Average Length of Stay

Average Length of Stay in Calculation and Bed-Day Costing

Average length of stay (ALOS) matters to health economists as an efficiency indicator and as the quantity that turns a bed-day price into the cost of an admission. The OECD, NHS England and Australian conventions for calculating it differ, and the average cost of a bed day overstates what a shorter stay saves. Skewed stays call for trim points, comparisons call for case-mix adjustment, and a falling ALOS can hide costs that have moved to readmissions or other settings, as an illustrative worked example comparing two hospitals shows.

How ALOS is calculated from bed-days and discharges

The calculation divides the days spent in hospital by the number of stays that generated them. With patient-level records, each stay runs from admission to discharge. The mean is then taken over all discharges in the period, as follows.

$$\bar{L} = \frac{1}{N} \sum_{i=1}^{N} L_i, \qquad L_i = d_i - a_i$$

where $\bar{L}$ is the average length of stay in days, $N$ is the number of discharges in the period, $L_i$ is the length of stay of discharge $i$, $d_i$ is its discharge date and $a_i$ is its admission date.

Routine statistics often divide a period's occupied bed-days by its discharges or admissions instead, which matches the patient-level figure only when activity is stable. Choices about the counting unit matter as much, including whether stays are counted per consultant episode, per hospital or across transfers, whether same-day cases are included, and whether a same-day stay counts as zero days or one.

OECD, NHS and Australian definitions compared

International and national bodies use the same basic ratio but differ in scope and in what counts as a stay. The differences are large enough to change rankings, so figures should be compared only within one definition.

The OECD's Health at a Glance describes ALOS as the average number of days patients spend in hospital, generally measured by dividing the total days stayed by all inpatients during a year by the number of admissions or discharges, with day cases usually excluded. The 2023 edition covered all inpatient cases for most countries and reported 7.7 days across 36 OECD countries in 2021; a footnote added that Japan, whose chart figure covered curative (acute) care only, had an average of 28 days for all inpatient care. The 2025 edition reports curative (acute) care for most countries, with 6.5 days across 36 OECD countries in 2023, with Türkiye and Mexico shortest at 4.7 days and Japan longest at 15.7 days. Because the scope changed, the fall from 7.7 to 6.5 days cannot be read as a trend.

In England, Hospital Episode Statistics record finished consultant episodes, which are joined into spells. The NHS Data Model and Dictionary defines a Hospital Provider Spell as the total continuous stay of a patient using a hospital bed on premises controlled by a health care provider, starting with an admission and ending with a discharge. A transfer to another provider ends the spell, so analysts link spells into continuous inpatient spells to capture a whole stay. For payment, Annex D of the 2025/26 NHS Payment Scheme counts the number of changes of calendar day from admission to discharge, after deducting days in critical care, rehabilitation, specialist palliative care and delayed transfers of care, so a same-day stay counts as zero days.

The Productivity Commission's Report on Government Services 2026 defines length of stay in Australian public hospitals as the period from admission to separation less any leave days. A separation can be a portion of a stay that ends with a change in the type of care, such as acute to rehabilitation, so one continuous stay can produce two separations.

ALOS as an efficiency indicator

The OECD treats ALOS as an indicator of efficiency in service delivery. All else equal, a shorter stay reduces the cost per discharge and moves care to less expensive settings, while long stays can signal patients waiting for rehabilitation or long-term care. It also notes that some patients may be discharged too early, when a longer stay might have improved outcomes or reduced the chance of readmission. ALOS is a driver of throughput, and it says something about technical efficiency only when outcomes are held constant.

The link to capacity follows from an accounting identity. In a steady state, annual discharges depend on beds, occupancy and the average stay, as follows.

$$D = \frac{365 \times B \times o}{\bar{L}}$$

where $D$ is annual discharges, $B$ is the number of staffed beds, $o$ is average bed occupancy as a proportion and $\bar{L}$ is the average length of stay in days. Halving $\bar{L}$ doubles potential throughput, which is why bed turnover and ALOS are read together. The OECD's 2023 edition adds that global budgets and payments based on diagnosis-related groups give hospitals a financial incentive to reduce the cost per hospitalisation, to which a shorter stay contributes.

Average versus marginal bed-day costs

In costing, ALOS multiplied by a cost per bed day gives a cost per admission. That shortcut, a form of top-down costing, spreads staffing, buildings and overheads evenly across every day of a stay. The marginal cost of one day more or less is usually far lower, because most resources are used early in an admission and many hospital costs are fixed in the short run.

Taheri and colleagues, using cost-accounting records for 12,365 surviving patients with stays of four days or more at a United States academic medical centre in fiscal year 1998, found that the incremental cost of the last full day averaged US$420, or 2.4% of the mean total cost of US$17,734; in their trauma centre about 40% of variable costs fell in the first three days. Page, Barnett and Graves argued that the opportunity cost of a bed day, its value in treating another patient, is the better value for resource decisions. In their contingent valuation, 37 Australian hospital chief executives would pay on average A$216 to release a ward bed day and A$436 for an intensive care bed day, about four and seven times lower than the accounting costs often used.

For economic evaluation, the saving from shorter stays depends on the freed capacity. Cash is saved only on variable costs or by closing beds, and otherwise the gain is the value of extra activity. Where a reduced stay drives a cost-effectiveness analysis, the bed-day value belongs in the sensitivity analysis, and micro-costing of the days actually saved is more defensible than an average daily rate.

Skewed stays, outliers and trim points

Lengths of stay are right-skewed. Most patients stay a few days and a few stay for weeks, so the mean sits above the median. Thompson and Barber argued that for costing the arithmetic mean is the most informative measure despite skewness, because only the mean scaled by the number of patients gives the total cost of treating them. A geometric mean or median describes the typical stay better but cannot be scaled up to a budget.

Payment systems handle very long stays by trimming rather than ignoring them. Annex D of the 2025/26 NHS Payment Scheme sets a trim point for each spell healthcare resource group (HRG), as follows.

$$T = \max\{5,\; Q_3 + 1.5 \times (Q_3 - Q_1)\}$$

where $T$ is the trim point in days, rounded to the nearest whole number, and $Q_1$ and $Q_3$ are the first and third quartiles of adjusted length of stay for spells in that HRG. Each spell's excess bed days equal its adjusted length of stay minus $T$, or zero if it does not exceed $T$, and are paid at a separate daily long-stay rate. Each outlier day is funded separately, so a few extreme cases do not drive the price per admission.

Case-mix adjustment for comparing like with like

A crude ALOS comparison penalises hospitals that treat more complex patients. Case-mix adjustment compares observed bed-days with those expected if each patient group stayed as long as a reference average. This is indirect standardisation, the approach behind the standardised mortality ratio, and the calculation is as follows.

$$E_h = \sum_{g} n_{hg} \, \bar{L}_g^{\,\mathrm{ref}}, \qquad R_h = \frac{O_h}{E_h}$$

where $E_h$ is the expected bed-days for hospital $h$, $n_{hg}$ is its number of discharges in case-mix group $g$, $\bar{L}_g^{\,\mathrm{ref}}$ is the reference ALOS for group $g$, $O_h$ is its observed bed-days and $R_h$ is the case-mix adjusted ratio, where a value above 1 means longer stays than expected. Australia's Report on Government Services publishes a relative stay index on this basis, dividing actual patient days for acute care separations in selected diagnosis-related groups by expected patient days adjusted for casemix, excluding, among others, patients who died or were transferred within 2 days and stays over 120 days. Such adjustment is a form of risk adjustment, and its quality depends on the grouping system, whose overall complexity a case mix index summarises.

Worked example with two hospitals and two case-mix groups

The figures below are illustrative. Two hospitals each discharge 400 patients in a year from two groups, elective hip replacement, with a reference ALOS of 5 days, and stroke, with a reference ALOS of 15 days.

1. Crude ALOS. Hospital A has 3,050 bed-days over 400 discharges, an ALOS of 7.6 days (7.625). Hospital B has 4,800 bed-days over 400 discharges, an ALOS of 12.0 days, so on crude figures B looks far less efficient.

2. Expected bed-days. For A, $E_A = 300 \times 5 + 100 \times 15 = 3000$ bed-days; for B, $E_B = 100 \times 5 + 300 \times 15 = 5000$ bed-days.

3. Adjusted ratio. For A, $R_A = 3050 / 3000 = 1.017$, stays about 1.7% longer than expected. For B, $R_B = 4800 / 5000 = 0.96$, stays 4% shorter than expected. B's higher crude ALOS reflects its larger share of stroke patients, and the ranking reverses.

4. Valuing a shorter stroke stay. Suppose B cuts its stroke ALOS from 14.0 to 13.0 days, releasing 300 bed-days a year. At an illustrative average cost of £450 per bed day the apparent saving is as follows.

$$\Delta C = 300 \times 450 = 135000$$

where $\Delta C$ is the change in cost in pounds, 300 is the bed-days released and 450 is the cost per bed day in pounds. At an illustrative marginal cost of £100 per bed day the cash saving is only £30,000 (300 × 100). The other £105,000 is fixed cost, saved only if beds close or turned into value only if the freed beds treat other patients.

5. Checking readmissions. If the shorter stay raised the stroke readmission rate by 2 percentage points, the 300 patients would generate 6 extra readmissions. At an illustrative 8 days each, these use 48 bed-days, offsetting 16% of the 300 released, and each brings the costly early days of a new admission. The net gain is smaller than the change in ALOS suggests.

Where a shorter stay can mislead

A falling ALOS is evidence of efficiency only if outcomes and total costs are tracked alongside it. Several mechanisms shorten stays without saving the health system anything.

Shorter stays can move costs rather than remove them. Care passed to rehabilitation units, community services, care homes or family carers still uses resources, and an evaluation from a health system or societal perspective has to count it. Spell-based counting compounds this. Busby, Purdy and Hollingworth found that stroke length of stay in English data was 16.5 days using continuous inpatient spells but 12.6 days using spells, and that areas transferring many patients for rehabilitation could look efficient for that reason alone.

Readmissions are the second check, using a measure such as the readmission rate. Martin, Street, Han and Hutton studied English admissions for hip replacement, hernia repair and stroke between 2002/3 and 2007/8. Conditional on survival, stroke patients with a shorter original stay were more likely to have an emergency readmission within 28 days, while hip replacement showed no relationship; overall, shorter stays were not generally associated with more readmissions, so the risk is condition specific.

Composition also moves ALOS. When simple cases move to day surgery and leave the inpatient count, the remaining inpatients are more complex and ALOS can rise as care improves. Deaths in hospital count as discharges in the OECD definition, so early deaths shorten the average. In a Markov model or other decision model, hospital time is best taken from the trial or cohort that informs the model, on a stated definition, with its skewness reflected through a distribution such as the gamma distribution in probabilistic analysis.

Sources

• Briggs A, Claxton K, Sculpher M. Decision Modelling for Health Economic Evaluation. Oxford: Oxford University Press; 2006. Chapter 4.
• Busby J, Purdy S, Hollingworth W. Calculating hospital length of stay using the Hospital Episode Statistics; a comparison of methodologies. BMC Health Services Research. 2017;17:347. https://doi.org/10.1186/s12913-017-2295-z
• Martin S, Street A, Han L, Hutton J. Have hospital readmissions increased in the face of reductions in length of stay? Evidence from England. Health Policy. 2016;120(1):89-99. https://pubmed.ncbi.nlm.nih.gov/26703897/
• NHS England. 2025/26 NHS Payment Scheme, Annex D: Prices and cost adjustments. London: NHS England; 2025, updated June 2025. Appendix 1 and Appendix 3. https://www.england.nhs.uk/long-read/25-26-nhsps-annex-d-prices-and-cost-adjustments/
• NHS England. NHS Data Model and Dictionary: Hospital Provider Spell. Accessed 1 October 2026. https://www.datadictionary.nhs.uk/nhs_business_definitions/hospital_provider_spell.html
• OECD. Health at a Glance 2023: OECD Indicators. Paris: OECD Publishing; 2023. Chapter 5, Hospital activity. https://doi.org/10.1787/7a7afb35-en
• OECD. Health at a Glance 2025: OECD Indicators. Paris: OECD Publishing; 2025. Chapter 5, Hospital activity. https://www.oecd.org/en/publications/2025/11/health-at-a-glance-2025_a894f72e/full-report/hospital-activity_8d89d2a3.html
• Page K, Barnett AG, Graves N. What is a hospital bed day worth? A contingent valuation study of hospital Chief Executive Officers. BMC Health Services Research. 2017;17:137. https://doi.org/10.1186/s12913-017-2079-5
• Productivity Commission. Report on Government Services 2026, Section 12: Public hospitals, key terms. Canberra: Productivity Commission; 2026. https://www.pc.gov.au/ongoing/report-on-government-services/health/public-hospitals/
• Taheri PA, Butz DA, Greenfield LJ. Length of stay has minimal impact on the cost of hospital admission. Journal of the American College of Surgeons. 2000;191(2):123-130. https://pubmed.ncbi.nlm.nih.gov/10945354/
• Thompson SG, Barber JA. How should cost data in pragmatic randomised trials be analysed? BMJ. 2000;320:1197-1200. https://doi.org/10.1136/bmj.320.7243.1197