Chinese Physicians Weigh Financial Losses Heavily When Deciding Which Patients to Admit Under DIP Payment Reform

When China rolled out its Diagnosis-Intervention Packet, or DIP, payment system, the goal was to rein in medical costs while keeping care quality intact. But a new study suggests the reform's financial signals are landing on physicians with surprising force: doctors at a large tertiary hospital in central China were far more deterred by the prospect of losing money on a patient case than they were attracted by the prospect of making money on one. The finding, published in Frontiers in Public Health, shows how frontline clinicians translate payment policy into admission decisions, and it carries a clear behavioral signature: losses loom much larger than gains.

The research team, led by Yuanming Song, Siyue Zheng, and Mingkai Zhao of the First Affiliated Hospital of Henan Medical University, surveyed 339 physicians using a discrete choice experiment, a survey method that asks respondents to repeatedly choose between hypothetical options whose characteristics are systematically varied. In this case, each choice task presented two hypothetical patient admission scenarios alongside an option to decline admission altogether. The hospital where the study took place has applied DIP payment to all inpatient cases since 2023, covering more than 3,000 DIP groups, making it a setting in which the system's incentives are already embedded in daily clinical judgment.

DIP itself is worth understanding. Unlike traditional fee-for-service reimbursement, which pays providers for each drug, test, and procedure and is widely blamed for encouraging overprovision, DIP is a case-based, point-based model operating within a regional global budget. Each hospitalization is categorized by a diagnosis-procedure combination, assigned a point value, and settled against a locally determined point value.

China's National Healthcare Security Administration piloted DIP in 71 cities between 2019 and 2021, and a national action plan set the goal of covering the entire country with DRG or DIP payment by 2025. Because China runs the world's largest basic medical insurance system, and because expenditure growth in its employee medical insurance fund outpaced revenue growth in 2024, the stakes of getting provider incentives right are enormous. To build the experiment, the researchers first reviewed 16 national policy documents issued between 2016 and 2025 and then interviewed eight key informants, including senior clinicians from pediatrics, urology, and oncology and five hospital administrators overseeing insurance, medical records, quality, medical affairs, and operations.

From this groundwork they distilled six attributes of an admission decision: clinical complexity, length of stay relative to the institutional benchmark, whether treatment followed a conventional or innovative pathway, the share of case revenue coming from professional services rather than drugs and tests, the expected DIP margin per case, and the compliance risk attached to coding and documentation. Each attribute was assigned realistic levels, for example a financial loss defined as payment falling 2,000 yuan short of cost, and the scenarios were pretested with 15 physicians for clarity and realism. A D-efficient design generated 36 choice sets distributed across four questionnaire versions of nine tasks each, plus a consistency-check task used to screen out inattentive responses.

The statistical results were striking. Using mixed logit models on 5,596 observations, the team found that an expected DIP loss was associated with a coefficient of minus 1.81, while an expected profit of the same magnitude carried a coefficient of only plus 0.45. A formal Wald test rejected the hypothesis that these effects were symmetric, confirming that the negative response to losses significantly exceeded the positive response to gains.

This asymmetry is a textbook signature of loss aversion, the behavioral economics principle that losses loom larger than equivalent gains, and it suggests that DIP's design, which frames every case as a financial unit under a fixed budget, makes physicians especially sensitive to cases expected to cost the hospital money. Expected DIP margin dominated the preference structure, accounting for 41.94 percent of relative attribute importance calculated from utility ranges. Clinical complexity came second at 20.11 percent, with high-complexity cases carrying a coefficient of minus 1.08, and compliance risk third at 16.01 percent, with moderate and high compliance risk associated with coefficients of minus 0.66 and minus 0.86 respectively.

A higher share of service revenue from professional fees such as consultation and surgery was positively associated with admission preference at 0.65, as was the innovative-technology scenario at 0.30, while a length of stay above the institutional benchmark carried a smaller negative weight of minus 0.23. Significant random-parameter standard deviations for most attributes revealed that physicians differed substantially in how much each factor mattered to them. The opt-out analysis added another layer.

Physicians declined to admit the hypothetical patient in 8.29 percent of the 3,051 valid choice sets, and the tendency to walk away rose when scenarios involved expected DIP loss, high clinical complexity, high or moderate compliance risk, or above-benchmark length of stay. Conversely, higher service revenue share, expected profit, and innovative technology reduced the likelihood of opting out. Taken together, the pattern is consistent with risk selection: in payment systems that transfer financial risk to providers, physicians may shy away from high-cost or medically complex patients when the payment formula does not adequately adjust for their resource needs.

Latent class modeling uncovered five distinct preference profiles among the physicians. The largest group, 28.5 percent, was labeled economically driven, favoring profitable and break-even cases, innovative technology, and higher service revenue while disliking high compliance risk. A 21.7 percent risk-averse group avoided complex cases and preferred lower service revenue shares.

A 20.6 percent group, labeled high-return and compliance-tolerant, showed positive utility for the high-compliance-risk scenario, a finding the authors caution reflects preferences for the complete questionnaire scenario and should not be read as willingness to engage in noncompliant practice. A 16.4 percent technology-preferring group and a 12.8 percent conservatively compliant group rounded out the picture. Notably, exploratory subgroup analyses found no significant overall differences by professional title, clinical role, or department, meaning the heterogeneity cannot be easily explained by observable professional characteristics.

The compliance-risk findings carry particular policy weight. DIP, like DRG legislation in the United States and audit mechanisms in Germany, relies on data-driven audits, inspections, and penalties, and the study suggests these regulatory pressures are now salient enough to shape which patients physicians prefer to admit. The authors argue that excessive reliance on retrospective, penalty-driven audits may increase risk avoidance without improving clinical value, and they propose instead a graduated approach combining real-time alerts, standardized coding guidance, department-specific training, and corrective feedback for first-time, non-malicious documentation problems.

On the payment side, they recommend refined risk adjustment for complex patients, supplementary compensation for unusually high-cost cases, and periodic review of recurrent loss-making case groups to check whether point values reflect actual resource use. At the hospital-management level, they suggest internal performance evaluation should avoid excessive reliance on departmental financial surplus and incorporate indicators of clinical appropriateness, patient safety, and quality of care. The study's limitations are clearly acknowledged.

It was conducted at a single tertiary hospital with a cross-sectional design, so the preferences captured reflect the local DIP implementation environment rather than the causal effect of the reform nationwide. The experiment measured stated preferences in hypothetical scenarios, not observed admissions, and social desirability bias may have influenced responses on financially or ethically sensitive attributes. Some attributes, particularly compliance risk, were composite scenarios whose individual components cannot be separated.

The authors call for multicenter and longitudinal studies linking stated preferences to actual admission behavior, work that will determine whether the loss-averse calculus observed here shapes real-world access to hospital care as China's payment reform reaches full national coverage.

Subject of Research: Physician admission preferences under China's Diagnosis-Intervention Packet hospital payment system, studied with a discrete choice experiment

Article Title: Physician admission preferences under the Diagnosis-Intervention Packet payment system: a discrete choice experiment in a Chinese tertiary hospital

Article References: Song, Y., Zheng, S., Zhao, M., Nie, Z., Zhu, S., Li, K., Duan, X., & Feng, Y. (2026). Physician admission preferences under the Diagnosis-Intervention Packet payment system: a discrete choice experiment in a Chinese tertiary hospital. Frontiers in Public Health, 14, Article 1946819. https://doi.org/10.3389/fpubh.2026.1946819

Image Credits: AI Generated

DOI: 10.3389/fpubh.2026.1946819

Keywords: Diagnosis-Intervention Packet, DIP payment reform, discrete choice experiment, physician preferences, loss aversion, provider payment incentives, health economics, China healthcare, tertiary hospital, latent class analysis, compliance risk, case-based payment