Mark Pennington1, Paul McCrone2. 1. King's Health Economics, PO24 David Goldberg Centre, Institute of Psychiatry, Psychology and Neuroscience, King's College London, De Crespigny Park, London, SE5 8AF, UK. mark.w.pennington@kcl.ac.uk. 2. King's Health Economics, PO24 David Goldberg Centre, Institute of Psychiatry, Psychology and Neuroscience, King's College London, De Crespigny Park, London, SE5 8AF, UK.
Abstract
INTRODUCTION: Medication non-adherence is a serious barrier to treatment of schizophrenia. Understanding the impact of non-adherence on costs is essential to the assessment of the cost effectiveness of interventions in which adherence to treatment is a concern. OBJECTIVES: We undertook a comprehensive review of the available literature on the impact on costs of non-adherence to antipsychotics in the treatment of schizophrenia. METHODS: We performed a search on multiple databases (MEDLINE, Embase, PsycINFO and Health Management Information Consortium) for any study reporting the impact of adherence to antipsychotics on costs in patients with schizophrenia up to February 2018. We included trials of behavioural interventions but excluded comparisons of different pharmacological therapies. Studies were included if at least one-third of the study population had schizophrenia and costs were reported. RESULTS: Thirty-four publications on 28 studies met the inclusion criteria. Twenty studies reported analyses of administrative databases, primarily Medicaid. Findings on healthcare costs were mixed but suggested that lower pharmacy costs in non-adherent patients may outweigh increased hospitalisation costs where drug costs are relatively high. A few studies published analysis of prospective cohort data, or trials of behavioural interventions intended to influence adherence, mainly in a European setting. Findings were again mixed but indicate that increasing adherence does not reduce overall costs. CONCLUSIONS: Inference from analysis of administrative data is limited by the risk of selection bias. Inference from trials is limited by small sample sizes. The literature does not consistently support an assumption that non-adherence increases healthcare costs.
INTRODUCTION: Medication non-adherence is a serious barrier to treatment of schizophrenia. Understanding the impact of non-adherence on costs is essential to the assessment of the cost effectiveness of interventions in which adherence to treatment is a concern. OBJECTIVES: We undertook a comprehensive review of the available literature on the impact on costs of non-adherence to antipsychotics in the treatment of schizophrenia. METHODS: We performed a search on multiple databases (MEDLINE, Embase, PsycINFO and Health Management Information Consortium) for any study reporting the impact of adherence to antipsychotics on costs in patients with schizophrenia up to February 2018. We included trials of behavioural interventions but excluded comparisons of different pharmacological therapies. Studies were included if at least one-third of the study population had schizophrenia and costs were reported. RESULTS: Thirty-four publications on 28 studies met the inclusion criteria. Twenty studies reported analyses of administrative databases, primarily Medicaid. Findings on healthcare costs were mixed but suggested that lower pharmacy costs in non-adherent patients may outweigh increased hospitalisation costs where drug costs are relatively high. A few studies published analysis of prospective cohort data, or trials of behavioural interventions intended to influence adherence, mainly in a European setting. Findings were again mixed but indicate that increasing adherence does not reduce overall costs. CONCLUSIONS: Inference from analysis of administrative data is limited by the risk of selection bias. Inference from trials is limited by small sample sizes. The literature does not consistently support an assumption that non-adherence increases healthcare costs.
Authors: Catherine Henderson; Martin Knapp; Ksenija Yeeles; Stephen Bremner; Sandra Eldridge; Anthony S David; Nicola O'Connell; Tom Burns; Stefan Priebe Journal: PLoS One Date: 2015-10-08 Impact factor: 3.240