Oasis Score

OASIS Score Calculation

The Oxford Acute Severity of Illness Score (OASIS) is a parsimonious severity score that achieves comparable predictive accuracy to APACHE using fewer variables. It does not require laboratory values, making it useful when lab data is missing.

When to Use This Skill

• Mortality prediction when lab data is incomplete
• Quick severity assessment with minimal variables
• Real-time severity scoring (no lab turnaround time)
• Research requiring a validated, simple severity metric
• Comparison with APACHE/SAPS scores

Score Components (First 24 Hours)

Total Range: 0-67 (theoretical maximum)

Pre-computed Table

SELECT
    subject_id,
    hadm_id,
    stay_id,
    oasis,
    oasis_prob,  -- Predicted in-hospital mortality
    age, age_score,
    preiculos, preiculos_score,
    gcs, gcs_score,
    heartrate, heart_rate_score,
    meanbp, mbp_score,
    resprate, resp_rate_score,
    temp, temp_score,
    urineoutput, urineoutput_score,
    mechvent, mechvent_score,
    electivesurgery, electivesurgery_score
FROM mimiciv_derived.oasis;

Critical Implementation Notes

1. No Laboratory Values Required: OASIS uses only vital signs, urine output, and administrative data - no labs needed.

2. Pre-ICU LOS Scoring: Time from hospital admission to ICU admission in minutes. Scoring is non-linear:

   • < 10.2 min: 5 points (immediate ICU)
   • 10.2-297 min: 3 points
   • 297-1440 min: 0 points (optimal)
   • 1440-18708 min: 2 points

   • 18708 min: 1 point

3. Mechanical Ventilation: Binary flag - any invasive ventilation during first 24 hours scores 9 points.

4. Elective Surgery: Requires BOTH:

   • Elective admission type AND
   • Surgical service (identified from first service transfer)

5. Ventilation Flag Cannot Be Missing: Unlike other components, ventilation defaults to 0 (no ventilation) if no data found.

6. Mortality Probability:

   oasis_prob = 1 / (1 + exp(-(-6.1746 + 0.1275 * oasis)))
   

Advantages Over APACHE/SAPS

• Simpler to calculate (10 variables vs 15-17)
• No laboratory data required
• Can be calculated earlier in admission
• Similar predictive accuracy

Example: Quick Severity Assessment

SELECT
    stay_id,
    oasis,
    oasis_prob,
    CASE
        WHEN oasis < 20 THEN 'Low Risk'
        WHEN oasis < 30 THEN 'Moderate Risk'
        WHEN oasis < 40 THEN 'High Risk'
        ELSE 'Very High Risk'
    END AS risk_category
FROM mimiciv_derived.oasis
ORDER BY oasis DESC;

Example: Compare OASIS vs SAPS-II Predictions

SELECT
    o.stay_id,
    o.oasis,
    o.oasis_prob AS oasis_mortality,
    s.sapsii,
    s.sapsii_prob AS sapsii_mortality,
    ABS(o.oasis_prob - s.sapsii_prob) AS prediction_difference
FROM mimiciv_derived.oasis o
INNER JOIN mimiciv_derived.sapsii s
    ON o.stay_id = s.stay_id
ORDER BY prediction_difference DESC;

References

• Johnson AEW, Kramer AA, Clifford GD. "A new severity of illness scale using a subset of Acute Physiology And Chronic Health Evaluation data elements shows comparable predictive accuracy." Critical Care Medicine. 2013;41(7):1711-1718.