Health systems globally face increasing morbidity and mortality from chronic diseases, yet many — especially in low- and middle-income countries — lack strong primary care. Among recent contributions to understanding how economic incentives can be harnessed to address this challenge is a study in which we analyze China’s efforts to promote primary care management for rural residents with chronic disease. Utilizing unique panel data and the plausibly exogenous variation in management intensity generated by administrative and geographic boundaries, we find that villagers with hypertension or diabetes residing in a township with more intensive primary care management have more primary care visits and better medication adherence with fewer specialist visits and hospitalizations. The resource savings from avoided inpatient admissions more than offset the costs of the program.
Low- and middle-income countries face increasing morbidity and mortality from chronic diseases. Globally, noncommunicable chronic diseases (NCDs) account for about two-thirds of deaths and this increasing burden of morbidity that reduces productivity, increases medical spending, and shortens lives (Bloom et al. 2012).
China is no exception. Indeed, given its aging population and relative success in controlling infectious disease, NCDs have long been the primary cause of ill health and premature death in China (Yang et al. 2008). Additionally, NCDs loom large in the policy goals for Healthy China 2030 and in efforts to harness AI and other new technologies for improving health. Evidence suggests widespread under-diagnosis and poor management of prevalent conditions such as hypertension and diabetes, especially in rural areas (Yang et al., 2008; Zhao et al., 2016; Lu et al., 2017). For example, high blood pressure was already the leading preventable risk factor for premature mortality in China in 2005 (He et al., 2009). China is also home to about one-quarter of the world’s population with diabetes, half or more of whom are undiagnosed (Zhao et al., 2016). Although the prevalence of diabetes increases with age and urbanization, but mortality rates are higher for rural Chinese (Bragg et al., 2017).
These factors underscore the need for strong primary care. Yet China, like many low- and middle- income countries, lacks adequate primary care resources both in terms of numbers and quality. Patients benefiting from China’s universal health coverage flock to hospital outpatient departments instead. As Professor Meng Qingyue and colleagues emphasize, “from 2005 to 2015, the proportion of healthcare services provided by primary care decreased by 7 percent” despite efforts to strengthen primary care, in part because “in 2010, [only] 5.6 percent of the doctors in township health centers had a formal medical education” (five years of medical school). This percentage has increased to only 10 percent in 2017 (Meng et al., 2019). China’s hospital-based healthcare system suffers from other inefficiencies as well (Zhang et al., 2017; Hu, 2019), including clear evidence that even well-qualified physicians do not have incentives always aligned with the best interests of patients (Currie et al., 2014).
Facing these challenges, the Chinese government has recognized this pressing issue and indeed has set a high priority on strengthening primary care services and providing public chronic disease management services during the past ten years of healthcare reform in China (Hu, 2019; Meng et al., 2019). For example, China’s Equalization of Basic Public Health Services (EBPHS) policy has started to provide more basic public health services and programs since 2009, including establishing health records for all citizens and launching community-level programs for monitoring and managing individuals suffering from hypertension and diabetes. These services are delivered by primary healthcare organizations and are free to the patient at the point of use. They are funded jointly by the central and local government, with a minimum funding for the basic package of ￥15 (￡1.70; €2; $2) per person in 2009 and ￥55 in 2017 (Yuan et al., 2010).
Unfortunately, there is not yet much rigorous evidence about the impact of improving primary care in China. To help fill this gap, in a recent study (Chen et al., 2019), we analyze the impact of a program that gives primary care physicians financial and reputational incentives to identify patients and enroll them in primary care management for their conditions. Specifically, our paper analyzes the effect of this policy in Tongxiang, China, which is a mostly rural county in Zhejiang province. Teaming up with researchers at the Zhejiang and Tongxiang Centers for Disease Control and Prevention, we assemble a unique dataset linking administrative and health data between 2011 and 2015 at the individual level for all of the more than 70,000 rural Chinese patients diagnosed with hypertension or diabetes by the end of 2015.
Our research strategy utilizes variation in management intensity generated by administrative and geographic boundaries to study the program’s effects. We focus on nearby villages that are within two kilometers of each other, but have primary care services managed by different townships (see Figure 1). The 14 pairs of boundary villages are similar across population characteristics such as age, gender, and educational attainment, and their residents enjoy identical insurance coverage and hospital access; but as shown in Figure 2, they differ in intensity of primary care management (partially explained by the pre-existing stock of primary care physicians in each township).
Figure 1: Illustration of Neighboring Villages
Figure 2. Illustrating the Study Design: Discontinuity of Village-level Hypertension Management Intensity, by Distance to Township Boundary
The stock of physicians before the reform was determined by higher-level officials as part of China’s rather rigid “headcount quota system” for all public service units. According to a 2016 report by the World Bank, World Health Organization, and three PRC ministries (of finance, health, and human resources), “by introducing rigidities and inefficiencies in the recruitment and management of health workers and limiting the mobility of health professionals, such a system distorts the labor market and compromises its ability to deliver quality healthcare services,” especially for rural primary care (see Note 1). Townships can hire temporary workers outside of the quota system, which is one response they undertook to leverage the pre-existing stock of physicians to scale up outreach and services with the funding and incentives under the program.
We utilize this plausibly exogenous variation to compare patients of neighboring villages in different townships. The regression results in Table 1 show how a different management intensity affects healthcare utilization (within a village-pair). The key explanatory variable, rank (the higher the better), captures the management intensity of disease d in township k for patient i. We find that patients residing in a village within a township with more intensive primary care management, compared to neighbors with less intensive management, had more primary care visits, fewer specialist visits, fewer hospital admissions, and lower inpatient spending in 2015 (see Table 1). For example, a greater primary care management intensity (proxied by the difference between the most- and least-intensely managed townships in our data) reduces the probability of hospitalization by 13 percent (a 0.018 percentage point lower likelihood of a hospital admission, from a base hospitalization rate of 0.14) for residents suffering from hypertension. These results suggest that primary care chronic-disease management reduces avoidable hospital admissions for ambulatory-care sensitive conditions such as hypertension and diabetes, indicating better health outcomes in terms of chronic disease control.
No such effects are evident in 2011, shortly after the launch of the program (2009). This suggests that the chronic disease management component of the EBPHS policy started to show its effectiveness a few years after its launch and baseline differences probably do not explain different program effects across villages.
We re-run our main analysis (Table 1) with controls on the number of physicians at baseline and find that our results are stable to this robustness check. We also find that reputational incentives from the rankings of primary care management of each township perhaps promote management intensity.
Exploring the mechanism for reduced specialist and hospital utilization, we find that patients with more intensive primary care management exhibit better drug adherence as measured by medication-in-possession (e.g., the number of days in which the patient had a filled anti-hypertensive prescription in 2015). As shown in Columns 8 and 9 of Table 1, comparing the townships that are ranked first and last, the difference in the number of days in possession of an anti-hypertensive drug is about 25 days, which is 29 percent of the average. For anti-diabetic medication, the difference is 14 percent of the average anti-diabetic drug possession days. These empirical results suggest that a higher-intensity management by village doctors may significantly increase medication adherence among patients. This finding is important, given substantial evidence that drug adherence is crucial for preventing complications from chronic conditions such as hypertension and diabetes and narrowing disparities in health outcomes (Conn et al., 2016; Simeonova, 2013; Wong et al., 2014; Yue et al., 2014).
ReferencesBloom, D. E., E. Cafiero, E. Jané-Llopis, S. Abrahams-Gessel, L. R. Bloom, S. Fathima, and D. O’Farrell, “The Global Economic Burden of Noncommunicable Diseases.” Working Paper No. 8712. Program on the Global Demography of Aging, 2012. URL: https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1288/2013/10/PGDA_WP_87.pdf
Zhao, Y., E. M. Crimmins, P. Hu, Y. Shen, J. P. Smith, J. Strauss, J., and Y. Zhang, “Prevalence, Diagnosis, and Management of Diabetes Mellitus among Older Chinese: Results from the China Health and Retirement Longitudinal Study,” International Journal of Public Health 61:3 (2006): 347-356. URL: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4880519/