ICU · respiratory
Oxygen Therapy and Oxygen Delivery — Comprehensive ICU Physiology
Also known as Oxygen therapy · Oxygen delivery · DO2 · VO2 · Oxygen cascade · Alveolar gas equation · Oxygen content · High-flow nasal cannula · Oxygen extraction ratio
Oxygen therapy and delivery — the physiological principles governing oxygen movement from the atmosphere to the cell, and the ICU interventions to optimise it. The oxygen cascade: atmospheric PO2 (159 mmHg at FiO2 21%) → tracheal PO2 (149, after humidification) → alveolar PO2 (PAO2 ~100, after CO2 mixing — calculated by the ALVEOLAR GAS EQUATION: PAO2 = FiO2(Patm - PH2O) - PaCO2/RQ) → arterial PO2 (PaO2 ~95, after shunt/VQ mismatch) → capillary PO2 (40, after tissue extraction) → mitochondrial PO2 (1-3, for oxidative phosphorylation). Oxygen content: CaO2 = (1.34 × Hb × SaO2) + (0.003 × PaO2) — the HAEMOGLOBIN component dominates (99% of total O2 content — dissolved O2 is negligible at normal pressures). Oxygen delivery: DO2 = CO × CaO2 × 10 — normal ~1000 mL/min. Oxygen consumption: VO2 = CO × (CaO2 - CvO2) × 10 — normal ~250 mL/min. O2 extraction ratio (ER) = VO2/DO2 — normal ~25% (the body extracts 25% of delivered O2 at rest). ICU management of hypoxaemia: increase FiO2 → increase Hb (transfuse) → increase CO (inotrope/fluid) → reduce shunt (PEEP, proning, recruit alveoli). Hyperoxia is HARMFUL (ROS, absorption atelectasis, coronary vasoconstriction) — titrate FiO2 to lowest setting maintaining SpO2 92-96%.
On this page & tools
Your progress
Saved locally on this device.
Target exams
Red flags
Overview
The oxygen cascade — from atmosphere to mitochondrion
The oxygen cascade — each step reduces PO2
| Location | PO2 (mmHg) | Mechanism of reduction |
|---|---|---|
| Atmosphere | 159 | FiO2 21% × Patm 760 mmHg |
| Trachea | 149 | Humidification: subtract PH2O 47 mmHg → FiO2 × (760-47) |
| Alveolus (PAO2) | ~100 | CO2 mixing: subtract PaCO2/RQ (40/0.8 = 50) → the ALVEOLAR GAS EQUATION |
| Artery (PaO2) | ~95 | Shunt + V/Q mismatch: A-a gradient = PAO2 - PaO2 (normal <15 mmHg in young, <25 in elderly) |
| Capillary | ~40 | Tissue oxygen extraction — O2 diffuses from capillary to mitochondrion |
| Mitochondrion | 1-3 | Critical for oxidative phosphorylation — if <1 → anaerobic metabolism → lactate |
| Mixed venous (PvO2) | ~40 | SvO2 ~75% — reflects the balance between DO2 and VO2 |
The oxygen-haemoglobin dissociation curve
Regions and landmarks of the oxyhaemoglobin dissociation curve
| Region / landmark | PO2 (mmHg) | Saturation | Significance |
|---|---|---|---|
| Plateau (top) | 60-100 | 90-100% | Above PaO2 60, large changes in PO2 cause tiny changes in SaO2 — a SAFETY MARGIN. This is why SpO2 is an INSENSITIVE early warning: it stays >90% down to PaO2 60, then falls off the cliff |
| P50 | 27 mmHg (37 °C, pH 7.40) | 50% | The affinity benchmark. LOWER P50 = higher affinity (curve LEFT). HIGHER P50 = lower affinity (curve RIGHT) |
| Steep portion | 20-60 | 40-90% | The WORKING range in tissue capillaries. Small drop in PO2 releases large O2 — efficient unloading. Mixed venous point (PvO2 40, SvO2 75%) lives here |
| Mixed venous point | 40 | 75% | Tissues extract ~25% of O2; the remaining 75% is the venous reserve |
P50 — the single-number summary of Hb affinity
| Condition | P50 (mmHg) | Direction | Cause | Effect |
|---|---|---|---|---|
| Normal (pH 7.40, 37 °C) | 27 | — | — | Balanced uptake and release |
| Acidosis / hypercapnia | >27 | RIGHT | Bohr effect | Better unloading |
| Alkalosis / hypocapnia | <27 | LEFT | Bohr effect | Poorer unloading |
| Fever / exercise | >27 | RIGHT | ↑Temperature | Unloading to muscle |
| Stored blood (banked RBC) | <27 | LEFT | ↓2,3-BPG | Poor unloading for ~24 h |
| CO poisoning | <27 | LEFT | Carboxyhaemoglobin | Uptake AND release both impaired |
| Methaemoglobinaemia | <27 | LEFT | Fe³⁺ Hb | Cannot bind + left shift |
| Chronic altitude / anaemia | >27 | RIGHT | ↑2,3-BPG | Compensatory unloading |
| HbF (fetal) | ~19 | LEFT | γ-chains, low 2,3-BPG binding | Placental O2 theft |
Oxygen content, delivery, and consumption — the key equations
Oxygen transport equations — the critical calculations
| Parameter | Equation | Normal value | Clinical significance |
|---|---|---|---|
| Oxygen content (CaO2) | (1.34 × Hb × SaO2) + (0.003 × PaO2) | 20 mL O2/dL blood | The Hb component is >99%. Dissolved O2 (0.003 × PaO2) is negligible at normal pressure. Increasing Hb is the MOST effective way to increase CaO2 |
| Arterial O2 delivery (DO2) | CO × CaO2 × 10 | ~1000 mL/min | The total O2 delivered to tissues per minute. Depends on BOTH cardiac output AND arterial O2 content |
| O2 consumption (VO2) | CO × (CaO2 - CvO2) × 10 | ~250 mL/min | The O2 actually used by tissues. Normal extraction ratio = VO2/DO2 = 25% |
| Extraction ratio (ER) | VO2 / DO2 | ~25% | At rest, tissues extract 25% of delivered O2. In exercise/shock → ER increases (up to 70%). If DO2 falls below critical threshold → ER cannot increase further → VO2 falls → anaerobic metabolism → lactate |
| Mixed venous O2 (SvO2) | Measured from PA catheter | 65-75% | SvO2 <65% = inadequate DO2 (either CO low, Hb low, or SaO2 low). SvO2 <50% = severe O2 debt. SvO2 >80% = shunting (arteriovenous shunt in sepsis — blood bypasses capillaries) |
Causes of hypoxaemia — the 5 mechanisms
Five causes of hypoxaemia — and which respond to oxygen
| Mechanism | A-a gradient | Response to 100% O2 | Examples | Management |
|---|---|---|---|---|
| V/Q mismatch (#1 cause) | Elevated | IMPROVES (well-ventilated alveoli compensate) | Pneumonia, COPD, asthma, pulmonary embolism (mild), atelectasis | Increase FiO2 + treat cause (antibiotics, bronchodilators) |
| Shunt | Elevated | DOES NOT IMPROVE (blood bypasses ventilated alveoli — 100% O2 cannot reach shunted blood) | ARDS, severe pneumonia, hepatopulmonary syndrome, intracardiac shunt (PFO), atelectasis (complete) | PEEP (recruit collapsed alveoli → reduce shunt), proning, ECMO |
| Diffusion impairment | Elevated | IMPROVES (increased FiO2 increases diffusion gradient) | Pulmonary fibrosis, pulmonary oedema, emphysema | Increase FiO2 |
| Hypoventilation | NORMAL (<15) | IMPROVES (but PaCO2 may remain elevated) | Opioid overdose, neuromuscular disease (GBS, MG), CNS depression | VENTILATE (NIV or intubation) — oxygen alone is insufficient |
| Low FiO2 (high altitude) | NORMAL | IMPROVES | High altitude (low Patm → low PAO2) | Increase FiO2 |
Oxygen therapy devices — the escalation ladder
Oxygen delivery devices — escalating FiO2
| Device | FiO2 range | Flow rate | Advantages | Disadvantages |
|---|---|---|---|---|
| Nasal cannula | 24-44% | 1-6 L/min | Comfortable, allows eating/talking, patient can titrate. Rule of thumb: FiO2 ≈ 24% + (flow L/min × 4%) | Low FiO2 only. Nasal drying/epistaxis at high flow. FiO2 varies with respiratory pattern |
| Simple face mask | 40-60% | 5-10 L/min | Higher FiO2 than nasal cannula. Covers nose + mouth | MUST have ≥5 L/min flow (to wash out CO2 from mask dead space → prevent rebreathing). Cannot eat. Claustrophobic |
| Venturi mask | 24-60% (precise) | Variable (set by valve) | PRECISE FiO2 (24%, 28%, 31%, 35%, 40%, 50%, 60% valves) — oxygen entrainment is fixed. Used for COPD patients who need controlled FiO2 | Less comfortable. FiO2 is FIXED — cannot titrate without changing valve |
| Non-rebreather mask | 80-90% | 10-15 L/min | HIGHEST FiO2 without intubation. Reservoir bag provides O2 during inspiration. One-way valves prevent exhaled air rebreathing | Mask must fit tightly. Valve over reservoir must be functional. Still has some room air entrainment (not truly 100%) |
| High-flow nasal cannula (HFNC) | Up to 100% | 30-60 L/min | Delivers PRECISE FiO2 + FLOW (washes out dead space) + PEEP (3-5 cmH2O from high flow) + humidification + heating (improves mucociliary clearance). FLORALI: reduced intubation in hypoxaemic respiratory failure | Requires specialized equipment (Optiflow/Airvo). Cannot deliver PEEP >5. Patient must be spontaneously breathing |
| NIV (BiPAP/CPAP) | Up to 100% | — | Delivers POSITIVE PRESSURE (opens collapsed alveoli, reduces work of breathing, offloads LV in cardiogenic pulmonary oedema). 3CPO: CPAP improves outcome in cardiogenic pulmonary oedema | Claustrophobia, pressure ulcers, gastric insufflation. Cannot use if altered mental status (aspiration risk) |
| Mechanical ventilation | 21-100% | — | Full control of FiO2 + PEEP + ventilation. Can deliver ANY combination | Invasive (intubation risks: VAP, vocal cord damage, sedation) |
| ECMO (VV) | — | — | Bypasses the lungs entirely — oxygenates blood extracorporeally. For refractory hypoxaemia (EOLIA trial) | Invasive, anticoagulation, resource-intensive |
Venturi mask — entrainment physics (why FiO2 is PRECISE and FLOW is FIXED)
[1]Venturi valves — entrainment ratio and total flow at each FiO2
| Valve (colour) | FiO2 | Air : O2 ratio | Example total flow | Practical meaning |
|---|---|---|---|---|
| Blue | 24% | 25 : 1 | 1 L O2 + 25 L air = 26 L/min; 2 L O2 → 52 L/min | Maximal air entrainment → highest total flow, lowest FiO2. Total flow far exceeds resting peak inspiratory flow (~30 L/min) → FiO2 is LOCKED |
| White | 28% | 10 : 1 | 2 L O2 → 22 L/min | Still above typical adult peak inspiratory flow at rest |
| Yellow | 31% / 35% | 6 : 1 / 5 : 1 | 3-4 L O2 → ~18-21 L/min | Standard "precise moderate" range |
| Red | 40% | 3 : 1 | 4 L O2 → 16 L/min | Upper end of the controlled-oxygen range |
| Green | 50% | 1.7 : 1 | 8 L O2 → ~22 L/min | Less entrainment, FiO2 climbing |
| Pink / grey | 60% | 1 : 1 | 6-12 L O2 → 12-24 L/min | Minimal entrainment → FiO2 high but TOTAL FLOW LOW. At 60% the total flow may fall BELOW peak inspiratory flow in a tachypnoeic patient → the patient entrains extra room air around the mask → true delivered FiO2 DROPS below 60% |
High-flow nasal cannula — the four mechanisms, in physics terms
HFNC delivers up to 60 L/min of gas whose FiO2 is independently titrated (0.21-1.0), heated to 37 °C and humidified to 100% relative humidity. It is NOT "just a wet nasal cannula" — the high delivered flow fundamentally changes airway mechanics. [1]
The four mechanisms of HFNC benefit
| Mechanism | Physics / physiology | Measurable effect |
|---|---|---|
| 1. Precise, FIO2-independent FiO2 | Delivered flow (30-60 L/min) far exceeds peak inspiratory flow, so NO room air is entrained — the set FiO2 IS the delivered FiO2 (unlike nasal cannula / simple mask) | SpO2 stabilises; FiO2 becomes a true dial, not a guess |
| 2. Anisotropic PEEP (flow-dependent) | High flow creates continuous positive pressure in the nasopharynx: ~0.5-1 cmH2O per 10 L/min with the mouth CLOSED. PEEP is ANISOTROPIC — it rises with flow and FALLS when the mouth opens (pressure vents through the oral cavity). 60 L/min + closed mouth ≈ 5-7 cmH2O; open mouth ≈ 2-3 cmH2O | Reduced work of breathing, splinted alveoli, reduced LV afterload (cardiogenic pulmonary oedema) |
| 3. Dead-space washout | High nasal flow FLUSHES CO2-rich anatomical dead space (nasopharynx, oropharynx, large airways) between breaths. Each inspiration draws fresh gas, not the patient's own exhaled CO2. This effectively REDUCES anatomical dead space → a larger fraction of each Vt is alveolar ventilation → less minute ventilation is needed for the same CO2 clearance → lower RR, lower Vt, lower work of breathing | ↓RR, ↓PaCO2 (even in hypercapnic failure — the "NIV-sparing" effect in COPD), reduced WOB on oesophageal-pressure measurement |
| 4. Heated humidified gas | 37 °C, 44 mg H2O/L (fully saturated). Dry cold gas paralyses mucociliary clearance, thickens secretions, and steals energy to condition the gas. Humidification PRESERVES ciliary beat frequency and mucus rheology | Secretions mobilise, less atelectasis, reduced metabolic cost of gas conditioning, dramatically improved comfort and tolerance vs dry O2 |
Starting and titrating HFNC at the bedside — and predicting failure
- Set FiO2 to the patient's requirement (start 1.0 in severe hypoxaemia, wean to ≤0.4 as tolerated — target SpO2 92-96%).
- Set flow starting at 30 L/min and titrate UP to 50-60 L/min over 10-15 min (sudden high flow causes airway irritation, claustrophobia, and aerophagia). Higher flow = more dead-space washout + more PEEP.
- Confirm heat and humidity — circuit temperature 37 °C, chamber water level adequate. A dry circuit negates the mucociliary benefit and dries secretions.
- Apply the ROX index to predict success or failure: ROX = (SpO2/FiO2) / RR. ROX ≥4.88 at ≥2 h predicts HFNC success; ROX <3.85 predicts failure and should trigger intubation planning. Trend it serially — a falling ROX is an early warning.[2]
- Monitor for failure (any): persistent RR >30-35, SpO2 <90% on maximal settings, increasing accessory-muscle use / paradoxical breathing, agitation or altered mental status, rising PaCO2, acidosis (pH <7.3). Do NOT delay intubation for a "HFNC trial" that is failing — the delay-to-intubation itself worsens outcome.
- Wean by stepping FiO2 down first (to ≤0.4), THEN reducing flow to 30 L/min, then stepping to standard O2. A rising ROX with a comfortable patient supports continuing; a static or falling ROX does not.
Hyperoxia injury — the reactive oxygen species chemistry
[1]The hyperoxia injury cascade — ROS species, sources, and defences
| Species | Formula | Source | Reactivity / half-life | Enzymatic defence | Tissue marker |
|---|---|---|---|---|---|
| Superoxide | O2•− | Mitochondrial ETC (I, III), xanthine oxidase, NADPH oxidase, uncoupled NOS | Moderate, short (µs) | Superoxide dismutase (SOD) | Reacts with NO → peroxynitrite |
| Hydrogen peroxide | H2O2 | SOD | Stable, membrane-permeable | Catalase, glutathione peroxidase (GSH) | Substrate for Fenton |
| Hydroxyl radical | •OH | Fenton: Fe2+ + H2O2 | EXTREME, ns (diffusion-limited) | NONE — prevent formation | 8-oxoG (DNA), MDA / 4-HNE (lipid) |
| Peroxynitrite | ONOO− | O2•− + NO | Strong, ms | None effective | Nitrotyrosine |
Clinical pearls
Red flags
[1] [1]Prognosis
Oxygen therapy outcomes — the evidence
| Strategy | Outcome | Evidence |
|---|---|---|
| HFNC vs standard O2 | HFNC reduces intubation in PaO2/FiO2 <200 | FLORALI trial |
| Conservative vs liberal O2 | Conservative is SAFE (no excess mortality) | ICU-ROX, LOCO2 |
| Liberal O2 | INCREASES mortality (RR 1.21) | IOTA meta-analysis |
| CPAP for cardiogenic pulmonary oedema | Reduces intubation + mortality | 3CPO trial |
| NIV for COPD exacerbation | Reduces intubation + mortality | Plant trial |
Key trials and evidence
FLORALI trial — HFNC vs NIV vs standard O2 (PMID 22642477)
Study design
Randomised — 310 patients with hypoxaemic respiratory failure
Population
Adults with PaO2/FiO2 <300, non-hypercapnic
Intervention
HFNC vs NIV (BiPAP) vs standard face mask O2
Primary outcome
Intubation rate at 28 days: 38% (HFNC) vs 47% (standard) vs 50% (NIV) — overall not significant
Subgroup (PaO2/FiO2 <200)
HFNC SIGNIFICANTLY reduced intubation: 35% vs 53% (NIV) — p=0.009
Clinical bottom line
HFNC is PREFERRED for moderate hypoxaemic respiratory failure — especially PaO2/FiO2 <200
IOTA meta-analysis — Liberal vs conservative oxygen (PMID 27568755)
Study design
Meta-analysis — 25 RCTs, 16,037 patients
Population
Acutely ill adults (sepsis, trauma, cardiac, stroke, critical illness)
Intervention
Liberal oxygen (SpO2 >96%) vs conservative (room air or titrated O2)
Primary outcome
Mortality: liberal oxygen INCREASED mortality (RR 1.21, 95% CI 1.03-1.43)
Key finding
Hyperoxia is HARMFUL — every unnecessary unit of O2 increases risk without benefit
Clinical bottom line
Titrate FiO2 to SpO2 92-96% — do NOT give 100% O2 routinely — conservative oxygen is the new standard
ROX index — predicting HFNC success or failure (PMID 30576221)
Study design
Multicentre derivation and validation — adults with acute hypoxaemic respiratory failure on HFNC
Index
ROX = (SpO2 / FiO2) / respiratory rate
Thresholds
ROX ≥4.88 at ≥2 h predicts HFNC success (low intubation risk); ROX <3.85 predicts failure (high intubation risk)
Use
Measure serially at 2, 6, and 12 h. A falling ROX is an early warning of failure — pair with bedside work-of-breathing assessment
Clinical bottom line
ROX is a simple, bedside gauge of whether HFNC is working — but it supplements, never replaces, clinical judgement; never delay intubation for a borderline number
HOT-ICU — lower vs higher oxygenation targets in ICU (PMID 33471452)
Study design
Multicentre RCT — 2928 adults with acute hypoxaemic respiratory failure
Intervention
Lower target PaO2 60 mmHg (8 kPa) vs higher target PaO2 90 mmHg (12 kPa)
Primary outcome
90-day mortality: NO DIFFERENCE between groups
Key finding
A lower oxygenation target was SAFE but did not reduce mortality — alongside ICU-ROX and LOCO2, conservative targets are non-inferior, supporting titration to SpO2 92-96% and avoidance of hyperoxia
Clinical bottom line
There is no benefit to deliberately supranormal PaO2; target the lowest effective FiO2 (SpO2 92-96%, or 88-92% in CO2-retaining COPD)
Oxygen-target trials at a glance — conservative is safe, liberal is harmful
| Trial | Population | Comparison | Result | Take-home |
|---|---|---|---|---|
| IOTA (Chu 2018) | Acutely ill adults (25 RCTs) | Liberal vs conservative O2 | Liberal ↑ mortality RR 1.21 | Hyperoxia is HARMFUL — titrate down |
| ICU-ROX (Stewart 2019) | Mechanically ventilated ICU | Conservative (SpO2 88-92%) vs liberal | No outcome difference | Conservative is SAFE |
| LOCO2 (Barrot 2020) | ARDS | Conservative (PaO2 60-90) vs liberal (150-180) | Conservative SAFE (no excess mortality) | Conservative is SAFE in ARDS |
| HOT-ICU (Schjørring 2021) | Acute hypoxaemic RF | PaO2 60 vs PaO2 90 | No difference in 90-day mortality | No benefit to higher PaO2 |
| FLORALI (Frat 2015) | Hypoxaemic RF (PaO2/FiO2 <300) | HFNC vs NIV vs standard O2 | HFNC ↓ intubation in PaO2/FiO2 <200 | HFNC preferred for moderate hypoxaemia |
SAQ — Oxygen delivery ladder and conservative targets
10 minutes · 10 marks
A 62-year-old man with community-acquired pneumonia has SpO2 88% on room air, RR 32, and is starting to tire. You are asked to escalate oxygen therapy and set targets.
References
- [1]Pittman RN, et al. Commentary on providing guidance to patients: physicians' views about the relative responsibilities of doctors and religious communities South Med J, 2013.PMID 23820320
- [2]Roca O, et al. An Index Combining Respiratory Rate and Oxygenation to Predict Outcome of Nasal High-Flow Therapy Am J Respir Crit Care Med, 2019.PMID 30576221
- [3]Schjørring OL, et al. Lower or Higher Oxygenation Targets for Acute Hypoxemic Respiratory Failure N Engl J Med, 2021.PMID 33471452