Pathways to optimising antibiotic use in Anhui: Identifying key determinants in community and clinical settings

Widespread antibiotic use is considered to be a major contributor to the growing problem of antimicrobial resistance in many countries, but there is a lack of evidence regarding its key drivers in China, particularly in rural settings and outside hospitals, and whether these differ from those found elsewhere. This evidence is essential for identifying the health system, behavioural and policy interventions that have most potential for limiting the spread of antibiotic resistance in China, while ensuring equitable access to antibiotics for essential treatment of bacterial infections. This innovative interdisciplinary project aims to acquire greater understanding of key social, economic, cultural, systemic and other influences on antibiotic use by investigating and quantifying current use of both prescribed and non-prescribed antibiotics and other forms of care to treat common infections in rural parts of Anhui Province. Our study will document treatment-seeking practices for selected common infections from all sources (including informal, Traditional Chinese Medicine (TCM), folk, home-based, private and government medical care) among patients at local health facilities and in the general population. We will gather information on testing, clinical diagnosis and antibiotic treatment procedures at lower levels of the health system and on over-the-counter purchasing of antibiotics from pharmacies and medicine shops. We will ascertain the annual incidence of common respiratory tract infections in the population of Anhui Province and assess the effects of the different forms of treatment they use, including antibiotics, on reported severity and duration of illness. We will also ascertain the feasibility of assessing clinical diagnostic accuracy, relationship between patient-reported symptoms and clinical diagnosis, and burden of antibiotic resistance in non-hospitalised patients, and investigate possible population biases in existing laboratory data through microbiological sampling. This range of evidence will enable us to formulate recommendations for appropriate interventions to optimise the use of antibiotics in Anhui Province and other regions of China.

Technical Summary

1.1: Microbiological sampling (Objectives 2, 4, 5, 10): 1,000 patients will be recruited for the COPD study in order to collect 100 S.pneumoniae, and 278 (555) patients in order to collect 100 (200) E.coli for the cUTI study, based on these assumptions: 40% of patients recruited to cUTI study and 25% of patients recruited to COPD will yield a pathogen; 90% of pathogens isolated from urine will be E.coli and 40% of pathogens isolated from sputum will be S.pneumoniae. In E.coli, resistance to nitrofurantoin, fosfomycin and co-amoxiclav will be ≪5%, resistance to cephalosporins and fluoroquinolones 50-60% (Qiao L-D, Chen S, Yang Y et al, 2013); in S.pneumoniae, penicillin resistance will be 12.5% and erythromycin 90% (Kim S-H, Song J-H, Chung D-R et al, 2012)

2.4: Direct Observations at pharmacies (Objectives 2, 6, 11): An observer is at each site for two 7-day periods, each spaced over 4 consecutive seasons (92 person-days of observations over 16 periods in 2x2x4 settings). All encounters in observation days are eligible for observation, using a structured checklist.

3.1 Prospective population cohort survey (Objectives 1, 2, 8, 9): Cases will be households, selected through cluster-randomization with 3 steps: (1) divides all counties in the Province into north, middle & south regions; (2) randomly selects 6 counties from each region, then 1 township and community from each county & 1 administrative village & 1 street from each selected township or community; (3) randomly draws 1 household from each administrative village (or street) as a starting point, then every 5th household using a random walk method. Baseline survey uses a structured questionnaire. A local informant completes case detection in cohort households bi-weekly for 12 months using a symptom checklist. Cases are reported by text message for follow-up in person or by phone using structured questionnaires. Data analysis will use descriptive statistics & multivariate modelling.