Table 1.
Main features of the studies discussed in the text. ABG arterial blood gas; ACS acute coronary syndrome; AIS acute ischaemic stroke; AMI acute myocardial infarction; CI confidence interval; CPR cardiopulmonary resuscitation; ED emergency department; GCS Glasgow coma score; GOSE Glasgow outcome scale extended; ICU intensive care unit; IQR interquartile range; ICB intracranial bleeding; mo month; MV mechanical ventilation; OR odds ratio; RCT randomised controlled trial; ROSC return of spontaneous circulation; SAB subarachnoidal bleeding; SIRS systemic inflammatory response syndrome; SpO2 pulse oximetry haemoglobin O2 saturation; SOFA sequential organ failure assessment; SSI surgical site infection; STEMI ST segment elevation myocardial infarction; TBI traumatic brain injury; TWA time-weighted average
| Study name | Design/sample size | Setting | Oxygenation parameter | Major findings | Ref. no. |
|---|---|---|---|---|---|
| IOTA | Meta-analysis/25 RCT, n = 16,037 | General ICU | “Conservative” vs. “Liberal”, i.e. lower vs. higher target according to individual study design | Higher mortality risk (relative risk 1.21 [95%CI 1.0–1.43]) with “liberal” O2 strategy (median baseline SpO2 96% [IQR 96–98%]) | 38 |
| ICU-ROX | Multicentre RCT/n = 965 | General ICU; MV | “Conservative” (lowest FIO2 possible keeping SpO2 between 91 and 97%) vs. “Usual” (no limit) | No difference in day 28 ventilator-free days and day 90/180 mortality | 39 |
| PROSPERO | Meta-analysis + Trial Sequential Analysis/36 RCT, n = 20,166 | General ICU | “Lower” vs. “Higher”, i.e. lower vs. higher target according to individual study design | No difference in mortality or morbidity | 42 |
| O2-ICU | Multicentre RCT/n = 400 | General ICU; expected ICU stay > 2 days; ≥ 2 SIRS criteria | Oxygenation target: PaO2 8–12 vs. 14–18 kPa (≈ 60–90 vs. 105–135 mmHg) | No difference in SOFA score; limitation: PaO2 < target in “high-normal oxygenation” group | 43 |
| LOCO2 | Multicentre RCT/n = 205 | ARDS | “Conservative” (PaO2 55–70 mmHg, SpO2 88–92%) vs. “Liberal” (PaO2 90–105 mmHg, SpO2 ≥ 96%) until day 7 | Premature halt for higher mortality in “Conservative” group (day 28: 34.3 vs. 26.5%; day 90: 44.4 vs. 30.4%); limitation: > 50% patients had PaO2 > upper level | 63 |
| HOT-ICU | Multicentre RCT / n = 2,888 | General ICU; acute hypoxemic respiratory failure | “Lower” (PaO2≈60 ± 7.5 mmHg) vs. “Higher” (PaO2≈90 ± 7.5 mmHg) | No difference in day 90 mortality | 64 |
| LUNG SAFE | Sub-study of multicentre, prospective, cohort study/ n = 2,005 | ARDS | Presence of day 1 “hyperoxemia” PaO2 > 100 mmHg), “sustained” (day 1 and day 2) or “excessive” O2 (FIO2 ≥ 0.6 + PaO2 > 100 mmHg) | 30% hyperoxaemia day 1, 12% “sustained hyperoxaemia”, 20% “excessive O2” | 65 |
| IMPACT | Multicentre retrospective/n = 16,326 | CPR; ABG within 24 h | PaO2 < 60 (“hypoxia”), 60–300 (“normoxia”), ≥ 300 mmHg (“hyperoxia”) | PaO2 ≥ 300 mmHg significantly higher mortality 63(CI:60–66)% vs. normoxia 45[CI43-48]%) vs. hypoxia (57[CI56-59]%) | 68 |
| HYPER2S | Multicentre RCT/n = 442 | Septic shock within first 6 h; MV | FIO2 = 1.0 during first 24 h vs. “standard treatment” | Premature safety stop for higher mortality with “FIO2 = 1.0” (day 28: 43 vs. 35%, p = 0.12; day 90: 48 vs. 42%, p = 0.16); lower number of ventilator-free days, more serious adverse events despite lower SOFA at day 7 | 75 |
| HYPER2S | Post hoc analysis of multicentre RCT/n = 393 | Septic shock within first 6 h according to Sepsis-3; MV | FIO2 = 1.0 during first 24 h vs. “standard treatment” | Higher mortality with “FIO2 = 1.0” and lactate > 2 mmol/L (day 28: 57 vs. 44%); no effect lactate ≤ 2 mmol/L | 76 |
| ICU-ROX | Post hoc analysis of multicentre RCT/n = 251 | Sepsis; MV | “Conservative” (lowest FIO2 possible keeping SpO2 between 91 and 97%) vs. “Usual” (no limit) | Mortality day 90 “Conservative” 36.2 vs. “Usual” 29.2% (p = 0.24); “…point estimates of treatment effects consistently favoured usual O2 therapy…” | 77 |
| Multicentre, retrospective/n = 1,116 | TBI; MV | PaO2 < 10.0 kPa (≈ < 75 mmHg) or 10.0–13.3 kPa (≈ 75-100 mmHg) or PaO2 > 13.3 kPa (≈ > 100 mmHg) | PaO2 > 13.3 kPa no relationship to outcome | 86 | |
| Multicentre retrospective/n = 2,894 | MV; 19% AIS, 32% SAB, 49% ICB | PaO2 < 60, 60–300 or ≥ 300 mmHg | PaO2 ≥ 300 mmHg in-hospital mortality 57 vs. 46/47% (p < 0.001) | 87 | |
| Multicentre retrospective/n = 432 | SAB; MV | 24 h TWA PaO2: “low”/“intermediate”/“high” (< 97.5/97.5–150/ > 150 mmHg) | TWA-PaO2: survivors 118(IQR90-155) vs. non-survivors 137(IQR104-167)mmHg (p < 001); multivariate analysis no relation between TWA-PaO2 and outcome | 91 | |
| SO2S | Multicentre RCT/n = 7,635 | AIS | Continuous (2-3L/min) vs. nocturnal nasal O2 vs. control | No difference in mortality and neurological outcome | 92 |
| Multicentre retrospective/n = 24,148 | TBI; MV | PaO2 50 mmHg-increments; hyperoxia PaO2 > 300 mmHg | No relation PaO2 vs. mortality except for PaO2 < 60 mmHg and GCS > 12 | 93 | |
| Multicentre retrospective/n = 3,699 | TBI; MV | PaO2 < 60, 60–300 vs. PaO2 ≥ 300 mmHg | No relation PaO2 ≥ 300 mmHg vs. GOSE < 5 at 6 mo | 95 | |
| Single centre retrospective/n = 688 | ED; MV, normoxia (PaO2 60-120 mmHg) on day 1 ICU | Hypoxia/normoxia/hyperoxia PaO2 < 60, 60–120, > 120 mmHg | Hyperoxia present in 43%; mortality 29.7 vs. 19.4 (normoxia) and 13.2 (hypoxia) % (p = 0.021 vs. normoxia) | 109 | |
| Multicentre retrospective/n = 3,464 | Polytrauma; ICU within 24 h | Patient-hours with SpO2 90–96% (“normoxia”) vs. > 96% (“hyperoxia”); hyperoxia in 10%- FIO2 increments until d3 and d4-7 | Increased risk of mortality with higher FIO2 during hyperoxia | 114 | |
| IMPACT | Post hoc of multicentre retrospective/n = 4,459 | CPR; ABG within 24 h | Highest PaO2 24 h ICU | 100 mmHg PaO2-increments 24% mortality risk increase (OR1.24[CI1.18–1.31]) | 121 |
| Multicentre prospective/n = 280 | CPR; therapeutic hypothermia | PaO2 > 300 mmHg 1 or 6 h post-ROSC | 3% (OR1.03[CI1.02–1.05]) risk increase in poor neurological outcome per 1 h hyperoxia duration | 124 | |
| Multicentre retrospective/n = 12,108 | CPR; therapeutic hypothermia | PaO2 ≥ 300 mmHg within 24 h | PaO2 ≥ 300 mmHg mortality 59(CI56-61)% vs. 47(CI45-50% (60-300 mmHg)/58(CI57-58)% (< 60 mmHg) | 125 | |
| FINNRESUSCI | Multicentre prospective/n = 409 | CPR out-of-hospital | PaO2 < 75 (“low”), 75–150 (“middle”), 150–225 (“intermediate”), PaO2 > 225 mmHg (“high”) | No association between hyperoxia and neurological outcome | 126 |
| TTM | Post hoc analysis of multicentre RCT/n = 869 | CPR out-of-hospital; therapeutic hypothermia | PaO2, TWA PaO2 37 h post-ROSC; PaO2 > 40 kPa (≈PaO2 > 300 mmHg), 8 ≤ PaO2 ≤ 40 (≈60 ≤ PaO2 ≤ 300 mmHg), PaO2 < 8 kPa (≈PaO2 < 60 mmHg) | No association with 6-mo neurological outcome | 129 |
| Meta-analysis/7 RCT, n = 429 | CPR | “Higher” (“liberal”) vs. “lower” (“conservative”) O2 target | Mortality 50% liberal vs. 41% conservative, p = 0.04 | 130 | |
| ICU-ROX | Post hoc analysis of multicentre RCT/n = 166 | “Suspected hypoxic ischaemic encephalopathy”; MV | “Conservative” (lowest FIO2 possible 91 ≤ SpO2 < 97%) vs. “Usual” (no limit) | Day 180: mortality 43% conservative vs. 59% “usual” (p = 0.15); “unfavourable neurological outcome” 55% conservative vs. 68% usual (p = 0.15) | 134 |
| DETO2X-SWEDEHEART | Multicentre RCT/n = 6629 | AMI | 6L/minO2 6-12 h | No effect on 1-year outcome | 138 |
| Oxygen Therapy in Acute Coronary Syndromes | Multicentre crossover RCT/n = 40,872 | ACS | 6-8L/minO2 vs. SpO2 90–95% | No effect on day 30-mortality | 140 |
| PROXI | Multicentre RCT/n = 1,386 | Elective/acute laparotomy | FIO2 0.8 vs. 0.3 until 2 h post-op | FIO2 0.8 19.1% vs. FIO2 0.3 20.1% SSI (p = 0.64) | 143 |
| Supplemental Oxygen in Colorectal Surgery | Single centre prospective/n = 5,749 | Major intestinal surgery > 2 h | FIO2 = 0.8 vs. 0.3 every 2 weeks alternating intervention study | 30d-SSI FIO2 = 0.8 10.8 vs. 11.0% (p = 0.85) | 144 |
| Intraoperative Inspiratory Oxygen Fraction and Postoperative Respiratory Complications | Multicentre retrospective/n = 79,322 | General surgery | Quintiles FIO2 0.31, 0.41, 0.52, 0.79 | Dose-dependent association FIO2 vs. day 7 “Major respiratory complications composite” and vs. day 30-mortality | 151 |
| WHO Meta-analysis/12 RCT, n = 5,976 | General surgery | FIO2 0.8 vs. 0.30–0.35 | FIO2 = 0.8 reduces SSI risk vs. 0.30–0.35 (OR0.80[CI0.64–0.99], p = 0.043): only general anaesthesia with tracheal intubation | 153 | |
| Single centre RCT/n = 210 | Open surgery for appendicitis | FIO2 = 0.8 vs. 0.30 until 2 h post-op | FIO2 = 0.8 SSI 5.6 vs.13.6% (p = 0.04); hospital stay 2.51 vs. 2.92 (p = 0.01) | 156 | |
| Cochrane Perioperative Oxygen Review | Meta-analysis/10 RCT, n = 1,458 | General surgery | “Higher” vs. “lower” FIO2 | “Higher” vs. “lower” FIO2 “very low evidence” serious adverse event risk | 157 |
| Meta-analysis/12 trials, n = 28,984 | General ICU; MV | FIO2 “low” vs. “high” (as defined by authors) | FIO2 “high”; no impact on pneumonia, ARDS, MV duration; FIO2 ≥ 0.8 increased risk of: atelectasis | 158 |