Abstract
The fragmentation of job-based and community-based insurance plans inevitably undermines health care accessibility in China’s market-oriented health system, especially for uninsured and rural residents. Based on the 2014 China Family Panel Studies, this secondary data analysis examined whether socioeconomic indicators, health-related determinants, and particularly social health insurance status affect physician visits in the past 2 weeks and hospital admissions in the past 12 months among a representative sample of older adults (n = 6,570). Grounded in Andersen’s behavioral framework, 2 series of logistic regression analyses were performed: one was built in a hierarchical manner, assessing blocks of predisposing, enabling, health-need, and lifestyle-behavioral factors; the other was conducted in a cross-referencing manner, comparing uninsured populations with job-based and community-based insurance enrollees. Results show that, after full adjustment, the odds of physician visits were lower among urban insurance enrollees (OR = 0.67, 95% CI: 0.47–0.97) than rural residents. For hospital admissions, both uninsured elders (OR = 0.65, 95% CI: 0.48–0.87) and community-based insurance enrollees (OR = 0.67, 95% CI: 0.47–0.97) had lower use of inpatient care than job-based insurance enrollees, demonstrating inequitable access. This study suggests that policy efforts should unify the social health insurance system to combat existing insurance-related inequities in health care use for underserved aging populations.
Keywords
Inequities in health care access across stratified socioeconomic groups constitute major challenges to public health and social justice. 1 Understanding the relationship between patient characteristics and medical service use is essential to crafting viable health policies and directing resources appropriately in the health system. One common approach to tackling health inequities is through the implementation of social health insurance programs, 2 which can mobilize funds and pool risk to modify systematic differences in health status between socioeconomic groups. Given the complex medical needs of managing chronic illness, acute conditions, and preventive care, access to health care services among older adults is of critical concern to health care sustainability. 3 As the most populous country with 202 million older adults, China is undergoing a rapid epidemiological transition. 4 Nevertheless, due to the privatization of a health care system driven by a neoliberalism-induced economic reform since the 1980s, 5 China has been confronting the crisis of health care accessibility (kanbingnan) and affordability (kanbinggui). 6 , 7
The Segmented Health Insurance System in China
In the context of post-reform China, the socialist regime of “equal access for equal need” in the health sector has shifted to a market-oriented health care system. Consequently, in response to the widening disparities in health care access, 8 the Chinese government restructured 4 types of social health insurance plans tailored to diverse social groups: Urban Employee Basic Medical Insurance (UEBMI; launched in 1998), Urban Resident Basic Medical Insurance (URBMI; launched in 2007), New Rural Resident Cooperative Medical Scheme (NCMS; reformed in 2003), and Government Employee Medical Insurance (GEMI). Social health insurance, theoretically distinct from private insurance, is often characterized as any form of publicly managed insurance plan financed by government subsidies together with contributions from households and enterprises to promote equity. 9 In this article, “health insurance” refers to social health insurance, as it is the major health-financing channel in China. Under the umbrella of the social health insurance system, UEBMI and GEMI are mandatory job-based programs mainly sponsored by employee and employer payroll taxes, whereas NCMS and URBMI are voluntary community-based programs primarily funded by government subsidies. 9
Notwithstanding its aim to provide universal coverage and equitable access to health services, the impact of multiple insurance plans on health care utilization is controversial. 10 , 11 On the one hand, research shows the positive effect of some insurance plans on the improved access to medical services. Economic studies indicated that both UEBMI and URBMI increased outpatient care utilization. 12 , 13 Consistently, health literature also documented linkage between NCMS enrollment and enhanced access to hospitalization. 14 , 15 On the other hand, counter-arguments reveal an ineffectiveness or even a negative effect of health insurance. For example, NCMS was found to be unrelated to the utilization of formal medical services, 16 nor could it increase the likelihood of outpatient visits. 12 These mixed findings underscore striking disparities in health service use among people covered by different insurance programs. 7
Study Purpose
China’s health insurance plans vary in coverage and generosity; however, many studies have been limited to comparing health care access between uninsured and insured populations. 17 Although a growing body of research starts to address insurance-specific variations in health care access 7 , 18 and health outcomes, 8 there is a dearth of empirical investigations into a nationwide picture unveiling structural distinctions entrenched in these 4 health insurance plans and their associated diversified impacts on elderly patients’ health care–seeking behaviors. Moreover, most of these studies neglect micro and macro factors that simultaneously influence patients’ health care decision making.
This study sought to examine how different insurance-specific groupings of older adults (uninsured, GEMI, UEBMI, URBMI, NCMS) differ in the probability of physician visits and hospital admissions. The current study represents the largest examination of health care access comparing retired civil servants (i.e., GEMI enrollees) with members of other insurance programs to date. It provides evidence to Chinese health policymakers on whether uninsured and insured elders with different insurance plans have access to health care equitably, provokes insights into how health resource allocations to disadvantaged social groups can be optimized, and proposes approaches to eliminating institutional barriers.
Literature Review and Hypothesis Development
The inconsistent effect of insurance plans on health care access in China stems from the polarizing benefits between job-based and community-based plans, and between urban and rural areas. 19 There are currently 3 major insurance programs in urban China. First, UEBMI targets urban employees in all private enterprises. It is jointly financed by 6% of employers’ budgets for gross payroll and 2% of employees’ individual wages on a monthly basis. 8 Second, URBMI targets unemployed urban residents and is mainly sponsored by public subsidies from central and local governments plus a small amount of individual premiums. 12 Third, GEMI is the most generous insurance plan exclusively for government officials and employees of state-own enterprises. 10 For the rural population, NCMS is the only available public insurance for self-employed agricultural workers, which operates through a similar funding mechanism as URBMI. 8 , 11
Coinciding with the pay-as-you-go pensions system in China, retired elders can enjoy the benefit supported by working-age enrollees, without monthly contributions to the respective health insurance programs, once they have reached minimum participation years. 8 Among these 4 public insurance programs, GEMI offers the most generous benefit package (reimbursement rate, RR = 95%), followed by UEBMI (RR = 68%), whereas URBMI (RR = 48%) and NCMS (RR = 44%) mainly cover inpatient care. 11 , 20 Nevertheless, the participation of these “mutually exclusive” insurance plans is not attributed to self-selection but constrained by the employment status and rural-urban household registration. 8 The benefit disparities embedded in the systemic fragmentation of health insurance plans inevitably undermine health equity as envisioned in the Healthy China 2030 Blueprint. 21
Insurance-Related Inequities in Health Care Access
Researchers and practitioners have long been interested in what facilitates the use of health services and what influences people to behave differently to their health care needs. 22 Although many established theories of health behavior exist, the Andersen’s behavioral model was selected in this study as it is not only “the most amenable conceptualization for framing secondary analyses” 23 (p471) but also a pragmatic tool to examine both socio-structural forces and proximal health risk factors. In addition, scholarship on social determinants of health manifests how systemic socioeconomic barriers are detrimental to downstream health outcomes and underscore the fundamental role of investigating upstream socio-structural indicators. 24 The Andersen’s framework distinguishes individual demographics from socioeconomic determinants into 3 interrelated components: predisposing, health-need, and enabling factors. 25 Predisposing variables reflect demographic information (e.g., age, gender) that exists prior to the development of health need and illness. Enabling factors represent the mobilization of practical resources, such as income, regular source of care providers, and health insurance status, that are shaped by a broader socio-political environment in affecting care access. 26
The impact of medical insurance on improved care access has been well recognized in Western democracies.
17
Heavy financial burdens preclude seeking health services among the elderly population, whereas the role of health insurance could reduce out-of-pocket health expenditure and thus be conducive to affordable health care
27
; therefore, a general hypothesis (H1) was developed: H1 Positive effect hypothesis: Insured older adults have a higher likelihood of health care access, compared to their uninsured counterparts. H2 Inequitable access hypothesis: Insured older adults with community-based health insurance have a lower likelihood of health care access compared to elders with job-based health insurance.
Andersen’s behavioral model applied to the examination of insurance-related inequities in health care access.
H3 Robust association hypothesis: The association between health insurance status and health care access is robust after controlling for predisposing indicators, enabling resources, health-need characteristics, and health behaviors.
Methods
Data Source
The data were derived from the 2014 national survey of the China Family Panel Studies (CFPS) carried out by the Institute of Social Science Survey at Peking University. CFPS is an ongoing annual longitudinal survey of Chinese communities, families, and individuals launched in 2010 (wave 1), followed up in 2012 (wave 2), and 2014 (wave 3).
34
The 2014 CFPS survey adopted a multi-stage probability sampling, with more than 1,800 villages in 29 provinces of China as the primary sampling units, and recruited 13,946 households representing 95% of the national population,
35
including 37,137 adults (aged
Sample size determination (n = 6,570) from 2014 CFPS survey.
Main Dependent Variables
Health care access was operationalized as physician visits and hospital admissions. Specifically, respondents were asked whether they had (1) seen a doctor for outpatient treatment in the previous 2 weeks and (2) been hospitalized in the previous 12 months. Physician visits in China refer to outpatient care consultations received from medical specialists, general practitioners, or even barefoot doctors in village clinics. 6 Hospital admissions refer to a patient admitted to a medical facility overnight for either emergent or elective admissions.
Main Independent Variable
Health insurance status was measured by asking whether respondents had enrolled in any form of the social health insurance plan as below: (1) GEMI, (2) UEBMI, (3) URBMI, (4) NCMS, or (5) the uninsured group. These 4 insurance programs constitute the landscape of social health insurance system available to seniors in China. 10
Covariates
Predisposing factors included age (60–69/70–79/80–95), gender (male/female), marital status (with/without partner), educational level (illiterate/primary school/junior high school/finished senior high school or above), household registration (agricultural/non-agricultural) and Chinese health belief (yes/no). China’s household registration (hukou) system intrinsically exacerbates the urban-rural cleavage of welfare entitlement, including health insurance plans. As Traditional Chinese Medicine (TCM) is an instrumental component integrated in China’s health care system, 37 Chinese health belief remains crucial in determining health-seeking behaviors of older adults, and it was assessed by asking whether respondents would prefer TCM, if available, to Western medical services in any health care facility.
Enabling factors measured tangible resources of obtaining health services, such as regular health care providers, old-age pensions, and major caregiver. Monthly old-age pensions were categorized into 4 levels: (1) ¥1–¥500 (low), (2) ¥501–¥2,000 (medium), (3) ¥2,000–¥68,000 (high), or (4) non-pensioners (none). Major caregiver was classified into 4 dimensions: (1) caregiving spouses, (2) adult children or their spouses, (3) other family members, or (4) no caregiver that had supported the respondent’s medical situation in the previous 12 months. 38 Usual source of care was determined through the question: “where do you usually see a doctor when you are sick?” 39 with responses grouped into 5 categories: (1) specialty hospital, (2) general hospital, (3) township hospital, (4) community health center, or (5) community clinic.
Health-need characteristics were measured by perceived health status (poor or fair/good/very good or excellent), functional limitation (yes/no), moderate mental distress (yes/no), and chronic illness (yes/no). Functional limitation was flagged when participants reported needing help with at least 1 out of 7 activities of daily living (ADLs), including outdoor activities, taking a bus, eating, shopping and meal preparation, washing clothes, managing transportation, and house cleaning.40,41 Moderate mental distress in the previous 30 days was measured by a 6-item Kessler Psychological Distress Scale (K6), 42 with an internal consistency coefficient of 0.863. This K6 composite measure (range: 0–24) was a 5-point scale, from 0 (never) to 4 (almost every day) to assess feelings of nervousness, hopelessness, restlessness, depression, worthlessness, and everything being an effort. A cut-off score of ≥5 out of 24 was applied to identify those who had moderate mental distress. 43 The chronic condition was a dichotomous variable based on any self-reported physician diagnosis of chronic illness in the previous 6 months.
Health behaviors were captured by 3 binary variables measuring whether participants had (1) smoked tobacco cigarettes in the previous 30 days, (2) drunk alcoholic beverages more than 3 times every week in the previous 30 days, and (3) done any physical exercise in the previous week, including indoor exercise such as dancing or yoga and outdoor sports such as jogging, hiking, or swimming.
Statistical Models
Chi-square test was used to test the statistical differences at the bivariate level (Table 1). Multivariable logistic regression analysis performed with SPSS 24.0 was employed to examine factors associated with the probability of physician visits and hospitalization respectively. Two series of logistic regression analyses were undertaken. In the first series, logistic regression was adopted in a hierarchical manner to determine the independent contribution of each additional block, 44 assessed by Naegelkerke pseudo R2. In the second series, logistic regression was conducted by shifting the reference category of the key variable so that “each model setting out a different insurance status 1 by 1 to act as the reference group.” 8 (p7)
CFPS 2014 Sample Description and Bivariate Analysis: Weighted Percentages (%).
Source: China Family Panel Studies 2014.
Abbreviations: GEMI, Government Employee Medical Insurance; NCMS, New Rural Resident Cooperative Medical Scheme; UEBMI, Urban Employee Basic Medical Insurance; URBMI, Urban Resident Basic Medical Insurance.
Binary variables are presented in 1 category.
First, the hierarchical logistic regression began with health insurance, age, and gender as a core block in the first model (model 1) to test H1 regarding the associations between insurance status and health services. The subsequent 4 models were performed to test H3 on the degree to which inclusion of the following additional factors reduced the association tested in H1: predisposing factors were entered in the second block (model 2), followed by adding enabling measures in the third block (model 3), health-need characteristics in the fourth block (model 4), and health behavior measures in the final block (model 5).
Second, to test H2 on the insurance-related inequities in health care access, another series of logistic regression analyses were conducted by switching the reference category for different insurance status groupings (uninsured, GEMI, UEBMI, URBMI, NCMS) in 5 fully adjusted models, respectively (models 5a–5e). Cross-sectional individual weights provided by CFPS were normalized to produce population representative estimates corrected for the sample size and nonresponse bias.
Results and Discussions
Sample Characteristics and Bivariate Analysis
Table 1 lists the descriptive statistics of Chinese elders’ health care seeking practices and insurance-specific characteristics. In a sample of 6,570 older adults who experienced illness in the previous year, the prevalence of physician visits (previous 2 weeks) was 37.3%, and the prevalence of hospital admissions (previous 12 months) was 22.4%. Male participants were slightly fewer than female (48.2% vs 51.8%). The majority of the respondents were insured (93.8%), aged 60–69 years (65.5%), living with a partner (79%), illiterate (56.2%), from rural households (69.1%), with low-level pensions (45.7%), and poor/fair perceived health status (52.2%). Older adults were mainly covered by NCMS (67%), followed by UEBMI (14%), URBMI (7.7%), and GEMI (5.2%).
According to the Chi-square statistics, older adults in different insurance status differ significantly with respect to socioeconomic circumstances, health conditions, and behaviors (all Ps < .05), except for alcohol consumption (P = .07). Specifically, compared to the uninsured, seniors with job-based UEBMI tended to have a completed senior high school education (25.2% vs 5.7%), urban household registration (96.7% vs 38.5%), affluent (>¥2,000) old-age income (50.8%), and more physical exercise (69.8 vs 38%), and they were less likely to be mentally distressed (20.3% vs 41.7%) or physically impaired (12.5% vs 21.7%). In contrast, uninsured older adults and community-based insurance enrollees were more likely to be illiterate, non-pensioners, mentally distressed, and physically impaired, compared to job-based UEBMI participants. In other words, social and health-related disadvantages intertwine with the “gradients of social health insurance status,” from insured to uninsured, from community-based to job-based insurance.
Multivariate Logistic Regression
The statistical results for 2 health care use analyses are shown in Tables 2 to 4, respectively. Tables 2 and 3 present the first series of hierarchical logistic regression analyses, which assessed the independent contribution of health insurance plans (vs the uninsured) and a multidimensional examination of Andersen’s framework of health care utilization. Table 4 presents the second series of logistic regression analyses by setting out each insurance status as the reference category in each model.
Hierarchical Logistic Regression of Physician Visits Among Older Adults (Testing H1 and H3).
Abbreviations: GEMI, Government Employee Medical Insurance; NCMS, New Rural Resident Cooperative Medical Scheme; UEBMI, Urban Employee Basic Medical Insurance; URBMI, Urban Resident Basic Medical Insurance.
The reference group is given in the parenthesis (OR = 1.00).
*P < .05; **P < .01; ***P < .001. The exact P-values for significant variables are shown in Table S2 in the Supplemental Material available online.
Hierarchical Logistic Regression of Hospital Admissions Among Older Adults (Testing H1 and H3).
Abbreviations: GEMI, Government Employee Medical Insurance; NCMS, New Rural Resident Cooperative Medical Scheme; UEBMI, Urban Employee Basic Medical Insurance; URBMI, Urban Resident Basic Medical Insurance.
The reference group is given in the parenthesis (OR = 1.00).
P < .05; ** P < .01; *** P < .001. The exact P-values for significant variables are shown in Table S3 in the Supplemental Material available online.
Cross-Referencing Logistic Regression of Physician Visits and Hospital Admissions Among Older Adults (Testing H2).
Abbreviations: GEMI, Government Employee Medical Insurance; NCMS, New Rural Resident Cooperative Medical Scheme; UEBMI, Urban Employee Basic Medical Insurance; URBMI, Urban Resident Basic Medical Insurance.
Odds ratios are fully adjusted for all covariates.
*P < .05; **P < .01; ***P < .001. The exact P-values for significant variables are shown in Table S4 in the Supplemental Material available online.
Controversial Effect: Insured Versus Uninsured (H1)
As shown in Table 2, in terms of physician visits, H1 was contradicted that insured older adults in China were less likely to visit doctors than the uninsured. This robust association was not attenuated by any cluster of variables, particularly for URBMI participants, as it maintained significance regardless of any block entered. The odds ratio (OR) of using outpatient care for URBMI participants as opposed to those uninsured ranged from 0.68 (95% confidence interval [CI]: 0.52–0.90) controlling for age and sex, to 0.67 after full adjustments (95% CI: 0.48–0.93). Similarly, other urban-centric health insurance plans (i.e., GEMI, UEBMI) were negatively correlated with doctor visits. The finding was different from the evidence of earlier studies showing a positive relationship between health insurance and outpatient visits (e.g., UEBM, 15 URBMI 45 ), but in line with NCMS’s negative impact. 12
As shown in Table 3, H1 was supported in that insured seniors (i.e., UEBMI) were more likely to have hospital admissions than uninsured populations. The age-sex-adjusted OR of having inpatient care was 36% greater for UEBMI participants (OR = 1.36, 95% CI: 1.03–1.80) than those uninsured. This odds ratio was subsequently reduced to be nonsignificant by predisposing and enabling factors and returned to be significant when health-need and behavior factors entered, yielding to a larger OR of hospitalization in the final model (OR = 1.49, 95% CI: 1.03–2.14). It resonated with previous studies indicating UEBMI’s positive relationship with inpatient care utilization. 12
Inequitable Access: Job-Based Versus Community-Based Insurance (H2)
Table 4 shows the second series of logistic regression analyses conducted in a cross-referencing manner, comparing uninsured populations and job-based and community-based insurance enrollees. It includes 2 panels: panel A presents the results of doctor visits, while panel B represents hospital admissions. Each panel has 5 models that set out a specific insurance status as the reference category: model 5a with reference to the uninsured, model 5b with reference to GEMI, model 5c with reference to UEBMI, model 5d with reference to URBMI, and model 5e with reference to NCMS.
As shown in panel A, compared to the uninsured or NCMS participants, the seniors having urban-centric insurance (i.e., GEMI, UEBMI, URBMI) were less likely to consult a physician in the urban area (ORs = 0.67–0.74, Ps < .05), where the availability of retail pharmacies was more adequate and resulted in self-medication being a common practice. 45 Also, it may be due to the “advantageous selection” effect that healthier individuals have a greater propensity to purchase voluntary plans. 46 (p281) When compared to urban insurance plans, rural elders with NCMS were more likely to visit doctors (ORs = 1.36–1.49, P values < .05). This is consistent with previous studies that “rural respondents visited physicians more than urban respondents,” while more urban respondents treated illnesses by themselves using over-the-counter medicine without physician consultations. 45 (p769), 47
According to panel B, similar to prior literature, 48 job-based UEBMI participants were more likely to have access to hospital care, when compared to uninsured elders (OR = 1.49, 95% CI: 1.03–2.14) or relative to community-based URBMI participants (OR = 1.55, 95% CI: 1.14–2.10). Likewise, job-based GEMI participants also had higher use of inpatient care than URBMI participants (OR = 1.46, 95% CI: 1.01–2.03). In contrast, uninsured seniors (OR = 0.67, 95% CI: 0.47–0.97) and community-based URBMI enrollees (OR = 0.65, 95% CI: 0.48–0.87) were less likely to have access to inpatient care compared to job-based UEBMI enrollees. Therefore, in terms of access to inpatient care, a discernible pattern of institutional inequities stands out – the superiority of job-based insurance and the disadvantaged position of community-based insurance. H2 was supported in that older adults with community-based health insurance had a lower likelihood of inpatient care access, compared to elders with job-based health insurance.
Robust Association: Decreasing Physician Visits Versus Boosting Hospital Admissions (H3)
The fully adjusted models in Tables 2 and 3 explained 26% and 24% of the total variance of physician visits and hospital admissions, respectively. These final models retain nonsignificant parameters as they were significant in prior bivariate analyses or sequential logistic regression analyses. The health insurance plans all act independent of demographic factors, enabling resources, health-need characteristics, and health behaviors to affect elders’ health care–seeking practices. Hence, irrespective of the aforementioned confounders, the distinct associations between health insurances and 2 health services lent support to H3. To synthesize the results of 2 health services, we plotted the OR and 95% CI of each health insurance program, compared with the uninsured (see Figure 3).
Odds ratio and 95% confidence interval of physician visits and hospital admissions among older adults (model 5a). GEMI, Government Employee Medical Insurance; NCMS, New Rural Resident Cooperative Medical Scheme; UEBMI, Urban Employee Basic Medical Insurance; URBMI, Urban Resident Basic Medical Insurance.
Taken together, the results show that the odds for doctor visits were significantly lower among insured seniors, while the odds for hospital admissions were higher among UEBMI seniors compared to the uninsured. In other words, urban-centric insurance programs (e.g., GEMI, UEBMI, URBMI) were related to decreased physician visits whereas job-based health insurance programs were effective in boosting inpatient care access (e.g., UEBMI, GEMI). Such boosting effect concentrating on hospital admissions rather than primary care medical visits is induced by the pro-hospitalization design of premium refund incentives of China’s social health insurance – prioritizing the coverage for inpatient services while neglecting the reimbursement for outpatient services. 15
Other Factors Contributing to Health Service Access (H3)
Besides the robust association between health insurance and health care use, other factors significantly influence access to care (see Tables 2 and 3). Among predisposing measures, older age was associated with both increased outpatient and inpatient care use. Consistent with the previous studies,6,18 females (OR = 0.68; 95% CI: 0.58–0.80) without a partner (OR = 0.57; 95% CI: 0.47–0.69) were less likely to receive inpatient care, mostly due to lacking a caregiver for medical escort services. Those with trust in traditional Chinese medicine had 39% greater odds of visiting physicians than those without this belief (OR = 1.39; 95% CI: 1.24–1.56), but not for inpatient care. Higher educational level was initially linked with fewer doctor visits or hospitalization, yet it lost the significance after controlling for health-need indicators. Urban household registration was related to more hospital admissions, but this relation was no longer significant when controlling for enabling factors.
With respect to enabling factors, elders with a medium level of old-age income were more likely to access outpatient care (OR = 1.25; 95% CI: 1.03–1.52), which aligned with prior research,7,27 but they were less likely to be hospitalized (OR = 0.79; 95% CI: 0.63–0.99) compared to non-pensioners. This may be attributable to respondents’ awareness of diseases. For example, “high-income individuals tend to recognize more symptoms as signs of disease and are more likely to see a doctor.” 7 (p141) Situated in China’s weak primary care infrastructure, there was a gradient effect of regular care providers on care access, where older adults were more likely to be hospitalized in higher levels of health care institutions (ORs = 1.45–4.72, Ps < .05), compared to community clinics perceived as inferior health facilities. 30 Receiving care from a spouse, adult children, and other family members were all correlated to increased access to inpatient care (ORs = 2.19–3.85, Ps < .05), in contrast to those without any caregiver. It is worth noting that major caregiver and regular care providers were initially associated with more doctor visits; however, these enabling effects were reduced to nonsignificant after adjusting for health-need characteristics.
Health-need characteristics were the strongest determinants in the use of health services, 25 with the largest proportion of variance explained. Being distressed, ADL limitations, chronic diseases, and poor self-reported health status were all significantly related to increased outpatient and inpatient service utilization. 6 For example, the likelihood of physician visits was 4.95 times higher for an individual who reported poor or fair health status than those reporting excellent health (OR = 4.95; 95% CI: 4.04–6.08). In terms of health behaviors, seniors who were not binge drinking had somewhat higher odds of visiting doctors (OR = 1.56; 95% CI: 1.31–1.85), while those who had not smoked in the previous month were associated with an increased possibility of being hospitalized (OR = 1.28; 95% CI: 1.08–1.52). The plausible reason could be a reflection of adaptive behaviors after having health problems. 6 (p10)
Limitations
The present study is limited in several ways. First, the data used are self-reported and thus at risk of measurement error. Second, a rigorous causal relationship between predictors and outcomes cannot be interpreted by relying on cross-sessional data. More waves of CFPS data are recommended to be included in analytic models to investigate causality. Third, it is important to acknowledge that the measurements of 2 health care services and some confounding factors were constrained to binary measures because of severe shortcomings in the breadth and depth of data from the CFPS survey. Therefore, the complexity of these health-related indicators could not be captured thoroughly and their validity is not yet well established. For instance, information regarding the volumes, types, and purpose (e.g., specific illness episode) of physician visits and hospital admissions were omitted. For health behaviors, the frequency and type of smoking and alcohol consumption were unavailable. The limited time frame assessed regarding doctor visits (2 weeks) and chronic conditions diagnosed (6 months) also raised challenges to their validity. Future research entails valid measurements vis-à-vis a comprehensive approach in recording the nature and magnitude of health-related indicators.
Conclusion and Implications
This study sheds light on disparities in health care access among older adults enrolled in diverse health insurance plans, identification of vulnerable Chinese elderly, and evidence-based assessments of health insurance policy in China. In the endeavor to fight against health inequities, both job-based and community-based health insurance programs are supposed to improve access to health care services. This study manifests distinct associations between different types of health insurance plans and health care access, particularly on inpatient care. On the one hand, urban-centric insurance enrollees had lower use of outpatient care compared to the rural population, partly due to urban-rural disparities in self-treatment. On the other hand, both uninsured elders and community-based insurance enrollees had lower use of inpatient care than job-based insurance enrollees, demonstrating inequitable access.
Risking Pooling for All? The Paradox of Segmented Insurance Plans
Figure 4 illustrates the weighted percentage of socioeconomic status distribution, prevalence rate of health conditions and access to inpatient services by insurance status in this CFPS sample. In the light of Andersen’s framework, at the level of predisposing characteristics, the results of bivariate analyses indicate that recipients of the job-based insurance (GEMI or UEBMI) were, indeed, associated with higher socioeconomic status. Thus, they were richer in resources of knowledge, money, power, and beneficial social connections to handle medical conditions than elders with community-based programs (URBMI or NCMS) and the uninsured. 8 In terms of health care needs, the data reveal that uninsured elders and community-based insurance enrollees were more vulnerable to multiple risk exposure, such as mental distress, functional limitation, poorer perceived health status, inadequate pensions, or identified caregivers. Ironically, when it comes to the policy level, the multivariate analyses demonstrate that although these socioeconomically disadvantaged groups have a greater burden of disease, the systemic barrier, institutional ineligibility and financial inability to purchase better health insurance, hinders them from accessing inpatient care. This stratification consequence was not ameliorated but further exacerbated by the insurance system. Such stratifying nature of social health insurance plans, in turn, perpetuates existing unequal distributions of welfare provision. 49
The percentage of socioeconomic status distribution, prevalence rate of health conditions and access to inpatient services by insurance status (chi-square, P < .01). GEMI, Government Employee Medical Insurance; NCMS, New Rural Resident Cooperative Medical Scheme; UEBMI, Urban Employee Basic Medical Insurance; URBMI, Urban Resident Basic Medical Insurance.
Ultimately, without any affiliation to formal institutions, community-based insurance enrollees and uninsured elders who were previously unemployed, self-employed, or informally employed encountered not only the intersectionality of social marginalization but also structural exclusion from the job-based insurance system. 50 Despite limited financial risk protection against catastrophic health expenditures, these precarious conditions persist throughout their life course. As such, maintaining the boundary of the mandatory job-based plans and voluntary community-based plans is questionable. Considering massive rural populations and urban unemployed residents in China, reducing gaps in the benefit package and integrating segmented risk pooling across these programs will be a worthwhile policy change to achieve cross-subsidization from the healthy to the ill and from the well-off to the needy.
Call for Policy Change: Toward an Integrated Health Insurance System
Recently, the Chinese government has been exploring new health financing models in eastern coastal provinces to integrate health insurance plans. Policymakers, health practitioners, and researchers are increasingly reaching a consensus that “the key in consolidation of the health insurance plans is to unify the plans in terms of their funding levels, cost-sharing methods, and payment systems.” 19 These initiatives, however, have been constrained within voluntary community-based insurances by combining NCMS with URBMI.51,52(p272) Other policy practices have been limited to incorporating mandatory job-based plans by transferring GEMI to UEBMI. 8 The results of this study highlight that it is the job-based and community-based institutional divide that creates the insurance-related inequities in health care access. The Chinese government should move beyond a superficial strategy that merges homogeneous insurance programs.
In 2018, a newly established State Medical Insurance Administration (guojia yiliao baozhang ju) 53 took over UEBMI and URBMI plans from the Ministries of Human Resources and Social Security and the NCMS plan from the National Health and Family Planning Commission. This 2018 State Council Institutional Reforms provide an unprecedented window of opportunity for a top-down approach to unify the segmented insurance system. In the pursuit of universal insurance coverage by 2020, breaking the institutional divide of insurance plans is indispensable so that the underserved aging population can benefit from an equitable health care system.
Supplemental Material
Supplemental material for Inequities in Access: The Impact of a Segmented Health Insurance System on Physician Visits and Hospital Admissions Among Older Adults in the 2014 China Family Panel Studies
Supplemental Material for Inequities in Access: The Impact of a Segmented Health Insurance System on Physician Visits and Hospital Admissions Among Older Adults in the 2014 China Family Panel Studies by Shen (Lamson) Lin in International Journal of Health Services
Footnotes
Acknowledgments
The author would like to express gratitude to the anonymous reviewer for providing critical insights. The author is grateful to the financial support from the University of Toronto Fellowship in sponsoring his doctoral study and research. The author thanks Professor Chen Xiao, Professor Esme Fuller-Thomson, Ms. Roz Spafford, Mr. Richard Bingham, Mr. James O’Neal for their constructive comments on the early version of this paper. Certain parts of the results were presented in the International Federation on Ageing 14th Global Conference in Toronto and the Society for Social Work and Research (SSWR)–2019 Annual Conference in San Francisco, where he received the 2019 SSWR Doctoral Student Travel Award. The author appreciates the Institute of Social Survey at Peking University for granting access to the data of 2014 China Family Panel Studies.
Ethical Approval
This is an analysis of secondary data so ethics approval is not required.
Declaration of Conflicting Interests
The author declared no potential conflict with respect to the research, authorship, and/or publication of this article.
Funding
The author received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental Material for this article is available online.
Author Biography
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
