General practitioners' antibiotic prescribing decisions in the Northern Dutch-German cross-border region: a discrete choice experiment

Abstract Background Redundant antibiotic (ABX) use unnecessarily contributes to antimicrobial resistance, complicating infection treatment. ABX prescription rates are higher in Germany than in the Netherlands, with increasing rates near the Dutch-German border. The reason for this remains unclear. Most ABXs are prescribed in primary care, particularly for upper respiratory tract infections (URTIs). Understanding the determinants influencing general practitioners (GPs) prescribing behaviors is crucial for developing targeted interventions. This study assessed how subjective and objective determinants influence GPs' ABX prescription for URTIs, extending and quantifying the results of our previous qualitative study. Methods ABX prescribing decisions were analyzed through a discrete choice experiment (DCE). The DCE consisted of eight choice tasks, each depicting two fictional scenarios described by six attributes. Based on our prior qualitative study, the attributes assessed were the effects of time constraints, patient pressure, language proficiency, C-reactive protein (CRP) levels, comorbidity, and clinical presentation (fever) on GPs' decisions to prescribe ABX for URTIs. Participants chose the scenario where ABX prescription was deemed most appropriate. The choices for the preferred vignette were regressed on the attributes using a conditional logit model. Results From mid-July to mid-September 2024, 47 GPs practicing in the Dutch-German ‘Ems-Dollart’ border region (n = 19 on the German side and n = 28 on the Dutch side) consented to and completed an online survey. In this DCE, a higher CRP level (40 mg/L) significantly increased the likelihood of intended ABX prescribing for URTIs compared to the reference category of 10 mg/L (OR 6.13, 95% CI 4.20–8.94). Similarly, the presence of a comorbidity (type II diabetes) (OR 2.78, 95% CI 1.91–4.05) and having fever as clinical presentation (OR 2.07, 95% CI 1.42–3.02), also significantly raised the likelihood of hypothetical ABX prescribing compared to their respective reference categories, (no diabetes and no fever). Conversely, patient pressure, time constraints and language barriers showed weak, non-significant associations. CRP had the highest relative importance (47%), followed by comorbidity (26%) and clinical presentation (19%), while non-clinical factors had less than 5% importance in decision-making. Conclusions Our study demonstrated that clinical factors play a significant role in intended ABX prescribing decision-making compared to non-clinical factors.