PhD Student Institute of Health Policy, Management and Evaluation, University of Toronto Toronto, Canada
Background: Medications for the cardiovascular system (CVS) are often prescribed to reduce cardiovascular disease (CVD)-morbidity and prevent CVD-mortality. In England, GP practice (GPP) and Clinical Commissioning Group (CCG) characteristics impact prescribing.
Objectives: To explore within-GPP, between-GPP(/within-CCG) and between-CCG variation in the GPP annual prescribing rate of CVS medications in England; and to assess associations between GPP characteristics and the prescribing rate.
Methods: Ecological-longitudinal study on primary care GPP annual prescribing rate of CVS medications (2015-2019) using GPP-level prescribing data (does not include prescriptions issued but never dispensed) in England. Unadjusted/adjusted three-level Negative Binomial models were used to explore between-GPP/within-CCG and between-CCG variation in the prescribing rate, estimating variance partition coefficients (VPCs). Associations with GPP characteristics were also assessed: percentages of registered patients that are female, aged ≥65 years, have hypertension, are obese and have diabetes; and patient population-weight GPP area deprivation. Fully conditional specification multiple imputation approach was used to account for missing data for the adjusted analysis.
Results: In total, 7886 GPPs nested within 191 CCGs with prescribing data on CVS medications for at least 1 year between 2015 and 2019 were included. The prescribing rate remained relatively stable over the study period (median 5419, 5412, 5369, 5394 and 5385 per 1000 registered patients in 2015, 2016, 2017, 2018 and 2019, respectively). Most prescribing rate variation was estimated to lie between-GPPs (VPC-between-GPP, Unadjusted: 60.6%; Adjusted: 59.2%) and within-GPPs (VPC-within-GPP, Unadjusted: 22.5%; Adjusted: 31.1%). As expected given the large dataset, all GPP characteristics were associated with the prescribing rate before and after adjustment for other GPP characteristics. Notably, increases in percentages of registered patients that are aged ≥65 years [adjusted rate ratio: 1.031 (95% confidence interval: 1.029-1.033)], have hypertension [1.018 (1.015-1.021)] and have diabetes [1.042 (1.037-1.046)] were estimated to result in an increase in the mean prescribing rate. GPPs in urban [0.865 (0.844-0.886), vs. rural] and with registered patients living in the least deprived [0.820 (0.789-0.852), vs. most deprived] areas were estimated to have lower mean prescribing rates.
Conclusions: Substantial between- and within-GPP variation in the prescribing rate of CVS medications exists in England. These results may inform interventions both to improve public health and to promote more cost-effective prescribing practices. Future work will explore spatiotemporal modeling.
