The definition the association adopts, why the calculations in use diverge, and a calculator to check whether your figures add up.
Average length of stay is not one figure: it is four different calculations circulating in the sector under the same name.
The difference between them is not subtle. Applied to the same unit in the same month, they can return results an order of magnitude apart. Two institutions with the same clinical reality report figures that cannot be compared, and any comparison between them ends up measuring the choice of formula rather than the care delivered.
As long as the calculation is not the same across every unit, any sector average is the average of different quantities. This page sets out the definition ABRAHCT adopts and offers the tool to check whether your own figures are internally consistent.
Consistency calculator
Total patient-days for a period can be added up two ways: from the capacity side and from the movement side. Both have to give the same number. This is not a convention, it is accounting.
operational beds × days in period × occupancy
= discharges in period × average length of stay
This identity is what separates a coherent set of figures from one that only looks coherent. Fill in your unit's figures below: if the two sides do not match, one of the four fields is being measured with a different definition from the other three.
Beds able to receive a patient in the period, not installed beds
30 for a month, 90 for a quarter, 365 for the year
Average for the period
Discharges, deaths and external transfers. Internal transfers do not count
Optional. Fill it in to check whether it matches the other figures
Fill in beds, days, occupancy and discharges.
The figures stay in your browser. Nothing is sent to ABRAHCT.
The definition
Average length of stay is the sum of patient-days in the period divided by the number of discharges in the same period, counting discharges, deaths and external transfers — the formula of the Brazilian Ministry of Health's Hospital Census Nomenclature Standard (Ordinance SAS/MS 312/2002), the same used by the ANS QUALISS E-EFI-05 indicator and by the ANAHP Observatory.
The daily sum of patients admitted at 00:00, as the ANS indicator sheet requires. A patient admitted and discharged on the same day counts as one patient-day.
A transfer between beds or wards within the same institution is not a discharge and does not end the admission.
In the occupancy rate used for validation, the denominator excludes blocked beds and includes extra beds — beds genuinely able to receive a patient.
The duration of each completed admission (discharge minus admission; same day counts as 1) is the data behind the median, P25 and P75, always published alongside the mean.
The four calculations
Divides total patient-days in the period by the number of discharges in the same period (discharges, deaths and external transfers).
This is the Brazilian national standard — Ministry of Health Ordinance SAS/MS 312/2002, the ANS QUALISS E-EFI-05 indicator sheet and the ANAHP Observatory all calculate it this way — and the same construction as the OECD's average length of stay. It is the figure that can be compared across institutions.
Takes the admissions that ended within the period and measures the full duration of each one: discharge date minus admission date.
This is where the median and the percentiles come from, which the mean does not reveal. In a stable unit, the cohort mean converges to the official indicator.
Measures how long the patients present on the census date have been admitted.
Always overestimates: a patient who stays 200 days appears in 200 daily censuses, one who stays 10 days appears in 10. The long-stay patient enters the calculation twenty times more often.
Uses admissions instead of discharges in the denominator.
The most common spreadsheet error. It understates in a growing unit and overstates in a shrinking one.
How to publish it
The distribution of length of stay in transitional care is skewed to the right: a handful of very long stays pulls the mean away from the typical case. Two units with the same mean can have quite different clinical realities.
For that reason the association always publishes median, mean, 25th and 75th percentile together, with the number of completed admissions behind the statistic. The median is the typical case; the distance between it and the mean measures the skew; the percentiles show the range holding the middle half of cases.
Status of this document
Definition adopted in August 2026, applying to the association's surveys and to every length-of-stay indicator it may publish. It is not ABRAHCT's invention: it is the Brazilian national standard — Ministry of Health Ordinance SAS/MS 312/2002, the ANS QUALISS E-EFI-05 indicator sheet and the ANAHP Observatory use the same formula — applied with the explicit operational rules our segment was missing. The 57-day average length of stay circulating in the sector comes from a third-party survey released by Medicina S/A in April 2025, and the method used to produce it is not known, so it remains cited with attribution to the outlet in Sector data, and not as an ABRAHCT figure.
Comments and suggestions on the methodology: abrahct@abrahct.org.br