Anyanwu, P;
Pouwels, K;
Walker, A;
Moore, M;
Butler, C;
Majeed, A;
Holmes, A;
... Costelloe, C; + view all
(2020)
Do variations in primary care practice characteristics explain the effect of a financial incentive scheme on antibiotic prescribing? A longitudinal study of the Quality Premium intervention in NHS England.
BJGP Open
, 4
(3)
, Article bjgpopen20X101052. 10.3399/bjgpopen20X101052.
Preview |
Text
Hopkins_bjgpopen20X101052.full.pdf - Published Version Download (942kB) | Preview |
Abstract
Background: In 2017, approximately 73% of antibiotics in England were prescribed from primary care practices. It has been estimated that 9%–23% of antibiotic prescriptions between 2013 and 2015 were inappropriate. Reducing antibiotic prescribing in primary care was included as one of the national priorities in a financial incentive scheme in 2015–2016. Aim: To investigate whether the effects of the Quality Premium (QP), which provided performance-related financial incentives to clinical commissioning groups (CCGs), could be explained by practice characteristics that contribute to variations in antibiotic prescribing. Design & setting: Longitudinal monthly prescribing data were analysed for 6251 primary care practices in England from April 2014 to March 2016. Method: Linear generalised estimating equations models were fitted, examining the effect of the 2015–2016 QP on the number of antibiotic items per specific therapeutic group age–sex related prescribing unit (STAR-PU) prescribed, adjusting for seasonality and months since implementation. Consistency of effects after further adjustment for variations in practice characteristics were also examined, including practice workforce, comorbidities prevalence, prescribing rates of non-antibiotic drugs, and deprivation. Results: Antibiotics prescribed in primary care practices in England reduced by -0.172 items per STAR-PU (95% confidence interval [CI] = -0.180 to -0.171) after 2015–2016 QP implementation, with slight increases in the months following April 2015 (+0.014 items per STAR-PU; 95% CI = +0.013 to +0.014). Adjusting the model for practice characteristics, the immediate and month-on-month effects following implementation remained consistent, with slight attenuation in immediate reduction from -0.172 to -0.166 items per STAR-PU. In subgroup analysis, the QP effect was significantly greater among the top 20% prescribing practices (interaction p<0.001). Practices with low workforce and those with higher diabetes prevalence had greater reductions in prescribing following 2015–2016 QP compared with other practices (interaction p<0.001). Conclusion: In high-prescribing practices, those with low workforce and high diabetes prevalence had more reduction following the QP compared with other practices, highlighting the need for targeted support of these practices and appropriate resourcing of primary care.
Archive Staff Only
 |
View Item |
