The relationship of antidepressant prescribing concentration to treatment duration and cost.

BACKGROUND: Widely accepted treatment guidelines and performance measures encourage patients to stay on antidepressant medication beyond the acute phase of treatment in order to achieve full remission and reduce risk of relapse. However, many patients discontinue antidepressant medication treatment prematurely for various reasons, including side-effects or nonresponse to the initial medication prescribed. Customization of medications to differing patient profiles could potentially improve medication treatment duration, but for many diseases physicians tend to concentrate on a limited subset of available medications. Little is known about the effects of concentration in prescribing on medication treatment duration and expenditures. AIMS OF THE STUDY: To determine the extent to which prescribing for treatment of depression is concentrated, using data from a privately insured population. To evaluate the relationship between prescribing concentration and subsequent duration of medication treatment, expenditure on medications, and the number of distinct medications used. STUDY POPULATION: Individuals receiving antidepressant treatment paid for by a large private managed behavioral health organization, in the US.
METHODS: The study uses psychotropic pharmacy claims data for 2003-06 for plan members who received a depression diagnosis and had an antidepressant claim. The resulting sample includes 9,017 patients seen by 543 prescribers. For each prescriber, we compute prescribing concentration, using the Herfindahl index and the share for the three most-used medications. Treatment expenditure is computed as the sum of payments by plan and by patients. Regression analysis is used to identify the association of prescribing concentration with medication treatment duration, expenditures and other utilization measures.
RESULTS: For these physicians, the mean share of the physician's total antidepressant prescribing accounted for by their three most-used regimens was 72%. The mean value of the Herfindahl index was 0.27. Over the 180-day follow-up period, the average patient had 103 days covered by antidepressant prescriptions, resulting in mean expenditures of $286, or $2.25 per day of medication supplied. Regression analysis indicates that higher concentration in a physician's prescribing was associated with fewer days of antidepressant coverage, lower medication expenditures, and subsequent use of fewer distinct medications. DISCUSSION: Higher concentration in prescribing is associated with shorter observed duration of medication treatment and lower expenditures on medications. The lower expenditures appear to be due to earlier discontinuation and fewer different medications, not to a lower cost per day supplied. Limitations of this study include lack of data on medical visits or on reasons for medication discontinuation, as the study is based on pharmacy claims data, not medical claims or surveys. In addition, it is not known whether the patient's antidepressant use represents a new episode. Finally, lack of randomization implies that the associations identified may not be causal. IMPLICATIONS FOR HEALTH CARE PROVISION AND USE: Concentration of physicians on certain medications may run counter to the increasing calls for customization of medication selection. IMPLICATIONS FOR HEALTH POLICY: Insurer policies which limit physicians' choice of medications may be lowering expenditures in part by reducing patients' medication treatment duration. IMPLICATIONS FOR FURTHER RESEARCH: Additional studies are needed to understand what mechanisms may link concentration in prescribing to medication treatment duration and expenditures.