Defining Medication Discontinuation from Prescription Claims Records to Initiate Clinical Intervention in Schizophrenia & Bipolar disorder: A Systematic Review

Abstract Background: Digitally accessible prescription claims have potential utility in timely identification and active follow up interventions for unsupervised medication discontinuation in consumers with schizophrenia and bipolar disorder, that may prevent relapse and other adverse outcomes. We conducted a systematic review in order to identify and understand definitions of non-adherence from prescription claims records to serve as decision points for interventions. Methods: Electronic searches were conducted using Ovid MEDLINE, PubMed, the Cochrane Central Register of Controlled Trials and PsycINFO on 12 January 2020 to identify studies published in English. Data extraction was done by two reviewers independently. A narrative synthesis and tabulation of outcomes was performed. PROSPERO (CRD42020201708). Results: Thirty-one studies were eligible for analysis. Three common computing parameters were used to calculate non-adherence over 12 months, the Medication Possession Ratio (MPR) was used most frequently (21 studies - 68%) with a non-adherence threshold <0.8 (or <80% use of the full prescription of medication) in 55% of studies (n=17). All studies defined non-adherence retrospectively, and the main outcome reported by most studies was increased hospitalisation (24 of 31 studies – 77%). Discussion: Computing non-adherence from prescription claim records is a potential decision point for interventions. Better utilisation of this widely available data should be used to improve clinical practice and ensure targeted access to services. Conclusion: While there was variation in methods used to compute non-adherence parameters from prescription claims records, all approaches demonstrated that non-adherence was associated with adverse outcomes, the foremost of which was rehospitalisation and increased costs of care. Prospero Registration: PROSPERO CRD42020201708