The study used de-identified records from the traditional Chinese medicine department of a tertiary hospital, covering admissions from 2023 through 2025. Of 1,583 records initially identified, 1,532 remained after exclusions for missing key variables, implausible or extreme costs, and out-of-pocket proportions outside the logical range of 0 to 1. The median age of patients was 61 years, with an interquartile range of 47 to 71 years, and 61.4% were women. Median total traditional Chinese medicine cost was 3,446.72 yuan per hospitalization, with an interquartile range of 1,616.00 to 5,324.05 yuan.

Researchers compared seven regression algorithms: linear, ridge, lasso and elastic net models, a decision tree, random forest and XGBoost. The outcome was transformed as log(1 + cost) because medical expenditure was strongly right-skewed, and the two skewed cost-component predictors were transformed in the same way. Hyperparameters were tuned through repeated five-fold cross-validation with five repeats on the training data, while a separate hold-out set containing 20% of records provided internal test validation. Performance was assessed with root mean square error, mean absolute error and R-squared on the log scale, with selected errors also back-transformed into yuan. Analyses were performed in R version 4.4.1.

The optimized random forest performed best in the internal test set, reaching an R-squared of 0.820, a root mean square error of 0.666 and a mean absolute error of 0.353 on the log-transformed scale. Its back-transformed test-set root mean square error was 1,408.6 yuan. XGBoost was a close second, with an R-squared of 0.818 and a back-transformed root mean square error of 1,512.6 yuan. The regularized linear models produced R-squared values of about 0.78 to 0.79, while the single decision tree ranked last, with an R-squared of 0.732, a log-scale root mean square error of 0.813 and a mean absolute error of 0.441. The random forest therefore explained about 82% of the variance in log-transformed total costs, although this was an internal estimate rather than externally validated performance.

Cross-validation results were broadly consistent with the hold-out results. The random forest’s mean cross-validated root mean square error was 0.631, compared with 0.666 in the test set, while its training R-squared of 0.967 indicated some degree of overfitting. The study authors described the training-to-test gap as modest. Predicted and observed values clustered around the one-to-one line, with residuals distributed across the observed cost range and minor deviations at the high-cost end. These results concern internal validation at one center: the study did not test the model in independent hospitals or in a temporally separate cohort.

Two complementary methods were used to interpret the winning model. Permutation importance measured how much model accuracy deteriorated when each variable was shuffled, while SHAP, or Shapley additive explanations, assigned each feature a contribution for each prediction and averaged those contributions across the dataset. Both methods placed the total cost of traditional Chinese medicine treatment items first, length of stay second and Chinese herbal medicine costs third. Their combined SHAP importance was approximately 81%. The mean absolute SHAP values were 0.7508 for treatment-item costs, 0.1836 for length of stay and 0.0997 for herbal medicine costs, compared with 0.0868 for age and still smaller values for the remaining variables.

The leading treatment-item variable represented charges for services such as acupuncture, tuina, moxibustion and acupoint application. Because these charges are themselves components of total traditional Chinese medicine expenditure, their prominence partly reflects arithmetic structure rather than an independently discovered risk factor. The same caution applies to the herbal medicine cost variable. The model therefore functions as a cost-structure explanatory tool: it shows where spending variation is concentrated, but it cannot establish that reducing a particular service or shortening a stay would cause an equivalent reduction in cost.

Length of hospital stay nevertheless had substantial predictive importance in both interpretation analyses. It had a Spearman correlation of 0.566 with total traditional Chinese medicine cost and ranked second in the permutation and SHAP rankings. Longer admissions may coincide with accumulated nursing, bed, medication, retreatment and daily service charges, but the study stressed that stay length can also reflect disease severity, treatment complexity, delayed recovery or discharge-planning constraints. Payment method and the proportion of out-of-pocket expenses made smaller contributions, while detailed reimbursement ratios, benefit packages, income and other socioeconomic measures were unavailable. The reported relationships are therefore predictive associations, not causal effects.

Descriptive comparisons found differences in median costs by sex, marital status, ethnicity, occupation, payment method and critical illness status. For example, median total cost was 3,193.08 yuan for men and 3,604.80 yuan for women. Patients classified as critically ill had a median cost of 3,830.01 yuan, compared with 3,413.57 yuan among other patients. These group differences were exploratory and cannot establish independent contributions because the models lacked several major clinical variables, including diagnosis, disease severity, comorbidities and traditional Chinese medicine syndrome differentiation. Unmeasured socioeconomic, geographic, cultural and healthcare-access factors may also have influenced the observed patterns.

For hospital managers, the immediate implication is that detailed accounting of individual traditional Chinese medicine procedures may be more informative than focusing only on medicines or broad patient characteristics. The authors also argue that standardized management of treatment items could help identify unwarranted variation in service use, and that payment systems shaped by diagnosis-related group and diagnosis-intervention packet reforms may benefit from recognizing the distinctive cost profile of traditional Chinese medicine departments. These are operational possibilities rather than evidence that a specific intervention has already reduced expenditure or improved care.

The authors identified several limits on the findings: the retrospective data came from one center, the models were internally validated only, cost components were included as predictors, major clinical information was omitted, and machine-learning importance does not demonstrate causation. They also noted that repeated admissions were not explicitly excluded, which could affect sample independence despite the methods describing each admission record as an independent unit of analysis. A genuinely prospective administrative model would need to use information available at or shortly after admission, such as age, sex, payment method, admission diagnosis, critical illness status, comorbidities and planned treatment category, rather than charges generated during the hospitalization. Before clinical or administrative use, the researchers call for multicenter external validation, including a temporally distinct validation set where feasible.

Subject of Research: Interpretable machine-learning analysis of cost drivers in inpatient traditional Chinese medicine services

Article Title: Interpretable machine learning for identifying cost drivers of inpatient traditional Chinese medicine services in a tertiary hospital in China

Article References: Li, X., Chen, S., Gong, M., Chen, C., Chen, J., & Zhang, J. (2026). Interpretable machine learning for identifying cost drivers of inpatient traditional Chinese medicine services in a tertiary hospital in China. Frontiers in Public Health, 14, Article 1937712. https://doi.org/10.3389/fpubh.2026.1937712

Image Credits: AI Generated

DOI: 10.3389/fpubh.2026.1937712

Keywords: traditional Chinese medicine, health economics, random forest, machine learning, hospital costs, SHAP analysis, retrospective study, cost accounting, medical expenditure, Guizhou China, Interpretable, machine

Researchers analyzed electronic hospital records for patients with peptic ulcer disease identified by International Classification of Diseases, 10th Revision codes K25 through K29. The study used a prevalence-based, person-based approach because peptic ulcer disease is chronic and cure could not be assumed during the observation period. Costs included examinations, laboratory and imaging tests, functional investigations, medicines and infusions, hospital beds, blood, surgery, medical supplies, and selected non-medical expenses such as food and beverages.

All monetary amounts were initially recorded in Vietnamese dong. Annual costs were adjusted to 2024 using the consumer price index and converted to US dollars at the State Bank of Vietnam’s average exchange rate of VND 24,260 per USD. Patients transferred from gastroenterology to another department and those with incomplete records were excluded. Names, telephone numbers, addresses, and other identifiers had been anonymized before the research team received the data.

Patients’ mean age was 51.7 years, with a standard deviation of 17.4. The largest age groups were 30–54 years, representing 38.8% of the cohort, and 55–70 years, representing 35.2%. Mean hospital stay was 10.6 days, with a range of one to 63 days. Gastric ulcer was the most common diagnosis, while 79.3% of patients had at least one comorbidity. Half of the cohort resided in Ho Chi Minh City and 46.5% lived in other provinces.

Medical supplies were the largest component of the estimated cost, totaling USD 2,054,049, or 49.6% of the six-year total. Medicines and infusions accounted for USD 729,242, or 17.6%; hospital bed fees for USD 518,245, or 12.5%; surgery for USD 402,911, or 9.7%; laboratory tests for USD 257,241, or 6.2%; and diagnostic imaging for USD 129,338, or 3.1%. The share attributed to medical supplies was 1.6–29.4% during 2017–2020 before rising to 41.35% in 2021 and 65.3% in 2022.

Average cost per patient was USD 387.20 in 2017, USD 793.10 in 2018, USD 681.10 in 2019, USD 1,485.90 in 2020, USD 1,810.50 in 2021, and USD 3,784.30 in 2022. Annual numbers of included patients also varied, from 79 in 2017 to 601 in 2022, so the retrospective estimates do not establish that costs per admission alone caused the overall increase. The authors associated the late-period cost shift with more severe complications and comorbidities, which may have required more surgery, supplies, injections, and infusions.

Men accounted for USD 2,352,844, or 56.8%, of total costs, compared with USD 1,789,277, or 43.2%, for women; the Mann–Whitney U test showed a statistically significant difference in treatment costs between the groups. Civilian patients accounted for USD 3,906,872, or 94.3%, while military personnel accounted for USD 235,249, or 5.7%, with a significant difference between the groups. Patients with comorbidities generated USD 3,940,121, or 95.1%, of total costs, compared with USD 202,000, or 4.9%, for those without comorbidities.

Medicines specifically identified for peptic ulcer treatment totaled USD 88,640. Proton pump inhibitors represented 90.7% of that amount, or USD 80,350, with pantoprazole accounting for 64.5% of PPI costs and esomeprazole for 28.26%. Antibiotics represented 7.0%, mucosal protective agents 1.4%, and histamine-2 receptor antagonists 0.9%. In the sensitivity analysis, a hypothetical 10% reduction in medical-supply costs would lower total direct medical costs by 5.0%, compared with 1.8% for medicines and infusions, 1.3% for hospital beds, 0.97% for surgery, 0.6% for laboratory tests, and 0.3% for diagnostic imaging.

The findings have important boundaries. The study covered one final-tier military hospital that also provides insured and public services, rather than a nationally representative sample of Vietnamese patients with peptic ulcer disease. It estimated direct medical costs and part of the direct non-medical burden but did not include travel, accommodation for caregivers, lost income, or other indirect costs. The analysis also recorded only whether comorbidities were present, not their type or severity. Although the study was described as a cost-of-illness assessment rather than a clinical-effectiveness trial, its detailed service-level estimates provide a bounded measure of the direct resources used for inpatient peptic ulcer disease care at that hospital.

Subject of Research: Retrospective estimation of direct inpatient costs for peptic ulcer disease in a Vietnamese military hospital

Article Title: Analysis of direct medical costs for the treatment of peptic ulcer disease in inpatients at a military hospital: a retrospective study in Vietnam

Article References: Ong, T. D., Tran, V.-Q., Nguyen, P. N. M., Nguyen, H. M., Phan, N. M., Vo, B. V., Nguyen, T. D., & Vo, T. Q. (2026). Analysis of direct medical costs for the treatment of peptic ulcer disease in inpatients at a military hospital: a retrospective study in Vietnam. Pharmacia, 73, Article e201763. https://doi.org/10.3897/pharmacia.73.e201763

Image Credits: AI Generated

DOI: 10.3897/pharmacia.73.e201763

Keywords: peptic ulcer disease, Vietnam, cost-of-illness, retrospective study, hospital costs, medical supplies, military hospital, health economics, Analysis, direct, medical, costs

The DRG model is now the dominant logic of hospital payment worldwide. Rather than reimbursing hospitals item by item for every drug, test, and bed-day, the payer assigns each admission to a group of clinically similar cases expected to consume similar resources, and pays a fixed sum per group. The design rewards efficiency: a hospital that treats a case for less than the group payment keeps the difference, while one that runs over absorbs the loss. But the mechanism only works fairly when the grouping is well calibrated. If a group lumps together cases with wildly different true costs, hospitals treating the expensive end of the distribution are systematically underpaid, and the payment standard itself becomes a hidden subsidy or penalty depending on which patients walk through the door.

The Chinese researchers set out to test precisely this property, which they call cost homogeneity, in the pediatric context. Their data came from the National Healthcare Security Administration and covered 289,026 pediatric inpatient encounters from January to June 2025. The team compared medical costs across adjacent diagnosis-related groups, the intermediate categories that sit beneath the final payment groups, and looked specifically at whether children with common diseases and children with idiopathic diseases, conditions whose origins are poorly understood and whose treatment pathways are often less predictable, generated comparable expenses within the same group.

The headline finding is a consistent gradient tied to hospital tier. Three adjacent diagnosis-related groups, labeled ES3, DT1, and EX2, all showed the same pattern: the higher the level of the hospital, the higher the average hospitalization expense for each subgroup. In other words, a child treated at a top-tier tertiary children’s hospital typically cost more to treat than a child with the same grouped diagnosis at a lower-level facility. That might sound unsurprising, since referral systems channel the sickest children upward, but it collides directly with the way China’s DRG payment standards are currently set. Under the prevailing arrangement, the payment for the same DRG group is uniform across hospitals of different tiers, with no adjustment for the variations in treatment costs that different levels of hospital actually incur.

The mismatch matters because the DRG framework in use was largely built around adult medicine. The authors argue that the existing grouping for pediatric cases needs to fully account for the unique characteristics of pediatric diseases rather than merely relying on a unified framework shared with adult patients. Children are not small adults, statistically or clinically. Their conditions evolve rapidly, diagnostic workups differ, length of stay patterns diverge, and rare or poorly understood disorders cluster in specialist children’s hospitals, which sit at the top of the hospital hierarchy and therefore at the expensive end of every tier-based gradient the study detected.

Two conditions crystallized the problem. The average hospitalization costs for Kawasaki disease, an inflammatory illness of young children that affects the blood vessels and is a leading cause of acquired heart disease in children worldwide, were significantly higher than those of the adjacent diagnosis-related group to which such cases were assigned, a difference the researchers report as statistically significant. The same held true for anorectal malformation complicated with rectal malformation, a congenital surgical condition requiring complex operative repair and often prolonged postoperative care. Both are exactly the kinds of cases that concentrate at national children’s centers, and both are exactly the kinds of cases a flat payment rate punishes.

The technical vocabulary of the paper deserves unpacking, because it explains why the findings are more than a bureaucratic complaint. The authors evaluated dispersion using measures such as the coefficient of variation, the ratio of the standard deviation to the mean, which expresses how spread out costs are relative to their average. A well-designed DRG group should show low within-group variation and clear between-group variation: patients in the same group should cost about the same, and patients in different groups should differ. When a group’s coefficient of variation is high, the fixed payment attached to it is essentially a lottery ticket, and the hospitals drawing the costly tickets lose money on cases they cannot refuse. The study’s evidence that idiopathic pediatric conditions and complex surgical anomalies inflate costs within their assigned groups is direct evidence that some pediatric groups fail this homogeneity test.

Why does this matter beyond the accounting department? Payment systems shape clinical behavior. If a hospital consistently loses money on Kawasaki disease admissions because the DRG rate was calibrated on a cost distribution that does not reflect tertiary pediatric practice, the institution faces pressure to limit such admissions, transfer complex children elsewhere, or skimp on the intensive workup these conditions demand. Conversely, if the rate is set too generously for the low-cost end of the distribution, lower-tier hospitals are rewarded in ways that may not correspond to quality. The study’s tier gradient suggests that a single national or regional price for a pediatric group cannot simultaneously be fair to a county hospital and to a national children’s center, because the case mix and cost structures at the two levels differ systematically.

The authors’ prescription follows directly from the diagnosis. They conclude that the implementation of DRG payment should take pediatric clinical characteristics into account, and they call for refined grouping adjustments to be performed for disease groups with high cost heterogeneity to improve the adaptability of the DRG system to pediatric cases. In practical terms, that could mean splitting groups whose internal variation is excessive, creating pediatric-specific categories rather than folding children into adult frameworks, and introducing tier-adjusted payment weights that acknowledge the real cost differences the study documented across hospital levels. The research was supported by the National Healthcare Security Administration, the very agency that administers the payment system, which signals institutional interest in getting the calibration right.

The stakes are considerable. China’s DRG-based payment reform is one of the largest health financing experiments in the world, covering hundreds of millions of insured patients, and pediatric care is a domain where the consequences of mispricing fall on some of the most vulnerable people in the system. A study of this scale, nearly three hundred thousand encounters across two major cities with different economic profiles, offers the kind of empirical grounding that payment reform rarely gets before the rules are locked in. Its message is deceptively simple: a payment system is only as good as its statistics, and for children, the statistics currently say the groups do not hold together. Until the grouping is refined, every Kawasaki disease admission and every complex congenital malformation repair quietly tests how much financial strain the system is willing to place on the hospitals that care for the sickest children.

Subject of Research: Cost homogeneity of pediatric inpatient care under China's DRG hospital payment system

Article Title: Evaluation of cost homogeneity among pediatric inpatients under DRG system: a study of inpatient encounters in Chengdu and Hangzhou, China

Article References: Song, F., Na, C., Lv, Y., Zang, Z., & Zhao, C. (2026). Evaluation of cost homogeneity among pediatric inpatients under DRG system: a study of inpatient encounters in Chengdu and Hangzhou, China. BMC Pediatrics. https://doi.org/10.1186/s12887-026-07693-7

Image Credits: AI Generated

DOI: 10.1186/s12887-026-07693-7

Keywords: DRG payment, pediatrics, hospital costs, cost homogeneity, Kawasaki disease, anorectal malformation, China, health policy, hospital tiers, case-mix grouping, healthcare financing, inpatient care

The study, published in BMC Infectious Diseases, examined cardiovascular surgery patients treated between January 2023 and December 2024. The overall prevalence of nosocomial infection in this population was 12.27 percent, a figure that underscores how common these complications are in a surgical specialty where patients often undergo open-chest procedures, receive mechanical ventilation, and carry multiple comorbidities. Yet the headline number conceals a striking heterogeneity. When the researchers stratified patients by DRG, infection rates ranged from comparatively low figures in some groups to nearly four in ten patients in others.

The highest burden appeared in the coronary artery surgery group designated FC19, where 39.57 percent of patients acquired a nosocomial infection. Heart valve surgery, grouped under FB29, followed with a rate of 34.49 percent, while complicated operations on the great vessels, coded FE19, saw infections in 31.09 percent of cases. These are not marginal differences from the overall average; they represent subgroups in which roughly one in three patients experienced an infection during hospitalization, an order of magnitude that demands targeted rather than generic prevention strategies.

Methodologically, the study tackled a persistent challenge in health economics research: infected and uninfected patients differ systematically in ways that bias simple cost comparisons. Patients who develop infections tend to be sicker at baseline, undergo longer operations, and stay in intensive care longer before the infection even appears. To address this, the investigators employed propensity score matching or inverse probability of treatment weighting within each DRG subgroup, techniques that statistically balance the comparison groups on observed characteristics. This design allowed them to isolate the additional costs attributable to infection itself, rather than to the underlying severity of illness.

The microbiology added another layer of concern. Klebsiella pneumoniae emerged as the most common pathogen in the study population, a finding consistent with the growing global prominence of gram-negative organisms in hospital-acquired infections. The researchers also tracked drug-resistant threats within their framework, referencing carbapenem-resistant Acinetobacter baumannii, methicillin-resistant Staphylococcus aureus, and carbapenem-resistant Pseudomonas aeruginosa as categories of concern. The dominance of Klebsiella is particularly notable given the rising prevalence of extended-spectrum beta-lactamase-producing and carbapenem-resistant strains of this organism worldwide, which complicates empirical antibiotic choices and raises both treatment costs and clinical risk.

The dominant infection type was equally striking. Non-ventilator-associated lower respiratory tract infections accounted for 90.57 percent of nosocomial infections in the cohort. That such a large share of infections involved the lower airways without being attributable to ventilator use points to a distinct pathophysiological and operational problem. Cardiac surgery patients arrive in the postoperative period with anesthesia-related respiratory depression, pain-limited coughing, immobility, and often preexisting pulmonary compromise, all of which create fertile conditions for pneumonia even after ventilators are removed. Prevention efforts aimed solely at ventilator bundles may therefore miss the bulk of the respiratory infection burden in this population.

The economic analysis quantified what those infections cost. Among patients in the heart valve surgery group FB29, a non-ventilator-associated lower respiratory tract infection significantly increased total hospitalization costs, with a cost ratio of 1.17, meaning infected patients incurred costs roughly 17 percent higher than statistically comparable uninfected peers, at a confidence interval of 1.01 to 1.36 and a P value of 0.037. More consistent effects appeared in the medicine cost category. Infections raised medication expenses by 50 percent in the FB29 valve surgery group, 27 percent in the FE19 great vessel group, and 30 percent in the FC19 coronary artery group, with all confidence intervals excluding unity and P values below 0.05.

One of the study’s most important conclusions is that the primary drivers of infection-attributable expenses differed across DRG groups. In some groups the excess costs flowed through pharmaceutical spending, reflecting prolonged or escalated antibiotic therapy and supportive medications. In others, longer intensive care stays, additional diagnostic testing, or extended ward admissions played larger roles in inflating total hospitalization costs. This granularity matters because it tells hospital administrators and infection-control teams not just that infections are expensive, but where in the cost structure those expenses accumulate for each category of cardiac operation.

The DRG lens is central to the study’s relevance for health policy. Under DRG-based payment systems, hospitals receive a fixed, prospectively determined payment per case grouped by diagnosis and procedure complexity. When a patient acquires an infection and costs rise, much of the additional expense is absorbed by the hospital rather than reimbursed, converting preventable infections from a payer problem into a direct institutional financial liability. The study’s findings suggest that hospitals performing high volumes of valve surgery, coronary artery procedures, or great vessel operations face particularly concentrated exposure, and that infection prevention in these DRGs functions as economic self-protection as much as clinical stewardship.

The authors conclude that further research is urgently needed to identify clinically recognizable, infection-related risk factors specific to each DRG, work that would enable precise infection management and better-informed resource allocation. In the interim, the study offers a template: rather than treating all cardiac surgical patients as a single population, hospitals can use DRG stratification to identify which procedure groups carry the highest infection risk, which pathogens dominate, and which cost categories bear the brunt of each preventable infection. As diagnostic-related group payment systems spread globally, that kind of subgroup-specific evidence is likely to become the standard against which infection-control investment is judged.

Subject of Research: Economic burden of nosocomial infections in cardiovascular surgery patients analyzed by DRG payment groups

Article Title: Direct economic burden of nosocomial infection in cardiovascular surgery patients based on DRG payment management

Article References: Wang, Y., Song, X., Jin, M., & Dong, Q. (2026). Direct economic burden of nosocomial infection in cardiovascular surgery patients based on DRG payment management. BMC Infectious Diseases. https://doi.org/10.1186/s12879-026-14448-3

Image Credits: AI Generated

DOI: 10.1186/s12879-026-14448-3

Keywords: nosocomial infection, cardiovascular surgery, diagnosis-related groups, lower respiratory tract infection, hospital costs, Klebsiella pneumoniae, health economics, cardiac surgery, infection control, DRG payment, hospital-acquired pneumonia, cost ratio

The research team systematically searched three databases for comparative studies conducted in OECD countries, screening records in duplicate and extracting data with a standardised template. Methodological quality was assessed using relevant National Institutes of Health evaluation tools. The window of evidence stretched from 2013 to 2025, capturing more than a decade of experience with these units as they proliferated under various names and organisational models. In the end, twenty-eight studies met the inclusion criteria, forming the evidentiary backbone for an assessment of mortality, length of stay, hospital readmissions, ICU readmissions, costs, and adverse events in adult patient populations.

The first striking finding is how inconsistent the evidence is. Mortality patterns did not point in a single direction: some studies suggested potential benefits of intermediate care, others reported no clear differences, and still others highlighted potential risks for certain patients. According to the review, these discrepancies were often linked to differences in case-mix and admission criteria, meaning that who gets admitted to an intermediate care unit may matter as much as the unit itself. This is a familiar problem in health services research, where the intervention is not a drug with a fixed dose but a flexible organisational arrangement that varies from hospital to hospital and country to country.

Length of stay outcomes were similarly inconsistent. Some studies associated intermediate care units with shorter hospital stays, plausibly because enhanced monitoring allows earlier de-escalation from intensive care or faster recognition of deterioration on general wards. Others reported minimal or no improvement at all. The heterogeneity here mirrors the mortality findings and underscores a central methodological challenge: without standardised definitions of what constitutes an intermediate care unit, comparing results across studies becomes an exercise in approximation rather than precision. Units described under the same label may differ in nurse-to-patient ratios, physician coverage, available organ support, and the acuity of patients they accept.

Readmission outcomes offered somewhat more consistency, though not the kind advocates of these units might have hoped for. Rates of hospital readmission generally did not differ substantially between patients managed in intermediate care units and those treated in alternative levels of care. Advantages in reducing ICU readmissions, one of the theoretical selling points of step-down and step-up intermediate care, were uncommon across the reviewed literature. In other words, the evidence does not currently support the idea that routing patients through intermediate care reliably prevents them from bouncing back to intensive care or returning to hospital shortly after discharge.

Where the evidence does converge is on money. Cost savings were more consistently observed, with many studies indicating lower hospital expenditure when patients were managed in intermediate care units rather than in intensive care. The economic logic is straightforward: intensive care beds are among the most expensive resources in any hospital, dominated by high staffing ratios, continuous monitoring, and expensive technology. If clinically appropriate patients can be safely managed one step down, the marginal cost per patient falls. Yet the review adds an important caveat, noting that these savings were occasionally accompanied by unfavourable clinical trade-offs in specific patient groups, a reminder that financial efficiency and clinical safety do not automatically travel together.

Adverse event rates, by contrast, were generally comparable across settings. This is arguably reassuring: it suggests that, on balance, moving appropriately selected patients into intermediate care does not systematically expose them to more harm than keeping them in intensive care or on general wards. But comparability is a modest benchmark. The review’s overall verdict is that effects on mortality and length of stay remain inconclusive, largely because of methodological and organisational heterogeneity across the evidence base, while the case for cost savings is the most consistent signal in the literature.

The pattern that emerges is one of selective benefit. Intermediate care units appear most useful for moderate-risk populations, patients too unstable for a general ward but not sick enough to require full intensive care. For these patients, enhanced monitoring at lower staffing intensity represents a rational middle ground. For high-risk patients, however, the suitability of intermediate care is uncertain, and the review suggests that admitting the wrong patients into these units could carry risks. Patient selection, in other words, is not a bureaucratic detail but the decisive variable that determines whether an intermediate care unit helps, does nothing, or harms.

For hospital planners and policymakers, the implications are significant. Many health systems face simultaneous pressures of ageing populations, rising critical care demand, and constrained budgets. Intermediate care units have been promoted as a way to relieve ICU congestion, improve patient flow, and reduce costs simultaneously. The new review partially vindicates that promise on the economic front while cautioning against overclaiming clinical benefits that the current evidence cannot support. It also highlights the danger of treating these units as a one-size-fits-all solution, since their effects appear to depend heavily on local case-mix, admission protocols, and the specific clinical purpose each unit is designed to serve.

The authors argue that standardised definitions, clearer admission criteria, and improved reporting are essential to strengthen future evidence. Until the field agrees on what an intermediate care unit actually is, comparative research will continue to produce the kind of mixed results this review documents. In the meantime, the practical takeaway for clinicians and administrators is a measured one: intermediate care units are a promising and generally cost-saving component of hospital architecture, best deployed for selected moderate-risk patients, and their expansion should be guided by explicit criteria rather than enthusiasm. As these units continue to spread across OECD health systems, the review provides both a warning against uncritical adoption and a roadmap for the research needed to settle the remaining questions.

Subject of Research: Effects of intermediate care units on patient outcomes and hospital resource use in OECD countries

Article Title: The effect of intermediate care units on patient and hospital utilisation outcomes: a systematic literature review

Article References: Detollenaere, J., Van de Voorde, C., Castanares-Zapatero, D., Bouckaert, N., Lefèvre, M., & Van den Heede, K. (2026). The effect of intermediate care units on patient and hospital utilisation outcomes: a systematic literature review. BMC Health Services Research. https://doi.org/10.1186/s12913-026-15737-y

Image Credits: AI Generated

DOI: 10.1186/s12913-026-15737-y

Keywords: intermediate care units, intensive care, systematic review, patient outcomes, hospital costs, length of stay, readmissions, mortality, health services research, patient selection, OECD countries, critical care

The research team, led by Minting Huang of Aerospace Center Hospital in Beijing and Lau Nga Wun of Macao Polytechnic University, assembled a provincial panel covering all 31 of China’s province-level jurisdictions across the full twenty-year window. Rather than relying on a single analytical lens, they deployed a multi-method framework in which each tool answered a distinct question. A data envelopment analysis combined with the Malmquist index, known as DEA-Malmquist, measured how efficiently provinces converted healthcare inputs into services over time and whether changes came from technological improvement or from catching up to best practice. The Dagum Gini coefficient decomposed inequality in allocation outcomes to reveal where spatial disparities actually originated. Finally, a GM(1,1) grey prediction model, a technique designed for forecasting with limited data, projected future trajectories of supply and demand. Treating efficiency and equity as related but independent dimensions, the design allowed the authors to track each separately across a period that included both deepening healthcare system reforms and the shocks of public health emergencies.

The efficiency findings carry the sharpest edge. Total factor productivity in healthcare allocation is a ratio-style measure: when it exceeds 1, the system is producing more health services per unit of beds, personnel and expenditure than it did before; when it falls below 1, growth in inputs is outpacing growth in outputs. The study found the index fluctuating downward across the two decades, and critically, it identified the technological progress component, abbreviated Techch, as the weak link, with a value of 0.929 indicating an average annual shrinkage of the technology frontier of more than 7 percent in the periods where it declined. In efficiency analysis, this frontier represents the best achievable performance given current knowledge and practice. When the frontier itself contracts rather than expands, no amount of managerial effort by individual provinces can fully compensate, because the ceiling of what is possible is dropping.

Geography mattered enormously. The analysis revealed a regional heterogeneity best summarized as high in the East, low in the West, with the technological change gap between the eastern provinces and the rest of the country statistically distinguishable under province-level bootstrap testing, a resampling technique that guards against the possibility that observed differences are artifacts of small sample sizes. Perhaps more telling was where decreasing returns to scale concentrated: in the richer regions. In production theory, decreasing returns to scale means that adding inputs yields progressively smaller output gains, a signal that simply building more capacity no longer pays off. The fact that this pattern clustered in China’s wealthiest, most resource-dense provinces suggests that scale expansion there has reached a point of diminishing returns, and that further growth in those areas requires new technology and better organization rather than more beds.

The equity analysis told a more nuanced story. Using the Dagum Gini coefficient, a decomposition method that splits overall inequality into within-region, between-region, and trans-variation density components, the researchers examined inequality in the provincial productivity change measure itself. Trans-variation density, which captures the degree to which regional distributions overlap rather than sit cleanly apart, has long dominated the decomposition, contributing 52.96 percent of overall inequality in 2022. The authors caution that this overlap reflects one of several possible sources of disparity, but they identify the gradient gap between the Eastern and Western provinces as the root cause of overall inequality. In other words, the deepest fault line in Chinese healthcare productivity runs between the wealthy coastal belt and the interior, even as the distributions of individual provinces blur across regional boundaries.

Interestingly, the study’s conclusions note a divergence between two kinds of equity. Inequality in physical resources per head, such as beds and staff relative to population, declined over the study period, indicating genuine progress in distributing the hardware of healthcare more evenly across the country. Yet inequality in productivity change did not decline, meaning that even as provinces converged in what they possess, they diverged in how effectively they converted those resources into services. This distinction matters for policy because it implies that the remaining inequity is not primarily about where the buildings and machines are, but about the human capital, management practices, and technological adoption that determine what those assets actually deliver to patients.

The forecasting component adds a financial warning sign. The GM(1,1) grey prediction model projects that healthcare resource supply will continue expanding, but that the average hospitalization cost per visit would reach approximately 15,587 CNY in nominal terms by 2029, based on an illustrative extrapolation of the 2003 to 2022 trend. The authors frame this as pointing to a risk of rigid growth in government health expenditure, a scenario in which costs climb steadily regardless of efficiency gains, squeezing public budgets. For a system serving 1.4 billion people and aging rapidly, the gap between expanding supply and rising unit costs is precisely the combination that strains fiscal sustainability.

The authors are careful about what their study can and cannot claim. They describe the work as a descriptive assessment and explicitly state that it does not identify causal effects. The Malmquist framework measures changes in relative performance against a shifting frontier, not the impact of any specific reform or event. Still, the patterns they document point toward identifiable directions for policy. The study indicates frontier-expanding investment, meaning spending aimed at pushing out the technological frontier rather than merely adding capacity, alongside scale consolidation in provinces operating under decreasing returns to scale. It also points to cross-regional coordination to address the East-West gradient and to value-based payment reform, which ties reimbursement to outcomes rather than volume, as directions suggested by the observed patterns.

Beyond China, the findings speak to a dilemma facing many developing countries. The conventional development playbook is to build capacity fast: more hospitals, more beds, more staff. China executed that playbook at extraordinary scale over two decades, and by the measure of physical resources per person, it narrowed gaps between regions. But the study suggests that the playbook has a shelf life. Once resource stocks are broadly in place, further progress depends on the harder, less visible work of technological diffusion, institutional design and payment incentives, and those dimensions showed no such convergence. A country can equalize what it owns while still unequal in what it achieves.

The timing of the productivity slowdown, with the index below 1 from 2018 through 2022, spans both the later phase of China’s healthcare reforms and the COVID-19 pandemic years, though the descriptive design cannot attribute the decline to either. What the data do establish is a clear empirical baseline: a system whose input growth has outrun its output growth, whose technological frontier has been contracting, whose deepest inequality follows an East-West gradient rooted in productivity change rather than resource counts, and whose costs are projected to keep rising into the next decade. For policymakers in Beijing and for health economists watching from abroad, the study reframes the central question of Chinese healthcare from how much the system has, to how well it uses what it has. The answer, on current evidence, is not yet well enough, and the trajectory suggests that the next two decades will demand a fundamentally different kind of investment than the last two.

Subject of Research: Efficiency and equity of healthcare resource allocation across Chinese provinces from 2003 to 2022

Article Title: Efficiency, equity, and future trajectories of healthcare resource allocation in China: a multi-method provincial panel analysis, 2003–2022

Article References: Huang, M., Wun, L. N., Huang, S., Wu, J., Deng, T., Gong, H., Wang, Q., & Wu, Z. (2026). Efficiency, equity, and future trajectories of healthcare resource allocation in China: a multi-method provincial panel analysis, 2003–2022. BMC Health Services Research. https://doi.org/10.1186/s12913-026-15751-0

Image Credits: AI Generated

DOI: 10.1186/s12913-026-15751-0

Keywords: healthcare resource allocation, China, total factor productivity, DEA-Malmquist index, Dagum Gini coefficient, grey prediction model, regional disparity, health equity, health economics, health policy, provincial panel data, hospital costs