Cost

Cost in Economic Evaluation

In health economic evaluation, cost means the benefit given up when staff time, medicines, beds and other inputs are committed to one use rather than the best alternative. That meaning differs from the figure in a hospital ledger or on an invoice, and the gap between them explains many disagreements about what a treatment "really costs". Every incremental cost-effectiveness ratio, budget impact estimate and cost-effectiveness model rests on how costs are identified, measured and valued, so errors here pass directly into decisions.

Cost in this sense is an economic concept, distinct from the accounting or pricing sense of the word. Estimating it involves identifying, measuring and valuing resources, choosing between average and marginal cost, and fixing the perspective that decides which costs count. In England the expected methods are set largely by the NICE methods manual (PMG36), with unit costs drawn from national sources and adjusted to a common price year.

Opportunity cost, accounting cost and charges

Economic evaluation starts from scarcity, since a resource used for one patient group is unavailable to another. The economic cost of a resource is therefore its opportunity cost, the value of what it would have produced in its best alternative use. In a health system with a fixed budget, the relevant alternative is usually other health care, so money spent on a new treatment stands for health that the displaced care would have produced.

Accounting cost is a different construct. Financial accounts record expenditure actually incurred, usually at historical prices, and spread overheads and capital across departments by accounting conventions. Such records are a common starting point for costing, but they can miss resources that carry no invoice (such as donated equipment, volunteer time or unpaid care) and can allocate shared costs in ways unrelated to how services consume them.

Charges and prices are a third construct. A charge is the amount a provider bills, and a list price is what a manufacturer asks; either can sit well above or below the resources used, because each reflects negotiation, cross-subsidy and market power. In practice, evaluations for a payer value purchased inputs at the prices the payer actually pays, on the reasoning that this money is what the payer forgoes. The NICE manual (paragraph 4.4.4) asks for prices that reflect as closely as possible those paid in the NHS, including price reductions known to be available across the NHS, such as those under patient access schemes and commercial access agreements.

Identifying, measuring and valuing resource use

Costing in economic evaluation follows three linked steps. Identification lists the resources that differ between the alternatives within the chosen perspective. Measurement counts how much of each resource each patient or pathway uses. Valuation attaches a unit cost to each resource, and the cost of an option is the sum of quantities multiplied by unit costs.

$$C = \sum_{i=1}^{n} q_i \, p_i$$

where $C$ is the total cost of an option, $n$ is the number of resource types, $q_i$ is the quantity of resource $i$ used (minutes of staff time, number of tests, bed days) and $p_i$ is the unit cost of resource $i$. Keeping quantities and unit costs separate is good practice, because it lets a reader see whether a cost difference comes from different resource use or from different prices, and it lets an analyst in another setting substitute local unit costs while keeping the measured quantities.

Resource use data come from trial case report forms, patient-level hospital records, routine administrative data, patient diaries and, where nothing else exists, structured expert opinion. The NICE manual expects evidence that resource use and cost data have been identified systematically, and asks for sources to be justified and alternatives tested in sensitivity analysis when more than one source is available (paragraphs 4.4.1 and 4.4.10).

Micro-costing, gross costing and time-driven activity-based costing

Costing methods differ in how finely resources are broken down. Micro-costing values each individual component of care, such as each drug dose, test and minute of staff time; the NICE manual describes it as costing each individual component of care related to a technology. Gross costing values larger bundles instead, such as an inpatient day, a clinic attendance or a whole episode classified to a healthcare resource group (HRG).

A second, separate distinction concerns where the cost data come from. Bottom-up costing records the resources each patient used and values them with unit costs, while top-down costing starts from spending recorded at departmental or organisational level and apportions it to units of activity. The two distinctions combine, so micro-costing and gross costing can each be done bottom-up or top-down. A Dutch study of appendectomy, normal delivery, stroke and acute myocardial infarction compared bottom-up micro-costing with top-down micro-costing and gross costing, and concluded that restricting bottom-up micro-costing to the components with the greatest effect on total cost, namely labour and inpatient stay, would likely give reliable estimates. The choice therefore trades precision against the effort of data collection, and precision matters most where the alternatives differ.

The NICE manual treats national HRG costs as a valuable source because they can reduce the need for local micro-costing, but it warns that an HRG may omit separately costed elements such as critical care or unbundled items. It also states that HRG data may be unsuitable when the new technology and its comparator fall in the same HRG, or when the HRG mean does not reflect the resource use of the new technology, and that micro-costing studies may then be more appropriate (paragraphs 4.4.9 and 4.4.10).

Time-driven activity-based costing (TDABC) is a structured form of micro-costing drawn from management accounting. Kaplan and Anderson described it as needing two estimates for each group of resources, namely the cost per unit of time of supplying the resource capacity, and the time taken to perform one unit of each activity. The method also accounts for idle or unused capacity. Kaplan and Porter later proposed it for health care, using the patient's medical condition over the full cycle of care as the unit for measuring costs.

Cost categories and why the labels need care

Cost categories help analysts check that nothing relevant has been missed and that nothing is counted twice. The conventional classification in the health economics literature separates four groups, although the boundaries and names vary between texts and guidelines.

Productivity costs can be valued in more than one way. The human capital approach values all output lost for the whole period of absence or lost working life, whereas the friction cost method counts only the output lost until an employer restores production, for example by replacing the worker. The original Dutch friction cost estimates were considerably lower than those from the human capital approach, so the method chosen can change results for conditions that affect people of working age.

The word "indirect" is a common source of confusion. Economic evaluation often uses indirect cost for productivity losses, as in the original friction cost paper, whereas cost accounting uses it for overheads and capital that cannot be traced to a single service. Hospital costing studies often use the accounting meaning, so a reader should always check which sense an analysis intends. Intangible costs raise a related risk of double counting. In a cost-utility analysis, pain and distress lower quality of life and so already reduce QALYs, and adding a monetary value for the same suffering to the cost total would count it twice.

Fixed, variable, average and marginal cost

Costs behave differently as activity changes, and this behaviour decides which unit cost is right for a given decision. Fixed costs, such as a scanner lease or salaried staff establishment, do not change with the number of patients treated over the relevant period. Variable costs, such as drugs and disposables, rise with each patient. Semi-fixed or step costs are constant up to a capacity limit and then jump, for example when an extra clinic session or ward must be opened.

Average cost is total cost divided by output, while marginal cost is the change in total cost from producing one more unit. The two are equal when there are no fixed costs and every unit uses the same resources, but otherwise they usually differ. With spare capacity, the marginal cost of an extra patient can be far below the average, because the fixed costs are already being paid. At full capacity, the marginal cost can exceed the average, because the next patient triggers a step cost.

The choice between average and marginal cost also depends on the time frame. Over a short period many resources are fixed, but over a longer period staff numbers, buildings and equipment can all be adjusted, so almost all costs become variable. Long-run marginal cost then equals long-run average cost only if average cost does not change with the scale of activity, but average cost is widely used as an approximation to it. The Unit Costs of Health and Social Care programme states that its estimates aim to reflect the long-run marginal opportunity cost, which is consistent with valuing resources for decisions that commit the health system over several years.

The NICE manual applies this distinction to technologies with several uses that are already in use in the NHS, such as a diagnostic platform that can test for several markers. The average cost should first be based on the expected use or throughput of the device for only the uses being evaluated. Where the technology is already recommended for another purpose and has enough spare capacity, an analysis using marginal cost may be supplied in addition to the average-cost analysis (paragraph 4.4.20).

Perspective decides which costs count

The perspective of an evaluation defines whose costs are included, and it should be fixed before resources are identified. A narrow payer perspective counts only what the payer funds, while a societal perspective adds costs falling on patients, families, employers and other public services.

For the NICE reference case, the perspective on costs is that of the NHS and personal social services (PSS), and productivity costs are excluded (paragraphs 4.2.7 and 4.2.9). Patient costs reimbursed by the NHS or PSS may be included. Some other costs may be presented only as separate analyses. These are unreimbursed patient costs; productivity costs, when they may be a critical component of the technology's value; the time of family members or friends providing care that the NHS or PSS might otherwise have provided; and, exceptionally and only when agreed with the Department of Health and Social Care, costs to other government bodies (paragraphs 4.4.22 to 4.4.24). In the United States, the Second Panel on Cost-Effectiveness in Health and Medicine recommended in 2016 that every cost-effectiveness analysis report two reference case analyses, one from a health care sector perspective and one from a societal perspective, a change from the single societal reference case set by the first Panel in 1996.

UK unit costs, price years and inflation

Valuation needs unit costs. For staff time and services in England, two national sources supply most of them, while medicine and device prices come from the sources the NICE manual lists, such as the drugs tariff, the Drugs and Pharmaceutical electronic Market Information Tool (eMIT) and NHS Supply Chain prices (paragraph 4.4.4). The Unit Costs of Health and Social Care manual, produced annually, gives costs for professional staff time, community and social care services and other items. Since 2022 it has been produced by a collaboration between the Personal Social Services Research Unit at the University of Kent (now the Care and Outcomes Research Centre) and the Centre for Health Economics at the University of York, funded by the National Institute for Health and Care Research. The 2025 manual, published in June 2026, reports costs in 2024/25 prices.

For hospital services, NHS England publishes the National Cost Collection for the NHS. It comprises aggregated costs, defined as the average unit cost of providing defined services to NHS patients in England, and patient-level costs; NHS providers submit costs annually, and the 2024/25 data are published as dashboards and data tables, including the national schedule of NHS costs. The NICE manual refers to this annual mandatory collection as reference costs and notes that it reports the national average unit cost of each HRG.

Every cost in an analysis should be expressed in a single price year, which should be stated. The NICE manual states that when current costs are not available, costs from earlier years should be adjusted using inflation indices appropriate to the cost perspective, such as the NHS cost inflation index and the PSS pay and prices index published in the unit costs report, or the Office for National Statistics consumer price index (paragraph 4.4.12). The NHS cost inflation index was developed by the Department of Health and Social Care with the Office for National Statistics, NHS England and the University of York; the 2025 manual reports a provisional overall increase of 4.02 per cent for 2024/25. On that basis, an illustrative unit cost of 500 pounds in 2023/24 prices becomes 500 × 1.0402 = 520.10 pounds in 2024/25 prices.

Costs that fall in future years are also discounted to present values, so that a pound spent in ten years counts for less than a pound spent now. The rate and its justification are covered on the discount rate page; the NICE reference case discounts costs and health effects at 3.5 per cent a year (paragraph 4.5.1).

What the NICE reference case requires on costs

Section 4.4 of the NICE manual, "Evidence on resource use and costs", sets out how costs are handled in technology appraisal and highly specialised technologies evaluations. Its rules on prices, HRG costs, inflation, and patient and productivity costs appear in the sections above; several further requirements shape the cost total directly.

• Resource use should include changes in infrastructure, use and maintenance, and staff training costs where appropriate (paragraph 4.4.2).
• For medicines mainly prescribed in primary care, prices are based on the drugs tariff (paragraph 4.4.7).
• Costs related to the condition and incurred in additional years of life gained because of the technology are included, while costs unrelated to the condition or technology are excluded (paragraph 4.4.11).
• Costs available only from other countries are converted to pounds sterling using an exchange rate from an appropriate and current source (paragraph 4.4.13).
• Value added tax is excluded from economic evaluations but included in budget impact calculations when the resources are liable for it (paragraph 4.4.19).

Worked example of micro-costing a day-case infusion

The figures below are illustrative and do not describe any real service. A hospital day unit gives a new intravenous treatment in single day-case episodes and treats 2,000 episodes a year. The analyst micro-costs one episode from resource use and unit costs, and then asks what the cost of one more episode would be. All amounts are in pounds.

Each line is a quantity multiplied by a unit cost, and the overhead line divides a fixed annual cost by annual activity, as the calculations below show.

• The drug costs 2 × 210 = 420
• Consumables cost 1 × 60 = 60
• Nurse time costs 90 × 1.00 = 90
• Consultant time costs 20 × 3.00 = 60
• Overheads come to 140,000 ÷ 2,000 = 70
• The cost per episode is 420 + 60 + 90 + 60 + 70 = 700

The next step separates fixed from variable costs. The drug and consumables vary with each episode, giving a variable cost of 420 + 60 = 480. Nurse and consultant time are provided by salaried sessions that are paid whatever the throughput, so their annual cost of 150 × 2,000 = 300,000 is fixed along with the overheads, giving fixed costs of 300,000 + 140,000 = 440,000 a year.

• The total annual cost is 440,000 + 2,000 × 480 = 1,400,000
• The average cost is 1,400,000 ÷ 2,000 = 700

If the unit has spare chair and staff capacity that has no other valuable use, one extra episode adds only the drug and consumables, so its marginal cost is 480, which is 220 less than the average. If instead the unit is full, 400 more episodes a year need an extra session costing 100,000 a year in staff, chair and overheads. Because the session cannot be bought in smaller pieces, its cost is spread over the 400 extra episodes, and the incremental cost per additional episode is then higher than the average.

• The incremental cost per additional episode is 480 + 100,000 ÷ 400 = 730
• The total annual cost at 2,400 episodes is 540,000 + 2,400 × 480 = 1,692,000
• The new average cost is 1,692,000 ÷ 2,400 = 705

The interpretation depends on the decision. An evaluation that uses the average of 700 overstates the cost of a small increase in use within spare capacity by 220 per episode, and understates the cost of an expansion that needs a new session by 30 per episode. For a decision that commits the service for many years, when staffing and space can be adjusted, a long-run average cost of about 700 per episode is usually the defensible choice, with the marginal figures tested as scenarios.

Using cost estimates in models and HTA decisions

Costs enter economic evaluation as differences between alternatives, so the incremental cost of a technology, not its total cost, is the numerator of the cost-effectiveness ratio. Items that are the same under both options cancel out and can often be omitted from the comparison, although they may still matter for a budget impact analysis. The decision is therefore most sensitive to costs that differ between the options, such as drug acquisition, monitoring and the downstream events that the technology prevents or causes.

For patient-level cost data from trials, the relevant summary is the arithmetic mean. Cost distributions are usually strongly skewed to the right because a few patients have complications, reoperations or long stays, and the median or geometric mean then falls below the arithmetic mean. Only the arithmetic mean, multiplied by the number of patients treated, gives the total cost that a health system will bear.

Cost results also carry uncertainty from sampling, from the choice of unit cost source and from assumptions about capacity and time horizon. The NICE manual asks for uncertainty in prices to be handled in the same way as other uncertain or variable parameters (paragraph 4.4.4).

Common misunderstandings about cost

A frequent error is to treat a charge, tariff or list price as if it were the resource cost. Prices and tariffs are set for payment purposes and can differ widely from the resources used; an evaluation should state what each unit cost represents and whether it reflects the price the decision maker actually pays.

A second error is to apply average cost when a decision changes activity only at the margin, or to apply short-run marginal cost to a decision that will commit resources for many years. The worked example shows that the direction of the error depends on whether capacity is spare or full.

A third is inconsistent scope. Mixing costs from different perspectives, counting productivity losses in a NICE reference case, or adding a monetary value for suffering that is already reflected in QALYs all produce results that do not answer the stated decision problem. The table of cost categories helps only when each category is tied to an explicit perspective.

A fourth is to leave price years and discounting implicit. Costs drawn from different years, unadjusted for inflation, or undiscounted future costs make an analysis impossible to reproduce and can shift results, especially for treatments with high upfront costs and long follow-up.

Sources

• Drummond MF, Sculpher MJ, Claxton K, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. 4th ed. Oxford: Oxford University Press; 2015.
• Gold MR, Siegel JE, Russell LB, Weinstein MC, editors. Cost-Effectiveness in Health and Medicine. New York: Oxford University Press; 1996.
• Russell LB, Gold MR, Siegel JE, Daniels N, Weinstein MC. The role of cost-effectiveness analysis in health and medicine. JAMA. 1996;276(14):1172-1177.
• Sanders GD, Neumann PJ, Basu A, Brock DW, Feeny D, Krahn M, et al. Recommendations for conduct, methodological practices, and reporting of cost-effectiveness analyses: Second Panel on Cost-Effectiveness in Health and Medicine. JAMA. 2016;316(10):1093-1103.
• National Institute for Health and Care Excellence. NICE technology appraisal and highly specialised technologies guidance: the manual (PMG36). London: NICE; published 31 January 2022, last updated 13 October 2026. Sections 4.2, 4.4 and 4.5.
• Jones K, Weatherly H, Barker A, Birch S, Castelli A, Chalkley M, Dargan A, Findlay D, Hinde S, Markham S, Smith D, Teo H. Unit Costs of Health and Social Care 2025 Manual. Canterbury and York: Care and Outcomes Research Centre (University of Kent) and Centre for Health Economics (University of York); 2026. doi:10.22024/UniKent/01.02.115569.
• NHS England. National Cost Collection for the NHS: 2024/25 National Cost Collection data. London: NHS England. Accessed 29 Sep 2026.
• Tan SS, Rutten FF, van Ineveld BM, Redekop WK, Hakkaart-van Roijen L. Comparing methodologies for the cost estimation of hospital services. European Journal of Health Economics. 2009;10(1):39-45.
• Tan SS, van Ineveld BM, Redekop WK, Hakkaart-van Roijen L. Comparing methodologies for the allocation of overhead and capital costs to hospital services. Value in Health. 2009;12(4):530-535.
• Špacírová Z, Epstein D, García-Mochón L, Rovira J, Olry de Labry Lima A, Espín J. A general framework for classifying costing methods for economic evaluation of health care. European Journal of Health Economics. 2020;21(4):529-542.
• Kaplan RS, Anderson SR. Time-driven activity-based costing. Harvard Business Review. 2004;82(11):131-138, 150.
• Kaplan RS, Porter ME. How to solve the cost crisis in health care. Harvard Business Review. 2011;89(9):46-61.
• Koopmanschap MA, Rutten FF, van Ineveld BM, van Roijen L. The friction cost method for measuring indirect costs of disease. Journal of Health Economics. 1995;14(2):171-189.
• Thompson SG, Barber JA. How should cost data in pragmatic randomised trials be analysed? BMJ. 2000;320(7243):1197-1200.