What is the ROX Index and How to Use It
The ROX index is a simple bedside calculation (SpO₂/FiO₂ divided by respiratory rate) that predicts whether high-flow nasal cannula (HFNC) therapy will succeed or fail in patients with acute hypoxemic respiratory failure, helping clinicians decide when to escalate to intubation or continue HFNC support.
Definition and Calculation
The ROX index is calculated as:
ROX = (SpO₂/FiO₂) / Respiratory Rate
Where:
- SpO₂ = oxygen saturation by pulse oximetry (%)
- FiO₂ = fraction of inspired oxygen (expressed as a decimal, e.g., 0.60 for 60%)
- Respiratory rate = breaths per minute
This is a completely non-invasive calculation using readily available bedside parameters 1.
Clinical Application and Timing
When to Measure
Calculate the ROX index at specific time points after initiating HFNC therapy:
- Baseline (at HFNC initiation)
- 2 hours after starting HFNC
- 6 hours after starting HFNC
- 12 hours after starting HFNC
The predictive accuracy improves over time, with measurements at 6-12 hours showing the best discrimination (AUROC 0.76-0.78) 1, 2.
Interpretation Thresholds
High-Risk for HFNC Failure (Consider Intubation):
- ROX < 2.85 at 2 hours
- ROX < 3.47 at 6 hours
- ROX < 3.85 at 12 hours
Low-Risk for HFNC Failure (Continue HFNC):
- ROX ≥ 4.88 at 2,6, or 12 hours consistently predicts HFNC success 1
Optimal Cutoff Range: The meta-analysis of 13 studies suggests the optimal cutoff range is 4.2-5.4, with most data centered between 4.5-6.0 2. A practical threshold of ROX > 4.4 at 6 hours demonstrates good sensitivity (91-94%) and specificity (87%) 3, 4.
Clinical Decision-Making Algorithm
- Initiate HFNC in patients with acute hypoxemic respiratory failure
- Calculate baseline ROX at time zero
- Reassess at 2 hours:
- If ROX < 2.85 → High risk of failure, prepare for intubation
- If ROX ≥ 4.88 → Likely success, continue HFNC
- Reassess at 6 hours:
- If ROX < 3.47 → Strongly consider intubation
- If ROX ≥ 4.88 → Continue HFNC with confidence
- Monitor trend: Patients who succeed show increasing ROX values over the first 12 hours, while those who fail show static or declining values 1
Evidence Supporting Clinical Use
Following ROX-guided decisions is associated with significantly better outcomes. In a 2024 study, patients receiving ROX-concordant care (clinical decisions aligned with ROX predictions) had dramatically lower rates of persistent organ dysfunction plus death (16% vs 62%), 28-day mortality (13.6% vs 38%), and in-hospital mortality (15% vs 50%) compared to ROX-discordant care 5. Premature intubation despite favorable ROX indices was particularly harmful, increasing odds of death by nearly 5-fold 5.
Important Caveats
Population Specificity: The ROX index was originally validated in pneumonia patients 1 and has been extensively studied in COVID-19 populations 2, 4. Its performance in heterogeneous ICU populations without pneumonia shows more limited discrimination (AUROC 0.56-0.64), and should not be used as a standalone decision tool in mixed populations 6.
Clinical Context Matters: The ROX index should guide but not replace clinical judgment. Consider the underlying etiology, trajectory of illness, and other physiological parameters. Do not delay intubation in patients with clear signs of respiratory exhaustion regardless of ROX values.
Serial Measurements Are Key: A single ROX measurement has limited value. The trend over time (particularly whether ROX is increasing or decreasing) provides more actionable information than isolated values 1.
Practical Implementation
The ROX index is most useful for:
- Identifying patients at low risk who can safely continue HFNC (ROX ≥ 4.88)
- Identifying patients at high risk who need close monitoring and preparation for intubation (ROX < 3.5-4.0)
- Avoiding premature intubation in patients responding well to HFNC
- Preventing delayed intubation in patients failing HFNC therapy
The index performs best when measured within 6-12 hours of HFNC initiation and should be used in conjunction with clinical assessment of work of breathing, mental status, and hemodynamic stability 1, 2.