Inability of Primary Care Providers to Predict Medication Fulfillment of New Prescriptions

Background/Aims: Physician prediction of patient medication adherence to chronic therapy is unreliable, but the accuracy of physician predictions is largely unstudied for new prescriptions. Our aim was to determine if provider perception of the likelihood a patient will pick up a medication is an accurate predictor of primary medication nonadherence. Methods: We conducted a prospective cohort study as part of a randomized clinical trial. Providers at 24 primary care and family medicine Geisinger clinics were asked to complete a “best practice alert” (BPA) within the electronic health record when placing an order for a new antihypertensive, antidiabetic, antihyperlipidemic or antiasthmatic medication. The BPA asked: “In your opinion, how likely is it that this patient will pick up this medication?” The provider could select from a 5-level Likert item with responses ranging from “very unlikely” to “very likely.” Provider response was correlated to the principle outcome variable, medication first fill after 14 days as identified from the records of the pharmacy to which the prescription was transmitted. Results: A total of 4,822 patients over 11 months were included, and 4,532 (94%) patients filled their prescription within 14 days. Providers answered the BPA 89% of the time. Among respondents, most felt their patients would be likely or very likely to pick up their new medication (90.6% vs. 86.8% of providers chose likely or very likely among adherent and nonadherent groups, respectively). Only 10 (3.9%) of new medication orders not filled (nonadherent) versus 110 (2.7%) filled (adherent) were suspected by providers to be unlikely or very unlikely to be picked up, resulting in only an 8.3% positive predictive value for primary medication nonadherence. Discussion: Our study suggests that physicians overwhelmingly believe their patients are likely/very likely to pick up their first prescription. A physician’s intuition about a patient’s likelihood of filling a new medication does not reliably identify patients who do not fill new prescriptions for chronic medications. Our study’s ability to assess prediction accurately is limited by the unusually high first-fill rate of patients in this trial. Our findings are congruent with other reports assessing physician perceptions of patient adherence to chronic medications