Abstract
Background and objectives
Recently, the financial burden on medical insurance funds has increased rapidly. The policy effects of paying for innovation await scientific measurement. This study uses DBS, a high-value medical product, as an example to quantify the causal effect of insurance coverage changes on healthcare utilization, informing global value-based payment strategies.
Methods
We identified 2,665 DBS-related records. After excluding cases with missing data, no implantation, diagnoses for comorbidity management or device adjustment, pre-surgery admissions, and abnormal length of stay, 2,471 cases were included for analysis. A nonparametric sharp regression discontinuity design was adopted to assess the causal effects of reimbursement strategy adjustments on eight measures of healthcare utilization. In addition, we performed robustness analyses.
Results
The reimbursement strategy adjustment significantly reduced the individual out-of-pocket expenditure (β=-109,051.30, P < 0.001), increased medical insurance reimbursement cost (β = 128,347.14, P < 0.001), and shortened both the average length of recovery (β=-28.23, P < 0.01)and the average length of stay (β=-37.41, P < 0.001). After accounting for covariates, the results remained significant, though the effect estimates were all attenuated, suggesting a strong association between demographic and medical characteristics and healthcare utilization. Additionally, the effects on total surgical cost, surgical consumable cost and proportion of drug costs were not statistically significant at the 0.05 level.
Conclusion
The medical insurance reimbursement strategy adjustment effectively reduces patients’ economic burden and improves healthcare utilization. However, it is crucial to consider strengthening cost monitoring for high-value medical technologies, optimizing clinical pathways, and balancing the interests of patients, hospitals, and insurance funds.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12913-025-13301-8.
Keywords: Medical insurance reimbursement, Healthcare utilization, Regression discontinuity design, China
Introduction
To reduce out-of-pocket medical expenses and improve overall population health, China has achieved near-universal health insurance coverage through two schemes: the Urban Employee Basic Medical Insurance and the Urban Rural Resident Basic Medical Insurance [1]. In 2024, the total revenue of the national basic health insurance fund was 3.48 trillion yuan, an increase of 4.4% year-on-year, while the participation rate remained above 95% [2].
The goals of Medicare include reducing the burden of high-value health care costs and improving health through better access to health care services [3, 4]. However, the actual impact of Medicare payments on healthcare utilization remains uncertain [5, 6]. Both common sense and empirical research suggest that health care utilization rises when individuals receive new or improved Medicare payments [7–9]. The inclusion of high-value medical technologies in health insurance has undeniably improved accessibility, yet it has simultaneously created new complexities in resource allocation and cost management within the public healthcare system [10]. Moreover, the precise impact of reimbursement on healthcare utilization remains a subject of debate [7], requiring further exploration.
In recent years, China has emphasized payment for innovation and value-based payment [11, 12]. However, over the past decade, the average growth rate of total health expenditures has consistently exceeded the annual growth rate of its gross domestic product [13]. These trends underscore the urgent need to understand the impact of access to innovative medicines and devices on overall healthcare utilization. Given the growth of the elderly population and the increased incidence of neurodegenerative diseases, we selected Deep Brain Stimulation (DBS), a modulating therapy that is more precise and customised in some areas than traditional neurological drugs, as an example of high-value medical products in this study. DBS was introduced in China in 1998, with the advantage of being efficient, minimally invasive, reversible and modifiable. Studies have shown that DBS has clinical benefits such as enhancing cognitive function and improving motor deficits [14, 15]. Furthermore, compared with medications, DBS can substantially reduce societal costs by decreasing medication expenditures and improving daily activity functioning [16]. To further enhance the accessibility and affordability of DBS, Shanghai, one of China’s most developed and populous cities, began covering DBS under its basic medical insurance on April 1, 2018, with a reimbursement amount of CNY 50,000. Subsequently, on June 15, 2022, the reimbursement policy changed, requiring patients to bear an initial 20% of the cost, with the remaining 80% covered by medical insurance [17]. Assessing DBS utilisation, coverage and equity has been the focus of national discussions [18–20]. No studies have yet examined the impact of DBS reimbursement strategy adjustments on healthcare utilization.
In this study, taking DBS as an example, we examined the causal effects of medical insurance reimbursement policy for high-value medical products on healthcare utilization. We employed nonparametric regression discontinuity designs, one of the quasi-experimental designs, to assess causal effects in environments where randomization is unfeasible. This study aims to provide empirical evidence on the relationship between reimbursement policy and healthcare utilization, and to inform healthcare policy reforms and resource allocation, thus contributing to the sustainability of insurance funds and the improvement of healthcare services.
Methods
Data source
The data used in this study were derived from the medical insurance settlement system in Shanghai, China. The database includes individual-level medical data for all participants covered by the Urban Employee Basic Medical Insurance and the Urban Rural Resident Basic Medical Insurance. Each record contained information on demographic characteristics (patient ID, sex, age, household origin, insurance type, etc.), clinical characteristics (hospital name, hospital level, medical treatment date, department of medical treatment, admission date, operating date, operating name, department of operating, discharge date, main diagnosis, diagnosis related groups, etc.), and cost information (cost components, insurance reimbursement cost, out-of-pocket expenditure, etc.).
We extracted records with the procedure name ‘Deep Brain Stimulation’ from the database for the period between January 1, 2021 and December 31, 2023. A total of 2,665 records were retrieved, of which 2,471 met the inclusion and exclusion criteria and were included in the final analysis. The exclusion criteria were: [1] missing values for patient ID or healthcare utilization information; [2] patients without confirmed DBS implantation, indicated by surgical disposable costs below CNY 50,000; [3] disease diagnosis related to comorbidity management or device adjustments; [4] records of repeated visits that represented admissions prior to the formal DBS surgery; and [5] abnormal length of stay (based on the Pauta criterion).
Variables
The outcome variables
We considered the following eight outcome variables to measure healthcare utilization: [1] Total surgical volume; [2] Average surgical cost; [3] Average out-of-pocket expenditure for surgery; [4] Average medical insurance reimbursement cost; [5] Average length of stay: average number of days a patient stayed in the hospital; [6] Average length of recovery: average number of days from surgery to discharge; [7] Surgical consumable cost: cost of disposable surgical consumables; and [8]Proportion of drug costs: proportion of surgical medication cost to total surgical cost.
Specifically, average surgical cost reflects the consolidated expenses associated with the operation. Average out-of-pocket expenditure and medical insurance reimbursement cost reflect the economic burden borne by two principal entities in the healthcare system. Average length of stay and average length of recovery represent the efficiency of healthcare services. Surgical consumable cost and proportion of drug costs capture the cost structure and resource utilization of healthcare services. All outcome variables were measured at the individual level and all costs were analyzed and reported in Chinese Yuan.
The independent variables and covariates
The key independent variable in our study was medical insurance reimbursement policy for DBS. Given that June 15, 2022, is the specific date of the adjustment in the medical insurance reimbursement strategy, we designate this date as the cutoff point.
Several covariates were included in our study to adjust the model, including demographic characteristics (age and sex), and medical characteristics (hospital level and disease type).
Statistical analysis
Univariate analysis
We first performed descriptive statistics for all variables, and conducted univariate analysis to test the differences between the group with and without the new reimbursement policy. Categorical variables were presented as frequencies (N) and percentages (%), while continuous variables were presented as means and standard deviations (SD).
Regression discontinuity design
We employed a regression discontinuity design (RDD) to explore the causal effects of the DBS reimbursement policy on inpatient healthcare utilization. RDD, a quasi-experimental approach particularly suitable for policy evaluation when randomized trials are impractical, which can be used to measure the impact of interventions by applying a treatment assignment mechanism based on a continuous variable (referred to as ‘running variable’ or ‘assignment variable’) [21]. RDD leverages clinical or policy decision rules in which individuals fall on either side of a threshold or cutoff for recommending treatment. RDD can be classified into sharp regression discontinuity design (SRDD) and fuzzy regression discontinuity design (FRDD). In FRDD, the probability of receiving the intervention jumps from ‘a’ to ‘b’ (0 < a < b < 1) at the threshold, whereas in SRDD, the probability deterministically jumps from 0 to 1 [22, 23].
In this study, the assignment variable was the date, with the threshold set as the date of the new DBS reimbursement policy implementation (June 15, 2022). On the day following the policy’s release, the medical insurance settlement system of hospitals in Shanghai implemented the latest policy. Once the date passed the cutoff point, all individuals were subject to the new policy, thus we adopted SRDD. We tested this assumption by graphically inspecting the density plot of the assignment variable and its components for each sample utilized in the analysis. For policymakers, our estimates reflect the local average treatment effects for patients receiving DBS near the policy transition, the direct financial impact per case under the new reimbursement scheme, and utilization changes specifically attributable to the payment adjustment.
We fitted regression discontinuity models using nonparametric local linear regression, as recommended by Imbens and Lemieux [24], to minimize the risks of data overfitting and misleading impact associated with higher-order polynomial models. Meanwhile, we used the data-driven optimal bandwidth method proposed by Imbens-Kalyanaraman (IK) to estimate the bandwidth around the threshold [25]. The optimal bandwidth used in each regression was reported in results.
Therefore, the following regression equation was used to estimate the causal effects of the reimbursement policy:
 |
where
represents the outcome variables, excluding surgical volume (since surgical volume was distributed continuously around the cutoff date). 
is a dummy variable, assigned a value of one after the threshold date and zero otherwise. 
represents the intercept and 
reflets the impact of the reimbursement policy on the outcomes. 
denotes the covariates included in our analysis, and the vector 
represents the respective coefficients. 
is the error term.
To test the robustness of results, we explored the sensitivity of our findings to different potential bandwidths [26, 27]. Additionally, we constructed higher-order (first, second, third, fourth) polynomial regression models to fit the data and capture time effects [27]. Furthermore, we employed global regression models and selected the optimal model based on the Akaike Information Criterion (AIC) to further assess robustness [28].
A two-sided P-value < 0.05 was considered to be statistically significant. All statistical analyses were performed using R software (version 4.5.0).
Results
Sample characteristics
As shown in Table 1, the final analysis included 2,471 surgical inpatients identified between January 2021 and December 2023. The age ranged from 2 to 81 years, with an average age of 54.60 years, and male participants accounted for 54.0% of the sample. The disease types were primarily consistent with the approved indications for DBS in China, with 67.6% diagnosed with Parkinson’s disease, and 11.1% diagnosed with other drug-refractory conditions such as bipolar disorder and cerebral infarction.
Table 1.
Sample characteristics between two groups
| Characteristics | Total (N = 2,471) |
Policy = 0 (N = 546) |
Policy = 1 (N = 1925) |
P value |
|---|---|---|---|---|
| Age, years (mean, SD) | 54.60 (17.46) | 53.18 (18.49) | 55.00 (17.14) | 0.009 |
| Sex (N, %) | 0.734 | |||
| Male | 1,333 (54.0) | 291 (53.3) | 1,042 (54.1) | |
| Female | 1,138 (46.1) | 255 (46.7) | 883 (45.9) | |
| Year of surgery (N, %) | < 0.001 | |||
| 2021 | 460 (18.6) | 460 (84.3) | 0 (0.0) | |
| 2022 | 603 (24.4) | 86 (15.8) | 517 (26.9) | |
| 2023 | 1408 (57.0) | 0 (0.0) | 1408 (73.1) | |
| Medical Insurance (N, %) | 0.098 | |||
| Urban employee basic medical insurance | 1,323 (53.5) | 275 (50.4) | 1,048 (54.4) | |
| Urban rural resident basic medical insurance | 1,148 (46.5) | 271 (49.7) | 877 (45.6) | |
| Household origin (N, %) | < 0.001 | |||
| Eastern | 1,502 (60.8) | 370 (67.8) | 1,132 (58.8) | |
| Central | 604 (24.4) | 109 (20.0) | 495 (25.7) | |
| Western | 240 (9.7) | 9.34 (51.0) | 189 (9.8) | |
| Northeastern | 122 (5.0) | 14 (2.6) | 108 (5.6) | |
| Disease type (N, %) | < 0.001 | |||
| Parkinson’s disease | 1,671 (67.6) | 353 (64.7) | 1,318 (68.5) | |
| Intractable epilepsy | 258 (10.4) | 67 (12.3) | 191 (9.9) | |
| Dystonia | 269 (10.9) | 52 (9.5) | 217 (11.3) | |
| Other | 273 (11.1) | 74 (13.6) | 199 (10.3) | |
| Diagnosis related group (N, %) | < 0.001 | |||
| Neurostimulator implantation or removal | 1,565(63.3) | 262 (48.0) | 1,303 (67.7) | |
| Craniotomy other than trauma | 478 (19.3) | 254 (46.5) | 224 (11.6) | |
| Surgery for psychiatric patients (without complications or comorbidities) | 67 (2.7) | 18 (3.3) | 49 (2.6) | |
| Other | 361 (14.6) | 12 (2.2) | 349 (18.1) |
SD standard deviation
Univariate analysis
As shown in Table 2, all outcome variables representing healthcare utilization showed statistically significantly differences (P < 0.05) before and after the reimbursement strategy adjustment. On average, after the implementation of the new reimbursement policy, the total surgical cost and medical insurance reimbursement cost increased by CNY 29,739.86 and CNY 145,842.92, respectively, while individual out-of-pocket expenditure decreased by CNY 116,103.06. Besides, the average length of stay and recovery were shortened by 0.88 days and 0.72 days, respectively, and the proportion of drug costs decreased by 0.8%.
Table 2.
Univariate analysis of outcome variables associated with reimbursement policy
| Outcome variables | Mean (SD) | P value | ||
|---|---|---|---|---|
| Policy = 0 | Policy = 1 | Difference | ||
| Total surgical volume (N) | 546 | 1925 | 1379 | < 0.001 |
| Average surgery cost (CNY) |
251,320.30 (58,774.29) |
281,060.16 (50,289.88) |
29,739.86 (2,764.16) |
< 0.001 |
| Average out-of-pocket expenditures (CNY) |
189,207.99 (52,469.05) |
73,104.93 (48,111.96) |
−116,103.06 (−2,498.92) |
< 0.001 |
| Average medical insurance reimbursed cost (CNY) |
62,112.31 (21,988.26) |
207,955.23 (67,739.34) |
145,842.92 (1,808.09) |
< 0.001 |
| Average length of stay (day) |
12.45 (9.12) |
11.57 (7.89) |
−0.88 (−0.43) |
0.039 |
| Average length of recovery (day) |
7.22 (6.89) |
6.50 (5.84) |
−0.72 (−0.32) |
0.017 |
| Surgical consumable cost (CNY) |
215,793.36 (60,758.59) |
247,899.15 (51,907.81) |
32,105.80 (2,856.73) |
< 0.001 |
| Proportion of drug costs (%) |
3.5% (2.4%) |
2. 7% (1.6%) |
−0.8% (−0.1%) |
< 0.001 |
SD standard deviation, CNY Chinese Yuan. The exchange rate is 1 CNY = 0.14 USD
Sharp regression discontinuity design
Manipulation of assignment variable
The regression discontinuity design requires that there must be no manipulation of the assignment variable around the threshold. Based on the formal test developed by McCrary [29], we failed to reject the null hypothesis of a smooth density across the threshold. As shown in Fig. 1, which provided further support that there was no manipulation of the assignment variable.
Fig. 1.
McCrary density test plot
Validity tests
The results of the regression model with covariates are shown in Fig. 2, which indicated the discontinuity of healthcare utilization outcomes at the cutoff date. Therefore, it is justifiable to conduct the causal analysis based on a SRDD.
Fig. 2.
Regression discontinuity plot
Causal effect estimates
Table 3 presents the nonparametric estimation results of models with and without covariates, indicating the causal effects of the reimbursement policy on healthcare utilization. As shown by the results, regardless of covariate adjustment, the total surgical cost, surgical consumable cost and proportion of drug costs did not exhibit significant changes (P > 0.05) under the given optimal bandwidth (column 7). In contrast, the DBS reimbursement strategy adjustment was associated with a decrease of CNY 109,051.30 (P < 0.001) in individual out-of-pocket expenditure, an increase of CNY 128,347.14 (P < 0.001) in medical insurance reimbursement cost, and a shorten of 28.23 days (P < 0.01) in the average length of recovery, although the effect estimates were attenuated after adjusting for covariates. Furthermore, the average length of stay only showed a significance change (
=−37.41, P < 0.001) in the unadjusted model.
Table 3.
Sharp regression discontinuity estimates
| Outcome variables | Model a | Coefficient | Standard Error | Z value | Pvalue | Bandwidth |
|---|---|---|---|---|---|---|
| Average surgery cost | Unadjusted | 27,900.82 | 33,274.46 | 0.84 | 0.402 | 157.31 |
| Adjusted | −23,415.62 | 49,259.41 | −0.48 | 0.635 | 130.75 | |
| Average out-of-pocket expenditures | Unadjusted | −109,051.30 | 22,897.87 | −4.76 | < 0.001 | 184.8 |
| Adjusted | −89,531.67 | 32,997.21 | −2.71 | 0.007 | 261.56 | |
| Average medical insurance reimbursed cost | Unadjusted | 128,347.14 | 19,134.92 | 6.71 | < 0.001 | 155.42 |
| Adjusted | 82,793.70 | 38,235.86 | 2.17 | 0.031 | 301.28 | |
| Average length of stay | Unadjusted | −37.41 | 12.15 | −3.08 | < 0.001 | 227.31 |
| Adjusted | −21.56 | 13.87 | −1.55 | 0.120 | 122.84 | |
| Average length of recovery | Unadjusted | −28.23 | 10.52 | −2.69 | 0.007 | 232.85 |
| Adjusted | −20.92 | 10.33 | −2.03 | 0.043 | 272.69 | |
| Surgical consumable cost | Unadjusted | 15,147.25 | 29,765.32 | 0.51 | 0.611 | 223.43 |
| Adjusted | 9,178.58 | 58,412.76 | 0.16 | 0.875 | 146.80 | |
| Proportion of drug costs | Unadjusted | −0.02 | 0.02 | −1.12 | 0.265 | 131.56 |
| Adjusted | −0.01 | 0.02 | −0.66 | 0.511 | 166.59 |
aAdjusted Models were adjusted for covariates, including age, sex, hospital level and disease type
Robustness tests
To test the robustness of our findings, we re-estimated the regression model using different bandwidth choices, including 50%, 80%, 120% and 150% of the optimal bandwidth used in the baseline regression. The results, presented in Supplementary Table S1, consistently demonstrated the statistical significance observed in the primary analyses, thereby confirming the robustness of our findings.
Besides, the results from fitting the data using higher-order polynomial models demonstrated that the fourth-order regression results were generally consistent with the results from the first-order regressions (see Supplementary Table S2 for details). Furthermore, the results of the global regression model, adjusted for covariates, are shown in Supplementary Table S3. These results consistently indicated that the reimbursement strategy adjustment had a significant negative impact on individual out-of-pocket expenditure, length of stay, and coverage, while demonstrating a significant positive impact on medical insurance reimbursement cost, thereby further confirming the robustness.
Discussion
This study, using DBS as an example, explored the causal effects of medical insurance reimbursement policy on healthcare utilization based on data derived from the medical insurance settlement system in Shanghai, China. By applying SRDD, our results demonstrate that the reimbursement strategy adjustment significantly reduced patients’ economic burden and improved healthcare services efficiency. Specifically, without considering the covariates, individual out-of-pocket expenditure decreased by approximately 43.4% of the pre-policy expenditure, and the average length of stay and recovery, which were indicators of healthcare services efficiency, were both significantly reduced, respectively. Besides, medical insurance reimbursement cost increased significantly, with an increment of about CNY 128,347.14 could be attributed to the reimbursement strategy adjustment. Theses findings highlight the importance of balancing medical insurance payments with disease medical costs. Sensitivity analyses using different bandwidths, along with global regression and higher-order polynomial regression models, confirmed the robustness of our results, which are similar to previous studies [16, 30].
Furthermore, we initially explored the differences in outcome variables before and after the new reimbursement policy implementation using univariate analysis. The results showed that all variables changed significantly. The observed increases in total surgical expenditures and insurance reimbursements may be primarily attributed to policy-driven incentives and an expansion in the utilization of medical services. While the reform has effectively alleviated patients’ financial burden, it may simultaneously introduce concerns regarding the long-term sustainability of the insurance fund and the potential for deviations in provider behavior.
However, these findings differed from the results of the SRDD, which indicated that the coefficients for total surgical cost, surgical consumable cost and proportion of drug costs were not statistically significant at the 0.05 level. Other studies have also suggested that results may differ between univariate analysis and regression discontinuity designs [31, 32]. One potential reason for this discrepancy could be that univariate analysis assumes the reimbursement strategy as the sole influencing factor, while SRDD, by nature, accounts for additional factors that may also contribute to the outcomes. For instance, surgical costs may also be affected by factors such as iterative product updates and suppliers’ market-oriented behavior, which partially explains why the SRDD results show an insignificant coefficient for surgical cost and opposing effects before and after adjusting covariates. Similarly, the proportion of drug costs and surgical consumable cost could be impacted by other policies, such as Diagnosis Related Group (DRG) payment adjustment, which could undermine the causal relationship between these outcomes and the reimbursement policy. These findings also indicate that there may be heterogeneity or dynamics in policy effects. The observed effects align with China’s systemic transition toward value-based healthcare delivery, particularly evident in Shanghai’s implementation of blended payment models combining innovative reimbursement incentives with cost-control measures. Evaluations should not simply zoom in on one policy tool [33, 34].
We conducted both unadjusted and covariate-adjusted models in the SRDD, with covariates including age, sex, hospital level and disease type. The results showed that, when considering covariates, all effect estimates decreased, suggesting that demographic and medical characteristics are strongly associated with healthcare utilization. This also indicates that the unadjusted model may somewhat overestimate the effects of the reimbursement strategy adjustment on the outcome variables. Notably, the significance of average length of stay was not consistent before and after controlling for covariates (P < 0.001 vs. P = 0.12), indicating that this outcome was strongly influenced by the included covariates. Some clinical studies have also confirmed that neurodegenerative diseases often exhibit different patterns of progression in men and women, leading to substantial heterogeneity in resource utilization between patients [35].
The findings of this study have important policy implications. Globally, the current DBS reimbursement policies vary widely across regions, such as Hong Kong, Taiwan, the United States, and Canada [36, 37]. However, all of these regions have relevant studies confirming the significant spillover effects and long-term health benefits of DBS reimbursement policies [38, 39]. While the current analysis suggests potential benefits for patients and improved financial accessibility, it is also important to acknowledge the possible shift in financial burden toward the medical insurance system, raising concerns about long-term fiscal sustainability. To ensure a more balanced and sustainable approach, policy design could consider targeted or tiered coverage strategies. For example, selectively expanding coverage for high-value interventions while limiting reimbursement for lower-value or marginally effective technologies could help manage resource allocation more effectively. Incorporating cost-effectiveness thresholds or outcome-based reimbursement models may also provide mechanisms to control expenditure while maintaining equitable access [40]. It is also essential to take action to improve surgical management capacity and develop appropriate medical insurance policies that address regional equity. Meanwhile, governments should establish cost-monitoring mechanisms for high-value medical products, optimize clinical pathways, and prevent cost increases resulting from overmedication or unreasonable market practices. Under the current medical insurance payment system, preoperative assessments can be strengthened, and standard payment units can be optimized by considering factors such as surgical complexity and disease severity. In the future, it may also be valuable to explore payment models based on treatment outcomes and health improvements. In addition, there should be increased emphasis on intelligent monitoring of medical insurance to effectively identify harmful practices such as avoiding responsibility of severe diseases, selecting treating mild diseases, decomposing hospital days, cost-shifting, upcoding services and lowering service quality. These actions will help further guarantee the sustainability of health insurance systems.
This study has several limitations. First, while our study utilizes a comprehensive database encompassing all beneficiaries of Shanghai’s basic social health insurance system, the generalizability of findings may be contextually constrained to developed regions. Considering socio-economic differences, factors such as healthcare financing capacity, patient cost-sharing burdens, and medical resource allocation efficiency must be systematically evaluated prior to cross-regional policy extrapolation. Second, we did not include data on disaggregated items, such as cost of examinations, beds and services, which could provide further insights into the main drivers of surgical cost increases. Further studies could explore these components to better understand the items contributing to cost escalation. Third, interventions like DRG payment adjustment may also impact the included outcome variables. However, due to varying implementation dates across healthcare institutions, we were unable to exclude their potential effects, which may result in an overestimation of the DBS reimbursement strategy’s impact. Fourth, given the limitations of the SRDD, particularly its external validity, the results are primarily applicable to groups near the cutoff of the running variable (i.e., the date of the new reimbursement policy implementation). Therefore, the findings may not be generalizable to broader populations or different scenarios.
Conclusion
Leveraging DBS-related data derived from the medical insurance settlement system in Shanghai, China, our study reveals that medical insurance reimbursement strategy adjustment has important implications for healthcare utilization. These include reducing the economic burden on patients, improving the efficiency of hospitalisation services and increasing the medical insurance reimbursement cost. To achieve the sustainable development of the healthcare system, actions should be taken to balance the interests of patients, hospitals, and medical insurance funds, while also adopting refined regulatory measures and optimising the service management mechanism.
Supplementary Information
Acknowledgements
Not applicable.
Authors’ contributions
Conception and design–Q.X., T.C., and H.W. Procurement of data–D.H. and T.C. Analysis of data–Q.X. and H.H. Drafting of the original manuscript–Q.X. Critical review of the original manuscript–D.H.，W.W., T.C., Y.Y., and H.W. All authors read and approved the final manuscript.
Funding
This study was funded by the Shanghai Municipal Science and Technology Commission (Grant No. 25692113900) under the 2025 High-Level Institution Development Program– Soft Science Research.
Author information.
Data availability
Data supporting this study were obtained from the Shanghai Municipal Medical Insurance Bureau but are subject to restrictions and not publicly available. Pending Bureau approval, data may be made available by the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
The protocol used in this study was approved by the Institutional Review Board of Shanghai Health Development Research Center (Shanghai Medical Information Center) (IRB number: SHDRC2025002). All methods were performed in accordance with the relevant guidelines and regulations (Declaration of Helsinki). The current study involved all the secondary data obtained from medical records, and the requirement of the informed consent for this study has been waived by the Institutional Review Board of Shanghai Health Development Research Center (Shanghai Medical Information Center).
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Ren Y, Zhou Z, Cao D, Ma BH, Shen C, Lai S, et al. Did the integrated urban and rural resident basic medical insurance improve benefit equity in china?? VALUE HEALTH. 2022;25(9):1548–58. [DOI] [PubMed] [Google Scholar]
- 2.Qiu Y. Development of China’s healthcare system in the past seventy years: review and analysis. Chin Soc Secur Rev. 2019;3(01):89–101. [Google Scholar]
- 3.Elshaug AG, Rosenthal MB, Lavis JN, Brownlee S, Schmidt H, Nagpal S, et al. Levers for addressing medical underuse and overuse: achieving high-value health care. Lancet. 2017;390(10090):191–202. [DOI] [PubMed] [Google Scholar]
- 4.Morenz AM, Liao JM, Au DH, Hayes SA. Area-level socioeconomic disadvantage and health care spending: a systematic review. JAMA Netw Open. 2024;7(2):e2356121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lichtenberg FR, edito. The effects of medicare on health care utilization and outcomes. Pub Place: De Gruyter; Year Published: Forum for Health Economics & Policy; 2002. [Google Scholar]
- 6.Ghijben P, Petrie D, Zavarsek S, Chen G, Lancsar E. Behavioral responses to healthcare funding decisions and their impact on value for money: evidence from Australia. Health Econ. 2025. 10.1002/hec.4958. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.McWilliams JM, Meara E, Zaslavsky AM, Ayanian JZ. Use of health services by previously uninsured medicare beneficiaries. NEW ENGL J MED. 2007;357(2):143–53. [DOI] [PubMed] [Google Scholar]
- 8.Zhou RA, Baicker K, Taubman S, Finkelstein AN. The uninsured do not use the emergency department more—they use other care less. HEALTH AFFAIR. 2017;36(12):2115–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Gaffney A, Himmelstein DU, Woolhandler S, Kahn JG. Pricing universal health care: how much would the use of medical care rise?? Study examines medical care use growth that could result from universal coverage. Health Aff. 2021;40(1):105–12. [DOI] [PubMed] [Google Scholar]
- 10.Javanmardi E, Maresova P, Xie N, Mierzwiak R. Exploring business models for managing uncertainty in healthcare, medical devices, and biotechnology industries. Heliyon. 2024. 10.1016/j.heliyon.2024.e25962. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Chen V, Shao W. Balancing medical innovation and affordability in the new healthcare ecosystem in China: review of pharmaceutical pricing and reimbursement policies. Health Care Science. 2023;2(6):381–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Lin K, Li Y, Yao Y, Xiong Y, Xiang L. The impact of an innovative payment method on medical expenditure, efficiency, and quality for inpatients with different types of medical insurance: evidence from a pilot city, China. Int J Equity Health. 2024;23(1): 115. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Yip W, Fu H, Chen AT, Zhai T, Jian W, Xu R, et al. 10 years of health-care reform in China: progress and gaps in universal health coverage. Lancet. 2019;394(10204):1192–204. [DOI] [PubMed] [Google Scholar]
- 14.Shashkin C. Complications of deep brain stimulation for movement disorders: literature review and personal experience. Complications Neurosurg. 2023;130:121–6. 10.1007/978-3-030-12887-6_15. [DOI] [PubMed]
- 15.Huang Z, Meng L, Bi X, Xie Z, Liang W, Huang J. Efficacy and safety of robot-assisted deep brain stimulation for Parkinson’s disease: a meta-analysis. Front Aging Neurosci. 2024;16:1419152. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Smilowska K, van Wamelen DJ, Pietrzykowski T, Calvano A, Rodriguez-Blazquez C, Martinez-Martin P, et al. Cost-Effectiveness of Device-Aided therapies in parkinson’s disease: A structured review. J Parkinson’s Disease. 2021;11(2):475–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Guo X, Feng C, Pu J, Jiang H, Zhu Z, Zheng Z, et al. Deep brain stimulation for advanced Parkinson disease in developing countries: a cost-effectiveness study from China. Neurosurgery. 2023;92(4):812–9. [DOI] [PubMed] [Google Scholar]
- 18.Stein A, Higgins N, Gajwani M, Gericke CA. The treatment gap for deep brain stimulation in Parkinson’s disease: a comparative analysis of cost and utilisation in high-income countries. Aust Health Rev. 2024;48(5):497–505. [DOI] [PubMed] [Google Scholar]
- 19.Lannon M. Comprehensive Needs Assessment for Deep Brain Stimulation in Canada. A Health Service Research Perspective; 2024. [Google Scholar]
- 20.Sarica C, Conner CR, Yamamoto K, Yang A, Germann J, Lannon MM, et al. Trends and disparities in deep brain stimulation utilization in the United States: a nationwide inpatient sample analysis from 1993 to 2017. The Lancet Regional Health–Americas. 2023. 10.1016/j.lana.2023.100599. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Villamizar Villegas M, Pinzon Puerto FA, Ruiz Sanchez MA. A comprehensive history of regression discontinuity designs: an empirical survey of the last 60 years. J Econ Surv. 2022;36(4):1130–78. [Google Scholar]
- 22.Venkataramani AS, Bor J, Jena AB. Regression discontinuity designs in healthcare research. BMJ. 2016;352:i1216. 10.1136/bmj.i1216. [DOI] [PMC free article] [PubMed]
- 23.Mi J, Han X, Cheng H, Pan Z, Guo J, Zhu H, et al. Effect of retirement on the body mass index in China: a nationwide study based on the regression discontinuity design. BMC Public Health. 2025;25(1):382. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Imbens GW, Lemieux T. Regression discontinuity designs: A guide to practice. J ECONOMETRICS. 2008;142(2):615–35. [Google Scholar]
- 25.Imbens G, Kalyanaraman K. Optimal bandwidth choice for the regression discontinuity estimator. Rev Econ Stud. 2012;79(3):933–59. [Google Scholar]
- 26.Calonico S, Cattaneo MD, Titiunik R. Robust nonparametric confidence intervals for regression-discontinuity designs. Econometrica. 2014;82(6):2295–326. [Google Scholar]
- 27.Calonico S, Cattaneo MD, Farrell MH, Titiunik R. Regression discontinuity designs using covariates. REV ECON STAT. 2019;101(3):442–51. [Google Scholar]
- 28.Kasali J, Adeyemi AA. Model-data fit using Akaike information criterion (AIC), Bayesian information criterion (BIC), and the sample-size-adjusted BIC. Square: Journal of Mathematics and Mathematics Education. 2022;4(1):43–51. [Google Scholar]
- 29.McCrary J. Manipulation of the running variable in the regression discontinuity design: a density test. J Econ. 2008;142:698–714. [Google Scholar]
- 30.Yong M, Xianjun X, Jinghu L, Yunyun F. Effect of health insurance on direct hospitalisation costs for in-patients with ischaemic stroke in China. Aust Health Rev. 2017;42(1):39–44. [DOI] [PubMed] [Google Scholar]
- 31.Boon MH, Craig P, Thomson H, Campbell M, Moore L. Regression discontinuity designs in health: a systematic review. Epidemiology. 2021;32(1):87–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Dai T, Jiang S, Liu X, Sun A. The effects of a hypertension diagnosis on health behaviors: A two-dimensional regression discontinuity analysis. HEALTH ECON. 2022;31(4):574–96. [DOI] [PubMed] [Google Scholar]
- 33.Wu J, He X, Feng XL. Can case-based payment contain healthcare costs?-A curious case from China. Soc Sci Med. 2022;312: 115384. [DOI] [PubMed] [Google Scholar]
- 34.Chen Y, Zhang X, Tang X, et al. How do inpatients’ costs, length of stay, and quality of care vary across age groups after a new case-based payment reform in China?? An interrupted time series analysis. BMC Health Serv Res. 2023;23(1): 160. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Li G, Ma J, Cui S, He Y, Xiao Q, Liu J, et al. Parkinson’s disease in China: a forty-year growing track of bedside work. Transl Neurodegener. 2019. 10.1186/s40035-019-0162-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Sharma VD, Safarpour D, Mehta SH, Vanegas-Arroyave N, Weiss D, Cooney JW, et al. Telemedicine and Deep brain stimulation - Current practices and recommendations. Parkinsonism Relat D. 2021;89(1):199–205. [DOI] [PubMed] [Google Scholar]
- 37.Zhang C, Ramirez-Zamora A, Meng F, Lin Z, Lai Y, Li D, et al. An international survey of deep brain stimulation utilization in Asia and Oceania: the DBS think tank East. Front Hum Neurosci. 2020;14: 162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Meng F, Hu W, Wang S, Tam J, Gao Y, Zhu XL, et al. Utilization, surgical populations, centers, coverages, regional balance, and their influential factors of deep brain stimulation for parkinson’s disease: a large-scale multicenter cross-sectional study from 1997 to 2021. INT J SURG. 2023;109(11):3322–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.McIntosh E, Gray A, Daniels J, Gill S, Ives N, Jenkinson C, et al. Cost-utility analysis of deep brain stimulation surgery plus best medical therapy versus best medical therapy in patients with Parkinson’s: economic evaluation alongside the PD SURG trial. Mov Disord. 2016;31(8):1173–82. [DOI] [PubMed] [Google Scholar]
- 40.Hefti L, Boëthius H, Loppow D, Serry N, Martin R, Rupalla K, et al. The Tango to modern collaboration and Patient-Centric value generation in health care–a real-world guide from practitioners for practitioners: A field analysis on value-Based health care of 12 leading institutions worldwide. CURR MED RES OPIN. 2025;41(1):31–41. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data supporting this study were obtained from the Shanghai Municipal Medical Insurance Bureau but are subject to restrictions and not publicly available. Pending Bureau approval, data may be made available by the corresponding author upon reasonable request.