Clinical background
The Horowitz index, also known as the PaO₂/FiO₂ ratio, is the most widely used measure for quantifying hypoxaemia in ventilated patients. Its clinical weight lies not in being a prediction instrument in itself, but in forming the backbone of the Berlin definition of ARDS, in which it alone determines the severity grading and thereby the therapeutic consequences to be drawn [2]. Without a standardised marker of oxygenation, intensive care would lack a common language for comparing patients between units, studies and countries, which was the main reason the ratio was incorporated into the ARDS definition as early as 1994 and retained in the Berlin revision of 2012 [2].
Calculating the Horowitz index
PaO₂ is the arterial oxygen tension measured in mmHg on an arterial blood gas. FiO₂ is the inspired oxygen concentration expressed as a fraction (0.21 for air up to 1.0 for 100% oxygen). The ratio has no unit of its own and is reported in mmHg, given that FiO₂ is dimensionless.
The Horowitz index was introduced in 1974 by Horovitz, Carrico and Shires as a simple measure of oxygenating capacity in trauma and shock, with the advantage that it normalises PaO₂ against the oxygen concentration actually delivered [1]. It took hold, however, only when it was incorporated into the American-European Consensus Conference definition of ARDS in 1994 and thereafter into the Berlin definition in 2012 [2].
The Berlin definition was developed by an expert panel under the European Society of Intensive Care Medicine with the support of the American Thoracic Society and the Society of Critical Care Medicine. The empirical evaluation comprised a patient-level meta-analysis of 4,188 ARDS patients from four multicentre studies together with 269 patients from three single-centre studies with physiological data [2]. Three mutually exclusive severity grades were defined, all requiring a PEEP or CPAP ≥ 5 cmH₂O:
| Severity | P/F ratio (mmHg) | Mortality in the Berlin cohort |
|---|---|---|
| Mild ARDS | 200–300 | 27% (95% CI 24–30) |
| Moderate ARDS | 100–200 | 32% (95% CI 29–34) |
| Severe ARDS | < 100 | 45% (95% CI 42–48) |
Interpretation in practice
The Horowitz index is always interpreted in the light of the current ventilator settings, particularly the PEEP. A P/F ratio of 150 at a PEEP of 5 cmH₂O represents a different lung injury from 150 at a PEEP of 15 cmH₂O: the latter patient has more recruitable lung tissue but is receiving more support, the former has poorer recruitability but less support. The ratio in itself cannot distinguish between them.
The severity grading translates into clinical action according to current guidelines:
- Mild ARDS (200–300): Early non-invasive ventilation or high-flow nasal oxygen may be considered. A standard PEEP strategy according to the low-PEEP protocol. The prognosis is relatively good, but monitoring is justified since some patients progress.
- Moderate ARDS (100–200): Invasive ventilation with a lung-protective strategy (tidal volume 6 mL/kg predicted body weight) and PEEP individualised according to a PEEP/FiO₂ table. Prone positioning is considered if there is no improvement.
- Severe ARDS (< 100): Prone positioning is recommended routinely. Assessment for extracorporeal membrane oxygenation (ECMO) or other extracorporeal lung support should be made in refractory hypoxaemia, particularly if the severity persists despite adequate PEEP optimisation [2,4].
Validation and performance
The discrimination of the Berlin definition for mortality was moderate in the derivation cohort, with an AUC of 0.577 (95% CI 0.561–0.593), significantly better than its predecessor the AECC definition (AUC 0.536) but far from a strong predictor [2].
In an external validation based on 3,442 patients from seven ARDS Network trials, the AUC of the P/F ratio for hospital mortality was 0.659 (95% CI 0.637–0.681) at a PEEP > 5 cmH₂O [3]. When PEEP was incorporated into the ratio (the P/FP ratio, that is, PaO₂ × 10 / [FiO₂ × PEEP]), the AUC rose to 0.710 (95% CI 0.691–0.730), and at a PEEP ≥ 18 cmH₂O the difference was even greater: 0.963 compared with 0.828 for the P/F ratio [3]. This underlines that the predictive value of the P/F ratio depends strongly on the PEEP the patient is receiving at the time of measurement.
In a prospective cohort of 100 consecutive ARDS patients, the mean AUC of the Berlin definition for 28-day mortality over four days was 0.604 in the subgroup with transpulmonary thermodilution, and lower in the whole group [4]. On day 1 after intubation the AUC was 0.697 in the subgroup but fell thereafter. The oxygenation index (OI = mean airway pressure × FiO₂ / PaO₂) consistently outperformed the P/F-based scores (AUC 0.689 on day 1, four-day mean 0.625) [4].
The global definition published in 2024 builds on Berlin and keeps the P/F thresholds unchanged, but also allows SpO₂/FiO₂ as an alternative when the SpO₂ is ≤ 97%, particularly in resource-limited settings [5]. It also allows high-flow nasal oxygen ≥ 30 L/min and ultrasound as an imaging modality, which broadens applicability but does not change the role of the P/F ratio during invasive ventilation.
Limitations
The P/F ratio has several well-known weaknesses that clinicians must keep in mind:
FiO₂ dependence. The ratio falls as FiO₂ rises even if PaO₂ is unchanged, since the relationship between PaO₂ and FiO₂ is not linear, particularly at high FiO₂ values. A patient on an FiO₂ of 1.0 has a lower ratio than the same patient on an FiO₂ of 0.5, all else being equal. This means the ratio is not a pure measure of lung injury but also a function of the oxygen setting chosen.
PEEP dependence. Two patients with an identical P/F ratio but different PEEP do not necessarily have lung injury of the same severity. A higher PEEP can recruit collapsed alveoli and raise PaO₂, so that the ratio underestimates severity [3]. Palanidurai et al. showed that 54% of patients were reclassified when PEEP was incorporated into the ratio, with 12.5% of patients with moderate ARDS moving to severe and 15% with mild ARDS moving to moderate [3].
Limited predictive value. With an AUC of around 0.58–0.66 in external cohorts, the P/F ratio alone is a weak predictor of mortality [2,3,4]. It should not be used alone for prognostication but combined with other variables such as PEEP, respiratory compliance, the oxygenation index and organ failure.
The need for an arterial blood gas. The ratio requires an arterial blood gas, an invasive procedure that may be unavailable in resource-limited settings. The 2024 global definition therefore allows SpO₂/FiO₂ as a surrogate when the SpO₂ is ≤ 97%, but with limitations regarding reliability in darker skin tones and in circulatory shock [5].
Does not apply to non-invasive oxygenation without PEEP/CPAP ≥ 5. The Berlin definition requires a PEEP or CPAP ≥ 5 cmH₂O for the ratio to classify ARDS. Patients on high-flow nasal oxygen can now be included under the global definition, but the P/F ratio is then not measured in the same way and the thresholds are validated differently [5].
References
- Horovitz JH, Carrico CJ, Shires GT. Arch Surg 1974;108(3):349–55. PMID: 4813081
- ARDS Definition Task Force; Ranieri VM, Rubenfeld GD, Thompson BT et al. Acute respiratory distress syndrome: the Berlin Definition. JAMA 2012;307(23):2526–33. PMID: 22797452
- Palanidurai S, Phua J, Chan YH et al. P/FP ratio: incorporation of PEEP into the PaO₂/FiO₂ ratio for prognostication and classification of acute respiratory distress syndrome. Ann Intensive Care 2021;11(1):90. PMID: 34370116
- Huber W, Findeisen M, Lahmer T et al. Prediction of outcome in patients with ARDS: A prospective cohort study comparing ARDS-definitions and other ARDS-associated parameters, ratios and scores at intubation and over time. PLoS One 2020;15(5):e0232720. PMID: 32374755
- Matthay MA, Arabi Y, Arroliga AC et al. A New Global Definition of Acute Respiratory Distress Syndrome. Am J Respir Crit Care Med 2024;209(1):37–47. PMID: 37487152
