Borek et al’s quantitative study of locum GPs’ prescribing of antibiotics compared with other prescribers in general practice, published over 6 years after the events studied, is seriously flawed both in the design of the study and in the authors’ interpretation of the outcome.1
The retrospective study design is flawed because there can be no randomisation of patients to either locum or non-locum consultation. The implied premise is that patients who attend locums and emerge with a particular diagnosis are exactly comparable with those attending others with the same diagnosis, and that the proportion of patients for whom antibiotics are appropriate is the same for both groups.
Patients often exercise choice in their booking of appointments. Many elect for continuity and seek an appointment with their regular doctor. Those whose symptoms are most severe may settle for an appointment with whoever is available soonest, which may be more likely to be a locum. It is therefore inappropriate to assume that the severity of the illness, the patients’ level of risk, the exact nature of the illness in patients with the same diagnostic label, or its likelihood of responding to an antibiotic is equivalent in both groups.
The authors then go on to muddle statistical significance with clinical significance. I believe that the same prescriber (me, for instance) with similar patients actually prescribes in a statistically different way on a Monday morning and a Friday evening for patients with the same diagnostic label … and probably differently again the next Monday. A 4% difference is no difference at all in the context of complex human behaviours of this kind. The conclusion should surely be that 6–7 years ago GPs, prescribing nurses, and locum GPs all prescribed antibiotics similarly.
- © British Journal of General Practice 2022
