Abstract Background Most antibiotic prescribing occurs in primary care. Even within the same health facility, there may be differences between prescribers in their tendency to prescribe antibiotics, which may be masked by summary data. We aimed to quantify prescriber variability in antibiotic prescription to patients with acute fever in primary care clinics in Myanmar. Methods We conducted a secondary analysis of prescribing data from 1090 patient consultations with 40 prescribing doctors from a trial investigating the effect of point-of-care C-reactive protein (CRP) tests on antibiotic prescription for acute fever. We used multilevel logistic regression models to assess inter-prescriber variability in the decision to prescribe antibiotics. Results The median odds ratio (MOR) in the unadjusted model was 1.82 (95% CI: 1.47–2.56) indicating that when two prescribers from this population are randomly selected then in half of these pairs the odds of prescription will be greater than 1.82-fold higher in one prescriber than the other. The estimated variability from this sample of prescribers corresponds to a population of prescribers where the top 25% of prescribers will prescribe antibiotics to over 41% of patients while the bottom 25% will prescribe antibiotics to less than 23% of patients. Inter-prescriber variation in antibiotic prescribing remained after adjustment for patient characteristics and CRP information (P < 0.001). Conclusions Despite sharing the same management guidelines, there was substantial inter-prescriber variation in antibiotic prescription to patients with acute fever. This variation should be considered when designing trials and stewardship programmes aiming to reduce inappropriate antibiotic prescribing.
Oxford University Press (OUP)