The objectives of the online survey were to: Quantify the relative importance of clinical, social and economic factors influencing primary care providers’ prescription of antibiotics for the treatment of childhood and adult upper respiratory tract infections; Evaluate primary care providers’ knowledge about the diagnosis and management of upper respiratory tract infections; and Evaluate primary care providers’ knowledge about antibiotic resistance. The online survey was sent to a sample of nearly 4,000 South African doctors, and 433 responded to the survey.Because it increases the morbidity and mortality of bacterial infection, as well as the duration and cost of antibacterial treatment, antibiotic resistance constitutes a significant threat to global public health. The problem is even more critical in low- and middle-income countries which have higher infectious disease burdens, often higher rates of antibiotic resistance, less access to diagnostic tools, and fewer financial resources to purchase newer more effective antibiotics. In South Africa, antibiotic resistance is particularly high, despite an effective drug regulatory system and various initiatives to tackle the problem. Antibiotic resistance is accelerated by the overuse and over-prescription of antibiotics, which is the product of complex interactions between providers' decisions and knowledge, and patients' expectations. Yet research on the determinants of prescribing behaviours from LMICs in general, and South Africa in particular, is limited, and has been criticised for being too descriptive and superficial, with limited insight into the relative importance of different behavioural determinants to be able to prioritise interventions. In South Africa, most research efforts have focused on hospitals rather than primary care, despite the fact the majority of antibiotics are prescribed in primary care, mostly for respiratory infections. This study aims to explore how the interactions between providers and patients influence inappropriate antibiotic prescribing for URTIs in public and private primary care in South Africa. The study will include three components. First, drawing on medical anthropology, we will explore qualitatively providers' and patients' perceptions an experiences of antibiotic prescribing. This will be done through observations of consultations, interviews with providers and focus group discussions with patients. Second, building on the first part and drawing on methods from marketing research, we will design a survey consisting of a series of hypothetical clinical cases where clinical and patient characteristics will be systematically varied; for each case, the providers taking part will be asked indicate what drugs they would prescribe in a list of proposed drugs. The results will allow us to quantify the relative importance of the factors influencing antibiotic prescribing, with a view to inform policy-makers design future interventions. Finally, drawing on recent economics and medical education research, we will move beyond observational research and design a small randomised field study to test the impact of patients' knowledge and financial incentives on the prescribing practices of public and private primary care providers. This will be achieved with the use of standardised patients, who are healthy subjects trained to portray specific symptoms and disclose a rehearsed medical history. These patients will be sent to visit providers who agreed to take part in the research, at a time and under an identity unknown to them. The standardised patients will only differ in their expectations of antibiotics and the insurance status they will disclose. This will allow us to test the impact of these different characteristics on the likelihood of antibiotic prescription. We anticipate that the results will provide invaluable insights into our understanding of prescribing decisions in the public and private sector in South Africa, thereby informing the stewardship programmes for antimicrobial resistance in this country. Beyond this setting, these findings will be useful to other middle-income countries with a similar mix of public and private providers. More generally, we aim to produce high-quality research and develop innovative methods that could be replicated in other low-income settings to study antibiotic prescribing.

Data was collected through short online self-administered questionnaire (SAQ) administered on Qualtrics platform. Sampling frame was South African general practitioners with verified contact details and current email addresses in national database purchased from Medpages. General practitioners from Gauteng province were excluded from the sampling frame because of the overlap with other study components. Emails inviting participation in the survey were sent to 3952 general practitioners. 433 general practitioners (11.0%) ultimately responded and completed the survey. Six different clinical scenarios were presented to the respondents to evaluate the relative importance of different factors influencing antibiotic prescribing. Key aspects of the scenarios were varied using experimental design principles, in a type of discrete choice experiment (DCE). DCE design was presented in 2 blocks with 4 different orderings of questions. A small sum of money was paid for each correct answer of the knowledge questions to increase attention and performance.