more visits with depression diagnosis codes were significantly more likely to receive antidepressants than those with only one visit, and were more likely to receive therapeutic doses at each time period (1-5 months). The duration of the clean period was not related to conclusions about therapeutic dosing. Among persons receiving antidepressants, those receiving selective serotonin reuptake inhibitors (SSRIs) were more likely to receive therapeutic doses and to continue treatment for at least 5 months than were those prescribed other classes of antidepressants. In multivariate analysis, being prescribed an SSRI versus another class of antidepressants was significantly associated with receiving both 1 month (OR = 7.3 [5.7-9.3]) and 5 months (OR = 2.0 [1.6-2.5]) of therapeutic treatment. DISCUSSION: Conclusions regarding the appropriateness of antidepressant prescribing can vary markedly, depending on how the quality measure is specified. Given that administrative data are and will continue to be used for both monitoring and quality improvement purposes in the short run, it is critical that we understand how variations in measurement specifications influence the conclusions that are drawn about treatment of depression in health plans.