The impact of DRG payment reform on stroke inpatient costs: a propensity score matching and difference-in-differences analysis of a large public hospital in Northeast China Authors: Lingxiao Gao, Ruinan Li, Huaizhi Cheng, Xiaoyu Dong, Yanrui Liu, Ruotong Tian, Tingting Zhang, Bin Guo Published: 2025 ABSTRACT BackgroundThe advancement of China’s Diagnosis Related Groups (DRG) payment reform represents a crucial measure to deepen the healthcare system reform, address the persistent rise in medical costs, and counter the imbalance between medical insurance fund revenues and expenditures. This study aims to evaluate the impact of the DRG payment policy on hospitalization costs for stroke patients.MethodsMedical records of stroke patients discharged from a large public hospital in Northeast China between September 2018 and December 2024 were collected. Propensity Score Matching (PSM) and Difference-in-Differences (DID) models were employed to analyze the impact of DRG payment on stroke patients’ length of stay (LOS), cost structure, and out-of-pocket (OOP) expenditure share.ResultsAfter the implementation of DRG payment, the total hospitalization costs for stroke patients decreased by 19.31%. While treatment costs showed minimal change, all other individual cost items experienced significant reductions, with the largest declines observed in Western medicine costs (23.11%) and consumable costs (32.21%). The LOS and OOP expenditure proportions decreased by 8.90% and 2.83%, respectively. Heterogeneity analysis indicates that DRG payment reduced hospitalization costs by 18.48% for elderly patients, 22.68% for non-elderly patients, 21.97% for male patients, and 15.94% for female patients.ConclusionThe DRG payment plays a vital role in optimizing the structure of hospitalization costs and controlling unreasonable cost growth.However, further efforts are needed, including deepening salary system reforms, strengthening the development of DRG information platforms, refining basic medical insurance, and providing medical assistance for low-income populations. Special attention should be given to vulnerable groups such as the elderly and women, with improved green channels and enhanced humanistic care.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-025-25946-5. Introduction The rapid rise in healthcare costs is one of the most pressing public health issues currently facing countries worldwide [1]. Research indicates that from 1995 to 2016, the annual growth rate of healthcare expenditures in high-income countries was 3.61%, while in low-income countries it reached 4.25%, both significantly outpacing the average global GDP growth rate (2.96%) during the same period [2]. A WHO report further indicates that global healthcare expenditures are rising sharply, with middle- and low-income countries experiencing particularly pronounced growth at an average annual rate of 6%, compared to 4% in high-income countries. By 2019, global healthcare spending exceeded 10% of GDP, growing at an average annual rate of 3.9% [3]. Driven by multiple factors including overmedicalization, shifts in disease patterns, and accelerated population aging, China’s healthcare expenditure surged from $78.5 billion in 2000 to $1.1 trillion in 2020, which represents a compound annual growth rate of approximately 14.8% [4]. The persistent surge in healthcare expenditures poses a dual threat to the sustainability of healthcare service provision and the affordability of medical care for residents: On the one hand, the unreasonable growth of medical expenses continues to intensify the financial pressure on healthcare insurance funds and public finances. Over the long term, this threatens the financial sustainability of the health system and exacerbates structural imbalances in medical resources [5]. On the other hand, rising healthcare costs also increase the risk of households experiencing catastrophic health expenditure (CHE). Research indicates that among households affected by CHE, medical expenses account for an average of 64% of their non-subsistence expenditures, with the rate of medical impoverishment remaining particularly high [6]. To mitigate the impact of rising medical costs on global healthcare systems, countries worldwide are establishing multi-stakeholder cost control mechanisms through various approaches. These include promoting reforms in healthcare payment systems, improving performance evaluation frameworks, and refining risk-sharing mechanisms. Such efforts aim to ensure the long-term stable development of healthcare systems. One such payment reform model is the Diagnosis Related Groups (DRG) system. This model categorizes cases with similar clinical pathways and resource consumption based on primary diagnosis, treatment modality, and patient characteristics, assigning differentiated reimbursement weights to each group [7]. Since its initial implementation in the United States in 1983, DRG has gained widespread international recognition as an advanced and scientific payment method, now adopted in over 50 countries and regions worldwide [8]. Different countries have adapted the DRG system to align with their unique healthcare system characteristics, ultimately forming distinct local payment models. Since introducing the G-DRG system in 2003, Germany has developed it into a core pillar of its healthcare system over two decades, with approximately 80% of hospital reimbursements settled through this system [9]. Japan integrated DRG with its traditional point-based system to create the Diagnosis Procedure Combination (DPC) framework. This system allocates reimbursement based on patient diagnosis, care type, and length of stay(LOS), effectively enhancing treatment quality and resource allocation efficiency [10]. South Korea introduced the K-DRG system, centered on LOS as the primary grouping variable, to address overtreatment under fee-for-service models. This significantly reduced hospitalization durations and readmission rates [11]. Thailand’s Thai-DRG developed 2,450 DRG groups, comprehensively covering all disease types under both the Civil Servant Welfare Scheme and Universal Health Coverage. This system plays a crucial role in controlling healthcare costs and achieving budget savings [12]. DRG payment reforms worldwide demonstrate their irreplaceable role in optimizing healthcare resource allocation, enhancing service efficiency, and promoting the sustainable development of healthcare systems. Studies on the effectiveness of DRG implementation across different countries and regions vary in focus and conclusions. Aragón MJ et al. utilized 15 years of microdata from England to conclude that DRG has a substantial and time-increasing impact on shortening hospital stays [13]; Messerle R et al. noted that DRG can increase patient discharge rates and inpatient activity, suggesting the reform incentivizes hospitals to proactively control medical costs, but cautioned against the risks of fragmented hospital stays and reduced service quality [9]; Khadivi Y et al. confirmed that the DRG payment system effectively controls drug expenditures and enhances nursing efficiency, thereby optimizing inpatient cost structures and boosting healthcare workers’ motivation [14]. China’s DRG payment system has undergone over a decade of development, achieving a major leap from localized pilot programs to nationwide implementation. In 2011, Beijing launched the nation’s first DRG payment pilot program. By 2019, the National Healthcare Security Administration and three other ministries has designated 30 national DRG payment pilot cities, a move that marked the reform’s entry into a substantive phase. The 2021 Three-Year Action Plan for DRG/DIP Payment Reform explicitly set the goal of extending DRG/DIP coverage to all designated medical institutions providing inpatient services [15]. China’s adoption of a dual-system healthcare supply structure, in which public hospitals play a dominant role and private hospitals serve as a supplement, is distinctly characteristic of the Chinese approach. However, challenges such as uneven distribution of medical resources between urban and rural areas and insufficient integration of healthcare services have prompted the Chinese government to gradually explore a DRG payment reform path tailored to national conditions. As the reform continues to advance, the academic community has begun to pay widespread attention to and evaluate the actual effects of DRG payment. Yu Q et al. calculated inpatient service scores across 49 hospitals, revealing that DRG plays an important role in enhancing financial operational efficiency and adjusting hospital revenue sources [16]; Luo M et al. found through empirical research that DRG payment reform significantly reduces total hospitalization costs for complex surgical patients, particularly in medication and material expenses [17]. However, while the DRG payment reform has yielded positive outcomes, it has also triggered a series of issues and challenges. In a longitudinal study conducted by Wang JY et al. at two sample hospitals in western China, it was found that patients’ LOS significantly increased after DRG implementation. Furthermore, cost-control effects varied across different types of medical insurance: inpatient costs decreased by 4.9% for employee medical insurance and 3.8% for resident medical insurance, deviating to some extent from the fundamental principles of fairness and inclusiveness inherent in the medical insurance system [18]. Zhang L et al. suggested that DRG payment implementation reduced the utilization of novel complex technologies, thereby inhibiting medical technological innovation [19]. Research indicates that healthcare system reform can reduce public hospitals’ heavy reliance on pharmaceutical sales and advanced technology utilization for fiscal revenue, thereby lowering medical expenditures and optimizing income structures [20].As key drivers of rising healthcare costs and a major component of health expenditure, public hospitals operate primarily with medical insurance funds as their main payer. This financing model necessitates that reform uses economic incentives to guide behavior. Given variations in disease severity, resource consumption levels, and reimbursement rates across different conditions, this study focuses solely on analyzing the impact of DRG on inpatient costs for a specific set of diseases. When conducting DRG payment effectiveness research in Northeast China, selecting stroke, which is a disease highly prevalent in cold regions, as a representative case study holds both regional significance and clinical research value. On the one hand, influenced by winter winds and cold air activity, the region experiences over five months of severe winter cold. Low temperatures stimulate sympathetic nervous system activity and increased adrenaline secretion, inducing vasoconstriction and blood pressure fluctuations, resulting in some of the highest stroke incidence and mortality rates nationally [21–23]. Simultaneously, the prolonged rehabilitation and complex complications impose significant economic burdens on stroke patients and their families. On the other hand, as a chronic disease requiring multidisciplinary collaborative care, stroke management through clinical pathways spans multiple stages, including acute diagnosis and treatment, rehabilitation training, and complication prevention, thereby providing a comprehensive research environment to examine the impact of DRG payment systems on disease cost control and healthcare quality. Despite growing research and practice on DRG payment reforms globally, few studies have focused on changes in actual economic burdens at the individual patient level, particularly regarding stroke in Northeast China. Furthermore, due to China’s relatively late adoption of DRG payment, most prior empirical studies have concentrated on short-term pre- and post-reform comparisons [24], making it challenging to fully evaluate the long-term dynamic effects of policy adjustments. Therefore, this study employs Propensity Score Matching (PSM) and Difference-in-Differences (DID) models to analyze changes in stroke patients’ LOS, inpatient cost structure, and proportion of out-of-pocket payments (OOP) before and after DRG reform at a large tertiary public hospital in the sample city. This analysis provides scientific evidence and practical guidance for alleviating patients’ medical burdens and refining medical insurance payment reform. Materials and methods Data sources The sample city in this study was among the first batch of DRG national pilot cities selected in Northeast China in 2019, and officially launched its DRG actual payment settlement scheme on 1 November 2021. A large tertiary-level Class A general hospital in this city was selected as the sample institution. This hospital operates two campuses with over 1,800 beds, handles more than 700,000 outpatient visits annually, and admits nearly 70,000 inpatients each year. Among these, it treats over 9,000 patients with cerebrovascular diseases annually, providing ample data for this empirical study. Data were sourced from the sample hospital’s information system, encompassing records of discharged patients including personal characteristics (gender, age, medical payment methods), clinical information (actual LOS, admission route, surgical status), and cost details (total hospitalization expenses and itemized breakdowns). All collected data excluded personal privacy and sensitive information, thus requiring neither ethical approval nor informed consent. Sample selection Based on the International Classification of Diseases, 10th Revision (ICD-10) codes for stroke (I60-I64), we identified hospitalized patients with a principal diagnosis within this range and a discharge date between 1 September 2018 and 31 December 2024 [25]. To ensure the accuracy and reliability of the study results, after excluding cases with missing key variables or illogical cost information, a total of 27,334 hospitalized patients were included. Among these, 11,532 patients were discharged before the reform (from 1 September 2018 to 31 October 2021), and 15,802 discharged after the reform (from 1 November 2021 to 31 December 2024). Outcome variables The outcome variables were the LOS, hospitalization costs, and OOP ratio for stroke patients. Hospitalization costs encompassed total costs, comprehensive medical service costs (including general medical service costs, general therapeutic procedure costs, nursing costs, and other costs), diagnostic costs (including pathological diagnosis costs, laboratory diagnosis costs, imaging diagnosis costs, and clinical diagnostic item costs), treatment costs (including non-surgical treatment costs and surgical treatment costs), western medicine costs, traditional Chinese medicine (TCM) costs, and consumable costs (including examination medical supplies costs, treatment medical supplies costs, and surgical medical supplies costs). The OOP ratio was defined as the proportion of total costs covered by personal cash payments, reflecting the direct financial burden borne by patients. Statistical analysis To eliminate the impact of inflation and price fluctuations on hospitalization costs across different years and ensure comparability of hospitalization data, all cost data from 2018 to 2024 were adjusted using the healthcare component of the Consumer Price Index (CPI) published by the Provincial Bureau of Statistics of Province H (where the sample hospitals are located). The adjustment used 2024 as the base year. To satisfy the core assumptions of the DID model, Kolmogorov-Smirnov tests were conducted on hospital stay duration and hospitalization costs. The results indicated that both variables showed skewed distributions (P < 0.05). After natural logarithmic transformation, both variables approximately followed a normal distribution. Propensity score matching-difference in differences (PSM-DID) This study employs PSM-DID to estimate the net effect of the DRG payment reform. This integrated approach constructs a counterfactual framework to control for selection bias. First, since DRG payment was only applied to patients enrolled in the National Basic Medical Insurance program, the sample was divided into treatment and control groups based on the type of health insurance coverage. The treatment group consisted of patients enrolled in basic medical insurance, including the Urban Employee Basic Medical Insurance and the Urban and Rural Resident Basic Medical Insurance (the latter has integrated the former Urban Resident Basic Medical Insurance and the New Rural Cooperative Medical Scheme; detailed information on China’s basic medical insurance is provided in Supplementary Material 1). The control group comprised patients under other payment models not subject to the DRG payment reform during the same period, including those covered by commercial health insurance, full self-payment, or full public funding. Second, covariates including gender, age, number of hospitalizations, employment, marital status, residence type, admission route, admission condition, other diagnoses, and surgical status were selected to estimate propensity scores using a Logit model. Finally, a 1:2 nearest-neighbor matching without replacement within a caliper of 0.05 was performed to improve statistical power and matching quality. On this basis, the DID model was introduced to examine the impact of DRG payment on the LOS, hospitalization costs, and actual economic burden for stroke patients. The specific model is as follows:\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\begin{aligned} \:\mathrm{ln}\left({Y}_{it}\right)&={\beta\:}_{0}+{\beta\:}_{1}{Time}_{t}+{\beta\:}_{2}{Treated}_{i}+\\&{\beta\:}_{3}{Time}_{t}*{Treated}_{i}+{\beta\:}_{4}{X}_{it}+{\epsilon\:}_{it} \end{aligned}$$\end{document} In the equation, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Y}_{it}$$\end{document} represents a series of hospitalization-related outcome variables for patient i at time t; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Time}_{t}$$\end{document} is a time dummy variable, coded as 0 before the implementation of the DRG policy and 1 after its implementation; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Treated}_{i}$$\end{document} is a group dummy variable, coded as 0 for the control group and 1 for the treatment group; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{Time}_{t}*{Treated}_{i}$$\end{document} is the interaction term between the two, and its coefficient \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\beta\:}_{3}$$\end{document}​ represents the net impact of the DRG payment reform; \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{X}_{it}$$\end{document}​ denotes control variables, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\epsilon\:}_{it}$$\end{document} represents the random error term. All statistical analyses were performed using STATA MP 18.0. Results Descriptive statistics Descriptive statistical analysis of key variables was conducted using the mean to measure central tendency and the standard deviation to measure dispersion, as shown in Table 1. The mean hospitalization costs were higher for male patients than for females both before and after DRG implementation. The hospitalized population was predominantly elderly, consistent with the epidemiological characteristics of stroke. Furthermore, patients with ≥ 4 diagnoses and surgical patients continued to exhibit higher mean hospitalization costs after the reform. Additionally, the mean values of LOS, total hospitalization costs, detailed cost components (excluding treatment costs), and OOP ratio all decreased. Table 1Descriptive statistics of hospitalization costs by patient characteristicsVariablePre_DRGPost_DRGMeanSDMeanSDGenderFemale14142.4014972.369134.5310119.73Male14906.9615187.2210004.8811233.08Age67.6311.7167.8710.68Readmission statusNo14889.1516481.4610296.3113084.78Yes14149.5812855.718888.677695.01EmploymentNo14936.9113883.909455.6310279.53Yes14547.6215250.969857.0511609.36Marital statusUnmarried15513.5214605.7810872.3514440.99Married14518.8915139.729522.1210509.31Residence typeRural13396.8214291.8010243.0314805.28Urban14742.9015196.659451.339708.21Admission routeOutpatient13754.4013850.478897.568529.92Emergency16856.5518063.2812043.4715457.21Other15690.1113996.3418276.4527564.60Admission conditionClear14603.4215425.769619.6410903.72Unclear14445.3310358.539141.798244.96Other diagnoses<411544.978475.767727.586049.87≥ 416852.0218222.9210758.4912684.53Surgical statusNo13225.949272.368348.104441.24Yes69467.7153505.2231905.8036216.98Length of stay9.584.487.903.20Total hospitalization costs14591.5115102.989581.0010713.82Comprehensive medical service costs2186.493528.721344.632020.78Diagnostic costs4313.232640.753243.551599.65Treatment costs324.041451.98347.151588.47Western medicine costs5931.375016.153535.283001.68Traditional Chinese medicine costs50.04107.9827.6273.00Consumable costs1306.995006.38847.935647.98OOP ratio27.16%28.07%25.29%22.54% Table 1Descriptive statistics of hospitalization costs by patient characteristicsVariablePre_DRGPost_DRGMeanSDMeanSDGenderFemale14142.4014972.369134.5310119.73Male14906.9615187.2210004.8811233.08Age67.6311.7167.8710.68Readmission statusNo14889.1516481.4610296.3113084.78Yes14149.5812855.718888.677695.01EmploymentNo14936.9113883.909455.6310279.53Yes14547.6215250.969857.0511609.36Marital statusUnmarried15513.5214605.7810872.3514440.99Married14518.8915139.729522.1210509.31Residence typeRural13396.8214291.8010243.0314805.28Urban14742.9015196.659451.339708.21Admission routeOutpatient13754.4013850.478897.568529.92Emergency16856.5518063.2812043.4715457.21Other15690.1113996.3418276.4527564.60Admission conditionClear14603.4215425.769619.6410903.72Unclear14445.3310358.539141.798244.96Other diagnoses<411544.978475.767727.586049.87≥ 416852.0218222.9210758.4912684.53Surgical statusNo13225.949272.368348.104441.24Yes69467.7153505.2231905.8036216.98Length of stay9.584.487.903.20Total hospitalization costs14591.5115102.989581.0010713.82Comprehensive medical service costs2186.493528.721344.632020.78Diagnostic costs4313.232640.753243.551599.65Treatment costs324.041451.98347.151588.47Western medicine costs5931.375016.153535.283001.68Traditional Chinese medicine costs50.04107.9827.6273.00Consumable costs1306.995006.38847.935647.98OOP ratio27.16%28.07%25.29%22.54% Descriptive statistics of hospitalization costs by patient characteristics Propensity score matching results The matching results indicate that sample balance was significantly improved, satisfying the prerequisite for causal inference. Figure 1 shows that a total of 27,310 samples were successfully matched, with only 24 samples excluded due to propensity scores falling outside the common range (9 from the control group and 15 from the treatment group), demonstrating strong comparability across samples. In terms of balance, Fig. 2 reveals that the absolute values of all standardized differences dropped below 10% after matching, indicating effective control of intergroup disparities. Meanwhile, Fig. 3 further demonstrates a significant increase in the overlap of propensity score distributions between the treatment and control groups post-matching. Statistical metrics further support these findings: the pseudo-R²value decreased significantly from 0.051 to 0.000 after matching, and both the weighted bias index and variance ratio fell within the ideal range (B = 4.0%, R = 1.03). These results confirm that PSM effectively reduced selection bias attributable to observable factors, thereby providing robust evidence for subsequent causal effect estimation. Fig. 1Common support distribution in propensity score matching Common support distribution in propensity score matching Fig. 2Comparison of standardized differences in covariates before and after propensity score matching Comparison of standardized differences in covariates before and after propensity score matching Fig. 3Propensity score kernel densities before and after matching. A Before Matching, B After Matching Propensity score kernel densities before and after matching. A Before Matching, B After Matching Difference-in-differences results Table 2 reports the coefficient estimates for the impact of DRG reform on stroke-related outcome variables among hospitalized patients. The coefficient for the LOS is significantly negative, with an effect size of−8.90% (derived from 1-e−0.0933), indicating that patient hospital stays decreased by 8.90% following DRG implementation. Regarding cost indicators, the results show total costs decreased significantly by 19.31%. Among the detailed cost categories, treatment costs showed no significant change (P=0.323), while significant reductions were observed in comprehensive medical service costs (−18.53%), diagnostic costs (−11.80%), western medicine costs (−23.11%), TCM costs (−10.89%), and consumable costs (−32.21%). Furthermore, the patients’ OOP ratio decreased significantly by 2.83% after DRG implementation, indicating that the reform effectively reduced patients’ actual medical expenditures. Table 2Impact of DRG payment on hospitalization cost metrics for stroke patientsLn(LOS)Ln(THC)Ln(CMSC)Ln(DC)Ln(TC)Ln(WC)Ln(TMC)Ln(CC)OOP ratiodid−0.0933***−0.2146***−0.2049***−0.1256***−0.036−0.2628***−0.1154***−0.3888***−0.0287***−0.007−0.007−0.011−0.008−0.037−0.013−0.031−0.014−0.003Gender0.0090.0574***0.0795***−0.0160**0.0683**0.1307***−0.2630***0.1087***−0.0211***−0.006−0.006−0.009−0.007−0.032−0.011−0.027−0.012−0.003Age0.0031***0.0052***0.0127***−0.0025***−0.0069***0.0089***0.0092***0.0071***−0.0022***−3E-04−3E-04−5E-04−4E-04−0.002−5E-04−0.001−6E-04−1E-04Readmission status0.0674***−0.0123*−0.0528***−0.0252***0.030.0241**0.2078***−0.1875***−0.0698***−0.007−0.007−0.01−0.008−0.034−0.012−0.029−0.013−0.003Employment0.0594***0.1521***0.1466***0.0816***0.0778**0.2114***0.2167***0.3053***0.0136***−0.008−0.007−0.011−0.009−0.038−0.013−0.032−0.015−0.004Marital status0.0826***0.001−0.0947***0.0637***0.1040.010.2131***−0.1739***−0.0384***−0.017−0.017−0.025−0.02−0.086−0.029−0.072−0.033−0.008Residence type0.1232***0.1544***0.1402***0.1655***0.0940*0.1864***0.2060***0.1384***−0.2146***−0.01−0.009−0.014−0.011−0.048−0.016−0.041−0.019−0.005Admission route−0.0585***0.0953***0.1898***0.0427***0.0643*0.1119***−0.1535***0.3258***0.0533***−0.007−0.007−0.011−0.008−0.036−0.012−0.031−0.014−0.003Admission condition0.0110.0160.0522***0.008−0.0060.0557**0.048−0.0210.006−0.013−0.013−0.02−0.015−0.067−0.023−0.056−0.026−0.006Other diagnoses0.0635***0.2162***0.2553***0.2527***0.3733***0.1591***0.2411***0.2580***−0.0168***−0.007−0.007−0.01−0.008−0.033−0.011−0.028−0.013−0.003Surgical status0.1803***1.1150***1.1269***0.3425***5.6900***0.7138***−0.5360***2.3885***0.0811***−0.018−0.018−0.027−0.021−0.092−0.031−0.077−0.035−0.009cons1.5771***8.1212***5.3514***7.5841***0.7966***6.7716***−0.0134.5032***0.6329***−0.035−0.035−0.052−0.041−0.176−0.06−0.149−0.068−0.016N241312413124131241312413124131241312413124131Adj. R20.04330.26870.18430.08530.14740.10630.02270.28860.1922Standard errors are reported in parentheses. *p<0.1, **p<0.05, ***p<0.001LOS Length of stay, THC Total hospitalization costs, CMSC Comprehensive medical service costs, DC Diagnostic costs, TC Treatment costs, WC Western medicine costs, TMC Traditional Chinese medicine costs, CC Consumable costs Table 2Impact of DRG payment on hospitalization cost metrics for stroke patientsLn(LOS)Ln(THC)Ln(CMSC)Ln(DC)Ln(TC)Ln(WC)Ln(TMC)Ln(CC)OOP ratiodid−0.0933***−0.2146***−0.2049***−0.1256***−0.036−0.2628***−0.1154***−0.3888***−0.0287***−0.007−0.007−0.011−0.008−0.037−0.013−0.031−0.014−0.003Gender0.0090.0574***0.0795***−0.0160**0.0683**0.1307***−0.2630***0.1087***−0.0211***−0.006−0.006−0.009−0.007−0.032−0.011−0.027−0.012−0.003Age0.0031***0.0052***0.0127***−0.0025***−0.0069***0.0089***0.0092***0.0071***−0.0022***−3E-04−3E-04−5E-04−4E-04−0.002−5E-04−0.001−6E-04−1E-04Readmission status0.0674***−0.0123*−0.0528***−0.0252***0.030.0241**0.2078***−0.1875***−0.0698***−0.007−0.007−0.01−0.008−0.034−0.012−0.029−0.013−0.003Employment0.0594***0.1521***0.1466***0.0816***0.0778**0.2114***0.2167***0.3053***0.0136***−0.008−0.007−0.011−0.009−0.038−0.013−0.032−0.015−0.004Marital status0.0826***0.001−0.0947***0.0637***0.1040.010.2131***−0.1739***−0.0384***−0.017−0.017−0.025−0.02−0.086−0.029−0.072−0.033−0.008Residence type0.1232***0.1544***0.1402***0.1655***0.0940*0.1864***0.2060***0.1384***−0.2146***−0.01−0.009−0.014−0.011−0.048−0.016−0.041−0.019−0.005Admission route−0.0585***0.0953***0.1898***0.0427***0.0643*0.1119***−0.1535***0.3258***0.0533***−0.007−0.007−0.011−0.008−0.036−0.012−0.031−0.014−0.003Admission condition0.0110.0160.0522***0.008−0.0060.0557**0.048−0.0210.006−0.013−0.013−0.02−0.015−0.067−0.023−0.056−0.026−0.006Other diagnoses0.0635***0.2162***0.2553***0.2527***0.3733***0.1591***0.2411***0.2580***−0.0168***−0.007−0.007−0.01−0.008−0.033−0.011−0.028−0.013−0.003Surgical status0.1803***1.1150***1.1269***0.3425***5.6900***0.7138***−0.5360***2.3885***0.0811***−0.018−0.018−0.027−0.021−0.092−0.031−0.077−0.035−0.009cons1.5771***8.1212***5.3514***7.5841***0.7966***6.7716***−0.0134.5032***0.6329***−0.035−0.035−0.052−0.041−0.176−0.06−0.149−0.068−0.016N241312413124131241312413124131241312413124131Adj. R20.04330.26870.18430.08530.14740.10630.02270.28860.1922Standard errors are reported in parentheses. *p<0.1, **p<0.05, ***p<0.001LOS Length of stay, THC Total hospitalization costs, CMSC Comprehensive medical service costs, DC Diagnostic costs, TC Treatment costs, WC Western medicine costs, TMC Traditional Chinese medicine costs, CC Consumable costs Impact of DRG payment on hospitalization cost metrics for stroke patients Standard errors are reported in parentheses. *p<0.1, **p<0.05, ***p<0.001 LOS Length of stay, THC Total hospitalization costs, CMSC Comprehensive medical service costs, DC Diagnostic costs, TC Treatment costs, WC Western medicine costs, TMC Traditional Chinese medicine costs, CC Consumable costs Heterogeneity analysis results According to the WHO definition, the study sample was stratified by age into two groups for baseline regression analysis: non-elderly (< 60 years) and elderly (≥ 60 years) [26]. The DRG policy was associated with a reduction in treatment costs of 17.38% for non-elderly patients and a decrease in TCM costs of 12.29% for elderly patients. For variables such as total hospitalization costs, LOS, OOP ratio, and other cost sub-items, the non-elderly population was the most significantly affected by the DRG reform, showing a pronounced downward trend across all metrics. In the gender stratification analysis, male patients experienced greater reductions than females in LOS, OOP ratio, total costs, and most cost components after DRG implementation, suggesting that male stroke patients were generally more responsive to the reform. However, treatment costs for both genders and TCM costs for females did not change significantly (P > 0.05). Regarding occupational stratification, the DRG payment reform primarily impacted employed patients. Treatment costs and TCM costs for employed stroke patients decreased by 10.87% and 23.93%, respectively, while the cost control effects for unemployed patients were not significant. Moreover, for other hospitalization cost variables, the reform effects were generally significantly greater for employed individuals than for unemployed individuals, as detailed in Table 3.Table 3Results of Heterogeneity Analysis for the DRG Payment ReformGroupsVariableLn(LOS)Ln(THC)Ln(CMSC)Ln(DC)Ln(TC)Ln(WC)Ln(TMC)Ln(CC)OOP ratioPanel A: Different age groups<60did−0.0993***−0.2575***−0.2348***−0.1769***−0.1911**−0.3010***−0.093−0.5167***−0.0687***−0.015−0.014−0.021−0.017−0.075−0.026−0.059−0.028−0.007cons1.6794***8.4542***6.2143***7.3233***0.4177.3457***0.7125**5.1386***0.6402***≥60did−0.0915***−0.2044***−0.2092***−0.1042***0.021−0.2548***−0.1312***−0.3499***−0.0129***−0.008−0.008−0.012−0.01−0.042−0.015−0.034−0.016−0.003cons1.7947***8.4511***6.1410***7.4642***0.3345**7.3281***0.5054***4.8963***0.4347***Panel B: Different sex groupsFemaledid−0.0845***−0.1738***−0.1754***−0.0869***0.013−0.2122***−0.01−0.3472***−0.0235***−0.01−0.01−0.015−0.012−0.053−0.02−0.045−0.021−0.004cons1.5985***8.1252***5.2052***7.7686***0.8421**6.6616***−0.194.2433***0.5555***Maledid−0.1005***−0.2480***−0.2310***−0.1549***−0.072−0.3064***−0.2060***−0.4257***−0.0339***−0.01−0.01−0.014−0.011−0.051−0.017−0.039−0.019−0.004cons1.5684***8.1846***5.5492***7.4329***0.8857***6.9921***−0.164.8270***0.6757***Panel C: Different career groupsEmployeddid−0.0442***−0.1086***−0.1453***−0.0410***0.028−0.1232***0.035−0.1963***−0.0256***−0.01−0.01−0.014−0.012−0.049−0.018−0.041−0.02−0.004cons1.5313***8.1032***5.2702***7.6891***0.9235**6.6985***−0.4899*4.1855***0.5473***Unemployeddid−0.1481***−0.3324***−0.2859***−0.2095***−0.1151**−0.4224***−0.2735***−0.6127***−0.0412***−0.011−0.011−0.016−0.012−0.057−0.019−0.043−0.021−0.005cons1.6788***8.3212***5.5595***7.6534***0.8979***7.0617***0.5891**5.0518***0.6903***Standard errors are reported in parentheses.*p<0.1, **p<0.05, ***p<0.001LOS Length of stay, THC Total hospitalization costs, CMSC Comprehensive medical service costs, DC Diagnostic costs, TC Treatment costs, WC Western medicine costs, TMC Traditional Chinese medicine costs, CC Consumable costs Table 3Results of Heterogeneity Analysis for the DRG Payment ReformGroupsVariableLn(LOS)Ln(THC)Ln(CMSC)Ln(DC)Ln(TC)Ln(WC)Ln(TMC)Ln(CC)OOP ratioPanel A: Different age groups<60did−0.0993***−0.2575***−0.2348***−0.1769***−0.1911**−0.3010***−0.093−0.5167***−0.0687***−0.015−0.014−0.021−0.017−0.075−0.026−0.059−0.028−0.007cons1.6794***8.4542***6.2143***7.3233***0.4177.3457***0.7125**5.1386***0.6402***≥60did−0.0915***−0.2044***−0.2092***−0.1042***0.021−0.2548***−0.1312***−0.3499***−0.0129***−0.008−0.008−0.012−0.01−0.042−0.015−0.034−0.016−0.003cons1.7947***8.4511***6.1410***7.4642***0.3345**7.3281***0.5054***4.8963***0.4347***Panel B: Different sex groupsFemaledid−0.0845***−0.1738***−0.1754***−0.0869***0.013−0.2122***−0.01−0.3472***−0.0235***−0.01−0.01−0.015−0.012−0.053−0.02−0.045−0.021−0.004cons1.5985***8.1252***5.2052***7.7686***0.8421**6.6616***−0.194.2433***0.5555***Maledid−0.1005***−0.2480***−0.2310***−0.1549***−0.072−0.3064***−0.2060***−0.4257***−0.0339***−0.01−0.01−0.014−0.011−0.051−0.017−0.039−0.019−0.004cons1.5684***8.1846***5.5492***7.4329***0.8857***6.9921***−0.164.8270***0.6757***Panel C: Different career groupsEmployeddid−0.0442***−0.1086***−0.1453***−0.0410***0.028−0.1232***0.035−0.1963***−0.0256***−0.01−0.01−0.014−0.012−0.049−0.018−0.041−0.02−0.004cons1.5313***8.1032***5.2702***7.6891***0.9235**6.6985***−0.4899*4.1855***0.5473***Unemployeddid−0.1481***−0.3324***−0.2859***−0.2095***−0.1151**−0.4224***−0.2735***−0.6127***−0.0412***−0.011−0.011−0.016−0.012−0.057−0.019−0.043−0.021−0.005cons1.6788***8.3212***5.5595***7.6534***0.8979***7.0617***0.5891**5.0518***0.6903***Standard errors are reported in parentheses.*p<0.1, **p<0.05, ***p<0.001LOS Length of stay, THC Total hospitalization costs, CMSC Comprehensive medical service costs, DC Diagnostic costs, TC Treatment costs, WC Western medicine costs, TMC Traditional Chinese medicine costs, CC Consumable costs Results of Heterogeneity Analysis for the DRG Payment Reform Standard errors are reported in parentheses.*p<0.1, **p<0.05, ***p<0.001 LOS Length of stay, THC Total hospitalization costs, CMSC Comprehensive medical service costs, DC Diagnostic costs, TC Treatment costs, WC Western medicine costs, TMC Traditional Chinese medicine costs, CC Consumable costs Validation of the DID model Parallel trends test A key assumption for applying the DID model to estimate the impact of the DRG policy on hospitalization costs is satisfying the parallel trends test, meaning that the outcome variables in the treatment and control groups exhibit identical trends prior to policy implementation [24]. Using the policy enactment date (1 November 2021) as the intervention point, the study constructed interaction terms between dummy variables for periods before and after the DRG policy and the treatment group dummy variable, then plotted the dynamic effects. Figure 4 shows that the 95% confidence intervals for coefficients in all periods prior to the DRG policy implementation included 0, indicating consistent trends between the treatment and control groups. This result supports the validity of the subsequent estimates of the policy effect. Fig. 4Results of the parallel trends test Panels (A) through (I) show the results of the parallel trends assumption test for each outcome variable before and after the intervention. The horizontal axis indicates different time points relative to the policy implementation, and the vertical axis represents the estimated coefficients of the interaction terms for each period. (A) LOS, Length of Stay (B) THC, Total hospitalization costs (C) CMSC, Comprehensive medical service costs (D) DC, Diagnostic costs (E) TC, Treatment costs (F) WC, Western medicine costs (G) TMC, Traditional Chinese medicine costs (H) CC, Consumable costs (I) OOP ratio, Out-of-pocket ratio Results of the parallel trends test Panels (A) through (I) show the results of the parallel trends assumption test for each outcome variable before and after the intervention. The horizontal axis indicates different time points relative to the policy implementation, and the vertical axis represents the estimated coefficients of the interaction terms for each period. (A) LOS, Length of Stay (B) THC, Total hospitalization costs (C) CMSC, Comprehensive medical service costs (D) DC, Diagnostic costs (E) TC, Treatment costs (F) WC, Western medicine costs (G) TMC, Traditional Chinese medicine costs (H) CC, Consumable costs (I) OOP ratio, Out-of-pocket ratio Robustness tests The dependent variable was winsorized at 1% in both tails to mitigate the influence of outliers on the regression results, thus improving the reliability and robustness of the findings, as presented in Table 4. The trends of all outcome variables were consistent with the DID model. With the exception of treatment costs, which were statistically non-significant (P = 0.346), all other outcomes exhibited significant negative effects, and the results proved robust. Table 4Results of Robustness Tests for the DRG Payment ReformLn(LOS)Ln(THC)Ln(CMSC)Ln(DC)Ln(TC) Ln(WC)Ln(TMC)Ln(CC)OOP ratiodid-0.0917***-0.2114***-0.2020***-0.1284***-0.0346-0.2503***-0.1141***Ln(CC)-0.0287***(0.0072) (0.0068) (0.0068)(0.0073)(0.0367)(0.0111)(0.0309)(0.0134)(0.0034) Control variablesYesYesYesYesYes YesYesYesYescons1.5891***8.1377***5.3753***7.6444***0.7899***6.8460***-0.00424.4900***0.6329***(0.0346)(0.0328)(0.0502)(0.0352)(0.1758)(0.0533)(0.1482)(0.0643)(0.0163) N241312413124131241312413124131241312413124131Adj. R20.04210.26730.18270.09800.14280.11910.02280.2946 0.1922Standard errors are reported in parentheses. *p<0.1, **p<0.05, ***p<0.001LOS Length of stay THC Total hospitalization costs, CMSC Comprehensive medical service costs, DC Diagnostic costs, TC treatment costs, WC western medicine costs, TMC traditional Chinese medicine costs, CC Consumable costs Table 4Results of Robustness Tests for the DRG Payment ReformLn(LOS)Ln(THC)Ln(CMSC)Ln(DC)Ln(TC) Ln(WC)Ln(TMC)Ln(CC)OOP ratiodid-0.0917***-0.2114***-0.2020***-0.1284***-0.0346-0.2503***-0.1141***Ln(CC)-0.0287***(0.0072) (0.0068) (0.0068)(0.0073)(0.0367)(0.0111)(0.0309)(0.0134)(0.0034) Control variablesYesYesYesYesYes YesYesYesYescons1.5891***8.1377***5.3753***7.6444***0.7899***6.8460***-0.00424.4900***0.6329***(0.0346)(0.0328)(0.0502)(0.0352)(0.1758)(0.0533)(0.1482)(0.0643)(0.0163) N241312413124131241312413124131241312413124131Adj. R20.04210.26730.18270.09800.14280.11910.02280.2946 0.1922Standard errors are reported in parentheses. *p<0.1, **p<0.05, ***p<0.001LOS Length of stay THC Total hospitalization costs, CMSC Comprehensive medical service costs, DC Diagnostic costs, TC treatment costs, WC western medicine costs, TMC traditional Chinese medicine costs, CC Consumable costs Results of Robustness Tests for the DRG Payment Reform Standard errors are reported in parentheses. *p<0.1, **p<0.05, ***p<0.001 LOS Length of stay THC Total hospitalization costs, CMSC Comprehensive medical service costs, DC Diagnostic costs, TC treatment costs, WC western medicine costs, TMC traditional Chinese medicine costs, CC Consumable costs Placebo tests The study randomly generated a pseudo-treatment group while ensuring the treatment group proportion matched the original proportion. The PSM-DID model was then used to iteratively estimate the effects for each of the nine outcome variables, with this process repeated 500 times. As shown in Fig. 5, the gray vertical dashed lines represent the coefficient estimates from the primary DID model, while the red horizontal dashed lines indicate the significance threshold (p < 0.1). With the exception of treatment costs, the coefficient estimates for the remaining eight outcome variables fell well outside the distribution of effects under the null hypothesis. Moreover, most P-values exceeded 0.1, failing to meet the 10% significance standard. These results indicate that these variables passed the placebo test, validating the accuracy of the prior DID results. In contrast, for treatment costs (β = −0.0363, p > 0.05), which were already statistically insignificant in the main model, the placebo test also failed to yield a significant effect, as shown in Fig. 5(E).These mutually corroborating results suggest that treatment costs may be less influenced by DRG policy, consistent with theoretical expectations. Fig. 5Results of the placebo tests Panels (A) through (I) present the results of placebo tests obtained by randomly assigning pseudo-treatment groups (500 simulation iterations). The horizontal axis represents the estimated treatment effect coefficients from the simulations, and the vertical axis shows the kernel density of the corresponding coefficients. (A) LOS, Length of Stay (B) THC, Total hospitalization costs (C) CMSC, Comprehensive medical service costs (D) DC, Diagnostic costs (E) TC, Treatment costs (F) WC, Western medicine costs (G) TMC, Traditional Chinese medicine costs (H) CC, Consumable costs (I) OOP ratio, Out-of-pocket ratio Results of the placebo tests Panels (A) through (I) present the results of placebo tests obtained by randomly assigning pseudo-treatment groups (500 simulation iterations). The horizontal axis represents the estimated treatment effect coefficients from the simulations, and the vertical axis shows the kernel density of the corresponding coefficients. (A) LOS, Length of Stay (B) THC, Total hospitalization costs (C) CMSC, Comprehensive medical service costs (D) DC, Diagnostic costs (E) TC, Treatment costs (F) WC, Western medicine costs (G) TMC, Traditional Chinese medicine costs (H) CC, Consumable costs (I) OOP ratio, Out-of-pocket ratio Discussion The DRG payment reform contributes to optimizing the cost structure for stroke hospitalization, yet room for improvement remains A key objective of the DRG payment reform is to control healthcare costs and promote the rational allocation of medical resources through a prepayment mechanism, while ensuring quality and safety of care, which is to achieve “cost control and efficiency enhancement” [27]. The results indicate that following the implementation of DRG payment, total hospitalization costs decreased by 19.31%. This reform also significantly curbed costs across individual cost categories, including comprehensive medical service costs, diagnostic costs, western medicine costs, TCM costs, and consumable costs. Among these, western medicine and consumable costs experienced the most substantial reductions, declining by 23.11% and 32.21%, respectively. Two primary factors may explain these outcomes. First, DRG employs a bundled payment approach wherein hospitals retain savings if actual costs fall below the medical insurance payment standard but must cover losses if costs exceed it. This “retain savings, cover overruns” principle incentivizes hospitals to reduce the use of ancillary drugs and prophylactic antibiotics [19]. Second, the national centralized volume-based procurement policy for drugs and medical supplies has substantially lowered acquisition costs for hospitals. Under the DRG system, hospitals increasingly prefer domestically produced alternatives from centralized procurement, which offer significant advantages in both price and quality [28]– [29]. Although implemented for only three years, the policy has demonstrated growing effectiveness in curbing the practice of subsidizing hospital operations through drug and material sales. Both total hospitalization costs and individual item costs have shown marked reductions. The comprehensive medical service costs and treatment costs reflect the technical and labor value of healthcare professionals. However, results indicate that neither category experienced the anticipated increase following DRG implementation (comprehensive medical service costs decreased by 18.53%, while treatment costs showed no significant change). This discrepancy may be attributed to the relatively short time since the policy’s implementation, suggesting that the labor value of healthcare personnel remains inadequately recognized and reflected during this phase. Research indicates that the undervaluation of healthcare professionals’ labor has long been a persistent issue in China’s disease cost accounting [30]. As a specialized group safeguarding public health, healthcare workers undergo extensive academic training, rigorous qualification assessments, and continuous on-the-job education, investing substantial time, financial resources, and energy [31]– [32]. Their daily work also exposes them to occupational hazards, including medical disputes, which may lead to serious personal injury [33]. However, healthcare workers in China generally receive low compensation and lack sufficient societal recognition [34]. This phenomenon reflects the reality of incommensurate compensation relative to their contributions, adversely affecting the attractiveness of healthcare professions and workforce stability. In Northeast China, the brain drain of medical talent has been exacerbated by multiple factors, including limited local fiscal capacity, lower regional wage levels, and relatively narrow professional development platforms [35]. If this situation persists, it will dampen enthusiasm for pursuing medical careers and reduce job satisfaction, ultimately hindering the sustainable development of the healthcare industry [36]. In contrast, developed nations regard medical professionals as an elite class, granting them top-tier wages and social status, where professional dignity aligns closely with the intrinsic value of their work [37]. The current payment reform represented by DRG offers an opportunity to address this dilemma. However, its incentive orientation must shift from the traditional volume-based approach (such as fee-for-service) to a value-oriented framework [38], thereby avoiding potential issues like overutilization associated with the former. Therefore, compensation system reforms should be continuously deepened. Measures such as increasing the proportion of fixed salaries and adjusting DRG weighting coefficients based on nursing intensity and treatment complexity should be implemented to enhance the value-based remuneration share for healthcare providers. Ultimately, a value-driven, multidimensional performance evaluation system should be established, incorporating key outcome indicators such as complication rates and readmission rates, to promote the structure of inpatient costs toward a sustainable development model grounded in skill-based compensation [39]. Concurrently, strengthening the development of DRG information management platforms, which involves the optimization, upgrading, and interconnection of medical record, medical information, clinical outcome monitoring and financial management systems, will establish a crucial pillar for public hospitals to implement cost control strategies effectively. This will provide essential information technology support for further reducing patient hospitalization costs [40]. The DRG payment reform significantly reduces OOP ratio and length of stay(LOS) for stroke patients Results indicate that after implementing DRG payment, the OOP ratio and LOS for stroke patients decreased by 2.83% and 8.90%, respectively. This suggests that the DRG reform offers significant advantages in reducing patients’ financial burden and enhancing the efficiency of healthcare resource allocation, consistent with previous domestic and international findings [41]– [42]. The reduction in OOP ratio highlights the economic leverage of the DRG reform, countering the perverse incentives of previous reimbursement systems that shifted costs to patients to sustain hospital revenues [43]. This not only lowers healthcare expenditures for stroke patients and their families but also improves the accessibility and affordability of medical services. However, the decrease in the OOP ratio was relatively modest compared to that of other dependent variables and may be insufficient to fundamentally alleviate the heavy burden on impoverished populations. Therefore, China must further refine its basic medical insurance system, gradually expand the coverage of medical insurance, increase reimbursement rates, and impose a cap on the self-payment limit for low-income groups in combination with the DRG payment weight [44].Additionally, it should provide medical assistance and secondary reimbursement to these groups. Concurrently, residents should be encouraged to enroll in commercial health insurance (such as Huiminbao, a complementary commercial insurance widely promoted in China since 2020) to ensure seamless integration with basic coverage and deliver multi-tiered, multidimensional healthcare protection [45]. The average length of stay (ALOS) is a key indicator for measuring the turnover efficiency of medical resources and the comprehensive service capacity of hospitals. China has long regarded it as a primary metric for evaluating healthcare quality and operational efficiency [46]. The DRG payment system, by establishing payment standards for stroke diagnosis-related groups, encourages hospitals to strengthen clinical pathway management to standardize and normalize diagnostic and therapeutic processes, thereby reducing unnecessary hospital stays [17, 47]. However, reducing LOS is a double-edged sword. While it enhances medical efficiency and lowers hospitalization costs, it also poses potential threats to medical quality and patient safety. On the one hand, shorter stays may increase the likelihood of hospitals refusing severe cases and engaging in hospitalization fragmentation. To avoid financial risks and control costs for efficiency improvement, some hospitals might decline to admit complex, high-cost, resource-intensive cases or split continuous treatment processes into multiple admissions to shift the risk of overspending [48]. On the other hand, the decline in LOS has led to a significant increase in patient admissions per unit time. The resulting compressed schedules and surging workloads exacerbate healthcare workers’ occupational burnout and operational risks, thereby posing potential threats to medical safety [49]– [50]. Therefore, as the core healthcare providers and primary cost controllers under the DRG system, hospitals must prioritize healthcare workers’ mental health. Establishing routine psychological support systems through measures such as anonymous stress surveys and mental health training can mitigate medical risks stemming from psychological fatigue [51]. Additionally, specialized institutions such as management committees should be actively established to supervise and regulate physicians’ deviant behaviors under the DRG payment system [24]. Simultaneously, it is recommended to promote Enhanced Recovery After Surgery (ERAS) protocols. Based on evidence-based medicine, these protocols integrate multidisciplinary perioperative interventions to develop DRG clinical pathways incorporating core ERAS elements. This approach ultimately achieves dual objectives: enhancing patient outcomes while optimizing hospital operational efficiency [52]. The DRG payment reform is effective in controlling costs for stroke patients, but exhibits variations by age and sex Results indicate that DRG payment reduced hospitalization costs by 18.48% for elderly patients and 22.68% for non-elderly patients, while male and female patients saw reductions of 21.97% and 15.94%, respectively. This suggests that elderly and female patients, as traditionally vulnerable populations, require special attention in future DRG reforms. Compared to the elderly, non-elderly patients generally have better physical fitness and recovery capacity, making them more likely to meet discharge criteria and satisfy DRG requirements for shortening hospitalization duration [53]. With advancing age, aging becomes the most significant non-modifiable risk factor for stroke. From a clinical perspective, elderly stroke patients often suffer from multiple comorbidities such as hypertension and diabetes, coupled with limited vascular patency and tolerance, significantly increasing their LOS and treatment complexity [54]. Research indicates a positive correlation between patient LOS and healthcare resource consumption [55]. Although China’s DRG payment system provides higher reimbursement rates for high-resource-consumption cases, these incremental payments still fail to fully compensate hospitals for actual costs [56]. Therefore, it is recommended to enhance green channels to optimize the entire treatment process through rapid identification, prioritized examinations, and multidisciplinary collaboration, thereby reducing waiting times for elderly patients and providing convenient and efficient care. Additionally, the DRG grouping mechanism should be optimized by developing specialized geriatric groups that incorporate disease severity and functional prognosis in older adults, thereby increasing reimbursement rates [57]. Furthermore, economic incentives should be employed to guide rational resource allocation, adjust DRG weightings to better support primary care facilities, and standardize two-way referrals for geriatric stroke patients to promote coordinated reform of the tiered medical system and DRG payment [41, 58]. Following the DRG reform, the reduction in total costs was more pronounced among male patients than female patients. This gender disparity stems from multiple contributing factors. First, male patients had a higher baseline total cost prior to the reform, which provided greater potential for reduction under cost containment policies. This base effect implies that the DRG payment policy may yield more pronounced marginal effects for male patients. Second, Female patients exhibit unique pathophysiological characteristics potentially linked to pregnancy and menopausal hormonal changes. However, current DRG-standardized clinical pathways do not adequately account for these factors, resulting in treatment plans that are less tailored to the actual needs of female patients [59]– [60]. Moreover, female patients exhibit lower rates of formal treatment and continuous care, along with poorer post-stroke quality of life compared to males. This suggests that social determinants such as economic constraints and educational levels often cause delayed medical care or poor rehabilitation adherence among women after illness onset, thereby increasing hospitalization costs and readmission risks [61]. Women play a significant role in both family and society, serving as not only the emotional bond connecting family members but also the core force driving social development and civilization progress. Their physical and mental health impacts not only individual quality of life and family harmony but is also inextricably linked to sustainable social development. Therefore, enhanced humanistic care for female patients is essential, emphasizing privacy protection and informed consent during treatment, along with providing social reintegration guidance to help them adapt to post-illness role changes [62]. Concurrently, grassroots organizations such as communities and women’s federations should collaborate with medical institutions to offer stroke-related health lectures, consultations, and guidance for women. This can help prevent treatment delays and disease progression caused by insufficient disease awareness, thereby reducing critical care costs under the DRG payment system [63]– [64]. Regularly organizing free health screenings can facilitate early identification of high-risk female stroke populations for standardized treatment, ultimately achieving the dual goals of lowering stroke incidence and enhancing women’s health awareness. Limitations and future directions This study has several limitations. First, as a single-center study, all data were sourced from one tertiary hospital, resulting in limited sample breadth, representativeness, and a potential risk of information bias. Future research should be conducted as multicenter studies with expanded sample sources and broader geographic coverage to obtain more generalizable conclusions. Second, owing to constraints in data availability, this study focused solely on inpatient costs for stroke patients, and the evaluation metrics were relatively limited. Future research should incorporate additional healthcare quality indicators, such as readmission rates and complication incidence, to allow for a more comprehensive assessment of the overall impact of the DRG payment system. Conclusion This study analyzed the policy effects of DRG payment reform on stroke patients. Our findings indicate that DRG payment significantly reduced patients’ LOS, OOP ratio, total hospitalization cost, and all other cost components other than treatment costs. The hospitalization cost structure was markedly optimized, though room for further adjustment remains. Furthermore, this empirical study may offer valuable insights for China’s future healthcare service provision and rational allocation of health resources and could also serve as a reference for healthcare system reforms in other countries with similar contexts. Supplementary Information Supplementary Material 1 Supplementary Material 1