Skip to content
ScienceQuest

A-a gradient by age: (age + 10) ÷ 4, age ÷ 4 + 4 and Mellemgaard, with alveolar oxygen in high US cities

At 70 the expected A-a gradient is 20.0 mmHg by (age + 10) ÷ 4, the calculator’s estimate, 21.5 by age ÷ 4 + 4 and 17.2 by Mellemgaard’s 2.5 + 0.21 × age. Moving from sea level to Colorado Springs lowers the expected PaO2 more than ageing from 20 to 90: room-air PAO2 falls 31.7 mmHg, while (age + 10) ÷ 4 rises 17.5.

Line chart of the expected A-a gradient in mmHg against age from 20 to 90 years, by three estimates. The (age + 10) ÷ 4 line, solid with circles, rises from 7.5 mmHg at 20 to 25.0 at 90. The age ÷ 4 + 4 line, dashed with squares, runs 1.5 mmHg above it all the way, from 9.0 to 26.5. The 2.5 + 0.21 × age line, dotted with triangles, rises from 6.7 to 21.4. The lines never cross: Mellemgaard’s runs under the other two at every age, 0.8 mmHg below (age + 10) ÷ 4 at 20 and 3.6 below at 90. A hatched band from 75 to 90 marks ages older than the 80 subjects Mellemgaard measured, who were 15 to 75. Line chart of the expected A-a gradient in mmHg against age from 20 to 90 years, by three estimates. The (age + 10) ÷ 4 line, solid with circles, rises from 7.5 mmHg at 20 to 25.0 at 90. The age ÷ 4 + 4 line, dashed with squares, runs 1.5 mmHg above it all the way, from 9.0 to 26.5. The 2.5 + 0.21 × age line, dotted with triangles, rises from 6.7 to 21.4. The lines never cross: Mellemgaard’s runs under the other two at every age, 0.8 mmHg below (age + 10) ÷ 4 at 20 and 3.6 below at 90. A hatched band from 75 to 90 marks ages older than the 80 subjects Mellemgaard measured, who were 15 to 75.
The expected A-a gradient by the two estimates the A-a gradient calculator prints and by Mellemgaard’s regression, 20 to 90 years. Hatched: older than Mellemgaard’s subjects, who were 15 to 75. Every plotted point is a row of the table. Image © ScienceQuest, CC BY 4.0: free to reuse with credit and a link to this page, as the image licence explains.

Expected A-a gradient by age

The expected A-a gradient in mmHg by (age + 10) ÷ 4 and age ÷ 4 + 4, the two the calculator prints, and by Mellemgaard’s 2.5 + 0.21 × age, from 20 to 90 years, with the difference between the calculator’s main estimate and Mellemgaard’s
Age (years) (age + 10) ÷ 4 age ÷ 4 + 4 2.5 + 0.21 × age Difference
20 7.5 9.0 6.7 0.8
30 10.0 11.5 8.8 1.2
40 12.5 14.0 10.9 1.6
50 15.0 16.5 13.0 2.0
60 17.5 19.0 15.1 2.4
70 20.0 21.5 17.2 2.8
80 22.5 24.0 19.3† 3.2†
90 25.0 26.5 21.4† 3.6†

The difference is (age + 10) ÷ 4 less 2.5 + 0.21 × age. At every age, age ÷ 4 + 4 runs 1.5 mmHg above (age + 10) ÷ 4. † Older than Mellemgaard’s 80 subjects, who were 15 to 75.

Alveolar oxygen in the high US cities

Barometric pressure, inspired oxygen (PiO2) and alveolar oxygen (PAO2) on room air at a PaCO2 of 40 mmHg and R of 0.8, at sea level and in the 12 of the 50 largest US cities above 1,000 ft, highest first
Place Elevation ft Pressure mmHg PiO2 mmHg PAO2 mmHg
Sea level 0 760.0 149.7 99.7
Colorado Springs, CO 6,009 608.9 118.0 68.0
Aurora, CO 5,405 622.9 120.9 70.9
Denver, CO 5,276 625.9 121.6 71.6
Albuquerque, NM 4,955 633.5 123.2 73.2
El Paso, TX 3,713 663.4 129.4 79.4
Tucson, AZ 2,481 694.3 135.9 85.9
Las Vegas, NV 1,996 706.8 138.6 88.6
Mesa, AZ 1,242 726.5 142.7 92.7
Oklahoma City, OK 1,196 727.7 143.0 93.0
Phoenix, AZ 1,088 730.6 143.6 93.6
Omaha, NE 1,056 731.4 143.7 93.7
Atlanta, GA 1,050 731.6 143.8 93.8

Pressure is the 1976 US Standard Atmosphere at the ground of each city’s USGS point, as on the boiling point by city page. FiO2 21%, water vapour 47 mmHg. A lower PaCO2, usual at altitude, raises PAO2 above these figures.

PaO2 at the expected gradient

Arterial oxygen (PaO2) in mmHg at the expected A-a gradient, PAO2 less the expected gradient, for a 30-year-old and a 70-year-old by each of the three age estimates, on room air at a PaCO2 of 40 mmHg
Place Age 30 Age 70
(age + 10) ÷ 4 age ÷ 4 + 4 2.5 + 0.21 × age (age + 10) ÷ 4 age ÷ 4 + 4 2.5 + 0.21 × age
Sea level 89.7 88.2 90.9 79.7 78.2 82.5
Colorado Springs, CO 58.0 56.5 59.2 48.0 46.5 50.8
Aurora, CO 60.9 59.4 62.1 50.9 49.4 53.7
Denver, CO 61.6 60.1 62.8 51.6 50.1 54.4
Albuquerque, NM 63.2 61.7 64.4 53.2 51.7 56.0
El Paso, TX 69.4 67.9 70.6 59.4 57.9 62.2
Tucson, AZ 75.9 74.4 77.1 65.9 64.4 68.7
Las Vegas, NV 78.6 77.1 79.8 68.6 67.1 71.4
Mesa, AZ 82.7 81.2 83.9 72.7 71.2 75.5
Oklahoma City, OK 83.0 81.5 84.2 73.0 71.5 75.8
Phoenix, AZ 83.6 82.1 84.8 73.6 72.1 76.4
Omaha, NE 83.7 82.2 84.9 73.7 72.2 76.5
Atlanta, GA 83.8 82.3 85.0 73.8 72.3 76.6

PAO2 from the table before, less the expected gradient at that age: 10.0, 11.5 or 8.8 mmHg at 30; 20.0, 21.5 or 17.2 mmHg at 70, worked before rounding. Mellemgaard’s 2.5 + 0.21 × age is an average, and healthy people fall on both sides of it.

Why the estimates and the cities differ

In healthy lungs the gap between alveolar and arterial oxygen widens with age, and all three lines follow it. StatPearls’ current chapter on the gradient gives (age + 10) ÷ 4, the estimate the calculator reads a gradient against. Its archived chapter gives age ÷ 4 + 4, which has the same slope and sits 1.5 mmHg higher at every age. Both are bedside rules: neither chapter names a sample they were fitted to. Mellemgaard fitted 2.5 + 0.21 × age to 80 healthy seated subjects aged 15 to 75, so his line is their average and healthy people fall on both sides of it: Gold and Koth put the SD around his predicted PaO2 at about 6 mmHg. Against his line, (age + 10) ÷ 4 sits 0.8 mmHg higher at 20 and 3.6 higher at 90.

The alveolar gas equation takes 47 mmHg of water vapour off the barometric pressure, applies the 21% of oxygen to what is left, then subtracts PaCO2 ÷ 0.8. At sea level that gives a PiO2 of 149.7 mmHg and a PAO2 of 99.7. The standard pressure in Denver, 625.9 mmHg, gives a PAO2 of 71.6. Type 760 for a patient in Denver and you invent a gradient of about 28 mmHg in a healthy lung.

At the expected gradient, a 70-year-old in Denver has a PaO2 of 51.6 mmHg by (age + 10) ÷ 4, 50.1 by age ÷ 4 + 4 and 54.4 by Mellemgaard. At sea level the same person has 79.7 by (age + 10) ÷ 4, and a 30-year-old 89.7. People who live high breathe harder and run a lower PaCO2, 30 to 40 mmHg in Salt Lake City against 36 to 44 at sea level by Gold and Koth, so their real PAO2 sits above the table’s.

The formulas

Alveolar gas equation [1]
PAO2=FiO2 (Patm−47)−PaCO2RP_AO_2 = F_iO_2\,(P_{atm} - 47) - \dfrac{P_aCO_2}{R}
A-a gradient [1]
A-a=PAO2−PaO2\text{A-a} = P_AO_2 - P_aO_2
Expected gradient, (age + 10) ÷ 4 [1]
A-aexpected=age+104\text{A-a}_{\text{expected}} = \dfrac{\text{age} + 10}{4}
Expected gradient, age ÷ 4 + 4 [2]
A-aexpected=age4+4\text{A-a}_{\text{expected}} = \dfrac{\text{age}}{4} + 4
Expected gradient, Mellemgaard [5]
A-aexpected=2.5+0.21×age\text{A-a}_{\text{expected}} = 2.5 + 0.21 \times \text{age}

FiO2 is the inspired fraction, 0.21 on room air; Patm the barometric pressure; 47 mmHg the water vapour pressure at 37 °C; PaCO2 and PaO2 the arterial CO2 and oxygen; R the respiratory quotient, 0.8; age in years. Patm here is the 1976 US Standard Atmosphere at each city’s elevation, as the boiling point by city page gives it: 101.325 kPa, 760.0 mmHg, at sea level.

Where each formula holds

Each formula’s sample, the people it holds for, and where it does not apply
Formula Based on Holds for Not for
(age + 10) ÷ 4 No sample reported: StatPearls gives it with no reference beside it A quick bedside figure for an adult breathing room air, and the one the calculator reads a gradient against Supplemental oxygen, which widens the expected gradient (Dhont 2020 adds 50 × (FiO2 − 0.21) to age ÷ 4 + 4); altitude, which is no term in it
age ÷ 4 + 4 No sample reported: the archived StatPearls chapter cites Singh 2022, which, like Dhont 2020, uses it without one A quick bedside figure for an adult breathing room air, 1.5 mmHg more generous than (age + 10) ÷ 4 Supplemental oxygen, which widens the expected gradient (Dhont 2020 adds 50 × (FiO2 − 0.21)); altitude, which is no term in it
2.5 + 0.21 × age (Mellemgaard 1966) A regression on 80 healthy seated subjects aged 15 to 75, so their average, with an SD of about 6 mmHg around the predicted PaO2 (Gold and Koth) Healthy people aged 15 to 75, seated Ages over 75, marked † in the table; children under 15, younger than anyone in the sample; altitude, which is no term in it
Alveolar gas equation The clinical approximation of the alveolar air equation, with R of about 0.8 (Gold and Koth) Steady breathing, with PaCO2 standing in for alveolar CO2 Exact gas exchange work: the precise form, with its correcting factor for R, gives a slightly higher PAO2 on room air; a patient whose R is far from the 0.8 used here, or whose breathing or FiO2 has not settled
US Standard Atmosphere, 1976 An average atmosphere, as on the boiling point by city page The average pressure at each city’s USGS point Today’s weather, or a hospital above or below that point: enter the day’s barometric pressure in the calculator

Calculators and simulators

Sources

  1. Hantzidiamantis PJ, Amaro E. Physiology, Alveolar to Arterial Oxygen Gradient. StatPearls. Treasure Island (FL): StatPearls Publishing; updated . NCBI Bookshelf NBK545153, PMID 31424737. It prints PAO2 = (Patm − PH2O) FiO2 − PaCO2/RQ, and “The expected A-a gradient can be estimated with the following equation: A-a gradient = (Age + 10) / 4”. https://www.ncbi.nlm.nih.gov/books/NBK545153/, read
  2. Pesola GR, Sankari A. Oxygenation Status and Pulse Oximeter Analysis (archived). StatPearls. Treasure Island (FL): StatPearls Publishing; updated . NCBI Bookshelf NBK592401, PMID 37276303. “For older individuals, there are formulas to adjust for age to calculate the A-a difference, such as [A-a difference adjusted for age= A-a difference (age/4) + 4]”, citing Singh 2022. https://www.ncbi.nlm.nih.gov/books/NBK592401/, read
  3. Singh A, Soni KD, Singh Y, Aggarwal R, Venkateswaran V, Ashar MS, Trikha A. Alveolar arterial gradient and respiratory index in predicting the outcome of COVID-19 patients; a retrospective cross-sectional study. Archives of Academic Emergency Medicine 2022;10(1):e28. Methods: “The expected A-a oxygen gradient for age was derived using the following formula (Age/4) + 4”, with 760, 47 and 0.80 in the gas equation. doi:10.22037/aaem.v10i1.1543, read
  4. Dhont S, Derom E, Van Braeckel E, Depuydt P, Lambrecht BN. The pathophysiology of ‘happy’ hypoxemia in COVID-19. Respiratory Research 2020;21:198. It relates the gradient to age as age/4 + 4, plus 50 (FiO2 − 0.21) on oxygen. doi:10.1186/s12931-020-01462-5, read
  5. Mellemgaard K. The alveolar-arterial oxygen difference: its size and components in normal man. Acta Physiologica Scandinavica 1966;67(1):10 to 20. Not readable online; its regression and sample are as Gold and Koth report them. doi:10.1111/j.1748-1716.1966.tb03281.x
  6. Gold WM, Koth LL. Pulmonary function testing. In: Broaddus VC, Mason RJ, Ernst JD, et al., eds. Murray and Nadel’s Textbook of Respiratory Medicine, 6th edition. Philadelphia: Elsevier Saunders; 2016, chapter 25, 407 to 435. Mellemgaard’s 2.5 + 0.21 × age in 80 healthy seated subjects aged 15 to 75, an SD of about 6 mmHg, the alveolar air equation in its precise and clinical forms, and PaCO2 of 36 to 44 mmHg at sea level against 30 to 40 in Salt Lake City. doi:10.1016/B978-1-4557-3383-5.00025-7, read
  7. NOAA, NASA and U.S. Air Force (1976). U.S. Standard Atmosphere, 1976. NASA TM-X-74335. https://ntrs.nasa.gov/citations/19770009539
  8. U.S. Geological Survey, National Geospatial Program (2026). Geographic Names Information System, Federal Codes file, FedCodes_AllStates_Text.zip (published ): each city’s populated place and its official location, NAD83. https://prd-tnm.s3.amazonaws.com/StagedProducts/GeographicNames/FederalCodes/FedCodes_AllStates_Text.zip, read
  9. U.S. Geological Survey. Elevation Point Query Service, version 1: ground elevation from the 3D Elevation Program at a point, in feet or meters. https://epqs.nationalmap.gov/v1/docs, read
  10. U.S. Census Bureau, Population Division (2026). Annual Estimates of the Resident Population for Incorporated Places of 20,000 or More, Ranked by July 1, 2025 Population (SUB-IP-EST2025-ANNRNK), and the dataset sub-est2025.csv. https://www.census.gov/data/tables/time-series/demo/popest/2020s-total-cities-and-towns.html, read

Updated

The values are facts and free to use; this table’s selection and presentation are © 2026 ScienceQuest.