Prescribing for the patient and the planet: a pragmatic feasibility study of a behaviour change intervention to reduce liquid amoxicillin prescribing in primary care.

BACKGROUND: Oral liquid medicines can be clinically appropriate but may be used when solid oral formulations are feasible, with implications for medicine costs and environmental sustainability. We evaluated the feasibility of GPILL, a theory-informed intervention using prescribing feedback, social norms and information on formulation-associated carbon dioxide equivalent (CO₂e) emissions to support reduced liquid amoxicillin prescribing in primary care.

METHODS: The intervention was called 'GPILL' which stands for a 'Green Prescribing Intervention to Lower Liquids' and was adapted from an existing prescribing feedback intervention based on social normalisation theory which incorporated social-norm feedback. Delivered as an emailed PDF of a letter, the intervention incorporated information about levels of prescribing of the practice in comparison to other practices in the area, the environmental impact of liquid formulations and guidance on supporting patients to swallow capsules rather than use a liquid formulation. A quasi-experimental before-and-after design was used for practices in one geographical area of England. Rates of amoxicillin liquid prescribing were compared to capsules for all practices to generate a ratio of liquid: solid prescribing. Practices in the top 50% received the intervention (n = 28); whilst the remaining practices acted as controls (n = 26). No patient-level inclusion or exclusion criteria were applied to the practice. Prescribing volumes were obtained from OpenPrescribing, with data extrapolated to calculate costs and carbon dioxide equivalents of a standard course of treatment, using published costs and UK conversion factors.The primary outcome was to assess the feasibility of the intervention. Secondary outcomes were number of liquid and solid prescriptions, prescribing costs and formulation-associated CO₂e. Outcomes were assessed using routinely collected prescribing data during winter months (December-February) from 2020 to 2024. No deviations from planned outcomes were reported.

RESULTS: It was feasible to identify high liquid-amoxicillin-prescribing practices using routine prescribing data and deliver a low-resource, theory-informed intervention through usual primary-care channels. A greater reduction in liquid prescribing was subsequently observed among intervention practices, providing a preliminary signal that warrants formal effectiveness evaluation. Liquid amoxicillin prescribing decreased by 23% in the intervention practices (from 3549 to 2746 prescriptions; Wilcoxin signed-rank p = 0.0287) while increasing by 7% in the control practices (from 2090 to 2234; p = 0.175). There was evidence of a statistically significant difference between groups (p = 0.01113). Liquid amoxicillin prescribing costs decreased by 50% in intervention practices compared with 29% in controls. Estimated liquid CO₂e emissions decreased by 15% in the intervention group, compared with a 20% increase in controls. This represented an absolute reduction of 131.1 kgCO₂e in the intervention group (z = 2.27, p < 0.05) equivalent to charging a smart phone over ten thousand times. These exploratory findings should be interpreted cautiously because allocation was non-random, intervention practices had higher baseline prescribing, and the study was not designed to establish causal effectiveness.

CONCLUSIONS: A targeted, theory-informed intervention was feasible and produced a preliminary signal of reduced amoxicillin liquid prescriptions, with associated reductions in costs and estimated formulation-associated CO2e. These findings support the feasibility of a low-cost behaviour change strategy to promote sustainable prescribing in primary care. A larger, prospective study using an appropriate controlled analysis is warranted to determine effectiveness, cost consequences and environmental impact.