Oxygen carriage

What it normally does

• Nearly all oxygen in blood travels bound to haemoglobin. Each gram of haemoglobin carries about 1.34 mL of oxygen, and each haemoglobin molecule holds four oxygen molecules on four iron-containing haem groups. Only about 3 mL per litre is dissolved free in the plasma at a normal PaO2 (dissolved oxygen), which is almost nothing.

  how much oxygen the blood carries (oxygen content) depends on how much haemoglobin there is, not just on how well the lung works — so a severely anaemic patient can have a perfect PaO2, perfect saturations, and starving tissues.

• Binding is cooperative: the first oxygen molecule changes the shape of the molecule and makes the next one bind more easily. So saturation plotted against oxygen tension is S-shaped (the sigmoid dissociation curve) — flat above a PaO2 of about 60 mmHg, then falling away steeply below it.

  saturation barely moves while the lung is quietly failing, and then collapses. A normal SpO2 is weak reassurance, and 90% is not 'a bit low' — it is the top of the cliff.

• Acid and carbon dioxide shift the curve to the right, so haemoglobin releases oxygen more readily exactly where tissue is acidic and working — that is the Bohr effect. Heat and 2,3-BPG shift the curve right as well, but by separate mechanisms rather than as part of the Bohr effect. The relationship runs the other way too: loading haemoglobin with oxygen lowers its affinity for carbon dioxide and H+, so it lets go of them (Haldane effect).

  a working muscle extracts more oxygen without the lung doing anything different — and, in the opposite direction, giving a chronic CO2 retainer too much oxygen pushes their carbon dioxide up.

• A pulse oximeter shines red (660 nm) and infrared (940 nm) light through the pulsing part of the signal and compares how much of each is absorbed. It was calibrated on healthy volunteers against two species only — oxyhaemoglobin and deoxyhaemoglobin — and it assumes nothing else is present and that there is a clean pulse to read.

  anything else stuck to haem, or any loss of pulsatile flow, makes the device confidently wrong rather than obviously broken.

What goes wrong

• Iron deficiency anaemia← from “Nearly all oxygen in blood travels bound to ha…”

  No iron means ferrochelatase cannot insert iron into protoporphyrin IX, so haem cannot be made. The erythroblast undergoes extra divisions while haemoglobin accumulates too slowly, and you end up with small pale cells (microcytic hypochromic) carrying less haemoglobin each. Total oxygen content per litre falls. The lung is untouched, so the blood leaving it is fully saturated — there is simply less to saturate. The body compensates by raising cardiac output and by making more 2,3-BPG, which shifts the curve right so the reduced load is unloaded more completely.

  Anaemia lowers oxygen CONTENT, not saturation or PaO2. Normal sats do not exclude it. In a man or postmenopausal woman, assume gastrointestinal blood loss.

  You would find: Tiredness, breathlessness on exertion, pale conjunctivae, a resting tachycardia and sometimes a soft flow murmur. Saturations 98% on air and a normal PaO2 — that normality is the teaching point, not reassurance. Blood film microcytic and hypochromic, ferritin low (though ferritin is an acute phase reactant and can be normal or high alongside iron deficiency in inflammation). In Australia this is commonest in menstruating and pregnant women, and falls disproportionately on Aboriginal and Torres Strait Islander communities, where low dietary iron and, in some remote communities, hookworm and Helicobacter pylori stack together, and childhood anaemia rates run well above the national figure.

• Carbon monoxide poisoning← from “A pulse oximeter shines red (660 nm) and infra…”

  CO binds haem with roughly 200 to 250 times the affinity of oxygen. Those sites now carry nothing useful, and CO on one subunit left-shifts the remaining three, so the little oxygen still bound is held on rather than released to tissue. Two hits: less carried, and what is carried will not come off. CO also binds cytochrome c oxidase, so the mitochondria cannot fully use what does arrive. And because carboxyhaemoglobin absorbs 660 nm light much like oxyhaemoglobin, the pulse oximeter counts it as oxygenated and reads high.

  Normal PaO2, high SpO2, dying patient. The oximeter cannot see carboxyhaemoglobin. Measure it.

  You would find: Winter, a faulty unflued gas heater, a house fire, or a generator run in a shed — and often more than one person in the house with headache, nausea and confusion. SpO2 99% alongside a lactic acidosis and a patient who looks unwell. PaO2 is normal too, because dissolved oxygen is unaffected. Cherry-red skin is a late and unreliable sign; do not wait for it. Diagnosis is co-oximetry on a blood gas, which measures carboxyhaemoglobin directly.

• Methaemoglobinaemia← from “A pulse oximeter shines red (660 nm) and infra…”

  An oxidising drug pushes haem iron from Fe2+ to Fe3+. Ferric haem cannot bind oxygen at all, and it left-shifts the normal subunits sitting beside it in the same molecule, so unloading fails as well. Methaemoglobin absorbs the red and infrared wavelengths almost equally, which drives the oximeter's ratio towards 1 — and a ratio of 1 is calibrated to read about 85%. The reading parks there and barely moves whatever oxygen you give.

  SpO2 fixed near 85%, cyanosis unresponsive to oxygen, chocolate-brown blood. Ask what oxidant they took.

  You would find: A blue patient who does not improve on oxygen, with SpO2 stuck around 85%. A saturation gap between the oximeter and the saturation calculated on the blood gas. Blood drawn is chocolate-brown and stays brown when shaken in air. Look for the trigger: dapsone, topical prilocaine or benzocaine spray, nitrites, some antimalarials.

• Hypoxaemia hidden by the flat top of the curve← from “Binding is cooperative: the first oxygen molec…”

  Between a PaO2 of 100 and 70 mmHg, saturation only slides from about 98% to 94%. So a pneumonia can wipe out nearly a third of the oxygen tension while the finger probe barely twitches. By the time SpO2 reads 90%, PaO2 is around 60 mmHg and the patient has arrived at the shoulder of the curve — from there a small further fall in tension causes a large fall in saturation. The reading was flattering, then it falls off a cliff.

  Flat above 60 mmHg, steep below. SpO2 90% is not 'nearly normal' — it is the edge. Count the respiratory rate.

  You would find: Respiratory rate rises before the saturation does — it is the earlier and better warning, and the one most often not counted. Watch also for a pigmented-skin over-read: oximeters tend to read a few points high in people with darker skin, hiding hypoxaemia (occult hypoxaemia). And any loss of a clean pulsatile signal — shock, cold peripheries, vasopressors, motion, and to a smaller and more variable degree dark nail polish — makes the number unreliable rather than obviously broken.

• Oxygen-induced hypercapnia in chronic CO2 retention← from “Acid and carbon dioxide shift the curve to the…”

  Three things happen when you flood a COPD patient with oxygen. The largest contributor is worsened ventilation–perfusion matching: high inspired oxygen releases hypoxic pulmonary vasoconstriction, so blood returns to poorly ventilated lung and physiological dead space rises. Second, deoxygenated haemoglobin is the form that carries CO2 as carbamino compounds and buffers H+; saturate it fully and it shifts that CO2 into the plasma (Haldane effect). A smaller contribution comes from blunted hypoxic respiratory drive. PaCO2 climbs, pH falls, and the patient goes quiet.

  Too much oxygen in a retainer raises CO2, mostly by worsened V/Q matching, with the Haldane effect second and reduced drive a minor player. Target 88 to 92%.

  You would find: Known COPD or obesity hypoventilation, given 15 L via a mask in the ambulance, arriving drowsy with a flapping tremor and a bounding pulse. Gas shows a respiratory acidosis — pH 7.25, PaCO2 80 mmHg — with a beautiful saturation of 100%. Australian practice (TSANZ acute oxygen guidance) targets 88 to 92% in patients at risk of hypercapnic respiratory failure for exactly this reason.

What we give, and how it works

Open a drug to see what it binds, what that does to the cell, and what you then see in the patient.