Prescribing antibiotics: Factors driving decision-making in general practice. A discrete choice experiment rights and content
Under a Creative Commons license
Open access


Antimicrobial resistance is a major threat to human health; we need rational antibiotic use.

Most antibiotics are prescribed in general practice.

We administered a discrete choice experiment survey to all GPs in Denmark.

We assessed the relative weight of factors that influence GPs' decision to prescribe antibiotics.

C-reactive protein level was the most important factor across five segments of GPs.



Antimicrobial resistance is a threat to human health. We need to strive for a rational use of antibiotics to reduce the selection of resistant bacteria. Most antibiotics are prescribed in general practice, but little is known about factors influencing general practitioners’ (GPs) decision-making when prescribing antibiotics.


To 1) assess the importance of factors that influence decisions by GPs to prescribe antibiotics for acute respiratory tract infections (RTIs) and 2) identify segments of GPs influenced differently when deciding to prescribe antibiotics.


A questionnaire survey including a discrete choice experiment was conducted. Danish GPs were asked to indicate whether they would prescribe antibiotics in six hypothetical choice sets with six variables: whether the GP is behind schedule, patient's temperature, patient's general condition, lung auscultation findings, C-reactive protein (CRP) level, and whether the patient expects antibiotics. Error component and latent class models were estimated and the probabilities of prescribing in different scenarios were calculated.


The questionnaire was distributed to every Danish GP (n = 3,336); 1,152 (35%) responded. Results showed that GPs were influenced by (in prioritised order): CRP level (Relative importance (RI) 0.54), patient's general condition (RI 0.20), crackles at auscultation (RI 0.15), and fever (RI 0.10). Being behind schedule and patient expectations had no significant impact on antibiotic prescription at the aggregate level. The latent class analysis identified five classes of GPs: generalists, CRP-guided, general condition-guided, reluctant prescribers, and stethoscopy-guided. For all classes, CRP was the most important driver, while GPs were heterogeneously affected by other drivers.


The most important factor affecting Danish GPs’ decision to prescribe antibiotics at the aggregate level, in subgroups of GPs, and across latent classes was the CRP value. Hence, the use of CRP testing is an important factor to consider in order to promote rational antibiotic use in the battle against antimicrobial resistance.


Anti-microbial medicine
Infectious diseases
Respiratory tract infections
Discrete choice experiment
Stated preference
Primary health care
General practice


General practitioner
C-reactive protein
Discrete Choice Experiment
Respiratory tract infection
95% confidence interval
Relative importance
Rank order

Cited by (0)