With a goal of improving health system quality and efficiency, reforms of China's health system over the past decade have sought to strengthen primary healthcare in lower-level clinics and health centers. Despite these wide-ranging reforms and initiatives, population-based studies have documented dramatic declines in patients' use of primary care facilities during this period. In this paper, we explore the determinants of this trend in China's rural areas using detailed longitudinal data following a nationally-representative sample of rural households and village clinics from 2011 to 2018. We estimate that between 2011 and 2018, the probability that individuals sought care at village clinics when ill dropped by 44%. At the same time, the utilization of outpatient services in county hospitals increased by 56% and patient self-treatment increased by 20%. Detailed Kitagawa-Oaxaca-Blinder decompositions suggest four primary drivers of this trend: the shifting burden of disease in rural areas, changes in how patients choose to seek care given different disease conditions, declining drug inventory in village clinics, and the decreasing importance of remoteness as a determinant of healthcare seeking behavior. Our results highlight the deteriorating role of village clinics in the rural healthcare system and the increasing importance of self-treatment and higher-tier primary care services.

Poor knowledge, scarce resources, and lack of or misaligned incentives have been widely documented as drivers of the irrational use of medicine (IUM), which significantly challenges the efficiency of health systems across the globe. However, there is limited understanding of the influence of each factor on IUM. We used detailed data on provider treatment of presumptive asthma cases in rural China to assess the contributions of provider knowledge, resource constraints, and provider behavior on IUM. This study enrolled 370 village providers from southwest China. All providers responded to a clinical vignette to test their knowledge of how to treat presumptive asthma. Resource constraints (“capacity”) were defined as the availability of the prescribed medicines in vignette. To measure provider behavior (“performance”), a subset of providers (104 of 370) were randomly selected to receive unannounced visits by standardized patients (SPs) who performed of presumptive asthma symptoms described in the vignette. We found that, 54% (201/370) of providers provided the vignette-based patients with prescriptions. Moreover, 67% (70/104) provided prescriptions for the SPs. For the vignette, only 10% of the providers prescribed the correct medicines; 38% prescribed only unnecessary medicines (and did not provide correct medicine); 65% prescribed antibiotics (although antibiotics were not required); and 55% prescribed polypharmacy prescriptions (that is, they prescribed five or more different types of drugs). For the SP visits, the numbers were 12%, 51%, 63%, and 0%, respectively. The lower number of medicines in the SP visits was due, in part, to the injections’ not being allowed based on ethical considerations (in response to the vignette, however, 65% of providers prescribed injections). The difference between provider knowledge and capacity is insignificant, while a significant large gap exists between provider performance and knowledge/capacity (for 11 of 17 indicators). Our analysis indicated that capacity constraints play a minor role in driving IUM compared to provider performance in the treatment of asthma cases in rural China. If similar findings hold for other disease cases, this suggests that policies to reduce the IUM in rural China have largely been unsuccessful, and alternatives for improving aligning provider incentives with appropriate drug use should be explored.