China’s Hospital Payment Reform Pays Some Patients More — And Hospitals Respond Differently

When China rolled out its Diagnosis-Intervention Packet, or DIP, payment reform, the policy’s architects promised a more transparent way to reimburse hospitals. Instead of paying hospitals fee-for-service, the system assigns every inpatient admission to a standardized packet based on diagnosis and treatment, gives each packet a relative weight, and multiplies that weight by a point value to determine the payment. But a new analysis of nearly 1.75 million hospital admissions reveals a hidden asymmetry baked into the system’s design: because the two main public insurance schemes maintain separate budgets and separate point values, patients with identical conditions can generate different payments depending solely on which insurance pool they belong to. And hospitals, the data show, respond to that difference in measurable ways.

The study, published in the International Journal for Equity in Health by researchers at Tongji Medical College of Huazhong University of Science and Technology, examined all 209 healthcare institutions in a single Chinese city between June 2021 and November 2023. The team analyzed 1,746,605 inpatient admissions, tracking four outcomes over time: decomposed admissions, seven-day readmissions, the share of admissions with high relative weights, and the share with low relative weights. Using an interrupted time-series design anchored to the reform’s budget-reconciliation breakpoints in May 2022 and May 2023, the researchers could isolate what changed when the payment rules changed.

The DIP reform’s structure is what creates the asymmetry. China’s Urban Employee Basic Medical Insurance, or UEBMI, covers the employed urban workforce and is generally funded at higher levels. The Urban-Rural Resident Basic Medical Insurance, or URRBMI, covers rural residents and other non-employee populations and typically pays less. Under DIP, each scheme runs its own budget pool and its own insurance-specific point value, so a patient in the same DIP group — same diagnosis, same intervention, same relative weight — receives a different total payment depending on which insurance type appears on the claim. The researchers found that this payment gap remained approximately 40 percent across the study phases.

What happened after the first budget reconciliation in May 2022 was striking. The rate ratio comparing URRBMI to UEBMI for decomposed admissions — admissions broken into multiple separate packets, a pattern often scrutinized as potential gaming — was 0.17, with a 95 percent confidence interval of 0.09 to 0.32 and an FDR-adjusted q-value below 0.001. In plain terms, decomposed admissions fell dramatically more for URRBMI patients than for UEBMI patients. Seven-day readmissions showed the same directional pattern: the URRBMI-to-UEBMI rate ratio was 0.55 (95 percent CI 0.43 to 0.71, q < 0.001), meaning short-interval readmissions declined markedly more for the lower-paying scheme’s patients.

The researchers were careful about causal inference. They used quasi-Poisson log-link models with an offset equal to log monthly admissions and hospital-clustered robust standard errors, aggregating outcomes to the hospital-month-insurance level. High- and low-RW indicators used fixed pooled pre-period thresholds — the 75th percentile at 975.76 and the 25th percentile at 321.00 — so the classification rules did not shift mid-study. Pooled models with insurance-by-policy interactions directly tested whether the two schemes responded differently to the same policy change. Residual-autocorrelation diagnostics, including Durbin-Watson and Ljung-Box tests, were non-significant, suggesting the models captured the time-series structure adequately.

Pre-trend checks revealed one modest wrinkle. Per-scheme pre-reconciliation slopes did not differ from zero for decomposed admissions or seven-day readmissions, meaning neither outcome was already moving before the policy breakpoint. However, a modest differential pre-trend — URRBMI relative to UEBMI — was detected for seven-day readmission, a caveat the authors acknowledge. Sensitivity analyses excluding October through December 2022, the period of China’s COVID-19 exit wave, produced changes of less than 10 percent for most key estimates, indicating the findings were not driven by that extraordinary period.

Where the differential responses concentrated is perhaps the most policy-relevant finding. The effects were concentrated in secondary and tertiary hospitals — the larger, better-resourced institutions that handle complex cases and have the most capacity to influence admission patterns. Primary hospitals showed far less differential response. This pattern is consistent with a mechanism in which larger hospitals, facing tighter budgets and more competitive pressure, are more attuned to the payment difference between insurance pools and more capable of acting on it, whether consciously or through operational drift.

The study’s authors frame the results as sparking concerns about hospitals potentially favoring higher-paying UEBMI inpatients. If decomposed admissions and short-interval readmissions — both markers of potentially aggressive admission or treatment patterns — decline more for URRBMI patients, the implicit worry is that lower-paying patients may be receiving less intensive care, fewer interventions, or less follow-up attention. The findings do not prove that individual patients were harmed or denied care, but they document a systematic, insurance-linked difference in how hospital behavior changed after the payment rules tightened.

Encouragingly, the story does not end with the asymmetry. The researchers found that the gaps between the two schemes narrowed over successive budget reconciliations as outcome-linked penalties and audit monitoring took effect. The DIP system’s design includes mechanisms that penalize hospitals for poor outcomes — including readmissions — and audit claims for signs of gaming. As those enforcement mechanisms matured, the differential response between the two insurance pools shrank, suggesting the system can correct its own payment-driven distortions when monitoring is active.

For China’s health policymakers, the study offers a technical but consequential lesson: when a payment system assigns different point values to different insurance pools, hospitals will notice, and their responses will differ by pool. The DIP reform’s separate-budget architecture may be administratively convenient, but it embeds a payment asymmetry that larger hospitals translate into differential admission and readmission patterns. The narrowing of gaps over successive reconciliations indicates positive effects in curbing payment-driven gaming and safeguarding equitable care — but the initial differential response shows how quickly payment incentives propagate through hospital operations. As China continues to refine its DIP rollout across provinces, the study suggests that unifying point values across insurance pools, or strengthening the outcome-linked penalties that appear to have narrowed the gaps, could be key to ensuring that a patient’s insurance type does not shape the care they receive.

Subject of Research: Health Policy

Article Title: Differential provider responses to insurance-specific point values under China’s diagnosis-intervention packet payment reform: an interrupted time-series analysis

Article References: Differential provider responses to insurance-specific point values under China’s diagnosis-intervention packet payment reform: an interrupted time-series analysis. (n.d.). https://doi.org/10.1186/s12939-026-03069-6

Image Credits: AI Generated

DOI: 10.1186/s12939-026-03069-6

Keywords: Differential, provider, responses, insurance-specific, point, values, under, China, diagnosis-intervention, packet, payment, reform