Dominic Hodgkin, Maureen T. Stewart, Elizabeth L. Merrick, Ye Zhang Pogue, Noreen A. Reilly-Harrington, Louisa G. Sylvia, Thilo Deckersbach, Andrew A. Nierenberg

Successful medication management for bipolar disorder requires clinicians to monitor and adjust regimens as needed, to achieve maximum effectiveness and patient adherence. This study aims to measure the prevalence of indications for medication adjustment at visits for bipolar disorder treatment; the frequency with which physicians recommend medication adjustments; and how strongly the indications predict the adjustments. Data included 3,094 visits for 457 patients in Bipolar CHOICE, a comparative effectiveness study that compared treatment with lithium versus quetiapine. A set of indications for adjustment was matched to reports of whether the physician recommended a medication adjustment at that visit, and what type. Associations between indication and adjustment were examined using bivariate tests and hierarchical logistic mixed effects models. Medication adjustment was recommended at 63% of the visits where one of the indications was present, and at 53% of all visits. In multivariable analyses, adjustment was more likely to be recommended if there was an indication of non-response or side effects, for patients who started on quetiapine rather than lithium, or for patients who were female, married, employed or more educated. The study's cross-sectional design implies that observed associations could result from confounding variables. Also, the CHOICE trial placed certain restrictions on physicians' medication choices, although this is not likely to have resulted in major alterations of prescribing patterns. Clinical inertia may help explain the lack of any adjustment recommendation at 37% of the visits where one of the indications was present. Other explanations could also apply, such as watchful waiting.