Arterial blood gas and acid-base result sheets

An arterial blood gas (ABG) and acid-base result sheet is not a single number. It is a small laboratory report that lists how acid or alkaline the blood is (pH), how much carbon dioxide it carries (PaCO2), how much bicarbonate it contains (HCO3-), how much oxygen it contains (PaO2), how much base is in excess or deficit (base excess), and often a lactate. Each result is printed beside its own reference interval and a flag, much like any other blood test. This page is about the result sheet itself: what each line means, the intervals on it, the patterns the lines form, reading a sheet step by step, and the mistakes that make a true gas look false. It builds on the topics on ABG interpretation and acid-base disorders without repeating them in full; use those pages for the physiology behind the numbers. The free NCBI Bookshelf text Clinical Methods (Walker, Hall and Hurst) was consulted alongside StatPearls for sampling and interpretation concepts. ^{[6] [3]}

A "result sheet" looks ordinary because the machine hides the algorithm. Behind the scenes the analyser directly measures pH and PaCO2 and calculates bicarbonate, base excess and the derived oxygenation indices from those two measurements and from haemoglobin, temperature and the inspired oxygen fraction. Serum chemistry, by contrast, measures total carbon dioxide content directly. The two can therefore differ by roughly 1.2 mmol/L, and a larger gap in a critically ill patient should prompt a repeat rather than an argument. ^[3]

Three points matter before any detail. First, the printed interval is the right one to use, because intervals vary by analyser, method, altitude, age and sex, and only the laboratory that ran the test knows its own. Second, the pH names the state of the blood (acidemia or alkalemia), while the processes that produce it are acidosis or alkalosis; compensation never returns the pH fully to normal and never overshoots, so an "over-corrected" pH tells you two processes are at work. ^[4] Third, no flag is a diagnosis on its own: read the whole sheet, then read the patient.

Each line of the report measures something different, and mixing them up is the most common beginner error.

• pH is the acidity of arterial blood. The analyser measures it. A low pH means acidemia; a high pH means alkalemia.
• PaCO2 is the arterial partial pressure of carbon dioxide, in mm Hg (or kPa). It is the respiratory part of the acid-base pair: high with hypoventilation, low with hyperventilation. Carbon dioxide is produced by tissues, carried to the lungs and exhaled; it is therefore also the best single number for the adequacy of ventilation.
• HCO3- (bicarbonate) is the metabolic part. On the gas report it is calculated from pH and PaCO2; on the serum chemistry panel it is measured directly. The Report's HCO3- window is the one generated with the gas.
• Base excess (BE), also called base deficit when negative, is the amount of base that would have to be added to, or removed from, a litre of blood to bring it to pH 7.40 at a PaCO2 of 40 mm Hg. It is a calculated quantity. A value near zero means no net excess or deficit of base; positive means an alkalotic load; negative means an acidotic load.
• PaO2 is the arterial partial pressure of oxygen. It measures oxygenation, not ventilation. Below a certain level it is defined as hypoxaemia.
• SaO2 is the arterial oxygen saturation calculated by the analyser from PaO2 and pH using the oxyhaemoglobin dissociation curve. It should agree with a simultaneous pulse oximetry reading (SpO2) when perfusion is good, but they are not the same measurement.
• Lactate is the end product of anaerobic glycolysis. It rises when tissue perfusion or oxygen use fails (shock, seizure, severe asthma, tissue ischaemia), and also with some drugs, toxins and hepatic failure. The analyser in your blood gas machine reports it in mmol/L; the clinical laboratory may use the same specimen or a different one.
• FiO2 is the fraction of oxygen in the inspired gas (0.21 on room air). It is not measured by the blood; it is written on the chart as the oxygen prescription, and it turns PaO2 into the more useful PaO2/FiO2 ratio.

Taken together, the first three (pH, PaCO2, HCO3-) usually settle the acid-base diagnosis; the next two (PaO2, SaO2) settle oxygenation; the last two (base excess, lactate) support the metabolic side. ^[1][3]

The PaO2 line does more than say "low". It is on this report because oxygenation can be assessed non-invasively with pulse oximetry and end-tidal carbon dioxide monitoring, but arterial blood gas analysis remains the gold standard. ^[3]

A result sheet rewards reading in the same order every time. The order of the analytes on the paper (pH, then PaCO2, then HCO3-, then oxygenation) is already a second-best order; base excess and lactate belong after bicarbonate, and oxygenation should not be left to last when the patient is breathless.

1. pH. Is the blood acidemic (below 7.35) or alkalemic (above 7.45), or within the normal interval? The pH establishes the primary process only if it is abnormal; remember that compensation wanes as the pH nears normal, so a normal pH is not proof of normal acid-base balance. ^{[4] [2]}
2. PaCO2. This is the respiratory half of the acid-base pair and the number for ventilation. Compare it with its own interval: low with hyperventilation, high with hypoventilation. If the pH is low and the PaCO2 is high, the primary process is respiratory acidosis. ^[2][4]
3. HCO3-. This is the metabolic half. A low bicarbonate with a low pH means metabolic acidosis; a high bicarbonate with a high pH means metabolic alkalosis. If the pH and one of the pair move together, that is the primary process; the other member, moving in the same direction, is compensation. ^[4]
4. Base excess, anion gap and delta gap. Base excess turns the directional impression into a quantity of acid or alkali. The anion gap then asks whether unmeasured acids have appeared, and the delta gap or delta ratio (the ratio of the rise in anion gap to the fall in bicarbonate, expressed as the delta gap added to the measured bicarbonate) asks whether a second metabolic process is hiding behind the first. A corrected bicarbonate that sits above the normal interval means a hidden metabolic alkalosis. ^[4]
5. PaO2 and SaO2. These are the oxygenation lines, and they are read against the inspired oxygen. On room air a PaO2 of 75 to 100 mm Hg (10.0 to 13.3 kPa) and an SaO2 of 94 to 100 per cent are expected. Hypoxaemia is a PaO2 below 60 mm Hg (8 kPa), and PaO2 over about 16 kPa (120 mm Hg) raises the saturation no further. ^[1][2]
6. Lactate. A high lactate points to tissue hypoperfusion, hypoxia or mitochondrial poisoning; a normal value in a shocked patient is reassuring but not decisive. Read it in the light of the clinical picture and the anion gap; an unexplained anion gap with a normal lactate should send you back to ketoacids and toxins. ^[4]

This order is deliberately slow. Most diagnostic errors on a gas come from glancing at the pH, deciding "acidosis", and stopping. Reading all six lines takes under a minute and prevents almost all of those errors.

Table 1 uses the intervals that the commonly cited free sources give for adults at sea level. They are a starting point and teaching values; your laboratory's printed interval is the one that matters, because intervals vary by method and population, and many analytes are method-dependent. ^[1]

Three details sit underneath this table.

PaCO2 changes with the geography. The BTS manual puts the normal arterial carbon dioxide tension at 4.5 to 6.0 kPa (34 to 45 mm Hg), and hypercapnia starts above 4.6 to 6.1 kPa (34 to 46 mm Hg), depending on which section and population is quoted. That small spread is the difference between "normal on my analyser" and "normal by guideline", and it is exactly why the report's own interval wins. ^[2]

PaO2 changes with altitude. At 3,000 feet (900 metres) and higher the oxygen value is lower. A result that would be flagged low on a coastal report can be expected at altitude, and comparing it with a textbook range written for sea level is a mistake. ^[1]

Base excess is not a measured analyte. It is a derived value, and its nominal interval is not printed by every analyser. Where it is reported, interpret the direction first (positive is a surplus of base, negative a deficit) and the exact number against the method used. ^[3]

The intervals themselves disagree between sources. MedlinePlus quotes arterial pH as 7.38 to 7.42, while its own ABG lab test and LITFL give 7.35 to 7.45, and the Merck Manual defines acidemia below 7.35 and alkalemia above 7.45. The same split runs through PaCO2 (38 to 42 mm Hg in the MedlinePlus encyclopedia against 35 to 45 mm Hg, 4.5 to 6.0 kPa in the BTS, and 4.7 to 6.0 kPa in standard sources) and bicarbonate (22 to 28 mmol/L in the encyclopedia, 22 to 26 in the lab test and LITFL, 22 to 29 in the Mayo catalog). These are different intervals for the same analyte, not different truths, and they are the reason a flag can fall on one report and inside the interval on another. Use the interval printed on the report that ran the test. ^{[1] [2] [4] [5]}

SaO2 intervals. The MedlinePlus encyclopedia gives 94 to 100 per cent and its lab test 95 to 100 per cent; both sit above the hypoxaemic threshold of 90 per cent used by the BTS. ^{[1] [2]}

The thresholds follow the interval in use. The Merck Manual defines metabolic acidosis as serum bicarbonate below 24 mmol/L and metabolic alkalosis as above 28 mmol/L, with respiratory acidosis above 40 mm Hg and respiratory alkalosis below 38 mm Hg; sources using the 35 to 45 mm Hg PaCO2 interval and 22 to 26 mmol/L bicarbonate interval call them abnormal beyond 45 and 35 mm Hg and below 22 and above 26 mmol/L. Hypercapnia is defined above 46 mm Hg (6.1 kPa) by the BTS guideline and above 45 mm Hg (6 kPa) in many texts. ^{[4] [3]}

The anion gap varies in the same way. Merck gives a typical value of 12 mEq/L and the albumin correction of 2.5 mEq/L per 10 g/L; Mayo lists 7 to 15 for people aged 7 years and over and LITFL quotes 12 plus or minus 4. ^{[4] [5]}

The delta gap is tested against the local normal bicarbonate interval, which may be 22 to 26, 22 to 28 or 22 to 29 mmol/L, so an 'abnormal' corrected bicarbonate shifts slightly with the interval used (as in the worked example, where the corrected bicarbonate of 37 sits above normal 22 to 29). ^{[4] [1]}

Table 2 summarises the common directional causes the free sources support. It is not exhaustive, and every cause must be read against the patient, the medication list and the clinical picture.

For the metabolic acidosis patterns themselves, the free Merck Manual account divides a raised anion gap into four leading causes: ketoacidosis, lactic acidosis, renal failure and toxic ingestions. A normal anion gap with a low bicarbonate and a raised chloride points instead to bicarbonate loss from the gut or kidney or to infusion of chloride-rich fluid. ^[4]

The high-versus-low frame also applies to oxygenation itself. The BTS guideline defines hypoxaemia as a PaO2 under 60 mm Hg (8 kPa), or an SaO2 under 90 per cent, and treats values above about 16 kPa (120 mm Hg) as hyperoxaemia, where saturation no longer changes even though a higher PaO2 still can cause harm. ^[2]

Once the six lines are read as a set, the combinations are recognisable. These are the patterns the sheet makes; they are the bridge between the piece of paper and the diagnosis on the acid-base disorders page.

Normal gas. pH, PaCO2, HCO3- and PaO2 all lie inside their intervals and the anion gap sits in range. This is the commonest sheet in a healthy person and the one to which every abnormal sheet is compared. It says nothing about rare disease, and it does not exclude a process compensated to within the normal interval.

Uncompensated metabolic acidosis. Low pH, low HCO3-, normal or low PaCO2, often a negative base excess. The respiratory system has not had time to compensate, or another process is overriding it. A low PaCO2 with a low pH is usually a compensatory hyperventilation, not a primary respiratory alkalosis. ^[4]

Compensated metabolic acidosis. pH may be near normal while HCO3- is low and PaCO2 is low. Winter's formula, PaCO2 = 1.5 x HCO3- + 8 (plus or minus 2), is the usual test of whether compensation is appropriate. ^[4] If the measured PaCO2 matches, the process is simple. If it does not, a second process is present.

Respiratory acidosis. Low pH, raised PaCO2. Acutely there is little rise in HCO3- (1 to 2 mmol/L for each 10 mm Hg rise in PaCO2); chronically the kidney raises HCO3- by 3 to 4 mmol/L for each 10 mm Hg rise. ^[4] A patient with chronic lung disease can have a near-normal pH on a markedly raised PaCO2 and a raised HCO3-, and a small further fall in pH signals an acute exacerbation on top of the chronic state.

Respiratory alkalosis. High pH, low PaCO2; HCO3- falls by 1 to 2 mmol/L acutely and 4 to 5 mmol/L chronically for each 10 mm Hg fall in PaCO2. ^[4]

Metabolic alkalosis. High pH, high HCO3-; PaCO2 rises modestly (0.6 to 0.75 mm Hg for each 1 mmol/L rise in HCO3-, and should not exceed about 55 mm Hg). ^[4]

Anion gap metabolic acidosis. A raised anion gap on top of a low HCO3- means unmeasured acids are present: ketoacids, lactate, renal failure acids, methanol, ethylene glycol or salicylate. The delta gap (or its ratio form, the delta ratio) tests for a second, hidden metabolic alkalosis: the anion gap rises by the same amount as the bicarbonate falls only if unmeasured acids are the only process, so the delta gap added to the measured bicarbonate should land inside the normal bicarbonate interval. If it lands above it, a metabolic alkalosis is also present, as in the vomiting drinker whose anion gap is 25 with a bicarbonate of 22, giving a corrected bicarbonate of 37. ^[4]

Normal anion gap (hyperchloraemic) metabolic acidosis. Low HCO3- with a normal anion gap and a raised chloride: bicarbonate has been lost from the gut or kidney, or chloride has been infused. This is the sheet of severe diarrhoea and of the renal tubular acidoses; the urine anion gap separates the two. ^[4]

Mixed disorder. Compensation cannot fully correct pH and never overshoots. When a calculated compensation predicts one pH and the actual pH is further in the same direction, or on the other side, a second primary process is in play. Two separate processes can cancel each other and leave a normal pH, which is why "normal pH" is never the same as "normal acid-base balance". ^[4]

Hypoxaemic sheet. A low PaO2 (under 60 mm Hg, or 8 kPa) with a normal or low PaCO2 is type 1 respiratory failure, from V/Q mismatch, shunt, diffusion limitation or hypoventilation. A low PaO2 with a raised PaCO2 is type 2 respiratory failure, usually from hypoventilation or advanced airways disease. ^[2]

Oxygenation sheet. PaO2 is usually read with the inspired oxygen. The PaO2/FiO2 ratio is PaO2 divided by the fraction of inspired oxygen: on room air (FiO2 0.21) a PaO2 of 63 mm Hg gives a ratio of about 300, and a PaO2 of 63 mm Hg on 40 per cent oxygen (FiO2 0.4) gives about 158. A lower ratio means worse oxygenation for the amount of oxygen being given. The alveolar-to-arterial (A-a) gradient is the difference between the oxygen tension in the alveolus and that in arterial blood; it rises when oxygen is not reaching the blood, as in V/Q mismatch, shunt or diffusion limitation, and it is largely unchanged when the only problem is pure hypoventilation. ^[2] For an A-a calculation the delivered FiO2 must be known exactly, which is one reason nasal cannulae make it hard to calculate (the delivered fraction is not predictable). ^[2]

The three sheets below are invented for teaching, are labelled as such in every line, and are not real patient data. They use the intervals in Table 1. Each question after each sheet is answered from the recorded numbers, not from the patient.

1. pH, PaCO2 and HCO3- are all inside their intervals, so there is no acidemia and no alkalemia and no acidosis or alkalosis to name. ^[4]
2. The PaCO2 of 39 mm Hg matches a normal PaCO2, so ventilation is adequate and there is no hypoventilation or hyperventilation. ^[2]
3. PaO2 92 mm Hg and SaO2 97 per cent are inside the interval, so oxygenation is satisfactory on room air; the PaO2/FiO2 ratio is 92 divided by 0.21, or about 438. ^[1][2]
4. Base excess 0 and anion gap 11 sit inside their intervals, and lactate is unremarkable, so there is no hidden metabolic process. ^[4]

One-sentence conclusion and next step: this is a normal arterial blood gas on room air, and no further acid-base investigation is needed unless the clinical picture changes.

1. pH 7.28 is below 7.35: acidemia. ^[2]
2. HCO3- 10 mmol/L is below 22 mmol/L, so the primary process is metabolic acidosis. ^[4]
3. PaCO2 22 mm Hg is low. Winter's formula gives an expected PaCO2 of about 1.5 x 10 + 8 = 23 mm Hg (plus or minus 2), so the low PaCO2 is the appropriate respiratory compensation, not a second disorder. ^[4]
4. Anion gap 22 is above the typical value of 12, so this is a high anion gap metabolic acidosis; the unmeasured acids in this setting are the ketoacids of diabetic ketoacidosis. ^[4]
5. Delta gap = 22 minus 10 = 12, and corrected HCO3- = 12 plus 10 = 22, which sits at the lower limit of normal; no hidden metabolic alkalosis is present. ^[4]
6. PaO2 104 mm Hg on 28 per cent oxygen (FiO2 0.28) gives a PaO2/FiO2 ratio of about 371, and SaO2 is 98 per cent, so oxygenation is currently adequate. ^[1][2]

One-sentence conclusion and next step: high anion gap metabolic acidosis (consistent with diabetic ketoacidosis) with appropriate respiratory compensation and preserved oxygenation, and the next step is treatment of the underlying acidosis with fluids and insulin according to the relevant protocol while repeating the gas to track the pH.

1. pH 7.31 is below 7.35: acidemia. ^[2]
2. PaCO2 62 mm Hg is above 42 mm Hg and above the hypercapnia threshold, so the primary process is respiratory acidosis; the kidney has had time to compensate, shown by a HCO3- of 31 mmol/L. ^[2][4]
3. PaO2 56 mm Hg is below 60 mm Hg (8 kPa) and SaO2 is 87 per cent, so there is also hypoxaemia; combined with a raised PaCO2 this is type 2 respiratory failure rather than type 1. ^[2]
4. HCO3- 31 mmol/L is about 7 mmol/L above the midpoint of normal; a rise of about 3 to 4 mmol/L per 10 mm Hg rise in PaCO2 is what chronic compensation produces, so this looks like an acute-on-chronic picture rather than a pure acute rise. ^[4]
5. The anion gap is normal and lactate is unremarkable, so a metabolic acidosis is not part of this picture. ^[4]
6. PaO2/FiO2 on 28 per cent oxygen is about 200, and the patient is hypoxaemic; with a PaCO2 this high, controlled oxygen (target saturation 88 to 92 per cent pending further gases) is preferred to high-concentration oxygen. ^[2]

One-sentence conclusion and next step: acute-on-chronic type 2 respiratory failure with hypoxaemia, and the next step is targeted low-concentration oxygen with repeat blood gases, bronchodilator therapy and reassessment for non-invasive ventilation if pH falls further.

A sheet can mislead even when the machine is perfect. These are the traps that produce a result that is true of the tube but not true of the patient.

1. Venous contamination or a fully venous sample. Venous blood has a lower PaO2, a higher PaCO2 and a slightly lower pH than arterial blood. In practice, venous PCO2 runs about 0.77 kPa (5.8 mm Hg) higher than arterial PCO2. A venous PCO2 below 6 kPa (45 mm Hg) with a good oxygen saturation makes significant hypercapnia unlikely and can spare an arterial puncture, but venous blood should never be used to judge oxygenation or a borderline acidosis. ^[2]
2. Capillary (earlobe) samples. These give accurate pH and PaCO2 but underestimate PaO2 by 0.5 to 1 kPa, so oximetry must be watched and an arterial specimen taken if the sample is doubtful. They should be used for the first gas only in shock or hypotension; in those settings take arterial blood. ^[2]
3. Air bubbles, delay and temperature. A sample left with an air bubble, or not analysed promptly, equilibrates with room air and drifts. Keep the sample on ice and run it within minutes, especially when oxygenation matters.
4. Heparin dilution and wrong tubes. Excess liquid heparin dilutes the sample and lowers PaCO2 and HCO3- slightly. Use the syringe's own dry heparin and mix it thoroughly but gently.
5. Oxygen therapy not recorded. The PaO2 cannot be interpreted without the FiO2, and the A-a gradient and PaO2/FiO2 ratio cannot be calculated without it. Write the oxygen device and flow rate on the request and on the chart. ^[2]
6. Calculated versus measured HCO3-. The gas analyser calculates HCO3- from pH and PaCO2; serum chemistry measures it directly. When they disagree, the directly measured value is considered more accurate. ^[4]
7. Altitude and age. PaO2 falls at altitude and many intervals shift with age and sex. Compare with the right expectation for the patient, not with a coastal 20-year-old. ^[1]
8. Over-reading a calculated base excess. Base excess is derived, and its interval is method-dependent; use the direction first and do not hang a precise diagnostic weight on the last digit. ^[3]
9. Treating the flag instead of the patient. The pH defines the state of the blood, not the disease. Compensation never fully corrects pH and never overshoots, so a normal pH can hide two opposing processes; check the anion gap and the corrected bicarbonate before reassuring. ^[4]

• An arterial blood gas result sheet reports pH, PaCO2, HCO3-, PaO2, base excess, SaO2 and often lactate, each beside its own interval and flag; no single line is the diagnosis. ^[1]
• The interval printed by the laboratory that ran the test is the one that counts; altitude, age, sex and method all move the numbers. ^[1]
• Acidemia is pH below 7.35 and alkalemia is pH above 7.45; acidosis and alkalosis are processes, and compensation never fully corrects the pH and never overshoots. ^[4]
• PaCO2 reflects ventilation, HCO3- the metabolic part, and PaO2 oxygenation; the machine calculates HCO3- and base excess from measured pH and PaCO2. ^[3][4]
• Winter's formula (PaCO2 = 1.5 x HCO3- + 8, plus or minus 2), the chronic respiratory compensations (3 to 4 mmol/L per 10 mm Hg in respiratory acidosis; 4 to 5 mmol/L per 10 mm Hg in respiratory alkalosis) and the metabolic alkalosis rule (0.6 to 0.75 mm Hg PaCO2 per 1 mmol/L HCO3-) are the standard compensation checks. ^[4]
• A raised anion gap means unmeasured acids are present; the delta gap added to the bicarbonate should land in the normal bicarbonate interval, and a higher value reveals a hidden metabolic alkalosis. ^[4]
• Hypoxaemia is a PaO2 below 60 mm Hg (8 kPa), or an SaO2 below 90 per cent; hyperoxaemia begins above about 16 kPa (120 mm Hg), where saturation stops rising. ^[2]
• Type 1 respiratory failure is a low PaO2 with normal or low PaCO2; type 2 is a raised PaCO2, by definition above 4.6 to 6.1 kPa, and can occur with normal oxygen saturation. ^[2]
• A venous PCO2 is about 0.77 kPa (5.8 mm Hg) higher than arterial, and a value below 6 kPa (45 mm Hg) excludes significant hypercapnia; venous blood cannot be used for oxygenation or for a borderline acidosis. ^[2]
• Capillary (earlobe) gases give accurate pH and PaCO2 but underestimate PaO2 by 0.5 to 1 kPa. ^[2]
• Always record the oxygen device and flow rate: without FiO2 the PaO2/FiO2 ratio, the A-a gradient and the oxygenation itself cannot be judged. ^[2]
• Related pages on MedGrasp (plain text, linked when published): acid-base disorders and renal tubular acidosis; ABG interpretation; common laboratory reference values; oxygen therapy; sepsis and shock.