Capacity Utilisation

Capacity Utilisation and the Cost of Spare Capacity in Health Services

Capacity utilisation in health care is the proportion of a service's available capacity that is actually used during a period, such as occupied bed-days out of available bed-days, theatre time used out of planned session time, or scans performed out of the scans a scanner could deliver. A low figure points to idle fixed resources, and a figure close to 100% points to a service with no room to absorb peaks in demand. The term here refers to health services, not the industry-wide capacity utilisation rate of macroeconomics. Very high occupancy harms patient flow and safety, and spare capacity changes the cost of a service in economic evaluation, where the NICE manual gives guidance on average and marginal costs.

Used capacity over available capacity

The measure is a ratio of two quantities counted in the same units over the same period. The numerator is the capacity used, and the denominator is the capacity that could have been used.

$$U = \frac{Q}{Q_{\max}} \times 100$$

where $U$ is capacity utilisation in per cent, $Q$ is the capacity used in the period (occupied bed-days, theatre minutes used, scans performed) and $Q_{\max}$ is the capacity available in the same units and period. Unused capacity is $Q_{\max} - Q$.

The denominator decides what the figure means. Physical capacity counts every bed or scanner that exists; staffed or available capacity counts only those that can be used with the staff rostered; and economic plant capacity, in the definition by Johansen quoted by Valdmanis and colleagues, is the maximal amount that can be produced per unit of time with existing plant and equipment when variable inputs are unrestricted. The same activity gives a higher utilisation against staffed capacity than against physical capacity, and closing beds raises utilisation with no change in the patients treated.

Beds, theatres and scanners

The bed form of the measure is the occupancy rate, meaning occupied bed-days divided by available bed-days. In England's KH03 collection, as the Nuffield Trust describes it, an overnight bed counts as occupied if it is occupied at midnight, and a day-only bed counts as occupied if at least one day case took place in it. The Nuffield Trust reports that in the fourth quarter of 2025/26 (January to March), 94,737 of 102,423 available overnight general and acute beds were occupied. Recomputed, 94,737 / 102,423 = 0.925, an occupancy of 92.5%, leaving 7,686 beds empty on an average night.

Theatre utilisation is usually measured in time. Kirk and colleagues describe touchtime utilisation as the time from the start of anaesthesia to the patient leaving theatre, for all cases on a list, as a percentage of available theatre time; the capped version counts only time within the planned session, while the uncapped version includes overruns. The Getting It Right First Time programme states that integrated care systems and providers are targeted to achieve a minimum of 85% capped theatre utilisation. Scanner and other equipment utilisation follows the same logic, with scans performed or scanner hours used set against the hours the equipment is staffed and open.

Why very high occupancy harms flow and safety

Emergency demand varies from day to day, so a hospital whose average occupancy sits close to its capacity has no buffer for busy days. The evidence on beds points to a level above which risks rise sharply, well below 100%.

Bagust, Place and Posnett simulated emergency admissions to a hypothetical English acute hospital with a discrete-event stochastic model. Risks became discernible when average occupancy exceeded about 85%, at which level a hospital could expect to be short of beds on four days a year; at 90% or more it could expect regular bed shortages and periodic bed crises. Because recovery took about twice the mean length of stay, at 85% occupancy, four days a year without beds could disrupt the hospital for up to eight weeks in total. The authors judged this rapid rise in risk unsurprising given queuing theory for systems with fixed or limited capacity. The bed turnover page shows how high occupancy shrinks the empty-bed interval between patients.

Keegan, writing in the Medical Journal of Australia, reported that high occupancy has been associated with greater risks of hospital-associated infection and access block and with harm to staff health, and that clinical observational data suggested occupancy above 85% could adversely affect safe, effective hospital function. The NICE guideline on emergency and acute medical care (NG94, 2018) recommends that providers plan capacity to minimise the risks associated with occupancy rates exceeding 90%. Its committee chose 90% as a pragmatic maximum and said higher levels could be permissible for efficiently managed elective pathways. The review identified no relevant economic evaluations; the one system model it included was the Bagust simulation. England's 92.5% sits above that figure.

Spare capacity in costing and economic evaluation

Much of the cost of a bed, a theatre or a scanner is fixed over a year, since equipment, buildings and staff rostered to sessions are paid for whether or not they are used. Average cost per unit therefore falls as utilisation rises, because the fixed cost is shared among more units.

$$AC = \frac{F}{Q} + v$$

where $AC$ is average cost per unit of activity, $F$ is the fixed cost in the period, $Q$ is the number of units delivered and $v$ is the variable cost per unit. Within capacity, the short-run marginal cost of one more unit is $v$, which is below average cost whenever $F$ is positive. The healthcare production function page shows how these costs follow from the inputs used.

Empty beds carry a cost, but total costs fall much less than in proportion when beds are removed. Gaynor and Anderson estimated a short-run cost function for over 5,000 US hospitals in 1983 to 1987 that allowed for uncertain demand. They put the cost of an empty bed at USD 36,443 in 1987 dollars, estimated that a 1% fall in beds would cut hospital costs by slightly over a third of 1%, and estimated that raising occupancy from the 1992 average of 65% back to the 1980 average of 76% would save the average hospital over USD 2 million, or 9.5% of costs.

The NICE manual for technology appraisal (PMG36, section 4.4.20) applies the same logic to technologies with several uses that are already used in the NHS, such as diagnostic tests that identify multiple markers. It says the average cost should initially be based on the expected use or throughput of the device for only the uses being evaluated. If the technology is already recommended for another purpose and enough spare capacity exists, an analysis using marginal costs may be supplied in addition. Section 4.4.21 allows sensitivity analyses that assign some fixed costs to other uses when there is evidence that those uses also give good value for money. When a service is full, an extra use displaces other patients, and its cost is the opportunity cost of the care forgone.

Worked example of costing a new use of a CT scanner

All figures in this example are illustrative. A CT scanner has a fixed annual cost of GBP 600,000 (equipment, maintenance and staffed sessions), capacity for 12,000 scans a year and a variable cost of GBP 40 per scan. It currently performs 7,200 scans for existing indications, and an appraisal considers a new indication adding 1,200 scans a year.

1. Current utilisation. Utilisation is 7,200 / 12,000 = 0.60, or 60%. Fixed cost per scan is 600,000 / 7,200 = 83.33, so average cost is 83.33 + 40 = GBP 123.33 per scan.

2. Utilisation with the new use. Scans rise to 7,200 + 1,200 = 8,400, so utilisation rises to 8,400 / 12,000 = 0.70, or 70%, and spare capacity remains. Spreading the fixed cost over all scans, average cost falls from GBP 123.33 to GBP 111.43 as utilisation rises from 60% to 70%.

3. Three costings of a scan for the new indication.

4. Interpretation. The cost per scan entered in the model ranges from GBP 40 to GBP 540, a 13.5-fold spread, from the same scanner and the same activity. Row 1 charges the whole scanner to the new indication, as if it had been bought for that use alone; PMG36 treats the shared-cost and marginal-cost analyses as additional to this starting point. The marginal costing is defensible here only because 3,600 scans of capacity would still be unused. Had the scanner already run at 95%, or 11,400 scans, the new indication would need 12,600 scans against a capacity of 12,000, so 600 scans would displace existing patients or require extra sessions, and GBP 40 would understate the cost.

Capacity utilisation in efficiency measurement

A utilisation ratio sets one used input against its capacity and says nothing about what was produced. Frontier methods separate unused capacity from waste in how inputs are used.

Valdmanis, Kumanarayake and Lertiendumrong used data envelopment analysis on 68 Thai public hospitals in 1999, treating beds, doctors, nurses and other staff as fixed inputs and spending on allowances, drugs and other operating costs as variable. Following methods developed by Färe and colleagues, they compared the output a hospital could reach with its observed inputs against the output attainable when only the fixed inputs constrain production and variable inputs are unrestricted, after first removing technical inefficiency; a ratio below 1 indicated unused plant capacity. Hospitals operated at 90 to 95% of capacity given their fixed inputs, possible increases amounted to 5% of capacity, and the authors concluded that further expansion would require plant investment. A hospital can therefore lack technical efficiency while having little spare capacity, or be efficient with much of its plant idle.

Limits and common misreadings

The ratio is simple to compute and easy to over-read. The main errors come from treating it as a target, a measure of value or a stable quantity.

• Higher is not always better. Green and Nguyen found that using target occupancy levels as the main basis for bed capacity is inadequate and may lead to excessive delays for beds, and that consolidating clinical services into single units may be counterproductive.
• Averages hide peaks. A quarterly mean of 85% can include days at full occupancy, and a midnight census misses beds used and freed during the day.
• The denominator moves. Closing staffed beds or cancelling sessions raises utilisation without any change in patients treated.
• Use is not value. A full theatre list says nothing about whether the procedures are worth doing, and Kirk and colleagues found that touchtime utilisation correlated with surgical cutting time but, used alone, gave a weak measure of how efficiently operating time was used.
• Population use is a different concept. A utilisation rate per head of population measures how much a service is used, not how full the service is.

Sources

• Bagust A, Place M, Posnett JW. Dynamics of bed use in accommodating emergency admissions: stochastic simulation model. BMJ. 1999;319(7203):155-158. https://doi.org/10.1136/bmj.319.7203.155
• Gaynor M, Anderson GF. Uncertain demand, the structure of hospital costs, and the cost of empty hospital beds. Journal of Health Economics. 1995;14(3):291-317. https://doi.org/10.1016/0167-6296(95)00004-2
• Getting It Right First Time (GIRFT), NHS England. Theatres and perioperative medicine. Accessed 3 October 2026. https://gettingitrightfirsttime.co.uk/medical_specialties/theatres-and-perioperative-medicine/
• Green LV, Nguyen V. Strategies for cutting hospital beds: the impact on patient service. Health Services Research. 2001;36(2):421-442. https://pmc.ncbi.nlm.nih.gov/articles/PMC1089232/
• Keegan AD. Hospital bed occupancy: more than queuing for a bed. Medical Journal of Australia. 2010;193(5):291-293. https://doi.org/10.5694/j.1326-5377.2010.tb03910.x
• Kirk W, Barter R, Seyed-Safi P, Davies A, Sabharwal S. The importance of surgical cutting time as a key performance indicator alongside touchtime utilisation in operating theatre efficiency optimisation. Cureus. 2024;16(9):e69660. https://doi.org/10.7759/cureus.69660
• National Institute for Health and Care Excellence. Emergency and acute medical care in over 16s: service delivery and organisation (NG94). Chapter 39, Bed occupancy. London: NICE; March 2018. https://www.nice.org.uk/guidance/ng94/evidence/39bed-occupancy-pdf-172397464704
• 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. Sections 4.4.20 and 4.4.21. https://www.nice.org.uk/process/pmg36/chapter/economic-evaluation-2
• Nuffield Trust. Hospital bed occupancy. QualityWatch indicator, using NHS England Bed Availability and Occupancy (KH03) data. Accessed 3 October 2026. https://www.nuffieldtrust.org.uk/resource/hospital-bed-occupancy
• Valdmanis V, Kumanarayake L, Lertiendumrong J. Capacity in Thai public hospitals and the production of care for poor and nonpoor patients. Health Services Research. 2004;39(6 Pt 2):2117-2134. https://doi.org/10.1111/j.1475-6773.2004.00335.x