Community antibiotic stewardship research implementation (uptake studies)
The problem: An extensive literature, much of it synthesised by our team, now details the efficacy of a wide range of different interventions (delayed prescribing, shared decision making, “nudge” poster, audit & feedback, near patient tests or prediction rules) that individually reduce prescribing by GPs. Their different mechanisms suggest that combined effects would be synergistic. Some would be ready for implementing if GPs’ engagement with reduced antibiotic prescribing was strong. Because it is not, the overall effectiveness is small.20 In other words, we know of effective interventions that reduce antibiotic use. However, it is difficult to get GPs to use them, and they receive little attention in GP registrar training programs.
Figure 2 illustrates our estimates of the community impact of several interventions, accounting for the two stages of (i) uptake by GPs and (ii) effect on patients’ antibiotic usage. For example, when GPs use delayed prescribing it reduces antibiotic dispensing to patients by 60%12, but only 1 in 5 GPs are using it, so the overall population impact is around a 12% reduction. The key problem is the GP uptake stage, rather than not having effective interventions.
The solution: To test methods to improve GP uptake of the effective community interventions in Australia, we plan to do two studies (one observational study, one randomised trial) which would start at the beginning of winter (Outline devised with AI Straus at a 3-day KT workshop in 2017).
Study 1: Pilot two phase study of priming followed by an intervention package
|Month||Intervention / study element||Measures|
|1||Local GP opinion leaders identified (via survey of local GPs) who then help to promote antibiotic stewardship ideas|
|2–3||Individual feedback to practices on antibiotic usage (compared to local GPs; compared to Australia and Sweden)||Antibiotic usage|
|4–5||Articles by GP opinion leaders (via PHN newsletter, email, etc) on antibiotic stewardship which also offer education sessions (online and evening face-to-face sessions with role plays of shared decisions and delayed prescribing) and materials (shared decision making tools; clinical prediction rule cards; "Nudge" poster; etc)||Uptake rates of education sessions and package|
|6||Qualitative study – interviews regarding barriers and facilitators of the uptake processes|
|7–8||Repeat previous 2 steps (modified from interviews)|
|9||Survey of all practices on awareness and uptake||% uptake sessions; antibiotic usage|
Study 2: Randomised trial 3 intensities of the 2-phase uptake process
After the first priming phase is conducted, we then plan to test more intensive interventions in other PHNs and Practice-Based Research Networks in Tasmania and Victoria (AI Radford), modified from the above pilot version, based on the experience and qualitative studies. PHNs and practices not involved in the pilot will be randomised to:
A. Minimum intervention: the steps above.
B. Plus more intensive Audit & Feedback: the above steps, plus provide monthly feedback to the practices on prescribing together with a newsletter (and or video) of “how to” stories and tips from practices that successfully implement the packages.
C. Plus near patient testing: the above package plus the practice would be offered a near patient testing kit for procalcitonin and strep test strips for patients with pharyngitis (both known to reduce prescribing, but we are interested in whether this also improves uptake).
Arms B and C will use the minimum intervention as a base, but examine whether either repeated audit and feedback or the offer of near patient tests will improve GP and practice uptake of the packages, and consequent increased reductions in antibiotic prescribing.
Outcomes: The primary outcome will be antibiotic usage (prescribing and dispensing), but supplemented by processes measures shown in Study 1, including uptake of each of the components of educational sessions, and the additional offerings of arms B and C.
(Note: We recognize that antibiotic prescribing rates are notoriously difficult to obtain in Australian general practice for two reasons: 1) the measure collected by GPs is unreliable because of ‘diagnosis drift’; 2) GPs are reluctant to consent to collection of these data, either collected passively from practice electronic records, or centrally via the Pharmaceutical Benefits Scheme (PBS), which requires a complex Dept Health consent form. Accordingly, we have derived 2 processes to derive GP prescribing outcomes: 1) dispensing data derived from the PBS, but managed in strict privacy conditions to prevent the identification of individual GPs or practices; 2) consenting practices to obtain both prescribing and dispensing data.