Question  – 

Does low-normal oxygenation compared with a high-normal oxygenation target range reduce organ dysfunction in critically ill patients?

Population and design

The study was conducted in 4 centres in the Netherlands. Adult patients, expected to stay in the ICU for >48 h with 2 or more SIRS criteria were eligible. Excluded were elective surgical patients, patients with pulmonary hypertension, severe ARDS, and severe COPD, CO or cyanide poisoning, methemoglobinemia, right to left shunt, and sickle cell disease. Also excluded were patients who were not for intubation and pregnant patients. Patients could be on any type of respiratory support including ventilation, nasal cannula, oxygen mask, high-flow oxygen. PEEP levels were used as appropriate.

Patients were randomized 1:1 to low normal PO2 group: 8–12 kPa (60–90 mm Hg) vs. high normal group: 14–18 kPa (105-135 mm Hg). The FiO2 was limited to a maximum of 0.6 and the PEEP level to 10 cm H2O unless otherwise indicated. The modality of respiratory support and use of prone positioning was based on clinician judgment. The FiO2 could be increased temporarily for procedures. Arterial blood gases were analyzed according to the usual practice. If the PO2 level was outside the target, the FiO2 was adjusted, and blood gases were repeated to confirm levels within the target range.

The non-respiratory SOFA score was calculated every day for 14 days. The respiratory SOFA was excluded to avoid change in the SOFA scores attributable to a change in the P/F ratios. The delta SOFA score, which is non-respiratory SOFA – baseline SOFA score from day 1 to day 14 was calculated from days 1–14.  A higher delta SOFA, would indicate more severe organ dysfunction. Death carried a maximum SOFA of 20, discharge from ICU carried a SOFA of 0. The delta SOFA was chosen as an endpoint because it was considered a good composite of the severity of organ failure, mortality, and length of stay in ICU over the first 14 days. The primary outcome was based on the delta SOFA from day 1 to day 14.

Designed to identify a difference of 0.33 SD of the primary endpoint, the delta SOFA. The sample size calculated was 385 patients for 85% power.

Enrollment

1832 patients were evaluated; 1258 did not meet inclusion or had exclusion criteria. 574 patients were randomized, 294 to the low normal arm and 280 to the high normal arm. After exclusion for various reasons (mostly lack of informed consent), 205 patients in the low normal arm and 195 patients in the high normal arm were included in the final analysis.

Both groups were well matched at baseline. A clear, statistically significant division was observed in the PO2 levels between groups, 81 (73–90) vs. 96 (81.8–111.8).

Primary outcome

Median SOFA rank score: –35 vs. –40 (not significantly different)

Secondary outcomes

No difference in the maximum non-respiratory SOFA score (median of 6 in both arms), maximum – baseline SOFA (0 in both arms), or the rate of SOFA score decline. The incidence of intubation, duration of mechanical ventilation, ventilator-free days, ICU-LOS, the total dose of noradrenaline, the highest dose of noradrenaline, ICU mortality, hospital mortality, and 90-d mortality were similar between groups. RRT was required in 20 vs. 21 patients, prone ventilation: 7 vs. 8 patients. The incidence of new MI, new-onset liver failure, new stroke, were all small numbers. The incidence of serious adverse events was also not different.

Strengths

• Specifically at patients with SIRS response and organ dysfunction
• Measured PO2 and not saturation levels 
• The upper limit for FiO2

Weaknesses

• Entirely based on PO2 levels, which may have varied from time to time
• The primary endpoint was novel, not a hardcore clinical outcome
• High FiO2 use for procedures of varying periods of time
• The PO2 levels prior to inclusion may have had an unknown impact
• The median PO2 level in the high normal group fell short of the target level (median level: 96; target: 105–135)
• The study was not powered for a possible smaller difference in the primary outcome 

Our Summary –

PaO2 remains a better target than SpO2.

Low oxygenation targets are safer.

We should think about oxygen as a drug and it should not be administered unless it’s absolutely necessary.