Cost Allocation

Cost Allocation in Health Care Costing and Economic Evaluation

Most of what a hospital spends cannot be traced to a single patient. Management, estates, finance, shared equipment and the nursing staff of a ward all serve many patients at once, so any cost per patient, per bed-day or per procedure has to share those costs out by a rule, a process known as cost allocation. Health economists meet the result whenever a model uses a hospital unit cost, a national average cost or a trial's patient-level costs, and the rule chosen can move the cost of one service a long way while the provider's total stays the same. An economic evaluation needs the costs that differ between the options compared, and the marginal cost of one more patient in a ward with spare capacity can sit far below the allocated average. The wider system of recording and classifying costs is covered under Cost Accounting.

Tracing first and allocating second

Mogyorosy and Smith, in their review of costing methods for the Centre for Health Economics at York, define overheads (indirect costs) as costs that cannot be traced to services in an economically feasible way. Elsewhere they give the example of a doctor who treats several patients in an intensive care unit, whose salary has to be apportioned between those patients before the cost of any one treatment is known. They stress that different but entirely reasonable methods can lead to significantly different results, and that overheads can account for more than 40% of the unit cost of hospital services.

The review quotes a three-level order of preference from the US Federal Accounting Standards Advisory Board. Costs should be traced directly whenever feasible and economically practicable, then assigned on a cause-and-effect basis, and only then allocated on a reasonable and consistent basis. Mogyorosy and Smith add that other criteria, such as benefit received, fairness and equity or ability to bear, are used in practice, and that the choice of base usually depends on which one most closely approximates the factors that generate overhead costs in the long run. In principle a strong cause-and-effect relationship should guide the choice, but the final decision is usually a matter of judgement.

The allocation chain from ledger to patient

Hospital costing moves money through a chain of cost objects. Mogyorosy and Smith summarise a seven-step World Health Organization method for the unit costs of hospital services, which defines the final product, sets up cost centres, identifies the full cost of each input and classes it as direct or indirect, assigns inputs to cost centres, allocates all costs to final cost centres, computes unit costs and reports the results. Each of the allocation steps uses the same proportional rule.

$$A_j = P \times \frac{d_j}{\sum_{i} d_i}$$

where $A_j$ is the cost allocated to cost object $j$, $P$ is the total of the shared cost pool, $d_j$ is the quantity of the driver or allocation base used by $j$ (bed-days, nursing hours, floor area or staff numbers) and the sum runs over every cost object $i$ that uses the pool. The shares always add up to the pool, so the rule decides who carries the cost, never how much is spent.

In England the chain is set nationally. The NHS England integrated costing standards for 2026 describe overheads as costs that cannot be directly linked to the treatment of a patient and that are incurred regardless of activity level. Overheads should be allocated to the cost centres that used the service by prescribed allocation methods, and locally defined allocation methods should not be used. Patient-facing costs then reach patients in two steps, first from resources to activities and then from costed activities to the correct patient event. The standards note that the prescribed methods mostly carry no relative weight value for acuity or intensity, although a trust already using one may continue and record it as a superior costing method.

Four ways to move support department costs

Support departments such as estates, finance and human resources also serve each other, and the methods differ in how far they recognise those flows. Mogyorosy and Smith describe four methods for moving support centre costs to the patient-facing (mission) centres, summarised in the table below. The common cost allocation page works one hospital through all four and shows how far the result for one clinic moves.

The review finds no consensus on the best method, and the common cost allocation page sets out the comparative evidence and the Medicare rules. NHS England requires reciprocal costing within the costing system and rules out a hierarchical method, which would let cost flow only one way between corporate support services.

Activity-based costing takes a different route. Instead of one or a few volume bases, it assigns resource costs to activities and then uses several cost drivers, ideally with a cause-and-effect link, to carry activity costs to services. Mogyorosy and Smith note that it does not eliminate all the difficulties and uncertainties of cost allocation. Time-driven activity-based costing uses the time each activity takes as the main driver.

Worked example with one ward and two patient groups (illustrative)

All figures in this example are illustrative. A mixed ward has a nursing pay pool of GBP 840,000 a year and treats two groups of patients. Elective hip replacement patients number 600 a year, stay 3 days and need about 4 nursing hours per bed-day. Stroke patients number 200, stay 12 days and need about 7 nursing hours per bed-day.

1. Measure the drivers. Hip patients use 600 × 3 = 1,800 bed-days and 1,800 × 4 = 7,200 nursing hours. Stroke patients use 200 × 12 = 2,400 bed-days and 2,400 × 7 = 16,800 nursing hours. The ward totals are 4,200 bed-days and 24,000 nursing hours.

2. Allocate by bed-days. The rate is 840,000 / 4,200 = GBP 200 per bed-day. Hip patients carry 1,800 × 200 = GBP 360,000, or GBP 600 per patient, and stroke patients carry 2,400 × 200 = GBP 480,000, or GBP 2,400 per patient.

3. Allocate by nursing hours. The rate is 840,000 / 24,000 = GBP 35 per nursing hour, and the stroke share follows from the proportional rule.

$$A_S = 840{,}000 \times \frac{16{,}800}{24{,}000} = 588{,}000$$

where $A_S$ is the nursing cost allocated to stroke patients in GBP. Hip patients carry the remaining 840,000 − 588,000 = GBP 252,000, or GBP 420 per patient (GBP 140 per bed-day), and stroke patients carry GBP 2,940 per patient (GBP 245 per bed-day).

Switching driver moves GBP 108,000 a year from hip to stroke patients. The stroke cost per patient rises by GBP 540 (22.5%) and the hip cost falls by GBP 180 (30%), with no change in what the ward spends.

4. Value a shorter stay. Suppose an early supported discharge service shortens each stroke stay by 4 days and costs GBP 700 per patient to deliver. Priced at the bed-day rate, the nursing cost avoided is 4 × 200 = GBP 800 per patient. Priced at the average stroke intensity, it is 4 × 7 × 35 = GBP 980. The days removed are the last days of the stay, though, and if those days need only 4 nursing hours each, the avoided cost is 4 × 4 × 35 = GBP 560.

5. Interpret. Set against the GBP 700 cost of the service, this gives a net saving of GBP 100 per patient under the bed-day rule, a net saving of GBP 280 under average intensity and a net cost of GBP 140 under the low-dependency assumption. The same service looks cost saving under two rules and cost increasing under the third. None of the three figures is a cash saving unless the nursing establishment actually shrinks; if it does not, the freed bed-days are worth what the ward does with them, which is an opportunity cost question that allocation cannot answer.

What allocated costs mean in economic evaluation

Allocation produces a fully absorbed average cost, in which every pound the organisation spends lands on some service. Fishman and Hornbrook, reviewing how health services research assigns resource use, describe step-down models as distributing actual incurred expenses so that every dollar spent is allocated, with the cost of any service changing as the number and intensity of services vary. They identify the main weakness as cost assignment that is specific to the organisation studied, set by its buildings, staffing mix and management model, which limits how far findings generalise to other settings.

An economic evaluation needs the costs that differ between the options compared. Mogyorosy and Smith note the argument that, when marginal analysis is used, overhead costs with no effect on the result could be omitted. They also warn that accounting costs can differ from opportunity costs, for example when a fully depreciated building has an accounting cost of zero but a substantial opportunity cost. The incremental cost of a technology therefore depends on which resources actually change, and the marginal cost of one more patient in a ward with spare capacity can sit far below the allocated average.

National unit costs already contain allocations. The NICE technology appraisal manual (PMG36, section 4.4.9) points to the national average unit cost of a healthcare resource group, reported through the annual mandatory collection of reference costs from all NHS organisations in England, as a source for costing resource use. A sensitivity analysis that varies the allocation driver, as in the example above, shows whether a conclusion rests on the rule.

Cost allocation beside nearby ideas

Cost allocation is one step inside larger costing approaches, and several neighbouring pages cover particular cases. Cost Accounting is the wider system that records, classifies and analyses an organisation's costs, of which allocation is one part. Common cost allocation is the case of a cost shared by several services, with the support department methods worked in full. Top-down costing starts from organisational totals and depends heavily on allocation, while micro-costing measures resource use item by item and leaves less to allocate, though its overheads still need a rule. Attributing costs to a disease in a cost-of-illness study is a different problem, which concerns whether a cost is caused by the condition, not how a shared provider cost is split.

Where allocated costs mislead

Most errors come from reading an allocated figure as if it had been measured. The points below follow from the sections above.

1. An allocated share is a convention. The worked example gives stroke care a ward nursing cost of GBP 2,400 or GBP 2,940 per patient from the same ledger, and both are defensible.
2. Full average cost is not the saving from doing less. Removing activity releases only the resources that actually change; the rest is reallocated to the services that remain.
3. Reimbursement rules can shape allocation. Mogyorosy and Smith report that when US hospital reimbursement was tied to reported cost, providers had an incentive to allocate more overhead to reimbursed services, often by changing allocation bases, and that allocation systems can act as an internal tax that alters behaviour.
4. Costs from different providers may not be comparable. The same review notes that arbitrary overhead allocation can undermine comparability and generalisability, so differences in unit cost between hospitals or countries can reflect accounting rules before efficiency.
5. Double counting is easy. An overhead already inside a national unit cost should not be added again as a separate item in the same model.

Sources

• Fishman PA, Hornbrook MC. Assigning resources to health care use for health services research: options and consequences. Medical Care. 2009;47(7 Suppl 1):S70-S75. doi:10.1097/MLR.0b013e3181a75a7f. https://pmc.ncbi.nlm.nih.gov/articles/PMC3588569/
• Mogyorosy Z, Smith P. The main methodological issues in costing health care services: a literature review. CHE Research Paper 7. York: Centre for Health Economics, University of York; 2005. https://www.york.ac.uk/media/che/documents/papers/researchpapers/rp7_Methodological_issues_in_costing_health_care_services.pdf
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). London: NICE; 2022, last updated 13 October 2026. Section 4.4.9. https://www.nice.org.uk/process/pmg36/chapter/economic-evaluation-2
• NHS England. Integrated information requirements, costing processes and costing methods 2026 (integrated costing standards; CP2 and CP3). London: NHS England; 2025, updated March 2026. Internet Archive capture, June 2026. https://web.archive.org/web/20260606171853/https://www.england.nhs.uk/long-read/integrated-costing-standards-2026/