Diagnosis-Related group payment and its impact on hospitalization costs and patient financial protection for neurointerventional stroke care in a Chinese tertiary hospital: a retrospective analysis Authors: Zhishui Chen, Tao Zhang, Kai Meng, Haijing Guan, Beibei Xu Published: 2026 ABSTRACT BackgroundStroke continues to impose a substantial economic burden globally. To mitigate escalating healthcare costs and enhance financial protection for patients, diagnosis-related group (DRG) payment systems have been widely implemented in China. Neurointerventional stroke care, characterized by its complexity and high costs, represents a critical area for evaluating the impact of such payment reforms. This study investigates the effects of the DRG payment system on hospitalization costs and patient out-of-pocket expenditures (OOPE) for neurointerventional stroke patients in a leading neurological hospital in China.MethodsWe conducted a retrospective cohort study using 2024 administrative data from Beijing Tiantan Hospital, China. The study included 5,205 inpatient stroke patients who underwent neurointerventional procedures (BE2 codes). Patients were categorized into a DRG payment group (Beijing residents, 18.87%) and a Fee-For-Service (FFS) payment group (non-Beijing residents, 81.13%) based on their medical insurance payment method. Pearson χ² tests and Mann-Whitney U tests were employed to compare total direct hospitalization costs (TDHC) and OOPE between the two groups. Quantile regression analysis further explored the impact of DRG payment across different expenditure levels to assess the heterogeneity of its financial protective effects.ResultsCompared to the FFS group, patients in the DRG payment group demonstrated significantly lower median TDHC (RMB 86,268 vs. RMB 102,885; p < 0.001), representing a 16.15% reduction. Similarly, the median OOPE for DRG patients was significantly lower (RMB 19,851 vs. RMB 30,266; p < 0.001), a 34.41% reduction. Quantile regression analysis confirmed consistent cost reductions across all expenditure quantiles (TDHC: 3.84%-5.16%; OOPE: 14.49%-36.32%), with more pronounced financial protection for patients with higher OOPE levels. Notably, centrally procured medical supplies were associated with increased costs, while higher insurance reimbursement, non-centrally procured medications, ICU transfers, and prolonged length of stay independently predicted higher expenditures.ConclusionsThis single-center study demonstrates a significant association between the DRG payment system and effective cost containment and enhanced patient financial protection for neurointerventional stroke care in a large Chinese neurological center. These findings underscore the necessity for future DRG payment policy development to include specialty-adjusted DRG weights and to ensure coordinated integration between payment reforms and medical supply procurement policies. Due to its single-center nature and reliance on single-year data, caution is warranted when generalizing these results to other institutions and policy contexts. Introduction The global burden of ischemic stroke poses a significant and escalating challenge to healthcare systems worldwide, particularly concerning financial sustainability [1].In 2021, stroke maintained its position as the second leading cause of death globally and the third leading cause of combined death and disability, incurring an annual economic cost exceeding US$890 billion, which represents nearly 0.7% of the global GDP [2]. A disproportionate 87% of stroke- related deaths and 89% of disability-adjusted life-years (DALYs) are observed in low- and lower-middle-income countries (LMICs), highlighting a critical global health inequity [3]. While advanced neuro-interventional therapies have significantly improved patient outcomes [4, 5], their inherent high cost and technological dependence exacerbate financial strain, especially within traditional fee-for-service (FFS) models that tend to reward volume over value. The accessibility of these cutting-edge neuro-interventions often remains restricted to patients with substantial financial resources [6], leading to pronounced disparities in healthcare delivery. This has driven policymakers to seek innovative payment reforms like diagnosis-related group (DRG) systems. Globally, FFS and DRG systems are the predominant inpatient payment models. FFS, with its retrospective reimbursement for individual services, incentivizes higher service volume and costs [7]. In contrast, DRG is a prospective system that provides predetermined, case-mix-adjusted payments, encouraging cost efficiency and care coordination [8]. Although international studies [9, 10] have demonstrated that the adoption of DRG payment systems can measurably influence hospital costs and clinical decision-making, there is a notable gap in research focusing on neuro-interventional stroke cases, particularly in non-Western countries like China, where DRG adoption is relatively recent. In China, DRG has been prioritized in healthcare reform to address rising costs, with Beijing fully implementing DRG reimbursement in March 2022. While initial evaluations [11, 12] have assessed the aggregate effects of this policy, evidence specifically pertaining to the costs associated with neuro-interventional procedures for neurological disorders remains conspicuously scarce. The substantial burden of ischemic stroke among Chinese adults, characterized by a prevalence of 2.6%, an annual incidence of 505.2 per 100,000 individuals, and a mortality rate of 343.4 per 100,000 [13], highlights the critical necessity for rigorous, contemporary evaluations of DRG payment mechanisms. To address this identified research gap, our study focused on neuro-interventional hospitalizations within a single healthcare institution. The primary aim was to gather and assess the total direct hospital costs (TDHC) and out-of-pocket expenditures (OOPE) for patients undergoing neuro-interventional treatment for ischemic stroke, with a particular emphasis on comparing the financial implications between DRG and FFS payment schemes. This comparative analysis is designed to elucidate the differential cost impacts of these payment models, thereby providing a more nuanced understanding of the cost-effectiveness of neuro-interventional therapies under the DRG system. This study operates under the theoretical premise that prospective payment systems, such as DRG, inherently align provider incentives with cost efficiency, whereas retrospective FFS models may inadvertently encourage volume-based resource utilization. We empirically test this premise by examining differential cost outcomes across these distinct payment systems in the context of neuro-interventional stroke care—a high-cost specialty area with limited existing evidence. Our investigation is structured around three core hypotheses: Hypothesis 1 DRG payment is associated with lower TDHC compared to FFS in neuro-interventional stroke care. Hypothesis 2 DRG payment reduces OOPE, with heterogeneous effects across expenditure quantiles. Hypothesis 3 Clinical and policy factors (e.g., centralized procurement, insurance type) modify the association between payment system and cost outcomes. This study fills a gap by focusing on neuro-interventional stroke—a high-cost specialty with limited DRG research—and uses quantile regression to analyze skewed cost data, offering nuanced insights into payment system effects across expenditure strata. Materials and methods Data sources and study population This investigation employed a single-center, retrospective analysis utilizing de-identified administrative data extracted from the electronic medical record system of Beijing Tiantan Hospital. This institution, a prominent tertiary referral center and home to the China National Clinical Research Center for Neurological Diseases (a national research platform jointly established by the Ministry of Science and Technology and the National Health Commission of China), serves a diverse patient population from all 31 provincial-level administrative regions of China. This broad patient catchment area offers a sample that is largely representative of nationwide neuro-interventional practice patterns for stroke care. To contextualize the payment mechanisms underpinning our study, it is important to elaborate on China’s basic healthcare insurance system and the DRG implementation in Beijing, which directly determined the payment methods for the two study groups. China’s basic medical insurance system comprises two primary schemes: the Urban Employee Basic Medical Insurance (UEBMI) and the Urban-Rural Resident Basic Medical Insurance (RURBMI), which collectively cover over 95% of the Chinese population. UEBMI is primarily for urban employees, with higher contribution rates and reimbursement levels, while RURBMI covers rural and non-employed urban residents, with lower contribution rates and corresponding reimbursement standards. In Beijing, the DRG payment system was fully implemented in March 2022 as part of national healthcare reform efforts to control medical costs and improve care efficiency; under this system, hospitals receive a predetermined, case-mix-adjusted payment based on the patient’s diagnosis and procedural codes rather than retrospective reimbursement for individual services. For hospital and patient payment mechanisms, Beijing residents are covered by the local UEBMI or RURBMI schemes, both of which adopt the DRG payment method for inpatient care at Beijing Tiantan Hospital. In contrast, non-Beijing residents, who are covered by their local basic medical insurance schemes (which have not yet fully implemented DRG payment), are subject to the traditional Fee-For-Service (FFS) reimbursement model when seeking medical treatment at Beijing Tiantan Hospital—meaning hospitals are reimbursed retrospectively based on the individual services and supplies provided, and patients pay the portion not covered by their local insurance according to FFS standards. This difference in payment mechanisms, determined by the patient’s administrative residence and corresponding local insurance policies, forms the basis of our quasi-experimental comparison between DRG and FFS groups. Patient cases were identified using the hospital’s DRG pre-grouping platform, which prospectively assigns each inpatient episode to an Adjacent Diagnosis-Related Group (ADRG) based on the principal diagnosis and primary procedure performed. For this study, we specifically selected ADRG “BE2 neuro-interventional procedures” as the primary analytical stratum. This particular grouping encompasses all acute ischemic stroke admissions that were managed with mechanical thrombectomy, stent placement, or balloon angioplasty, thereby ensuring a high degree of clinical homogeneity within the study cohort. During the study period, while non-local patients were subject to the traditional fee-for-service (FFS) reimbursement model, the hospital’s DRG information system simultaneously performed pre-grouping for all inpatient episodes, regardless of the patient’s payment mechanism. This dual classification capability allowed for a quasi-experimental design, wherein Beijing residents (whose care was exposed to DRG-based payment) were compared against non-Beijing residents (whose care remained under retrospective FFS reimbursement) within the identical clinical pathway of neuro-interventional stroke treatment. It is important to acknowledge that this assignment to a payment system was based on administrative residence rather than random allocation, a characteristic inherent to quasi-experimental designs, the limitations of which are addressed in the discussion. Patients were included in the study if they had a discharge date between January 1, 2024, and December 31, 2024, and were assigned the BE2 ADRG code. Exclusion criteria were applied to enhance the consistency and comparability of the data, removing cases that might involve atypical cost structures or clinical courses. Specifically, we excluded: (i) Patients admitted or discharged through international departments, due to their distinct fee structures; (ii) Cases exclusively covered by commercial insurance or fully government-reimbursed healthcare schemes, as these also present unique payment dynamics; (iii) Patients who experienced in-hospital mortality; and (iv) Patients who opted for voluntary discharge against clinical recommendations. Following the application of these criteria, a total of 5,205 eligible patients remained for analysis (Fig. 1). The exclusion of in-hospital mortality and voluntary discharges against medical advice was implemented to standardize the clinical pathways under investigation, acknowledging that these atypical cases often involve significantly different resource utilization patterns. We recognize that this approach might introduce a potential bias by excluding high-risk or potentially high-cost cases, and this limitation is thoroughly discussed in the study’s discussion section. Fig. 1Samples flow chart Samples flow chart The study protocol received approval from the Institutional Review Board of Beijing Tiantan Hospital (Protocol No. KY2025-106) and was conducted in strict adherence to the ethical principles outlined in the Declaration of Helsinki. Given the utilization of anonymized administrative data, the requirement for individual informed consent was waived by the institutional review board. Measures Leveraging China’s fragmented, region-based health insurance pooling system [14], this study compared two distinct patient cohorts: those enrolled in the Beijing municipal insurance pool, primarily reimbursed through DRG payments, and out-of-province patients, who remained under the traditional FFS model. Both cohorts received identical neuro-interventional procedures delivered by the same multidisciplinary stroke team within a single institution, thereby facilitating a robust within-institution comparison. The classification of DRG versus FFS enrollment, determined by the patient’s insurance registration location, is inherently non-random, potentially introducing selection bias. To mitigate this, multivariable models were employed, adjusting for key confounders including age, comorbidity burden, specific insurance type, and clinical severity proxies. It is acknowledged that unobserved factors, such as socioeconomic status and referral patterns, may still influence the results, thus limiting definitive causal inference. The reimbursement regime was operationalized as a binary exposure variable: patients enrolled in Beijing’s municipal insurance pool (local residents) were classified as DRG payees, whereas patients insured by other provinces (non-local residents) were categorized under retrospective FFS. This dichotomy accurately reflects the current Chinese healthcare financing system, where health insurance pooling is region-based, allowing the same hospital to provide identical neuro-interventional services under two distinct payment mechanisms. Variables influencing resource utilization and cost expenditure were systematically collected across three domains: baseline characteristics, clinical characteristics, and outcome measures. Variables were selected based on their theoretical relevance to cost outcomes (e.g., comorbidity quantified by the Charlson Comorbidity Index) and their implications within the prevailing policy context (e.g., centralized procurement status). A comprehensive set of baseline variables was assembled, encompassing demographic, social, and health-financing domains. Age was categorized based on clinical practice conventions into children and adolescents (< 18 years), adults (18–64 years), and elderly adults (≥ 65 years). Marital status was dichotomized into married versus other (single, divorced, widowed), with the latter representing potentially diminished social support networks. Insurance type [15] was classified into Urban Employee Basic Medical Insurance (UEBMI) and Rural and Urban Resident Basic Medical Insurance (RURBMI), reflecting China’s two primary insurance schemes, which possess distinct benefit structures. Disease-specific attributes directly shaping therapeutic decisions and resource utilization were documented. Comorbidity burden was quantified using the Charlson Comorbidity Index (CCI) and categorized as 0 points (none), 1–2 points (mild), 3–4 points (moderate), or ≥ 5 points (severe), adhering to the validated scoring system proposed by Ozgur et al. [16]. Cerebrovascular lesions were categorized into three pathologically distinct entities based on angiographic characteristics: malformation (abnormal vessel development), stenosis (focal luminal narrowing), or aneurysm (focal wall outpouching). Neuro-interventional procedures were coded as embolization, stent placement, or balloon angioplasty, based on standardized criteria [17]. Admission urgency differentiated scheduled (elective) from emergency presentations. Anesthesia modality was dichotomized as local (conscious sedation) or general. Mechanical ventilation and Intensive Care Unit (ICU) transfer were captured as binary indicators of intra- or post-procedural critical care utilization. Furthermore, to reflect China’s national volume-based procurement (VBP) policy [18], data were recorded on whether patients received centrally procured medical supplies (CPMS) and centrally procured medications (CPM). These procurement policies simultaneously reduce unit prices and guarantee supply volumes, thereby directly influencing total hospital costs. Two key post-discharge outcomes were captured to gauge immediate resource use and early clinical stability. First, unplanned 30-day readmission (defined as any inpatient stay within 31 days of the index discharge) was ascertained from routinely linked administrative claims, serving as an indicator of post-procedural complications or premature discharge. Second, length of stay (LOS) was operationalized as a three-level categorical variable: ≤7 days, 8–14 days, and > 14 days, reflecting short, intermediate, and prolonged hospitalizations based on prior health economics literature. Two principal cost outcomes were extracted for each index admission: (1) total direct hospitalization costs (TDHC), defined as the sum of all inpatient charges billed to the payer and/or patient on the date of service, and (2) out-of-pocket expenditures (OOPE), representing the portion of those costs directly borne by the patient after insurance adjudication. All monetary values were expressed in 2024 Chinese Renminbi (RMB) and were not discounted, given the single-year observation window. To delineate cost structure, total direct costs were further disaggregated into four mutually exclusive components: (a) medical consumables costs (including catheters, stents, coils, balloons, and other disposable devices); (b) pharmaceutical costs (all prescription and intravenous medications administered during hospitalization); (c) diagnostic and laboratory service costs (imaging, neurophysiology, blood work, and pathology); and (d) ancillary service costs (including bed charges, nursing services, and other professional fees). Cost proportions were calculated as a percentage of the total per-admission expenditure. This breakdown provided a clear and concise snapshot of how the study cohort’s hospital bills were distributed across major cost categories. Statistical analysis All statistical analyses were performed using Stata/MP 18.0. Baseline characteristics were rigorously compared between the DRG and FFS cohorts. For categorical variables, Pearson’s chi-squared ( χ²) tests were utilized to assess significant differences. Median total direct hospitalization costs (TDHC) and out-of-pocket expenditures (OOPE) were reported alongside their corresponding interquartile ranges (IQR: 25th − 75th percentiles). Consequently, between-group differences in cost distributions were evaluated using the non-parametric Mann-Whitney U test. For multivariable modeling, a two-step strategy was applied. First, TDHC and OOPE were log10-transformed to mitigate skewness, a common practice for highly skewed healthcare expenditure data. Payment type (DRG vs. FFS), along with the complete set of baseline, clinical, and outcome variables identified in the “Measures” section, were incorporated as independent predictors. Second, recognizing that ordinary least-squares regression primarily estimates mean effects, we employed quantile regression (QR) across the 10th, 25th, 50th, 75th, and 90th percentiles. Quantile regression is increasingly recognized as a robust analytical tool for modeling skewed healthcare expenditure data and is widely utilized in contemporary multivariable cost analyses [19, 20]. This method provides a robust, distribution-free approach to quantify the heterogeneous effects of reimbursement policy on low-, median-, and high-cost patient segments, thereby offering a more comprehensive understanding of financial risk heterogeneity. Coefficients derived from the QR models were subsequently used to generate quantile-specific marginal effects for payment type and other covariates. These effects were visually represented in coefficient plots, which displayed the change in log-cost per one-unit change in each predictor at each specific quantile. All statistical tests conducted were two-sided, and a P-value of less than 0.05 was considered statistically significant. Results Descriptive statistics Table 1 provides a comprehensive summary of the baseline demographic and clinical characteristics of the 5,205 patients who underwent neuro-interventional treatment for acute ischemic stroke under two distinct reimbursement regimes. Of this cohort, 4,223 patients (81.13%) were reimbursed via the traditional FFS model, while 982 patients (18.87%) were covered by the Beijing DRG scheme. Table 1Characteristics of participantsVariablesFFS (n = 4,223)DRG (n = 982)Chi2 P Gender11.8780.001 Female1868(44.23%)375(38.19%) Male2355(55.77%)607(61.81%)Age Group46.701< 0.001 Children & Adolescents58(1.37%)3(0.31%) Adult2683(63.53%)526(53.56%) Elderly1482(35.10%)453(46.13%)Marital Status1.2190.269 Other Status396(9.38%)81(8.25%) Married3827(90.62%)901(91.75%)Insurance Type212.617< 0.001 RURBMI1966(46.55%)207(21.08%) UEBMI2257(53.45%)775(78.92%)CCI Group46.665< 0.001 No2568(60.81%)486(49.49%) Mild1511(35.78%)442(45.01%) Moderate125(2.96%)43(4.38%) Severe19(0.45%)11(1.12%)Cerebrovascular Lesion37.615< 0.001 Malformation113(2.68%)16(1.63%) Stenosis1733(41.04%)507(51.63%) Aneurysm2377(56.29%)459(46.74%)Neuro-Interventional Procedure2.9530.228 Embolization1272(30.12%)269(27.39%) Stent Placement2226(52.71%)542(55.19%) Balloon Angioplasty725(17.17%)171(17.41%)Admission type7.8460.005 Elective3767(89.20%)845(86.05%) Emergency456(10.80%)137(13.95%)Ventilator Use1.0310.310 No4122(97.61%)953(97.05%) Yes101(2.39%)29(2.95%)ICU Transfer5.6550.017 No3826(90.60%)865(88.09%) Yes397(9.40%)117(11.91%)Anesthesia Type15.298< 0.001 Local1573(37.25%)432(43.99%) General2650(62.75%)550(56.01%)Use of CPMS28.859< 0.001 No1075(25.46%)333(33.91%) Yes3148(74.54%)649(66.09%)Use of CPM0.1680.681 No86(2.04%)18(1.83%) Yes4137(97.96%)964(98.17%)31-Day Readmission1.3780.240 No4044(95.76%)932(94.91%) Yes179(4.24%)50(5.09%)Length of stay4.4270.109 <= 7 days3795(89.87%)866(88.19%) 8–14 days375(8.88%)96(9.78%) > 14 days53(1.26%)20(2.04%) Table 1Characteristics of participantsVariablesFFS (n = 4,223)DRG (n = 982)Chi2 P Gender11.8780.001 Female1868(44.23%)375(38.19%) Male2355(55.77%)607(61.81%)Age Group46.701< 0.001 Children & Adolescents58(1.37%)3(0.31%) Adult2683(63.53%)526(53.56%) Elderly1482(35.10%)453(46.13%)Marital Status1.2190.269 Other Status396(9.38%)81(8.25%) Married3827(90.62%)901(91.75%)Insurance Type212.617< 0.001 RURBMI1966(46.55%)207(21.08%) UEBMI2257(53.45%)775(78.92%)CCI Group46.665< 0.001 No2568(60.81%)486(49.49%) Mild1511(35.78%)442(45.01%) Moderate125(2.96%)43(4.38%) Severe19(0.45%)11(1.12%)Cerebrovascular Lesion37.615< 0.001 Malformation113(2.68%)16(1.63%) Stenosis1733(41.04%)507(51.63%) Aneurysm2377(56.29%)459(46.74%)Neuro-Interventional Procedure2.9530.228 Embolization1272(30.12%)269(27.39%) Stent Placement2226(52.71%)542(55.19%) Balloon Angioplasty725(17.17%)171(17.41%)Admission type7.8460.005 Elective3767(89.20%)845(86.05%) Emergency456(10.80%)137(13.95%)Ventilator Use1.0310.310 No4122(97.61%)953(97.05%) Yes101(2.39%)29(2.95%)ICU Transfer5.6550.017 No3826(90.60%)865(88.09%) Yes397(9.40%)117(11.91%)Anesthesia Type15.298< 0.001 Local1573(37.25%)432(43.99%) General2650(62.75%)550(56.01%)Use of CPMS28.859< 0.001 No1075(25.46%)333(33.91%) Yes3148(74.54%)649(66.09%)Use of CPM0.1680.681 No86(2.04%)18(1.83%) Yes4137(97.96%)964(98.17%)31-Day Readmission1.3780.240 No4044(95.76%)932(94.91%) Yes179(4.24%)50(5.09%)Length of stay4.4270.109 <= 7 days3795(89.87%)866(88.19%) 8–14 days375(8.88%)96(9.78%) > 14 days53(1.26%)20(2.04%) Characteristics of participants A notable finding was the marked difference in sociodemographic distribution between the two groups. The DRG cohort presented as significantly older, with 46.13% of patients aged ≥ 65 years compared to 35.10% in the FFS group (p < 0.001). Furthermore, the DRG cohort exhibited a higher proportion of male patients (61.81% vs. 55.77%, p = 0.001). Distinct patterns in insurance coverage were also observed: DRG patients were more frequently insured under Urban Employee Basic Medical Insurance (UEBMI) (78.92% vs. 53.45%, p < 0.001), reflecting the alignment with regional policy. The prevalence of moderate or severe comorbidity (Charlson Comorbidity Index (CCI) ≥ 3) was approximately twice as high in the DRG group (5.50% vs. 3.41%), and stenosis, rather than aneurysm, constituted the predominant lesion type (51.63% vs. 41.04%, p < 0.001). Process-of-care indicators further differentiated the cohorts. Emergency admissions (13.95% vs. 10.80%), intensive care unit (ICU) transfers (11.91% vs. 9.40%), and the utilization of local anesthesia (43.99% vs. 37.25%) were modestly higher among DRG patients. Conversely, the use of centrally procured medical supplies was lower in the DRG group (66.09% vs. 74.54%, p < 0.001). No statistically significant differences were observed for ventilator use, centrally procured medications, 31-day readmission rates, or the distribution of length of stay. These baseline data indicate that patients enrolled in the DRG scheme were typically older, carried a heavier comorbidity burden, and were primarily covered by resident-based insurance, while FFS patients exhibited demographic and insurance profiles consistent with inter-provincial referral patterns. These identified imbalances underscore the critical necessity of in subsequent cost and outcome comparisons. Cost outcome and structure Table 2 illustrates the unadjusted distributions of per-admission hospitalization costs and their four pre-specified components under the two reimbursement regimes. Overall, patients reimbursed through DRGs incurred notably lower expenditures compared to those covered by FFS. Table 2Cost outcome and structure between DRG and FFSCost comparisonFFS (n = 4,223)DRG (n = 982)Z P Per-admission TDHC (RMB)102,885(62871,168903)86,268(58207,129170)6.77< 0.001Per-admission pharmaceutical cost (RMB)1064(763,1428)1017(703,1432)1.680.093Per-admission medical consumables cost (RMB)81,560(46026,147191)66,999(43588,107936)6.46< 0.001Per-admission diagnostic/laboratory cost (RMB)2409(1886,3060)2318(1748,3067)2.380.017Per-admission ancillary cost (RMB)9420(8677,10512)9364(8579,10412)1.710.088Per-admission OOPE (RMB)30,266(18746,48158)19,851(17318,43797)17.83< 0.001Pharmaceutical cost proportion (%)1.05(0.65,1.79)1.16(0.07,1.85)-3.78< 0.001Medical consumables cost proportion (%)85.06(78.59,91.00)83.09(77.52,88.29)6.46< 0.001Diagnostic/laboratory cost proportion (%)2.49(1.47,4.04)2.83(1.81,4.34)-4.56< 0.001Ancillary cost proportion (%)10.07(6.26,14.53)11.35(8.19,15.08)-6.59< 0.001OOPE proportion (%)31.57(24.65,38.31)20.72(19.36,25.96)24.70< 0.001 Table 2Cost outcome and structure between DRG and FFSCost comparisonFFS (n = 4,223)DRG (n = 982)Z P Per-admission TDHC (RMB)102,885(62871,168903)86,268(58207,129170)6.77< 0.001Per-admission pharmaceutical cost (RMB)1064(763,1428)1017(703,1432)1.680.093Per-admission medical consumables cost (RMB)81,560(46026,147191)66,999(43588,107936)6.46< 0.001Per-admission diagnostic/laboratory cost (RMB)2409(1886,3060)2318(1748,3067)2.380.017Per-admission ancillary cost (RMB)9420(8677,10512)9364(8579,10412)1.710.088Per-admission OOPE (RMB)30,266(18746,48158)19,851(17318,43797)17.83< 0.001Pharmaceutical cost proportion (%)1.05(0.65,1.79)1.16(0.07,1.85)-3.78< 0.001Medical consumables cost proportion (%)85.06(78.59,91.00)83.09(77.52,88.29)6.46< 0.001Diagnostic/laboratory cost proportion (%)2.49(1.47,4.04)2.83(1.81,4.34)-4.56< 0.001Ancillary cost proportion (%)10.07(6.26,14.53)11.35(8.19,15.08)-6.59< 0.001OOPE proportion (%)31.57(24.65,38.31)20.72(19.36,25.96)24.70< 0.001 Cost outcome and structure between DRG and FFS The median (interquartile range, IQR) total direct hospitalization costs (TDHC) for DRG patients was RMB 86,268 (58,207 − 129,170), significantly lower than the RMB 102,885 (62,871 − 168,903) observed for FFS patients (Mann-Whitney Z = 6.77, p < 0.001). This represents an absolute reduction of approximately 16.15%. Medical consumables, which represented the dominant cost driver in both groups, were substantially lower under DRG (median RMB 66,999 vs. 81,560; Z = 6.46, p < 0.001). Diagnostic and laboratory charges also showed a modest decrease (RMB 2,318 vs. 2,409; Z = 2.38, p = 0.017), whereas pharmaceutical and ancillary costs did not differ materially between the groups (both p > 0.05). The DRG system also induced a shift in the relative weight of expenditures. The share attributable to consumables declined from 85.06% to 83.09% (p < 0.001), while the proportions for diagnostics and ancillary services increased slightly. This suggests a compositional rebalancing of costs rather than an indiscriminate, across-the-board reduction. Most notably, the median out-of-pocket expenditures (OOPE) for patients decreased from RMB 30,266 (IQR 18,746–48,158) under FFS to RMB 19,851 (IQR 17,318–43,797) under DRG, representing a substantial reduction of approximately 34.41%. Furthermore, OOPE as a percentage of total cost decreased from 31.57% to 20.72% (p < 0.001). These unadjusted comparisons strongly indicate that DRG payment was associated with lower absolute spending and a marked reduction in the patient’s financial burden, an effect primarily achieved through the containment of high-value consumables. Univariate between-group comparisons of TDHC and OOPE We subsequently conducted a comparative analysis of TDHC and OOPE across distinct patient populations. Table 3 summarized median TDHC and their inter-quartile ranges by demographic, clinical, and process-of-care strata under each payment regime. Across nearly every subgroup, DRG patients exhibited lower expenditure than their FFS counterparts, with the magnitude and consistency of the difference underscoring the robustness of the DRG effect. Women, adults aged 18–64 years, and elderly patients all experienced significantly lower median costs under DRG (p < 0.001 for each), as did beneficiaries holding either Urban Employee or Rural/Urban Resident insurance. Among comorbidity categories, the largest absolute savings were observed among those with no or mild Charlson burden, while moderate comorbidity showed no significant difference. Lesion morphology further refined the pattern: aneurysm cases, which commanded the highest expenditures, saw an 17.72% reduction under DRG (p < 0.001), whereas malformation and stenosis subgroups did not. Patients who received centrally procured medical supplies saw a 15.85% cost reduction under DRG (p < 0.001). Table 3TDHC differentials across patient subgroupsVariablesFFS (n = 4,223)DRG (n = 982)Z P Gender Female133,246(89914,183724)112,523(77565,168682)4.89< 0.001 Male80,252(56258,107631)69,353(54623,107631)3.78< 0.001Age Group Children & Adolescents64,920(50484,119624)49,505(35290,81837)0.190.211 Adult114,204(70330,177052)96,121(64828,150859)4.97< 0.001 Elderly84,127(56444,133286)71,079(55552,116509)2.990.003Marital Status Other Status105,695(61501,173036)77,597(59122,131446)2.210.026 Married102,533(62928,167571)86,846(58099,128904)6.37< 0.001Insurance Type RURBMI100,814(63712,151307)76,431(54472,121305)4.75< 0.001 UEBMI104,422(62390,175169)88,644(59488,135349)5.52< 0.001CCI Group No117,061(71167,178019)96,353(63590,151004)5.45< 0.001 Mild82,612(56518,126307)72,347(56701,109766)2.380.017 Moderate96,083(68143,139965)85,838(60863,135555)0.690.488 Severe112,850(68767,156870)155083(126462,175999-2.130.033Cerebrovascular Lesion Malformation63,289(51092,92464)71,018(54734,90184)-0.570.567 Stenosis64,383(52709,87381)63,327(52776,84639)0.690.491 Aneurysm151,304(106072,190321)124,494(91807,174493)6.02< 0.001Neuro-Interventional Procedure Embolization132,159(92784,183397)121,994(90286,174550)2.230.026 Stent Placement96,588(60565,159055)77,799(56824,111001)6.45< 0.001 Balloon Angioplasty67,746(52473,113727)64,210(50507,94417)1.560.119Admission type Elective98,739(60984,171613)82,042(56959,121539)6.89< 0.001 Emergency123,800(92474,153048)111,919(77565,149037)1.620.105Ventilator Use No101,515(62364,168903)84,060(57929,125449)7.06< 0.001 Yes146,131(118794,168123)143,975(129170,178933)-0.950.343ICU Transfer No98,990(61068,171268)80,777(56961,121205)7.18< 0.001 Yes124,079(96153,154142)121,978(88351,159314)0.370.714Anesthesia Type Local66,615(52767,122998)61,117(52526,85258)3.51< 0.001 General119,951(84011,177044)105,894(79802,162017)3.77< 0.001Use of CPMS No60,773(52862,75997)59,239(52562,75255)0.280.78 Yes123,484(84869,178925)103,909(73062,158567)6.15< 0.001Use of CPM No92,140(64786,146028)136,055(79672,224132)-1.670.095 Yes103,118(62807,169510)86,031(58104,128305)7.07< 0.00131-Day Readmission No104,779(64182,170037)87,262(58104,129851)7.24< 0.001 Yes64,548(,50281,105251)81,500(58833,117027)-2.310.021Length of stay <= 7 days100,809(61772,168846)82,869(57415,124494)6.76< 0.001 8–14 days117,482(78681,165057)105,553(69215,145471)1.740.082 > 14 days146,132(109511,190527)125,570(95728,183236)0.320.747 Table 3TDHC differentials across patient subgroupsVariablesFFS (n = 4,223)DRG (n = 982)Z P Gender Female133,246(89914,183724)112,523(77565,168682)4.89< 0.001 Male80,252(56258,107631)69,353(54623,107631)3.78< 0.001Age Group Children & Adolescents64,920(50484,119624)49,505(35290,81837)0.190.211 Adult114,204(70330,177052)96,121(64828,150859)4.97< 0.001 Elderly84,127(56444,133286)71,079(55552,116509)2.990.003Marital Status Other Status105,695(61501,173036)77,597(59122,131446)2.210.026 Married102,533(62928,167571)86,846(58099,128904)6.37< 0.001Insurance Type RURBMI100,814(63712,151307)76,431(54472,121305)4.75< 0.001 UEBMI104,422(62390,175169)88,644(59488,135349)5.52< 0.001CCI Group No117,061(71167,178019)96,353(63590,151004)5.45< 0.001 Mild82,612(56518,126307)72,347(56701,109766)2.380.017 Moderate96,083(68143,139965)85,838(60863,135555)0.690.488 Severe112,850(68767,156870)155083(126462,175999-2.130.033Cerebrovascular Lesion Malformation63,289(51092,92464)71,018(54734,90184)-0.570.567 Stenosis64,383(52709,87381)63,327(52776,84639)0.690.491 Aneurysm151,304(106072,190321)124,494(91807,174493)6.02< 0.001Neuro-Interventional Procedure Embolization132,159(92784,183397)121,994(90286,174550)2.230.026 Stent Placement96,588(60565,159055)77,799(56824,111001)6.45< 0.001 Balloon Angioplasty67,746(52473,113727)64,210(50507,94417)1.560.119Admission type Elective98,739(60984,171613)82,042(56959,121539)6.89< 0.001 Emergency123,800(92474,153048)111,919(77565,149037)1.620.105Ventilator Use No101,515(62364,168903)84,060(57929,125449)7.06< 0.001 Yes146,131(118794,168123)143,975(129170,178933)-0.950.343ICU Transfer No98,990(61068,171268)80,777(56961,121205)7.18< 0.001 Yes124,079(96153,154142)121,978(88351,159314)0.370.714Anesthesia Type Local66,615(52767,122998)61,117(52526,85258)3.51< 0.001 General119,951(84011,177044)105,894(79802,162017)3.77< 0.001Use of CPMS No60,773(52862,75997)59,239(52562,75255)0.280.78 Yes123,484(84869,178925)103,909(73062,158567)6.15< 0.001Use of CPM No92,140(64786,146028)136,055(79672,224132)-1.670.095 Yes103,118(62807,169510)86,031(58104,128305)7.07< 0.00131-Day Readmission No104,779(64182,170037)87,262(58104,129851)7.24< 0.001 Yes64,548(,50281,105251)81,500(58833,117027)-2.310.021Length of stay <= 7 days100,809(61772,168846)82,869(57415,124494)6.76< 0.001 8–14 days117,482(78681,165057)105,553(69215,145471)1.740.082 > 14 days146,132(109511,190527)125,570(95728,183236)0.320.747 TDHC differentials across patient subgroups Table 4 presented the distribution of median OOPE and their inter-quartile ranges across all pre-specified subgroups under each payment regime. Consistent with the pattern observed for total direct hospital costs, DRG reimbursement was associated with lower patient-borne spending in virtually every stratum examined. Across age strata, adults aged 18–64 years registered the steepest DRG-related reduction in out-of-pocket expenditure (median fell from RMB 33,720 to 22,362), while children and adolescents showed no significant difference. Insurance type further stratified the effect: RURBMI beneficiaries experienced a 28.55% decrease (35,227 vs. 25,171), and UEBMI holders a 27.33% decrease (25,396 vs. 18,456). Among patients who received centrally procured medical supplies, DRG payment was associated with a 33.84% reduction in median out-of-pocket expenditure (36,145 vs. 23,913 RMB); for centrally procured medications, the corresponding reduction was 34.43% (30,171 vs. 19,784 RMB). These findings indicated that the cost-containment effect of DRG was broad-based but not uniform, thereby justifying multivariable quantile regression to dissect residual heterogeneity. Table 4OOPE differentials across patient subgroupsVariablesFFS (n = 4,223)DRG (n = 982)Z P Gender Female39,434(24472,56476)25,291(18531,33920)13.33< 0.001 Male25,034(16300,38937)17,172(11448,25420)12.87< 0.001Age Group Children & Adolescents25,503(20913,38905)18,442(5444,29188)1.410.161 Adult33,720(21071,51854)22,362(15616,32278)13.62< 0.001 Elderly24,984(15909,40507)18,202(11657,25539)10.37< 0.001Marital Status Other Status32,203(22283,52647)18,748(13192,31505)6.14< 0.001 Married30,064(18437,47806)19,999(13298,29932)16.70< 0.001Insurance Type RURBMI35,227(23988,52405)25,171(19109,34519)8.55< 0.001 UEBMI25,396(15586,43525)18,456(12544,28114)11.21< 0.001CCI Group No34,688(21244,53165)22,583(14855,32307)14.13< 0.001 Mild25,011(15946,39012)18,219(11977,24713)10.14< 0.001 Moderate27,137(17877,39503)19,751(11687,29112)2.640.008 Severe30,175(22936,39363)38,841(23965,106300)-1.440.149Cerebrovascular Lesion Malformation23,523(16937,31091)18,354(13924,23471)2.460.014 Stenosis21,080(14810,29438)15,396(11032,20940)11.38< 0.001 Aneurysm42,470(27602,58898)28,147(19434,34519)15.33< 0.001Neuro-Interventional Procedure Embolization38,961(24947,56465)27,384(18751,33993)9.83< 0.001 Stent Placement28,922(18448,45906)18,696(12552,26377)15.06< 0.001 Balloon Angioplasty21,926(14756,34299)16,400(10329,24477)5.51< 0.001Admission type Elective29,294(18077,47865)19,215(12743,28891)17.17< 0.001 Emergency36,371(25882,50410)25,831(18231,32556)6.120.105Ventilator Use No29,954(18572,48005)19,485(13095,29557)17.88< 0.001 Yes41,917(28982,51491)31,237(27958,36962)2.200.028ICU Transfer No29,411(18112,47955)19,288(12774,29101)17.28< 0.001 Yes36,645(26118,50061)26,934(18552,32556)5.57< 0.001Anesthesia Type Local22,433(14756,35439)14,801(10698,21116)11.59< 0.001 General35,915(23575,52320)24,682(18148,33266)13.95< 0.001Use of CPMS No20,338(14928,27170)14,122(10728,19456)10.11< 0.001 Yes36,145(22826,52807)23,913(17614,32851)16.19< 0.001Use of CPM No43,806(21280,78040)83,982(39517,224132)-2.200.028 Yes30,171(18739,47865)19,784(13204,29597)18.28< 0.00131-Day Readmission No30,714(19121,48570)20,095(13332,29982)17.98< 0.001 Yes19,733(13204,35363)17,811(12433,23371)1.010.314Length of stay <= 7 days29,850(18327,47664)19,446(12790,29753)17.02< 0.001 8–14 days33,500(22597,51599)23,711(15230,30141)5.63< 0.001 > 14 days33,345(25495,48833)31,425(21199,34710)1.710.087 Table 4OOPE differentials across patient subgroupsVariablesFFS (n = 4,223)DRG (n = 982)Z P Gender Female39,434(24472,56476)25,291(18531,33920)13.33< 0.001 Male25,034(16300,38937)17,172(11448,25420)12.87< 0.001Age Group Children & Adolescents25,503(20913,38905)18,442(5444,29188)1.410.161 Adult33,720(21071,51854)22,362(15616,32278)13.62< 0.001 Elderly24,984(15909,40507)18,202(11657,25539)10.37< 0.001Marital Status Other Status32,203(22283,52647)18,748(13192,31505)6.14< 0.001 Married30,064(18437,47806)19,999(13298,29932)16.70< 0.001Insurance Type RURBMI35,227(23988,52405)25,171(19109,34519)8.55< 0.001 UEBMI25,396(15586,43525)18,456(12544,28114)11.21< 0.001CCI Group No34,688(21244,53165)22,583(14855,32307)14.13< 0.001 Mild25,011(15946,39012)18,219(11977,24713)10.14< 0.001 Moderate27,137(17877,39503)19,751(11687,29112)2.640.008 Severe30,175(22936,39363)38,841(23965,106300)-1.440.149Cerebrovascular Lesion Malformation23,523(16937,31091)18,354(13924,23471)2.460.014 Stenosis21,080(14810,29438)15,396(11032,20940)11.38< 0.001 Aneurysm42,470(27602,58898)28,147(19434,34519)15.33< 0.001Neuro-Interventional Procedure Embolization38,961(24947,56465)27,384(18751,33993)9.83< 0.001 Stent Placement28,922(18448,45906)18,696(12552,26377)15.06< 0.001 Balloon Angioplasty21,926(14756,34299)16,400(10329,24477)5.51< 0.001Admission type Elective29,294(18077,47865)19,215(12743,28891)17.17< 0.001 Emergency36,371(25882,50410)25,831(18231,32556)6.120.105Ventilator Use No29,954(18572,48005)19,485(13095,29557)17.88< 0.001 Yes41,917(28982,51491)31,237(27958,36962)2.200.028ICU Transfer No29,411(18112,47955)19,288(12774,29101)17.28< 0.001 Yes36,645(26118,50061)26,934(18552,32556)5.57< 0.001Anesthesia Type Local22,433(14756,35439)14,801(10698,21116)11.59< 0.001 General35,915(23575,52320)24,682(18148,33266)13.95< 0.001Use of CPMS No20,338(14928,27170)14,122(10728,19456)10.11< 0.001 Yes36,145(22826,52807)23,913(17614,32851)16.19< 0.001Use of CPM No43,806(21280,78040)83,982(39517,224132)-2.200.028 Yes30,171(18739,47865)19,784(13204,29597)18.28< 0.00131-Day Readmission No30,714(19121,48570)20,095(13332,29982)17.98< 0.001 Yes19,733(13204,35363)17,811(12433,23371)1.010.314Length of stay <= 7 days29,850(18327,47664)19,446(12790,29753)17.02< 0.001 8–14 days33,500(22597,51599)23,711(15230,30141)5.63< 0.001 > 14 days33,345(25495,48833)31,425(21199,34710)1.710.087 OOPE differentials across patient subgroups Results of quantile regression analysis Having demonstrated broad, unadjusted reductions in both total direct hospital costs and out-of-pocket expenditures under the DRG regime, we next sought to isolate the independent effect of payment type while accounting for residual baseline heterogeneity. To this end, we estimated two complementary models: OLS regression for the mean effect and quantile regression across the 10th, 25th, 50th, 75th and 90th percentiles for the full cost distribution. The analyses for TDHC, as detailed in Table 5 and visualized in Fig. 2, demonstrated that the DRG payment was consistently linked with reduced costs across all quantiles. The coefficients for DRG payment varied from − 0.023 at the 25th percentile to -0.017 at the 90th percentile, all of which were statistically significant (p < 0.05). This corresponded to a decrease in TDHC ranging from 3.84% (calculated as 1–10^-0.017) to 5.16% (1–10^-0.023). The coefficient plots in Fig. 2 further substantiated this pattern, with the confidence intervals remaining below zero across nearly the entire spectrum of costs, thereby reinforcing the robustness of the DRG effect on cost reduction. Additional variables also demonstrated quantile-dependent effects. For instance, higher insurance status (UEBMI), ICU transfer, non-use of CPM, and longer length of stay were associated with higher costs at most quantiles. Notably, within the low-cost group (below the 50th percentile), age had no significant impact on cost differences, whereas in the high-cost group (above the 50th percentile), older age was associated with lower costs, indicating a reduction in cost differences with increasing age. In the higher expenditure groups (75th and 90th percentiles), the use of CPM demonstrated an enhanced negative effect on costs, indicating a more pronounced cost containment effect. Conversely, in the lower expenditure groups (10th and 25th percentiles), the impact of CPM use on costs was minimal, suggesting that the cost-saving benefits of centralized procurement medication were more evident among patients incurring higher costs. Table 5Quantile regression analysis for TDHCVariableOverall10th percentile25th percentile50th percentile75th percentile90th percentileCoeff95%CICoeff95%CICoeff95%CICoeff95%CICoeff95%CICoeff95%CIPayment Type FFSref.ref.ref.ref.ref.ref. DRG-0.022***[-0.034,-0.010]-0.015[-0.031,0.002]-0.023**[-0.038,-0.009]-0.028**[-0.046,-0.009]-0.026***[-0.034,-0.018]-0.017*[-0.030,-0.004]Gender Femaleref.ref.ref.ref.ref.ref. Male-0.041***[-0.052,-0.031]-0.050***[-0.067,-0.034]-0.044***[-0.057,-0.032]-0.046***[-0.057,-0.035]-0.016***[-0.025,-0.008]-0.009*[-0.017,-0.001]Age Group Children & Adolescentsref.ref.ref.ref.ref.ref. Adult0.055*[0.005,0.105]0.130**[0.045,0.216]0.069[-0.037,0.175]0.049[-0.051,0.149]0.035[-0.012,0.082]-0.014[-0.075,0.046] Elderly0.022[-0.029,0.073]0.108*[0.023,0.194]0.049[-0.059,0.157]0.008[-0.095,0.111]0.01[-0.037,0.057]-0.027[-0.089,0.034]Marital Status Other Statusref.ref.ref.ref.ref.ref. Married-0.018*[-0.035,-0.000]-0.017[-0.043,0.007]-0.004[-0.027,0.019]-0.022*[-0.041,-0.003]-0.008[-0.023,0.006]-0.01[-0.025,0.006]Insurance Type RURBMIref.ref.ref.ref.ref.ref. UEBMI0.021***[0.011,0.030]0.009[-0.004,0.023]0.013[-0.002,0.027]0.032***[0.017,0.047]0.019***[0.010,0.028]0.012*[0.001,0.024]CCI Group Noref.ref.ref.ref.ref.ref. Mild-0.035***[-0.046,-0.025]-0.025*[-0.048,-0.002]-0.034***[-0.047,-0.021]-0.039***[-0.051,-0.027]-0.021***[-0.030,-0.012]-0.017**[-0.028,-0.005] Moderate-0.041**[-0.069,-0.013]-0.057*[-0.113,-0.001]-0.045[-0.109,0.017]-0.041[-0.085,0.004]-0.022[-0.057,0.014]-0.014[-0.063,0.034] Severe-0.040[-0.103,0.023]0.018[-0.117,0.153]-0.074[-0.162,0.013]-0.096[-0.283,0.091]-0.006[-0.166,0.153]-0.004[-0.080,0.071]Cerebrovascular Lesion Malformationref.ref.ref.ref.ref.ref. Stenosis0.034[-0.001,0.070]0.020[-0.033,0.074]0.051*[0.007,0.094]0.058*[0.000,0.116]0.002[-0.034,0.039]-0.027[-0.078,0.023] Aneurysm0.234***[0.198,0.270]0.165***[0.119,0.212]0.219***[0.180,0.259]0.269***[0.213,0.325]0.247***[0.210,0.285]0.216***[0.160,0.272]Neuro-Interventional Procedure Embolizationref.ref.ref.ref.ref.ref. Stent Placement-0.002[-0.015,0.010]0.051***[0.027,0.075]0.006[-0.014,0.027]-0.025*[-0.046,-0.005]-0.020***[-0.030,-0.010]-0.01[-0.023,0.003] Balloon Angioplasty-0.059***[-0.075,-0.043]-0.080***[-0.120,-0.041]-0.067***[-0.097,-0.037]-0.060***[-0.084,-0.035]-0.042***[-0.055,-0.030]-0.020**[-0.036,-0.005]Admission type Electiveref.ref.ref.ref.ref.ref. Emergency-0.024*[-0.046,-0.002]-0.051*[-0.100,-0.002]-0.034*[-0.067,-0.002]-0.039**[-0.069,-0.010]-0.009[-0.036,0.018]0.005[-0.025,0.035]Ventilator Use Noref.ref.ref.ref.ref.ref. Yes0.176***[0.141,0.211]0.166***[0.093,0.240]0.192***[0.131,0.255]0.200***[0.141,0.258]0.192***[0.140,0.244]0.124***[0.080,0.169]ICU Transfer Noref.ref.ref.ref.ref.ref. Yes0.059***[0.034,0.084]0.050[-0.010,0.112]0.032*[0.002,0.062]0.041**[0.013,0.069]0.085***[0.037,0.132]0.148***[0.100,0.196]Anesthesia Type Localref.ref.ref.ref.ref.ref. General0.050***[0.040,0.061]0.102***[0.086,0.117]0.073***[0.063,0.085]0.036***[0.026,0.047]0.023***[0.013,0.033]0.021***[0.014,0.028]Use of CPMS Noref.ref.ref.ref.ref.ref. Yes0.094***[0.081,0.106]0.061***[0.041,0.081]0.074***[0.054,0.095]0.110***[0.097,0.124]0.137***[0.119,0.156]0.130***[0.112,0.147]Use of CPM Noref.ref.ref.ref.ref.ref. Yes-0.040*[-0.074,-0.007]0.011[-0.044,0.066]-0.019[-0.070,0.030]-0.029[-0.077,0.019]-0.04[-0.115,0.034]-0.150*[-0.284,-0.016]31-Day Readmission Noref.ref.ref.ref.ref.ref. Yes-0.042***[-0.065,-0.020-0.057**[-0.100,-0.014]-0.056***[-0.082,-0.030]-0.036***[-0.056,-0.015]-0.014[-0.047,0.018]-0.019[-0.042,0.005]Length of stay <= 7 daysref.ref.ref.ref.ref.ref. 8–14 days0.035***[0.018,0.053]0.038*[0.002,0.073]0.028*[0.001,0.054]0.026*[0.005,0.047]0.018[-0.001,0.037]0.040*[0.006,0.075] > 14 days0.098***[0.057,0.140]0.125*[0.019,0.230]0.106***[0.044,0.169]0.117***[0.068,0.165]0.086***[0.046,0.125]0.040*[0.007,0.073] Table 5Quantile regression analysis for TDHCVariableOverall10th percentile25th percentile50th percentile75th percentile90th percentileCoeff95%CICoeff95%CICoeff95%CICoeff95%CICoeff95%CICoeff95%CIPayment Type FFSref.ref.ref.ref.ref.ref. DRG-0.022***[-0.034,-0.010]-0.015[-0.031,0.002]-0.023**[-0.038,-0.009]-0.028**[-0.046,-0.009]-0.026***[-0.034,-0.018]-0.017*[-0.030,-0.004]Gender Femaleref.ref.ref.ref.ref.ref. Male-0.041***[-0.052,-0.031]-0.050***[-0.067,-0.034]-0.044***[-0.057,-0.032]-0.046***[-0.057,-0.035]-0.016***[-0.025,-0.008]-0.009*[-0.017,-0.001]Age Group Children & Adolescentsref.ref.ref.ref.ref.ref. Adult0.055*[0.005,0.105]0.130**[0.045,0.216]0.069[-0.037,0.175]0.049[-0.051,0.149]0.035[-0.012,0.082]-0.014[-0.075,0.046] Elderly0.022[-0.029,0.073]0.108*[0.023,0.194]0.049[-0.059,0.157]0.008[-0.095,0.111]0.01[-0.037,0.057]-0.027[-0.089,0.034]Marital Status Other Statusref.ref.ref.ref.ref.ref. Married-0.018*[-0.035,-0.000]-0.017[-0.043,0.007]-0.004[-0.027,0.019]-0.022*[-0.041,-0.003]-0.008[-0.023,0.006]-0.01[-0.025,0.006]Insurance Type RURBMIref.ref.ref.ref.ref.ref. UEBMI0.021***[0.011,0.030]0.009[-0.004,0.023]0.013[-0.002,0.027]0.032***[0.017,0.047]0.019***[0.010,0.028]0.012*[0.001,0.024]CCI Group Noref.ref.ref.ref.ref.ref. Mild-0.035***[-0.046,-0.025]-0.025*[-0.048,-0.002]-0.034***[-0.047,-0.021]-0.039***[-0.051,-0.027]-0.021***[-0.030,-0.012]-0.017**[-0.028,-0.005] Moderate-0.041**[-0.069,-0.013]-0.057*[-0.113,-0.001]-0.045[-0.109,0.017]-0.041[-0.085,0.004]-0.022[-0.057,0.014]-0.014[-0.063,0.034] Severe-0.040[-0.103,0.023]0.018[-0.117,0.153]-0.074[-0.162,0.013]-0.096[-0.283,0.091]-0.006[-0.166,0.153]-0.004[-0.080,0.071]Cerebrovascular Lesion Malformationref.ref.ref.ref.ref.ref. Stenosis0.034[-0.001,0.070]0.020[-0.033,0.074]0.051*[0.007,0.094]0.058*[0.000,0.116]0.002[-0.034,0.039]-0.027[-0.078,0.023] Aneurysm0.234***[0.198,0.270]0.165***[0.119,0.212]0.219***[0.180,0.259]0.269***[0.213,0.325]0.247***[0.210,0.285]0.216***[0.160,0.272]Neuro-Interventional Procedure Embolizationref.ref.ref.ref.ref.ref. Stent Placement-0.002[-0.015,0.010]0.051***[0.027,0.075]0.006[-0.014,0.027]-0.025*[-0.046,-0.005]-0.020***[-0.030,-0.010]-0.01[-0.023,0.003] Balloon Angioplasty-0.059***[-0.075,-0.043]-0.080***[-0.120,-0.041]-0.067***[-0.097,-0.037]-0.060***[-0.084,-0.035]-0.042***[-0.055,-0.030]-0.020**[-0.036,-0.005]Admission type Electiveref.ref.ref.ref.ref.ref. Emergency-0.024*[-0.046,-0.002]-0.051*[-0.100,-0.002]-0.034*[-0.067,-0.002]-0.039**[-0.069,-0.010]-0.009[-0.036,0.018]0.005[-0.025,0.035]Ventilator Use Noref.ref.ref.ref.ref.ref. Yes0.176***[0.141,0.211]0.166***[0.093,0.240]0.192***[0.131,0.255]0.200***[0.141,0.258]0.192***[0.140,0.244]0.124***[0.080,0.169]ICU Transfer Noref.ref.ref.ref.ref.ref. Yes0.059***[0.034,0.084]0.050[-0.010,0.112]0.032*[0.002,0.062]0.041**[0.013,0.069]0.085***[0.037,0.132]0.148***[0.100,0.196]Anesthesia Type Localref.ref.ref.ref.ref.ref. General0.050***[0.040,0.061]0.102***[0.086,0.117]0.073***[0.063,0.085]0.036***[0.026,0.047]0.023***[0.013,0.033]0.021***[0.014,0.028]Use of CPMS Noref.ref.ref.ref.ref.ref. Yes0.094***[0.081,0.106]0.061***[0.041,0.081]0.074***[0.054,0.095]0.110***[0.097,0.124]0.137***[0.119,0.156]0.130***[0.112,0.147]Use of CPM Noref.ref.ref.ref.ref.ref. Yes-0.040*[-0.074,-0.007]0.011[-0.044,0.066]-0.019[-0.070,0.030]-0.029[-0.077,0.019]-0.04[-0.115,0.034]-0.150*[-0.284,-0.016]31-Day Readmission Noref.ref.ref.ref.ref.ref. Yes-0.042***[-0.065,-0.020-0.057**[-0.100,-0.014]-0.056***[-0.082,-0.030]-0.036***[-0.056,-0.015]-0.014[-0.047,0.018]-0.019[-0.042,0.005]Length of stay <= 7 daysref.ref.ref.ref.ref.ref. 8–14 days0.035***[0.018,0.053]0.038*[0.002,0.073]0.028*[0.001,0.054]0.026*[0.005,0.047]0.018[-0.001,0.037]0.040*[0.006,0.075] > 14 days0.098***[0.057,0.140]0.125*[0.019,0.230]0.106***[0.044,0.169]0.117***[0.068,0.165]0.086***[0.046,0.125]0.040*[0.007,0.073] Quantile regression analysis for TDHC Fig. 2Quantile regressions predicting TDHC at the 0.05–0.95 quantile Quantile regressions predicting TDHC at the 0.05–0.95 quantile Quantile regression analyses of OOPE, as summarized in Table 6 and illustrated in Fig. 3, revealed a pattern consistent with our findings on TDHC. At the 10th percentile, the association between DRG payment and OOPE did not achieve statistical significance. However, from the 25th to the 90th percentiles, DRG was consistently associated with significantly lower OOPE, with coefficients ranging from − 0.068 to -0.196. This translated to a reduction in OOPE of approximately 14.49% (calculated as 1–10^-0.068) to 36.32% (1–10^-0.196), indicating that DRG effectively mitigated financial burden for patients across the mid to high OOPE strata. These findings underscored the potential of DRG to promote equity in healthcare resource allocation by addressing disparities in patient out-of-pocket payments. Additionally, analyses of OOPE revealed several meaningful findings beyond the payment type variable. Notably, the insurance type showed a diminishing effect on OOPE as the distribution shifted towards higher expenditures. Specifically, the cost difference between UEBMI and RURBMI beneficiaries decreased as OOPE increased. This suggests that UEBMI was more sensitive to cost reduction in the lower OOPE strata, offering greater relief to patients in this segment. However, as OOPE grew, the disparity in financial burden between these insurance statuses diminished, indicating that the relative impact of insurance type on OOPE was less pronounced among higher OOPE patients. We also observed that among patients who utilized CPMS, those in the upper OOPE strata exhibited higher OOPE. This finding suggests that patients with a higher propensity to pay were more inclined to opt for higher-value, centrally negotiated neuro-interventional consumables. Table 6Quantile regression analysis for OOPEVariableOverall10th percentile25th percentile50th percentile75th percentile90th percentileCoeff95%CICoeff95%CICoeff95%CICoeff95%CICoeff95%CICoeff95%CIPayment Type FFSref.ref.ref.ref.ref.ref. DRG-0.093***[-0.112,-0.075]-0.001[-0.033,0.032]-0.068***[-0.091,-0.045]-0.105***[-0.120,-0.090]-0.150***[-0.166,-0.133]-0.196***[-0.211,-0.180]Gender Femaleref.ref.ref.ref.ref.ref. Male-0.035***[-0.051,-0.020]-0.041*[-0.076,-0.005]-0.042***[-0.059,-0.024]-0.049***[-0.064,-0.033]-0.018**[-0.030,-0.007]-0.007[-0.023,0.010]Age Group Children & Adolescentsref.ref.ref.ref.ref.ref. Adult0.056[-0.019,0.131]0.113[-0.031,0.258]0.061[-0.072,0.194]0.033[-0.039,0.106]0.022[-0.038,0.082]-0.002[-0.048,0.043] Elderly0.001[-0.076,0.076]0.066[-0.090,0.221]0.010[-0.134,0.154]-0.020[-0.095,0.055]-0.021[-0.081,0.039]-0.051*[-0.101,-0.001]Marital Status Other Statusref.ref.ref.ref.ref.ref. Married-0.030*[-0.056,-0.005]-0.032[-0.071,0.007]-0.019[-0.049,0.011]-0.025*[-0.045,-0.004]-0.029***[-0.042,-0.017]-0.032[-0.063,0.001]Insurance Type RURBMIref.ref.ref.ref.ref.ref. UEBMI-0.136***[-0.151,-0.122]-0.219***[-0.246,-0.192]-0.164***[-0.187,-0.142]-0.128***[-0.144,-0.113]-0.102***[-0.114,-0.090]-0.069***[-0.084,-0.055]CCI Group Noref.ref.ref.ref.ref.ref. Mild-0.049***[-0.064,-0.033]-0.038*[-0.073,-0.004]-0.046***[-0.064,-0.027]-0.052***[-0.063,-0.041]-0.035***[-0.047,-0.023]-0.021***[-0.033,-0.009] Moderate-0.046*[-0.088,-0.003]-0.05[-0.104,0.005]-0.049*[-0.095,-0.004]-0.056**[-0.098,-0.015]-0.032[-0.088,0.024]-0.030[-0.079,0.019] Severe0.013[-0.081,0.108]-0.041[-0.152,0.069]-0.075[-0.161,0.012]-0.010[-0.152,0.133]0.029[-0.201,0.258]-0.016[-0.390,0.359]Cerebrovascular Lesion Malformationref.ref.ref.ref.ref.ref. Stenosis0.031[-0.023,0.084]-0.002[-0.105,0.102]0.033[-0.031,0.096]0.057[-0.005,0.119]0.021[-0.034,0.076]0.035[-0.030,0.100] Aneurysm0.205***[0.152,0.259]0.142**[0.035,0.250]0.189***[0.130,0.249]0.219***[0.151,0.288]0.221***[0.161,0.281]0.241***[0.169,0.313]Neuro-Interventional Procedure Embolizationref.ref.ref.ref.ref.ref. Stent Placement0.012[-0.007,0.030]0.072***[0.041,0.102]0.024*[0.003,0.045]-0.013[-0.033,0.008]-0.022*[-0.039,-0.004]-0.020*[-0.039,-0.002] Balloon Angioplasty-0.045***[-0.069,-0.021]-0.039*[-0.078,-0.001]-0.051***[-0.076,-0.026]-0.055***[-0.078,-0.032]-0.054***[-0.079,-0.029]-0.044***[-0.064,-0.023]Admission type Electiveref.ref.ref.ref.ref.ref. Emergency-0.014[-0.047,0.018]-0.054[-0.121,0.013]-0.036[-0.072,0.001]-0.042**[-0.072,-0.013]-0.018[-0.058,0.021]-0.002[-0.075,0.072]Ventilator Use Noref.ref.ref.ref.ref.ref. Yes0.141***[0.089,0.193]0.159***[0.077,0.241]0.165***[0.113,0.217]0.156***[0.099,0.214]0.159***[0.097,0.222]0.105*[0.015,0.195]ICU Transfer Noref.ref.ref.ref.ref.ref. Yes0.054**[0.017,0.092]0.089**[0.023,0.154]0.024[-0.003,0.051]0.037*[0.002,0.073]0.058**[0.019,0.097]0.071*[0.006,0.137]Anesthesia Type Localref.ref.ref.ref.ref.ref. General0.060***[0.043,0.076]0.112***[0.080,0.145]0.067***[0.043,0.090]0.041***[0.023,0.059]0.028***[0.014,0.043]0.032***[0.015,0.050]Use of CPMS Noref.ref.ref.ref.ref.ref. Yes0.061***[0.042,0.080]0.027[-0.008,0.063]0.055***[0.030,0.080]0.094***[0.073,0.115]0.114***[0.096,0.132]0.115***[0.096,0.134]Use of CPM Noref.ref.ref.ref.ref.ref. Yes-0.256***[-0.306,-0.207]-0.017[-0.107,0.074]-0.069[-0.156,0.018]-0.124[-0.280,0.032]-0.497***[-0.654,-0.339]-0.687***[-0.886,-0.487]31-Day Readmission Noref.ref.ref.ref.ref.ref. Yes-0.074***[-0.107,-0.040]-0.079[-0.178,0.019]-0.090***[-0.126,-0.055]-0.073***[-0.110,-0.037]-0.042*[-0.084,-0.001]-0.041**[-0.072,-0.011]Length of stay <= 7 daysref.ref.ref.ref.ref.ref. 8–14 days0.042**[0.016,0.068]0.054*[0.006,0.102]0.042***[0.021,0.064]0.026[-0.001,0.052]0.009[-0.026,0.043]0.052***[0.024,0.080]> 14 days0.071*[0.009,0.133]0.107***[0.056,0.158]0.094***[0.041,0.147]0.082*[0.018,0.146]0.056[-0.002,0.114]0.062*[0.014,0.111] Table 6Quantile regression analysis for OOPEVariableOverall10th percentile25th percentile50th percentile75th percentile90th percentileCoeff95%CICoeff95%CICoeff95%CICoeff95%CICoeff95%CICoeff95%CIPayment Type FFSref.ref.ref.ref.ref.ref. DRG-0.093***[-0.112,-0.075]-0.001[-0.033,0.032]-0.068***[-0.091,-0.045]-0.105***[-0.120,-0.090]-0.150***[-0.166,-0.133]-0.196***[-0.211,-0.180]Gender Femaleref.ref.ref.ref.ref.ref. Male-0.035***[-0.051,-0.020]-0.041*[-0.076,-0.005]-0.042***[-0.059,-0.024]-0.049***[-0.064,-0.033]-0.018**[-0.030,-0.007]-0.007[-0.023,0.010]Age Group Children & Adolescentsref.ref.ref.ref.ref.ref. Adult0.056[-0.019,0.131]0.113[-0.031,0.258]0.061[-0.072,0.194]0.033[-0.039,0.106]0.022[-0.038,0.082]-0.002[-0.048,0.043] Elderly0.001[-0.076,0.076]0.066[-0.090,0.221]0.010[-0.134,0.154]-0.020[-0.095,0.055]-0.021[-0.081,0.039]-0.051*[-0.101,-0.001]Marital Status Other Statusref.ref.ref.ref.ref.ref. Married-0.030*[-0.056,-0.005]-0.032[-0.071,0.007]-0.019[-0.049,0.011]-0.025*[-0.045,-0.004]-0.029***[-0.042,-0.017]-0.032[-0.063,0.001]Insurance Type RURBMIref.ref.ref.ref.ref.ref. UEBMI-0.136***[-0.151,-0.122]-0.219***[-0.246,-0.192]-0.164***[-0.187,-0.142]-0.128***[-0.144,-0.113]-0.102***[-0.114,-0.090]-0.069***[-0.084,-0.055]CCI Group Noref.ref.ref.ref.ref.ref. Mild-0.049***[-0.064,-0.033]-0.038*[-0.073,-0.004]-0.046***[-0.064,-0.027]-0.052***[-0.063,-0.041]-0.035***[-0.047,-0.023]-0.021***[-0.033,-0.009] Moderate-0.046*[-0.088,-0.003]-0.05[-0.104,0.005]-0.049*[-0.095,-0.004]-0.056**[-0.098,-0.015]-0.032[-0.088,0.024]-0.030[-0.079,0.019] Severe0.013[-0.081,0.108]-0.041[-0.152,0.069]-0.075[-0.161,0.012]-0.010[-0.152,0.133]0.029[-0.201,0.258]-0.016[-0.390,0.359]Cerebrovascular Lesion Malformationref.ref.ref.ref.ref.ref. Stenosis0.031[-0.023,0.084]-0.002[-0.105,0.102]0.033[-0.031,0.096]0.057[-0.005,0.119]0.021[-0.034,0.076]0.035[-0.030,0.100] Aneurysm0.205***[0.152,0.259]0.142**[0.035,0.250]0.189***[0.130,0.249]0.219***[0.151,0.288]0.221***[0.161,0.281]0.241***[0.169,0.313]Neuro-Interventional Procedure Embolizationref.ref.ref.ref.ref.ref. Stent Placement0.012[-0.007,0.030]0.072***[0.041,0.102]0.024*[0.003,0.045]-0.013[-0.033,0.008]-0.022*[-0.039,-0.004]-0.020*[-0.039,-0.002] Balloon Angioplasty-0.045***[-0.069,-0.021]-0.039*[-0.078,-0.001]-0.051***[-0.076,-0.026]-0.055***[-0.078,-0.032]-0.054***[-0.079,-0.029]-0.044***[-0.064,-0.023]Admission type Electiveref.ref.ref.ref.ref.ref. Emergency-0.014[-0.047,0.018]-0.054[-0.121,0.013]-0.036[-0.072,0.001]-0.042**[-0.072,-0.013]-0.018[-0.058,0.021]-0.002[-0.075,0.072]Ventilator Use Noref.ref.ref.ref.ref.ref. Yes0.141***[0.089,0.193]0.159***[0.077,0.241]0.165***[0.113,0.217]0.156***[0.099,0.214]0.159***[0.097,0.222]0.105*[0.015,0.195]ICU Transfer Noref.ref.ref.ref.ref.ref. Yes0.054**[0.017,0.092]0.089**[0.023,0.154]0.024[-0.003,0.051]0.037*[0.002,0.073]0.058**[0.019,0.097]0.071*[0.006,0.137]Anesthesia Type Localref.ref.ref.ref.ref.ref. General0.060***[0.043,0.076]0.112***[0.080,0.145]0.067***[0.043,0.090]0.041***[0.023,0.059]0.028***[0.014,0.043]0.032***[0.015,0.050]Use of CPMS Noref.ref.ref.ref.ref.ref. Yes0.061***[0.042,0.080]0.027[-0.008,0.063]0.055***[0.030,0.080]0.094***[0.073,0.115]0.114***[0.096,0.132]0.115***[0.096,0.134]Use of CPM Noref.ref.ref.ref.ref.ref. Yes-0.256***[-0.306,-0.207]-0.017[-0.107,0.074]-0.069[-0.156,0.018]-0.124[-0.280,0.032]-0.497***[-0.654,-0.339]-0.687***[-0.886,-0.487]31-Day Readmission Noref.ref.ref.ref.ref.ref. Yes-0.074***[-0.107,-0.040]-0.079[-0.178,0.019]-0.090***[-0.126,-0.055]-0.073***[-0.110,-0.037]-0.042*[-0.084,-0.001]-0.041**[-0.072,-0.011]Length of stay <= 7 daysref.ref.ref.ref.ref.ref. 8–14 days0.042**[0.016,0.068]0.054*[0.006,0.102]0.042***[0.021,0.064]0.026[-0.001,0.052]0.009[-0.026,0.043]0.052***[0.024,0.080]> 14 days0.071*[0.009,0.133]0.107***[0.056,0.158]0.094***[0.041,0.147]0.082*[0.018,0.146]0.056[-0.002,0.114]0.062*[0.014,0.111] Quantile regression analysis for OOPE Fig. 3Quantile regressions predicting OOPE at the 0.05–0.95 quantile Quantile regressions predicting OOPE at the 0.05–0.95 quantile Discussion This study addresses a critical gap in evidence on payment system effects in neuro-interventional stroke care, a high-cost specialty with limited targeted research. By employing quantile regression, we found that DRG payment was associated with lower TDHC and OOPE across most expenditure strata compared to FFS, consistent with international studies of appendicitis patients in South Korea [21], somatic patients in Switzerland [22], and colectomy patients in the United States [23]. These findings add to the literature by demonstrating differential effects in a specialized clinical context, highlighting the need for specialty-specific payment policy adjustments. China’s DRG reform operates within a government-led, universal basic medical insurance framework that differs fundamentally from market-oriented or single-payer systems in other countries. Unlike the United States’ multi-payer model, where Medicare Severity-DRG incorporates episode-based bundled payments spanning 90-day periods for conditions such as ischemic stroke, covering both inpatient and post-acute care [24, 25], China’s DRG system remains limited to inpatient services without outpatient integration. In contrast to Germany’s G-DRG system, which permits flexible clinical severity stratification with up to nine severity levels and provider-level adjustments [25], China emphasizes standardization of medical behavior and centralized procurement for high-cost supplies. While Germany adapted the Australian Classification of Health Interventions into its localized Operation and Procedure Classification, China utilizes ICD-10-CM with localized procedure codes suited to its public hospital-dominated delivery model. Compared with South Korea’s single-payer National Health Insurance, which relies on managed entry agreements and outcome-based pricing for high-cost innovative therapies [26], China employs separate centralized procurement policies for pharmaceuticals and devices rather than integrating these into DRG payment. Although both countries introduced DRG systems to address overtreatment under fee-for-service, South Korea retains fee-for-service for certain categories, whereas China has comprehensively implemented DRG payment in public tertiary hospitals. Furthermore, whereas the Netherlands initially established over 30,000 diagnosis treatment combinations before simplifying to approximately 5000 due to operational complexity [27], and England and Germany utilize numerous exclusion mechanisms for specialized services through additional payments or negotiated budgets, China’s DRG grouping maintains a more streamlined, administratively oriented structure with limited supplementary payment mechanisms. We note that DRG’s cost-lowering association may reflect unobserved confounders (e.g., better pre-hospital care for local patients), and we avoid causal claims, framing findings as associations rather than direct policy effects. Specifically, we identified a decrease in TDHC ranging from 3.84% to 5.16% and a reduction in OOPE of approximately 14.49% to 36.32%. Notably, the reduction in OOPE was more pronounced, indicating that DRG-based reimbursement had a stronger effect on alleviating patients’ financial burden compared to the earlier findings from the 2011 Beijing DRG payment pilot reform [28], which reported reductions of 6.2% and 10.5% in health expenditures and out-of-pocket payments, respectively. This finding holds particular relevance in China’ s healthcare landscape, where DRG payment policies aim to mitigate patients’ economic strain and enhance equitable healthcare access [29]. Despite extensive review of the extant literature, we identified limited comparable international evidence regarding DRG’s impact on OOPE. A Thai study [30] found no significant OOPE differences between DRG and FFS systems, contrasting with our findings. This divergence likely reflects fundamental differences in healthcare financing structures: Thailand’s Universal Health Coverage Scheme maintains OOPE at 11.3% of total health expenditures [31], whereas China’s system exhibits substantially higher patient cost-sharing (23.3% of health expenditures) [32]. These structural variations highlight the importance of context in interpreting payment system effects, limiting direct international comparability. Our findings underscore the high proportion of medical consumables in neuro-interventional treatment costs, which align with the rapid development of vascular stents and embolic materials [33], heralding a new era in cerebrovascular disease management. As medical technology evolves, associated costs tend to escalate. Our study revealed that neuro-interventional treatment is indeed a procedure with substantial consumable consumption. The proportion of medical consumables cost under FFS and DRG reached 85.06% and 83.09%, respectively. Although DRG payment led to a reduction in consumables costs, their proportion remained a major cost component (consistently above 80%). The regulation of high-value consumables has become a critical point in controlling the rapid growth of medical expenses in China [34], especially given the slower growth of health insurance funds compared to expenditure. Thus, policy oversight for neuro-interventional treatment in China necessitates further strengthening of DRG payment policies in conjunction with medical service pricing and cost accounting. Future research should evaluate consumables’ cost-effectiveness, integrating medical and economic considerations. In addition to the payment type, we identified several intriguing individual and clinical treatment factors influencing TDHC and OOPE. Our analysis revealed that the adult age group exhibited higher TDHC and OOPE compared to the elderly and pediatric populations, suggesting a greater economic burden and a greater propensity for costly interventions among middle-aged patients. This observation aligns with the composition of our study sample, where over 60% of the adult group comprised aneurysmal cases, often opting for high-value consumables such as stents and coils [35]. Our quantile regression analysis revealed a noteworthy pattern among the highest cost patient cohort: with each increment in age category(e.g., from middle-aged to elderly), TDHC and OOPE decreased. This suggests an inverse relationship between age and hospital expenditures, which contrasts with the positive association typically observed in conventional cost studies [36, 37]. This suggests middle-aged patients are a key demographic for cost containment, while DRG must avoid undertreatment of high-cost elderly patients. Future research should examine quality outcomes to ensure cost reductions do not compromise care. Our analysis revealed distinct patterns in the impact of insurance status on TDHC and OOPE. UEBMI and RURBMI constitute the two primary insurance populations in China, accounting for over 95% of the population [38]. For TDHC, UEBMI beneficiaries consistently incurred higher costs across all expenditure categories compared to RURBMI, reflecting the higher healthcare expenditure among Chinese employees insured under the UEBMI system. This aligns with the contributions and benefit coverage associated with these insurance types and is consistent with previous findings by Lin et al. [39]. The reimbursement rates for UEBMI were around 85%, while those for RURBMI were approximately 65% [40]. Given the better financial protection offered by UEBMI, patients are more likely to opt for higher-value medical consumables. However, the pattern for OOPE diverged from that of TDHC. As expenditure categories increased, the gap in OOPE between RURBMI and UEBMI narrowed, suggesting that the UEBMI system’s impact on patient financial burden was more pronounced among lower-cost patients, while the cost difference diminished as personal expenditure grew. However, this disparity was less pronounced among patients with higher OOPE because the cost-containment effect of UEBMI did not perform better than RURBMI in this cost bracket. The enhanced financial protection provided by UEBMI for patients with higher OOPE did not translate into a significant reduction in their financial burden compared to RURBMI beneficiaries. This suggests that while UEBMI may offer better overall financial protection, its impact on cost containment is not as effective among patients with higher OOPE as it is among those with lower OOPE. This highlights the need to reduce healthcare inequality between insurance types and evaluate UEBMI’s potential for overutilization. Unexpectedly, our study also revealed that intensive use of healthcare resources, such as ICU transfers and extended lengths of stay, consistently led to significant increases in both TDHC and OOPE. These cost implications align with the findings of Schouela [41] and Dykes [42] et al. Recently, the sustained growth in China’s healthcare expenditures has spurred widespread attention towards the assessment of efficiency and cost performance [43]. The ongoing refinement of DRG payment instruments is imperative and should be tailored according to the medical expenditures and costs associated with specific diseases, as well as patient conditions and individual circumstances [44]. Our study presents several implications for future research directions: Firstly, from a cost-containment perspective in the management of ischemic stroke, it is essential to scrutinize the critical junctures of high healthcare resource consumption, with a focus on enhancing the efficiency of each medical process and ensuring the rationality of medical expenditures. Secondly, to optimize DRG payment policies, there is a compelling need to ground these policies in the actual clinical practices for ischemic stroke, effectively distinguishing between varying degrees of disease severity and healthcare resource utilization within the same disease category. Given the backdrop of China’s National Volume-Based Procurement policy, which aims to centralize purchasing power and implement a pooled procurement system to encourage the use of cost-effective medications and medical supplies [45], our study delves into the practical implications of this policy on patient-level outcomes. By examining the use of CPMS and CPM among patients, we explore how these procurement strategies may influence healthcare costs and resource utilization within the context of neuro-interventional treatment for ischemic stroke. In line with existing research, our analysis of centrally procured medications aligns with previous findings: Wang et al. demonstrated that centralized procurement of cardiovascular drugs effectively reduced medication burdens for patients with cardiovascular diseases [46]. Yang et al. found that centralized procurement of tyrosine kinase inhibitors alleviated the economic pressure on patients with chronic myelogenous leukemia [47], and Lin et al. showed that centralized procurement of meropenem reduced the financial burden on hospitalized patients [48]. In our study sample, medications procured through CPM showed a significant downward burden trend, particularly among higher-cost patients. This confirms the national centralized drug procurement policy’s effectiveness in alleviating individual and socio-economic burdens, providing empirical evidence for its optimization. However, contrary to expectations, centralized procured medical supplies (CPMS) for neuro-intervention had an inverse cost-increasing effect, especially in higher-expenditure groups. This discovery not only contradicts the established conclusion that centralized procurement of artificial joints significantly lowers hospitalization costs [49] but also prompts a deeper reflection on the differential effects of VBP policies across various medical domains. Limitations Our study, while methodologically robust, acknowledges several inherent limitations that warrant careful consideration when interpreting the findings. Firstly, the single-center, retrospective design inherently constrains the generalizability of our findings. Although the dataset was derived from a prominent national neurological center that serves a diverse patient population from across China, patient selection was not entirely randomized. Specifically, out-of-province patients, who were treated under the traditional FFS model, may present with more complex conditions and consequently higher inherent costs compared to local Beijing residents reimbursed under DRG schemes. This potential selection bias could lead to an overestimation of the observed effects of DRG implementation, thereby limiting the causal inference of our findings. Therefore, future research would benefit significantly from multicenter studies, potentially leveraging national sampling data, to provide more scientifically rigorous validation and enhance the external validity of these results. Secondly, despite our comprehensive collection of baseline demographic, clinical, and outcome variables from both medical and billing records, the current analysis did not encompass behavioral economic factors. Specifically, patient-specific financial status and willingness to pay, or the deeper mechanisms influencing healthcare-seeking behaviors, were not explored. Understanding these factors is crucial, as they can significantly impact both the utilization and perceived value of healthcare services, and their exclusion may limit the comprehensiveness of our analysis of DRG’s impact on patient financial protection. Further investigations should delve into the associations between patient and physician behaviors and their impact on clinical outcomes, incorporating validated measures such as the National Institutes of Health Stroke Scale (NIHSS) and modified Rankin Scale (mRS) scores to provide a more holistic evaluation. Thirdly, this study primarily focused on direct medical costs associated with neuro-interventional treatment. It did not account for indirect costs, such as loss of work productivity due to illness or caregiving, which constitute a substantial economic burden on patients and society Additionally, our analysis did not differentiate between medical cost accounting and service pricing, a distinction that could influence the precise estimation of economic impacts. The exclusion of indirect costs and the lack of differentiation between cost accounting and pricing may lead to an incomplete understanding of the overall economic implications of DRG payment reform for neurointerventional stroke care. Subsequent health economic studies should adopt a broader perspective, integrating real-world data to conduct comprehensive cost-benefit or cost-effectiveness analyses. Such an approach would provide a more complete understanding of the overall economic implications of neuro-interventional treatments for neurological disorders. Conclusion This study provides a preliminary evaluation of the association between DRG and FFS payment systems, and their respective relationships with hospitalization costs and OOPE among patients undergoing neuro-interventional treatment for ischemic stroke at a single national neurological center in China. Our findings indicate that DRG is associated with significantly lower costs across most of the expenditure distribution, with particularly pronounced effects in higher OOPE strata. These observations imply that DRG payment system may have the potential to mitigate financial burdens on patients and contribute to more equitable access to neuro-interventional care, though this inference is limited by the study’s observational design. However, the unexpected increase in costs associated with high-value consumables under centralized procurement policies highlights the need for further research to understand the underlying mechanisms and optimize the balance between cost containment and quality of care. Limitations of this study—including non-random payment system assignment, single-center design, and exclusion of high-risk cases—restrict generalizability and preclude causal inference. Future studies should focus on multicenter designs to validate these findings and explore the broader implications of payment reforms on healthcare resource allocation and patient outcomes. Additionally, there is a clear need for further investigation into the cost-effectiveness of high-value medical consumables in the context of neuro-interventional treatments to inform policy decisions and ensure the sustainable advancement of value-based healthcare delivery.