Our results indicated that higher hospital bed density and physician density were significantly associated with lower levels of ABR. Surprisingly, higher GDP per capita was significantly associated with lower CRPA level, however, lower GDP per capita was associated with lower MRSA level. Higher OOP health expenditure was associated higher CRPA level, but no significant associations between OOP expenditure and other resistant bacteria were found. Also surprising was that ABR prevalence in central economic zone was significantly negatively associated with education, and positively associated with OOP health expenditure, however, these associations were not found in eastern economic zone and western economic zone. To our best of knowledge, it is the first cross-regional examination of the societal economic determinants of level of ABR using panel data modelling within countries, including China[15]. It is also the first study to compare the predictors of ABR in in different bacteria and three different economic zone in mainland China.
This study found that increasing the number of physicians could significantly reduce ABR prevalence, which was similar to the finding from previous studies[28]. It was reported that non-prescription antibiotic dispensing at pharmacies was an issue with 48.5% for diarrhoea and 70.1% for adult acute upper respiratory tract infection nationwide in China[29], therefore, increasing number of educated physicians might help curb non-prescription of dispensing of antibiotics, further decrease ABR. It was also reported that physician-targeted interventions were effective to decrease antibiotic prescription[30] and irrational use of antibiotics by physicians[31], therefore, the increased allocation of physicians could better optimize antibiotic therapy, which were conducive to containing of ABR.
We found that number of hospital beds was significantly associated with lower ABR rate, which was not reported in other studies. However, it was well known that hospital beds were potential reservoirs of bacteria in hospitals, and preventing bed contamination might help prevent dissemination of ABR[32]. It was noted that patients either colonized or infected with ABR bacteria must be isolated to prevent spread of the ABR to other patients and the closer the contact the easier acquired[32]. Larger number of hospital beds per 1,000 population denoted that more patients could timely receive health service with physicians’ prescription, and it was also associated with better environmental sanitation and better health infrastructure, which might reduce the prevalence of ABR.
There were no previous studies exploring the association between GDP per capita and ABR level within countries, but it was reported that a negative and significant correlation between GDP per capita and ABR prevalence existed at the global level[16]. Surprisingly, this study showed that there was a negative relation between GDP per capita and CRPA level but a positive relation between GDP per capita and MRSA level. Poor environmental sanitation, inappropriate antibiotic usage, low vaccination rates, poor laboratory and infection control capacity in the low/mid-income region might ignore the problem of ABR and lead to increased ABR level[16, 19]. However, some studies reported that population with higher GDP were associated with higher antibiotic usage, which further resulted in higher ABR level[13, 33]. These different findings might be related to antibiotic stewardship and surveillance of antibiotic use on different levels.
OOP health expenditure were positively associated with CRPA level in mainland China, which was similar with the findings that a higher proportion of OOP health expenditure was associated with higher levels of ABR across countries[13, 14, 17]. Unsurprisingly, supplier-induced demand was an important determinant factor for excess use of health care[34, 35]. Patients with higher OOP health expenditure might consume expensive and advanced antibiotics, such as carbapenems, which resulted in higher CRPA prevalence.
Education level could reduce the self-medication and over-the-counter antibiotic consumption, further to decrease ABR prevalence[36], which was consistent with our finding. However, why education was a positive factor for ABR only in central economic zone needed further exploration, in addition, the effect of access of water source on ABR also needed to be studied in the future.
There is no one comprehensive, universally agreed definition of governance[37], and practices defined as governance might vary depending on country and context[38]. Because of data gaps, the indicator of governance was difficult to measure, so we did not include it. In the previous studies, the focus had been mainly on antibiotic consumption as the most important factor contributing to ABR[14, 18]. Although we did not assess the association between antibiotic consumption and ABR in this study, because adequate data is lacking, the findings in other studies confirmed that many social and economic factors are of more importance than the antibiotic consumption in explaining the diversity in the level of ABR in different region[14, 18].
There are some limitations in our study. First, some confounding factors both in the human, agricultural sectors and the environment that might correlate with ABR level could not be collected from each province, such as temperature, antibiotic consumption, governance, due to inadequate data. Relevant studies including the above socioeconomic factors using real-world data are needed in the future. Second, it was presented with a relationship between ABR level and socio-economic factors, but not a definitive evidence of ABR mechanism. In addition, we acknowledge the ABR level from CARSS as the better data to represent the provincial level presently, however, it could not apply to primary care facilities.