Patient-Based Costing, DRGs, EHR Design, Process Mining, and Clinical Terminologies in Acute Hospitals

Abstract

In an acute hospital, Diagnosis-Related Groups (DRGs) and patient-based costing solve different, complementary problems. DRGs are a prospective payment and case-mix classification mechanism: they group inpatient episodes for reimbursement and resource allocation using coded diagnoses, procedures, severity, and other episode variables. Patient-based costing - especially Activity-Based Costing (ABC) and Time-Driven Activity-Based Costing (TDABC) - is an internal measurement method: it estimates the actual resources consumed by a given patient across the care cycle. The practical implication is that DRGs tell the hospital what it is likely to be paid, while patient-based costing tells the hospital what the case actually cost. An effective acute-hospital costing architecture therefore requires structured EHR data, reliable timestamps and event logs, linkage to HR/finance/supply systems, and a semantic layer that uses SNOMED CT and LOINC for point-of-care capture and WHO ICD-11 and WHO ICHI for aggregation, reporting, and case-mix analysis. Process mining is pivotal because it can reconstruct the “as-is” pathway from EHR and ancillary-system logs, expose pathway variation and bottlenecks, and help operationalize ABC/TDABC at scale - provided that mined pathways are still clinically validated.

1. Introduction

Modern hospital payment and management increasingly separate the question of how care is reimbursed from the question of what care really costs to produce. DRG systems were built to pay hospitals prospectively on a per-case basis, whereas ABC and TDABC were developed to measure resource consumption at the activity and patient levels. TDABC is generally treated as a simplified evolution of traditional ABC: instead of allocating costs through many activity cost drivers, it estimates the cost of care through process maps, time estimates, and capacity cost rates. In healthcare, TDABC has been widely promoted as a cost-accounting method compatible with value-based care because it produces patient-level cost information across full care cycles rather than broad departmental averages.

2. How patient-based costing relates to DRGs

DRG systems are patient classification systems (PCS) that use routinely collected discharge data to group inpatient episodes into clinically coherent and economically homogeneous classes, primarily for case-mix measurement and often for prospective reimbursement; by contrast, patient-based costing estimates the actual resources consumed by individual patients within and across those groups.

Under the U.S. Inpatient Prospective Payment System, CMS (Centers for Medicare & Medicaid Services) pays acute inpatient stays on a rate-per-discharge basis. The payment for a case is computed by multiplying the hospital’s base rate by the DRG weight, and each DRG weight is intended to reflect the average resources required for that group relative to all DRGs. CMS also states that grouping depends primarily on principal diagnosis, additional diagnoses, procedures, and, in some DRGs, age, sex, and discharge status.

WHO’s ICD-11 reference guide makes the same conceptual point in broader international terms: in casemix grouping systems such as DRGs, ICD-based data are used for reimbursement and resource allocation; assignment is based on algorithms that use coded diagnoses, coded procedures, and other variables. WHO ICD-11 was explicitly designed to accommodate the detail needed for diagnosis-related casemix groupings, although the exact grouper logic remains country-specific and may also be constrained by national legislation and capabilities.

Patient-based costing relates to DRGs, therefore, as an internal cost engine relates to an external payment classifier. DRGs estimate what a hospital should be paid for an average case in a category; patient-based costing measures what a particular patient actually consumed. The two only partially overlap. DRGs are built around cost homogeneity at the group level, but even well-designed DRGs can mask substantial within-group variation, especially for complex patients, unforeseen complications, and atypical pathways. That is why patient-level costing is indispensable for operational management: it reveals whether a specific DRG is profitable or loss-making locally, which phase of care drives the variance, and whether the problem lies in coding, pathway design, staffing mix, supply use, or avoidable length of stay (LOS).

The practical relationship can be stated simply. DRG reimbursement = expected revenue for the episode. Patient-based costing = observed cost of producing the episode. Margin = DRG payment minus patient-level cost. Once that comparison is available by ward, surgeon, pathway variant, and complication status, the hospital can move from generic cost control to targeted redesign. In this sense, patient-based costing does not replace DRGs; it makes DRG management clinically and managerially actionable.

Note: The essence of a patient classification system (PCS) is to reduce the complexity of individual patients and care episodes into a limited number of standardized groups that are clinically interpretable and resource-relevant. A patient classification system is a rule-based method that uses patient and episode data to assign cases to standardized groups that are sufficiently similar for purposes such as case-mix measurement, reimbursement, planning, benchmarking, and cost analysis.

Note: NordDRG (Nordic Diagnosis Related Groups) is a common, annual, and internationally standardized patient classification system used for healthcare production, cost benchmarking, and payment in Denmark, Finland, Iceland, Norway, Sweden, and the Baltic states.Based on WHO ICD-10, NOMESCO Classification of Surgical Procedures (NCSP), and the Health Care Financing Administration DRG (HCFA-DRG) system, it classifies inpatient/day surgery stays into roughly 20+ definition tables to ensure consistent, transparent reimbursement and clinical analysis. The HCFA-DRG system - now managed by the U.S. Centers for Medicare & Medicaid Services (CMS) as MS-DRGs - is a patient classification system introduced in 1983 for Medicare to pay hospitals a fixed, prospective fee for inpatient stays based on diagnosis, procedures, age, and comorbidities.

3. What data and procedures an EHR must support

A hospital EHR must enable structured,  ergonomically and user-friendly designed clinical documentation workflows for diagnoses, interventions, observations, and discharge abstraction, thereby improving completeness, semantic consistency, and downstream use for DRG assignment and patient-based costing.

The EHR must support more than clinical documentation. To implement TDABC, the hospital needs a data environment that can answer five operational questions for every step in care: what happened, who did it, when it happened, where it happened, and how often it happened. The published TDABC literature and implementation guides consistently treat these elements as foundational because TDABC depends on pathway mapping, time estimates, resource identification, capacity measurement, and total patient-cost calculation.

A hospital EHR and its connected systems should therefore capture, at minimum, the following episode-level data.

1. First, the hospital needs episode linkage: patient identifier, encounter identifier, admission and discharge times, unit transfers (ADT), bed movements, and discharge destination.
2. Second, it needs coded clinical content: main condition, secondary diagnoses, comorbidities, complications, procedures, and present-on-admission (POA) status where supported.
3. Third, it needs activity timestamps: triage, consult request and completion, imaging order and result, laboratory order/collection/result, medication administration, transport, operating-room in/out, anesthesia start/stop, PACU entry/exit, discharge order, and actual discharge.
4. Fourth, it needs resource attribution: staff role, performer ID, specialty, location, room, equipment, and device.
5. Fifth, it needs direct-consumption data: implants, consumables, blood products, pharmacy items, and quantities used.
6. Sixth, it needs financial master data outside the transactional EHR but linked to it: salaries, benefits, overhead rules, equipment depreciation, maintenance, and practical capacity denominators.
7. Finally, it needs outcome data to pair cost with value: length of stay, complications, readmissions, mortality, and ideally patient-reported outcomes (PROM) where relevant.

Audit logs are especially important because they capture time-sequenced clinician interactions with the EHR and can be aggregated into task and workflow measures. The literature also emphasizes that audit logs are already generated by EHRs for compliance and security purposes and may range from coarse event records to granular clickstream data. For costing, that matters because audit logs can supply objective timestamped evidence for pathway reconstruction, team participation, ordering behavior, and documentation burden.

In procedural terms, an acute hospital should implement six concrete EHR-enablement steps. It should:

1. standardize structured documentation for diagnoses, interventions, observations, and discharge abstraction;
2. make key operational timestamps mandatory or system-generated rather than optional free text;
3. capture performer, location, and order/result provenance for each event;
4. integrate EHR data with LIS, RIS/PACS, pharmacy, OR systems, bed management, ERP, HR/payroll, and materials management;
5. maintain a cost dictionary that converts resources into capacity cost rates; and
6. validate all extracted pathways with clinicians and finance staff before using them for managerial decisions. 

That last point is crucial because the systematic review on automated process-map discovery found that many TDABC studies still relied on observation, interviews, and multidisciplinary validation, and many published process maps were not validated by specialists.

4. The role of process mining in ABC and TDABC

Process mining analyzes event data generated by operational systems to reconstruct how processes actually ran, rather than how they were supposed to run. In healthcare, the literature describes its use for diagnostic, treatment, and organizational processes, and frames it as a way to extract knowledge from routinely logged data.

For costing, process mining matters because a usable event log needs, at minimum, a case identifier, an activity name, and a timestamp; richer logs may also contain resource, location, and cost attributes. That is almost exactly the information TDABC needs to build or refine process maps and time equations.

Its role in ABC/TDABC is fourfold:

1. First, it can automate pathway discovery, reducing the burden of manual whiteboard mapping.
2. Second, it can measure variation: loops, rework, waiting time, parallel work, bottlenecks, and non-standard pathways.
3. Third, it can stratify cost drivers by revealing which subpathways consume more time, staff, or resources even within the same diagnosis or DRG.
4. Fourth, it can turn TDABC from a one-off study into a continuous monitoring system. 

Recent work on “cost mining” explicitly extends this logic by using patient-level pathway data to identify high-cost subgroups and the timing of cost-generating decisions across pathways.

At the same time, process mining is not a complete substitute for direct observation or clinical validation. The TDABC process-map review found that direct observation, surveys, and interviews remained common, that only a minority of studies used hospital documents or healthcare records to integrate process maps, and that many maps were not validated by specialists. Process mining should therefore be understood as an accelerant and validation tool for TDABC - not as an excuse to ignore the clinical reality of work done outside digital systems.

5. The role of SNOMED CT, LOINC, WHO ICD-11, and WHO ICHI

SNOMED CT is the most useful standard for granular point-of-care clinical meaning. SNOMED International describes it as the most comprehensive multilingual clinical terminology and emphasizes that it enables consistent representation of clinical content in EHRs. In a costing architecture, that makes SNOMED CT the preferred layer for recording problems, findings, procedures, and clinical context as clinicians actually document them. Its further value is that SNOMED CT is designed to interoperate with classifications and code systems through mapping.

LOINC is the standard for observations, measurements, and clinical documents. Because TDABC and process mining depend heavily on reliable event definitions and trigger points, LOINC-coded laboratory orders, results, vital signs, assessments, and structured forms improve comparability and extraction quality. In practical terms, LOINC helps define when a laboratory-related activity occurred, what kind of observation it was, and how observation-intensive pathways differ in cost.

WHO ICD-11 is the standard for diagnostic classification, aggregation, and case-mix use. The World Health Organization (WHO) states that ICD-coded terms are a main basis for health recording and statistics in secondary and tertiary care, and its reference guide explicitly states that ICD data are used in DRG/casemix reimbursement and resource allocation. For patient-based costing, WHO ICD-11 is therefore not primarily the point-of-care vocabulary; it is the layer used to aggregate episodes, define cohorts, support morbidity analysis, and connect clinical episodes to case-mix and financing logic. WHO ICD-11’s coding rules for main condition, code clustering, and distinguishing diagnoses arising during stay from those present at admission are also directly relevant to costing validity because they affect severity adjustment and episode comparability.

WHO ICHI is the standard for intervention classification across diagnostic, medical, surgical, mental health, rehabilitation, allied health, and public health interventions. WHO explicitly notes that WHO ICHI can support redevelopment of national intervention classifications for casemix funding systems and that, together with WHO ICD-11, it provides a base for financing health services, including casemix financing. For hospital costing, WHO ICHI’s most important role is to provide a standardized intervention vocabulary when local procedure lists are fragmented, overly billing-oriented, or not clinically expressive enough. It is particularly valuable for cross-site benchmarking and for service lines in which interventions extend beyond classic surgical coding.

The most robust acute-hospital design is therefore layered. SNOMED CT and LOINC should dominate transactional clinical capture inside the EHR; WHO ICD-11 and WHO ICHI should dominate downstream aggregation, reporting, and case-mix analysis; and a mapping layer should connect the two. SNOMED International’s collaboration with WHO has explicitly pursued such alignment, including work on linkage with WHO ICD-11 Mortality and Morbidity Statistics (MMS).

Note: Transactional clinical capture refers to the structured, event-level recording of clinical activities, observations, diagnoses, and interventions within routine operational systems at the point of care, so that the resulting data can support care delivery, coding, analytics, process mining, and patient-level costing (time-stamped, operational, and granular).

Note: Transactional capture refers to the event-level recording of care activities in operational systems; clinical documentation is the clinician-authored record of the patient’s condition, reasoning, and treatment; abstraction is the selective extraction of relevant data elements from the record for reporting or analysis; and coding is the translation of documented and abstracted information into standardized terminologies (e.g. SNOMED CT) or classifications (e.g.WHO ICD, WHO ICHI) for reimbursement, statistics, and interoperability (Transactional capture records events, documentation explains care, abstraction selects facts, and coding standardizes them).

6. Practical example: implementing patient-based costing for emergency appendectomy in an acute hospital

A practical acute-hospital starting point is emergency appendectomy, because published TDABC studies have already shown that appendectomy pathways can be modeled with medical-record timestamps and then redesigned operationally. In pediatric appendicitis/appendectomy, published studies used timestamp-based process maps and later showed that targeted redesign - including standing orders, advanced-practice-provider involvement, and same-day discharge - changed pathway duration and cost structure; the post-intervention model included six phases of care, 33 processes, and 19 personnel types, and same-day discharge removed postoperative floor cost even though PACU (Post-Anaesthesia Care Unit) time increased. 

The example below is therefore an illustrative implementation blueprint, not a claim about one hospital’s unpublished results. It adapts the published appendectomy work, the TDABC seven-step framework, and consensus guidance into a realistic acute-hospital project.

For this service line, the hospital should extract: ED arrival and triage times; surgical consult time; imaging order/result times; laboratory order/collection/result times; antibiotic administration time; decision-to-operate time; OR room entry/exit; anesthesia start/stop; incision/closure if available; PACU entry/exit; ward transfer; discharge order and actual discharge; readmission and surgical-site infection events; all staff roles involved; consumables and instrument packs; pharmacy use; bed-day consumption; and financial inputs for each relevant resource pool.

That dataset is enough to produce patient-level cost by phase of care and to compare actual local cost with the relevant national DRG payment.

Phase	Indicative timing	Main work	Deliverables
1. Charter and scope	Month 1	Choose cohort, define episode boundaries, appoint clinical-finance-IT team	Project charter; governance map; target cohort definition
2. Pathway discovery	Months 2–3	Combine clinician workshops, chart review, and log extraction	Validated current-state process map; event catalog
3. Semantic and data model	Months 3–4	Normalize diagnoses, observations, interventions, and locations	SNOMED CT/LOINC/ICD-11/ICHI mapping sheet; event-log schema; data dictionary
4. Resource costing	Months 4–5	Build resource pools and practical capacity denominators	Capacity cost rate library; cost-center/resource crosswalk
5. Pilot TDABC engine	Months 5–7	Calculate patient-level cost for a pilot sample	Cost per case; cost by phase; cost-driver analysis; DRG margin view
6. Validation	Months 7–8	Compare algorithmic times with clinician review and sampled observation	Validation report; reconciliation with finance totals; exception log
7. Redesign	Months 8–10	Standardize order sets, reduce delays, redesign discharge pathway	Future-state pathway; intervention package; KPI dashboard
8. Scale-up	Months 10–12	Move from pilot to routine monthly reporting	Service-line dashboard; SOPs; training pack; monthly governance review

This timetable is an implementation inference, but it is consistent with published TDABC project guidance emphasizing condition selection, project planning, multidisciplinary teams, and reproducible reporting.

The decisive outputs are not merely “a cost number.” The hospital should expect, at minimum, five managerial products:

1. patient-level cost per episode;
2. cost by phase of care - ED, imaging, OR, PACU, ward, discharge;
3. cost drivers - for example OR time, postoperative floor stay, imaging, or staffing mix;
4. DRG margin analysis for the pathway; and
5. redesign KPIs such as median time-to-incision, postoperative bed-hours, same-day discharge rate, complications, readmissions, and infection rate.

In the published appendicitis work, the point of TDABC was precisely to identify redesign opportunities rather than merely to produce retrospective accounting.

Note: Precalculation and postcalculation are essential, sequential cost-management processes in (healthcare) production and (healthcare) process/project management used to compare anticipated costs against actual outcomes. Precalculation estimates costs beforehand for budgeting, while postcalculation analyzes real costs afterward to calculate profitability, identify deviations, and optimize future efficiency.

Note: Potentially Preventable Complications (PPCs) are harmful, in-hospital events occurring after admission - such as infections, injuries, or surgical errors - that result from care processes rather than natural disease progression. 

Note: Potentially Preventable Readmissions (PPRs) are unplanned return hospitalizations closely related to a prior admission, often caused by suboptimal care, premature discharge, or poor care transitions. Affecting nearly 1 in 5 surgical patients, these events are frequently linked to surgical complications, infections, or medication mismanagement, often targeted for reduction through improved quality measures.

Note: "Never events" events are considered "never" events because, with proper safety systems and adherence to evidence-based guidelines, they should not occur. Key "Never Events":

• Surgery on the wrong body part, patient, or wrong procedure: These are the most frequently reported never events, with wrong-site surgery accounting for 42% of cases.
• Unintentionally retained foreign objects: Surgical items left inside a patient after a procedure.
• Severe pressure ulcers: Stage 3 or 4 pressure ulcers acquired after admission.
• In-hospital complications: Such as pulmonary embolism, deep vein thrombosis (DVT) following orthopedic surgery, or catheter-associated urinary tract infections (CAUTIs).
• Medication errors: Such as administering the wrong drug, wrong dose, or wrong route.
• Mismatched blood transfusion: Transfusing the wrong blood type (ABO, Rh, ..).
• Environmental errors: Patient death or serious disability associated with a fall while in the hospital, or the use of restraints.

7. Conclusion

In an acute hospital, DRGs and patient-based costing should be treated as complementary layers of the same management system. DRGs provide the case-mix and reimbursement logic; ABC/TDABC provide the patient-level production logic. An EHR that is fit for patient-based costing must therefore support structured semantic coding, event timestamps, resource attribution, and linkage to finance and supply systems. Process mining is the operational bridge that turns raw logs into clinically interpretable pathways and ongoing cost intelligence. SNOMED CT and LOINC make point-of-care data computable; WHO ICD-11 and WHO ICHI make those data aggregable for reimbursement, comparison, and policy use. Hospitals that combine these elements can move from average-cost accounting to precise, pathway-level management of cost, efficiency, and value, while keeping an eye on the Quintuple Aim.

Note: The Quintuple Aim is a healthcare framework designed to improve system performance by simultaneously pursuing five key goals: improving the patient experience, enhancing population health, reducing costs, improving clinician well-being, and advancing health equity. It expands on the Triple Aim and Quadruple Aim, adding health equity as a critical, core component to address systemic disparities.

Bibliography

Centers for Medicare & Medicaid Services. MS-DRG Classifications and Software. CMS.

CKS DRG. DRG systems around the world and the ways of financing. CKS DRG; 2021.

EU. EuroDRG – Diagnosis-Related Groups in Europe: towards Efficiency and Quality. EU; 2024.

Organisation for Economic Co-operation and Development. Better Ways to Pay for Health Care. OECD; 2016.

World Health Organization. International Classification of Diseases (ICD). WHO.

World Health Organization. ICD-11 Reference Guide. WHO; 2026 draft/reference release.

World Health Organization. International Classification of Health Interventions (ICHI). WHO.

World Health Organization. ICHI Reference Guide. Geneva: WHO; 2023.

SNOMED International. What is SNOMED CT?

SNOMED International. SNOMED CT Maps.

SNOMED International. World Health Organization partnership page.

Regenstrief Institute. What LOINC is.

UHasselt - Business Process Analysis Made Easy (bupaR)

Casas, M. (1991). Issues for comparability of DRG statistics in Europe. Results from EURODRG. Health Policy, 17(2), 121-132.

De Weerdt, J., Schupp, A., Vanderloock, A., & Baesens, B. (2013). Process Mining for the multi-faceted analysis of business processes—A case study in a financial services organization. Computers in Industry, 64(1), 57-67.

Etges, A. P. B. D. S., Polanczyk, C. A., & Urman, R. D. (2020). A standardized framework to evaluate the quality of studies using TDABC in healthcare: the TDABC in Healthcare Consortium Consensus Statement. BMC health services research, 20(1), 1107.

France, F. H. R. (2003). Case mix use in 25 countries: a migration success but international comparisons failure. International journal of medical informatics, 70(2-3), 215-219.

Iachecen, F., Dallagassa, M. R., Portela Santos, E. A., Carvalho, D. R., & Ioshii, S. O. (2023). Is it possible to automate the discovery of process maps for the time-driven activity-based costing method? A systematic review. BMC health services research, 23(1), 1408.

Itchhaporia, D. (2021). The evolution of the quintuple aim: health equity, health outcomes, and the economy. Journal of the American College of Cardiology, 78(22), 2262-2264.

Janssenswillen, G., & Depaire, B. (2017). BupaR: business process analysis in R.

Janssenswillen, G., Depaire, B., Swennen, M., Jans, M., & Vanhoof, K. (2019). bupaR: Enabling reproducible business process analysis. Knowledge-Based Systems, 163, 927-930.

Kannampallil, T., & Adler-Milstein, J. (2023). Using electronic health record audit log data for research: insights from early efforts. Journal of the American Medical Informatics Association, 30(1), 167-171.

Keel, G., Savage, C., Rafiq, M., & Mazzocato, P. (2017). Time-driven activity-based costing in health care: a systematic review of the literature. Health policy, 121(7), 755-763.

Kobel, C., Thuilliez, J., Bellanger, M., & Pfeiffer, K. P. (2011). DRG systems and similar patient classification systems in Europe. Diagnosis-related groups in Europe, 37-58.

Leusder, M., Relijveld, S., Demirtas, D., Emery, J., Tew, M., Gibbs, P., ... & IJzerman, M. (2024). Toward value-based care using cost mining: cost aggregation and visualization across the entire colorectal cancer patient pathway. BMC medical research methodology, 24(1), 321.

Martin, J. A., Mayhew, C. R., Morris, A. J., Bader, A. M., Tsai, M. H., & Urman, R. D. (2018). Using time-driven activity-based costing as a key component of the value platform: a pilot analysis of colonoscopy, aortic valve replacement and carpal tunnel release procedures. Journal of clinical medicine research, 10(4), 314.

Martin, N., De Weerdt, J., Fernández-Llatas, C., Gal, A., Gatta, R., Ibáñez, G., ... & Van Acker, B. (2020). Recommendations for enhancing the usability and understandability of process mining in healthcare. Artificial Intelligence in Medicine, 109, 101962.

Munoz-Gama, J., Martin, N., Fernandez-Llatas, C., Johnson, O. A., Sepúlveda, M., Helm, E., ... & Zerbato, F. (2022). Process mining for healthcare: Characteristics and challenges. Journal of Biomedical Informatics, 127, 103994.

Nundy, S., Cooper, L. A., & Mate, K. S. (2022). The quintuple aim for health care improvement: a new imperative to advance health equity. Jama, 327(6), 521-522.

Rojas, E., Munoz-Gama, J., Sepúlveda, M., & Capurro, D. (2016). Process mining in healthcare: A literature review. Journal of biomedical informatics, 61, 224-236.

Van Der Aalst, W., Adriansyah, A., De Medeiros, A. K. A., Arcieri, F., Baier, T., Blickle, T., ... & Wynn, M. (2011, August). Process mining manifesto. In International conference on business process management (pp. 169-194). Berlin, Heidelberg: Springer Berlin Heidelberg.

Yangyang, R. Y., Abbas, P. I., Smith, C. M., Carberry, K. E., Ren, H., Patel, B., ... & Lopez, M. E. (2016). Time-driven activity-based costing to identify opportunities for cost reduction in pediatric appendectomy. Journal of pediatric surgery, 51(12), 1962-1966.

Yangyang, R. Y., Abbas, P. I., Smith, C. M., Carberry, K. E., Ren, H., Patel, B., ... & Lopez, M. E. (2017). Time-driven activity-based costing: a dynamic value assessment model in pediatric appendicitis. Journal of pediatric surgery, 52(6), 1045-1049.