1. ARDS & Mechanical Ventilation
Before 2000, ARDS was ventilated with high tidal volumes (10–15 mL/kg). ARMA proved low tidal volume ventilation reduces mortality, launching the modern era. Subsequent trials refined PEEP strategy, established prone positioning and VV-ECMO, and shifted paralysis from routine to rescue-only.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size (95% CI) | P | Bias/Weaknesses | External Validity | GRADE | PMID | Take-Home Message |
ARMA (ARDSNet) | 2000 | Brower | NEJM | 861 | 6 mL/kg TV | 12 mL/kg | In-hospital mortality | 31.0% vs 39.8% | RR ~0.78 | 0.007 | Stopped early; control TV higher than then-usual | High – foundational | High | 10793162 | Ventilate ARDS at 6 mL/kg PBW, keep Pplat <30 — the single most important ARDS trial ever done. |
PROSEVA | 2013 | Guérin | NEJM | 466 | Prone ≥16h | Supine | 28-day mortality | 16.0% vs 32.8% | HR 0.39 (0.25–0.63) | <0.001 | Unblinded; expert proning centers | Moderate–High | High | 23688302 | Prone position early and long (≥16h/day) in moderate-severe ARDS (P/F<150) — it saves lives. |
EOLIA | 2018 | Combes | NEJM | 249 | Early VV-ECMO | Conventional + rescue ECMO | 60-day mortality | 35% vs 46% | RR 0.76 (0.55–1.04) | 0.09 | Stopped for futility; 28% crossover | Moderate – ECMO centers only | Moderate | 29791822 | Frequentist "negative," but Bayesian reanalysis favors ECMO — reasonable rescue in very severe ARDS at expert centers. |
ACURASYS | 2010 | Papazian | NEJM | 340 | Cisatracurium 48h | Placebo | 90-day mortality | 31.6% vs 40.7% (adj sig) | Adj HR 0.68 (0.48–0.98) | 0.04 (adj) | Benefit only after adjustment; superseded by ROSE | Low–Moderate | Low–Moderate | 20843245 | Suggested early paralysis helps in severe ARDS — but do not treat as definitive; see ROSE. |
ROSE | 2019 | Moss (PETAL) | NEJM | 1006 | Early cisatracurium | Usual care, lighter sedation | 90-day mortality | 42.5% vs 42.8% | — | 0.93 | Refutes routine early paralysis | High | High | 31112383 | Routine early paralysis does NOT help when using modern light-sedation, high-PEEP strategy — reserve NMBA for rescue. |
ART | 2017 | Cavalcanti | JAMA | 1010 | Recruitment + PEEP titration | Low PEEP (ARDSNet) | 28-day mortality | 55.3% vs 49.3% (HARM) | HR 1.20 (1.01–1.42) | 0.041 | Bundled intervention confounds | High – argues against routine RM | High (harm) | 28973363 | Aggressive lung recruitment + decremental PEEP titration increases mortality — avoid routine aggressive recruitment maneuvers. |
ALVEOLI | 2004 | Brower (ARDSNet) | NEJM | 549 | Higher PEEP | Lower PEEP | In-hospital mortality | 27.5% vs 24.9% (NS) | — | 0.48 | Baseline imbalance | High | Moderate | 15269312 | Simply raising PEEP without individualizing to physiology does not reduce mortality. |
LOVS | 2008 | Meade | JAMA | 983 | Higher PEEP + recruitment | Lower PEEP | Hospital mortality | 36.4% vs 40.4% (NS) | RR 0.90 (0.77–1.05) | 0.30 | Open-lung bundle | High | Moderate | 18270352 | Open-lung ventilation strategy is safe but not clearly superior for mortality. |
FACTT | 2006 | Wiedemann (ARDSNet) | NEJM | 1000 | Conservative fluid | Liberal fluid | 60-day mortality | 25.5% vs 28.4% (NS) | — | 0.30 | Improved vent-free days | High | Moderate | 16714767 | Keep ARDS patients dry (conservative fluid strategy) — more ventilator-free days without a mortality penalty. |
CESAR | 2009 | Peek | Lancet | 180 | Transfer to ECMO center | Conventional | Death/severe disability, 6mo | 37% vs 53% | RR 0.69 (0.05–0.97) | 0.03 | Only 76% referred actually got ECMO | Low–Moderate | Low–Moderate | 19762075 | Referral to a specialist ECMO center improves outcome in severe respiratory failure — a pragmatic, not purely ECMO-specific, result. |
OSCILLATE | 2013 | Ferguson | NEJM | 548 | HFOV | Conventional | In-hospital mortality | 47% vs 35% (HARM) | RR 1.33 (1.09–1.64) | 0.005 | Stopped early for harm | High – ended HFOV use | High (harm) | 23339639 | Do NOT use HFOV routinely in adult moderate-severe ARDS — it kills. |
OSCAR | 2013 | Young | NEJM | 795 | HFOV | Conventional | 30-day mortality | 41.7% vs 41.1% (NS) | — | 0.85 | — | High | Moderate | 23339638 | Confirms HFOV offers no benefit over conventional ventilation. |
FLORALI | 2015 | Frat | NEJM | 310 | HFNC | Standard O2 / NIV | Intubation rate 28d | 38% vs 47% vs 50% (NS) | — | 0.18 | Intubation criteria subjective | Moderate | Moderate | 25981908 | HFNC is a reasonable first-line respiratory support in hypoxemic non-hypercapnic failure; may reduce mortality vs NIV. |
Driving pressure meta-analysis | 2015 | Amato | NEJM | 3562 | Lower driving pressure | Higher | 60-day survival | ΔP strongest mortality mediator | RR 1.41 per 1SD ΔP↑ (1.31–1.51) | <0.001 | Post-hoc mediation, not RCT | High – reframed VILI mechanism | Moderate | 25693014 | Target driving pressure (Pplat−PEEP) <15 cmH2O — it predicts mortality better than tidal volume or PEEP alone. |
2. Sepsis & Septic Shock
Rivers' 2001 EGDT trial drove a decade of protocolized resuscitation. By 2014–15, ProCESS/ARISE/ProMISe uniformly refuted mandatory EGDT. Steroids remain contested; vasopressin/angiotensin II added options; HES starches proven harmful.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size (95% CI) | P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
Rivers EGDT | 2001 | Rivers | NEJM | 263 | 6-h EGDT protocol | Standard care | In-hospital mortality | 30.5% vs 46.5% | — | 0.009 | Single-center; superseded | Low (superseded) | 11794169 | Historic proof that early aggressive resuscitation matters — but the SPECIFIC protocol (CVP/ScvO2 targets) is no longer required. |
ProCESS | 2014 | ProCESS Investigators | NEJM | 1341 | Protocol EGDT/standard | Usual care | 60-day mortality | 21.0% vs 18.2% vs 18.9% (NS) | — | 0.83 | Academic sepsis-aware centers | High | 24635773 | Rigid EGDT protocol (central lines, ScvO2 monitoring) is unnecessary if usual care already includes early fluids/antibiotics. |
ARISE | 2014 | ARISE Investigators | NEJM | 1600 | EGDT | Usual care | 90-day mortality | 18.6% vs 18.8% | RR 0.98 (0.80–1.21) | 0.90 | Lower than expected mortality | High | 25272316 | Confirms ProCESS — protocolized EGDT adds no benefit over modern usual care. |
ProMISe | 2015 | Mouncey | NEJM | 1260 | EGDT | Usual care | 90-day mortality | 29.5% vs 29.2% (NS) | — | 0.90 | — | High | 25776532 | Third confirmatory trial — EGDT protocol officially retired from guidelines. |
ANDROMEDA-SHOCK | 2019 | Hernández | JAMA | 424 | Capillary-refill guided | Lactate guided | 28-day mortality | 34.9% vs 43.4% | HR 0.75 (0.55–1.02) | 0.06 | Frequentist negative; Bayesian favors benefit | Moderate | 30772908 | Simple bedside capillary refill may guide resuscitation as well as (or better than) serial lactate. |
CENSER | 2019 | Permpikul | AJRCCM | 310 | Early norepinephrine | Standard (placebo) | Shock control at 6h | 76.1% vs 48.4% | — | <0.001 | Single-country; surrogate endpoint | Moderate | 30704260 | Starting norepinephrine EARLY (even peripherally) speeds shock resolution — don't wait for central access or "enough" fluids. |
VASST | 2008 | Russell | NEJM | 778 | Vasopressin + norepi | Norepinephrine | 28-day mortality | 35.4% vs 39.3% (NS) | — | 0.26 | Subgroup hypothesis-generating | Moderate | 18305265 | Vasopressin as a norepinephrine-sparing add-on is safe; possible benefit in less-severe shock (hypothesis only). |
ATHOS-3 | 2017 | Khanna | NEJM | 321 | Angiotensin II | Placebo | MAP response at 3h | 69.9% vs 23.4% | — | <0.001 | Surrogate endpoint; no mortality benefit shown | Moderate | 28528561 | Angiotensin II effectively raises BP in refractory vasodilatory shock — a third-line vasopressor option, not proven to save lives. |
ADRENAL | 2018 | Venkatesh | NEJM | 3800 | Hydrocortisone 200mg/d | Placebo | 90-day mortality | 27.9% vs 28.8% | OR 0.95 (0.82–1.10) | 0.50 | Faster shock reversal only | High | 29347874 | Steroids speed up shock reversal (fewer ICU days) but do NOT reduce mortality. |
APROCCHSS | 2018 | Annane | NEJM | 1241 | Hydrocortisone + fludrocortisone | Placebo | 90-day mortality | 43.0% vs 49.1% | RR 0.88 (0.78–0.99) | 0.03 | Factorial disrupted by drotrecogin withdrawal | Moderate | 29490185 | Hydrocortisone + fludrocortisone combo DID show a mortality benefit — conflicts with ADRENAL; steroid debate remains unresolved. |
CORTICUS | 2008 | Sprung | NEJM | 499 | Hydrocortisone | Placebo | 28-day mortality (nonresponders) | 39.2% vs 36.1% (NS) | — | 0.69 | Underpowered | Moderate | 18184957 | No mortality benefit from steroids; ACTH stim test does not predict who benefits — abandon it as a selection tool. |
VITAMINS | 2020 | Fujii | JAMA | 211 | Vit C+thiamine+hydrocortisone | Hydrocortisone | Vasopressor-free time (7d) | No difference | — | 0.83 | Open-label | Moderate | 31950979 | Adding vitamin C/thiamine to hydrocortisone adds no benefit — the "HAT therapy" hype is not supported. |
CITRIS-ALI | 2019 | Fowler | JAMA | 167 | High-dose vitamin C | Placebo | mSOFA at 96h | No difference (primary) | — | 0.86 | Underpowered mortality | Moderate | 31573637 | High-dose IV vitamin C does not improve organ dysfunction scores in sepsis-related ARDS. |
6S | 2012 | Perner | NEJM | 798 | HES 130/0.42 | Ringer's acetate | 90-day death/dialysis | 51% vs 43% (HARM) | RR 1.17 (1.01–1.36) | 0.03 | — | High (harm) | 22738085 | HES starches cause harm (death/dialysis) in septic patients — avoid entirely. |
CHEST | 2012 | Myburgh | NEJM | 7000 | HES 130/0.4 | Saline | 90-day mortality | 18.0% vs 17.0% (NS); more RRT with HES | RR 1.06 (0.96–1.18) | 0.26 | More RRT with HES | High | 23075127 | Confirms HES increases renal replacement therapy need — starches have no place in modern ICU resuscitation. |
SAFE | 2004 | SAFE Study | NEJM | 6997 | 4% albumin | Saline | 28-day mortality | 20.9% vs 21.1% (NS) | RR 0.99 (0.91–1.09) | 0.87 | Harm signal in TBI subgroup | High | 15163774 | Albumin is as safe as saline for general resuscitation — but avoid albumin in traumatic brain injury. |
3. Hemodynamics & Fluids
Saline was default for decades; SMART/SALT-ED shifted practice toward balanced crystalloids, later tempered by neutral PLUS/BaSICS. FEAST is a landmark cautionary negative trial (fluid boluses increased mortality in African children).
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size (95% CI) | P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
SMART | 2018 | Semler | NEJM | 15802 | Balanced crystalloids | Saline | MAKE30 | 14.3% vs 15.4% | OR 0.91 (0.84–0.99) | 0.04 | Single-center, cluster, unblinded | Moderate–High | 29485925 | Balanced crystalloids (LR/Plasma-Lyte) modestly reduce major adverse kidney events vs saline in general ICU patients. |
SALT-ED | 2018 | Self | NEJM | 13347 | Balanced crystalloids | Saline | Hospital-free days | No diff; MAKE30 4.7% vs 5.6% | OR 0.82 (0.70–0.95) | 0.01 | Non-ICU ED patients | Moderate | 29485926 | Same balanced-fluid benefit extends to non-critically-ill ED patients receiving IV fluids. |
PLUS | 2022 | Finfer | NEJM | 5037 | Balanced (Plasma-Lyte) | Saline | 90-day mortality | 21.8% vs 22.0% (NS) | — | 0.90 | Neutral – tempers SMART | High | 35041780 | Large confirmatory trial: fluid choice (balanced vs saline) probably does NOT change mortality — pick either reasonably. |
BaSICS | 2021 | Zampieri | JAMA | 11052 | Balanced (Plasma-Lyte) | Saline | 90-day mortality | 26.4% vs 27.2% (NS) | HR 0.97 (0.90–1.05) | 0.47 | Neutral | High | 34375394 | Second large neutral trial — reinforces that fluid type effect on mortality, if any, is small. |
65 trial | 2020 | Lamontagne | JAMA | 2600 | Permissive hypotension (MAP 60–65) | Usual vasopressors | 90-day mortality | 41.0% vs 43.8% (NS; adj sig) | Adj OR 0.82 (0.68–0.98) | — | Age ≥65 only; unblinded | Moderate | 32049269 | In patients ≥65, a lower MAP target (60–65) is safe and reduces vasopressor/fluid exposure. |
SEPSISPAM | 2014 | Asfar | NEJM | 776 | High MAP (80–85) | Low (65–70) | 28-day mortality | 36.6% vs 34.0% (NS) | HR 1.07 (0.84–1.38) | 0.57 | Chronic HTN subgroup: less AKI at high MAP | Moderate | 24635770 | Standard MAP goal of 65 is appropriate for most; consider higher target only in chronically hypertensive patients. |
CLASSIC | 2022 | Meyhoff | NEJM | 1554 | Fluid restriction | Standard fluid | 90-day mortality | 42.3% vs 42.1% (NS) | RR 1.00 (0.89–1.13) | 0.96 | — | High | 35709019 | A restrictive IV fluid strategy after initial resuscitation is safe in septic shock — supports "less is more" once stabilized. |
FEAST | 2011 | Maitland | NEJM | 3141 | Fluid bolus | No bolus | 48-h mortality | 10.5%/10.6% vs 7.3% (bolus HARM) | RR 1.45 (1.13–1.86) | 0.003 | Pediatric, resource-limited Africa | High | 21615299 | Landmark warning: fluid boluses can KILL in resource-limited pediatric febrile illness — do not extrapolate Western ICU fluid habits universally. |
ESCAPE (PAC) | 2005 | Binanay | JAMA | 433 | Pulmonary artery catheter | Clinical assessment | Days alive out of hospital | No difference; more PAC AEs | — | NS | Advanced HF only | Moderate | 16234501 | Routine PAC-guided therapy in decompensated heart failure adds risk without outcome benefit. |
4. Transfusion & Hematology
The "10/30 rule" dominated until TRICC (1999) established restrictive transfusion (Hb 7) as standard, extended by TRISS/FOCUS/TRIPICU. CRASH-2 established TXA in trauma; PATCH warned against platelets in antiplatelet-associated ICH.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size (95% CI) | P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
TRICC | 1999 | Hébert | NEJM | 838 | Restrictive (Hb 7) | Liberal (Hb 10) | 30-day mortality | 18.7% vs 23.3% (NS overall) | — (−0.84 to 10.2) | 0.11 | Underpowered; benefit in younger/less-ill subgroups | High | 9971864 | Transfusing at Hb 7 (not 10) is at least as safe — founded restrictive transfusion as the ICU default. |
TRISS | 2014 | Holst | NEJM | 998 | Restrictive (Hb 7) | Liberal (Hb 9) | 90-day mortality (septic shock) | 43% vs 45% (NS) | RR 0.94 (0.78–1.09) | 0.44 | — | High | 25270275 | Restrictive transfusion (Hb 7) is safe even in septic shock specifically. |
FOCUS | 2011 | Carson | NEJM | 2016 | Restrictive (Hb 8/sx) | Liberal (Hb 10) | Death/inability to walk (60d) | 35.2% vs 34.7% (NS) | OR 1.01 (0.84–1.22) | 0.90 | Elderly hip surgery w/ CV disease | High | 22168590 | Restrictive transfusion is safe even in elderly patients with cardiovascular disease after hip fracture surgery. |
TRIPICU | 2007 | Lacroix | NEJM | 637 | Restrictive (Hb 7) | Liberal (Hb 9.5) | New/progressive MODS | 12% vs 12% (NS) | — | NS | Stable pediatric ICU | High | 17409370 | Restrictive transfusion strategy extends safely to stable critically ill children. |
PROPPR | 2015 | Holcomb | JAMA | 680 | Plasma:platelet:RBC 1:1:1 | 1:1:2 | 24h & 30-day mortality | NS (12.7% vs 17.0%; 22.4% vs 26.1%) | — | 0.12 (30d) | Primary NS; hemostasis better 1:1:1 | Moderate | 25647203 | 1:1:1 ratio achieves faster hemostasis and fewer early deaths from bleeding, even though overall mortality was not statistically different. |
CRASH-2 | 2010 | CRASH-2 Collaborators | Lancet | 20211 | Tranexamic acid | Placebo | 28-day mortality | 14.5% vs 16.0% | RR 0.91 (0.85–0.97) | 0.0035 | Mostly resource-limited settings | High | 20554319 | Give TXA within 3 hours of major trauma bleeding — it saves lives; later administration may harm. |
CRASH-3 | 2019 | CRASH-3 Collaborators | Lancet | 12737 | TXA (TBI) | Placebo | Head-injury death (28d) | 18.5% vs 19.8% (overall NS) | RR 0.94 | — | Benefit in mild-moderate TBI subgroup | Moderate | 31623894 | TXA likely helps in mild-moderate TBI given early; benefit less clear in severe TBI. |
PATCH | 2016 | Baharoglu | Lancet | 190 | Platelet transfusion (antiplatelet ICH) | Standard care | Death/dependence (mRS shift, 3mo) | Worse with platelets | Adj OR 2.05 (1.18–3.56) | 0.0114 | Small; spontaneous supratentorial ICH only | Moderate (harm) | 27178479 | Do NOT give platelets for antiplatelet-associated intracerebral hemorrhage — it worsens outcome. |
5. Renal / AKI / CRRT
Single-center ELAIN suggested early RRT benefit; large multicenter trials (AKIKI, IDEAL-ICU, STARRT-AKI) refuted routine early/accelerated initiation. RENAL and VA/NIH ATN showed higher-intensity RRT dosing confers no survival benefit.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size (95% CI) | P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
AKIKI | 2016 | Gaudry | NEJM | 619 | Early RRT | Delayed RRT | 60-day mortality | 48.5% vs 49.7% | — | 0.79 | Unblinded; 49% of delayed never needed RRT | High | 27352532 | Waiting for a clear indication before starting RRT is safe and avoids unnecessary dialysis. |
AKIKI 2 | 2021 | Gaudry | Lancet | 278 | Delayed | More-delayed | 60-day RRT-free days | No benefit; possible harm signal | — | — | Second delay may harm | Moderate | 34023006 | Pushing delay even further (beyond standard indications) offers no benefit and may be harmful — there is a limit to "watchful waiting." |
STARRT-AKI | 2020 | STARRT-AKI Investigators | NEJM | 2927 | Accelerated RRT | Standard RRT | 90-day mortality | 43.9% vs 43.7% | RR 1.00 (0.93–1.09) | — | More RRT-dependence with accelerated | High | 32668114 | Largest trial confirms: accelerated RRT initiation offers no benefit and increases long-term dialysis dependence. |
IDEAL-ICU | 2018 | Barbar | NEJM | 488 | Early RRT (septic shock) | Delayed | 90-day mortality | 58% vs 54% (NS) | — | 0.38 | Stopped for futility | High | 30304656 | Early RRT specifically in septic shock-associated AKI shows no benefit — watchful waiting applies here too. |
ELAIN | 2016 | Zarbock | JAMA | 231 | Early RRT | Delayed | 90-day mortality | 39.3% vs 54.7% | HR 0.66 (0.45–0.97) | 0.03 | Single-center; surgical; contradicts AKIKI/STARRT | Low–Moderate | 27209269 | The single-center signal favoring early RRT did NOT replicate in larger trials — treat with caution. |
RENAL | 2009 | RENAL Study | NEJM | 1508 | Higher-intensity CRRT (40mL/kg/h) | Lower (25) | 90-day mortality | 44.7% vs 44.7% | — | 0.99 | — | High | 19846848 | Standard-dose CRRT (~25 mL/kg/h) is sufficient — higher intensity does not improve survival. |
ATN (VA/NIH) | 2008 | Palevsky | NEJM | 1124 | Intensive RRT | Less-intensive | 60-day mortality | 53.6% vs 51.5% (NS) | OR 1.09 (0.86–1.40) | 0.47 | — | High | 18492867 | Confirms RENAL — intensive RRT dosing does not reduce mortality; standard dosing is appropriate. |
6. Sedation, Analgesia, Delirium
Continuous deep sedation with benzodiazepines was standard until linked to delirium and prolonged ventilation. Daily interruption, paired SAT+SBT, light sedation, dexmedetomidine, and early mobilization (ABCDEF bundle) reshaped practice.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size/P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
SPICE III | 2019 | Shehabi | NEJM | 4000 | Early dexmedetomidine | Usual sedation | 90-day mortality | 29.1% vs 29.1% (NS) | HR 1.00; P=0.98 | More bradycardia; supplemental sedation often needed | High | 31112380 | Early dexmedetomidine as sole sedative does not improve survival — use it for specific indications, not routinely for all. |
MENDS | 2007 | Pandharipande | JAMA | 106 | Dexmedetomidine | Lorazepam | Delirium/coma-free days | 7.0 vs 3.0 days | P=0.01 | Small; single-domain | Moderate | 18073360 | Dexmedetomidine reduces delirium/coma days compared to benzodiazepine sedation. |
MENDS2 | 2021 | Hughes | NEJM | 422 | Dexmedetomidine | Propofol (septic, ventilated) | Days alive w/o delirium/coma | No difference | — | Neutral | Moderate | 33528922 | Dexmedetomidine is not superior to propofol for delirium-free days in ventilated septic patients — either is reasonable. |
SEDCOM | 2009 | Riker | JAMA | 375 | Dexmedetomidine | Midazolam | Time at target sedation | Similar; less delirium (54% vs 76.6%) | P<0.001 (delirium) | Industry-funded | Moderate | 19188334 | Dexmedetomidine reduces delirium prevalence vs midazolam while achieving similar sedation quality. |
ABC trial | 2008 | Girard | Lancet | 336 | Paired SAT+SBT | SBT + usual sedation | Ventilator-free days (28d) | 14.7 vs 11.6 days | P=0.02; 1-yr mortality benefit | Unblinded | High | 18191684 | Pair daily awakening trials WITH breathing trials — this combination reduces vent days and improves 1-year survival. |
Kress SI | 2000 | Kress | NEJM | 128 | Daily sedation interruption | Continuous | Duration of vent | 4.9 vs 7.3 days | P=0.004 | Single-center | Moderate | 10816184 | Daily sedation interruption shortens ventilator duration — the foundational "sedation vacation" trial. |
Schweickert (early mobilization) | 2009 | Schweickert | Lancet | 104 | Early PT/OT + SAT | SAT alone | Independent function at discharge | 59% vs 35% | P=0.02 | Small; single-center | Moderate | 19446324 | Early physical/occupational therapy combined with sedation interruption improves functional independence at discharge. |
HOPE-ICU | 2013 | Page | Lancet Respir Med | 141 | Haloperidol | Placebo | Delirium/coma-free days | No difference (8 vs 9) | P=0.66 | Single-center | Moderate | 23628572 | Prophylactic haloperidol does not prevent delirium in the critically ill. |
MIND-USA | 2018 | Girard | NEJM | 566 | Haloperidol/ziprasidone | Placebo | Days alive w/o delirium/coma | No difference | — | Refutes antipsychotics for ICU delirium | High | 30346242 | Antipsychotics (haloperidol or ziprasidone) do NOT treat or shorten ICU delirium — abandon routine use for this indication. |
7. Nutrition
Early aggressive nutrition was standard, but EPaNIC showed late PN improved outcomes. Trophic/permissive underfeeding matched full feeding; glutamine (REDOXS) proved harmful; NUTRIREA-2 flagged caution with early enteral nutrition in shock.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size/P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
EPaNIC | 2011 | Casaer | NEJM | 4640 | Early PN (day 3) | Late PN (day 8) | ICU discharge alive/earlier | Late PN better | Adj OR 1.06; P=0.04 | Belgian high-glucose context | High | 21714640 | Withholding parenteral nutrition until day 8 (allowing tolerated caloric deficit) speeds recovery vs early PN. |
EDEN | 2012 | Rice (ARDSNet) | JAMA | 1000 | Trophic enteral | Full enteral | Ventilator-free days | No difference | P=0.89 | ALI patients | High | 22307571 | Trophic (low-volume) enteral feeding is equivalent to full-dose feeding in early ALI — no need to push full calories immediately. |
PermiT | 2015 | Arabi | NEJM | 894 | Permissive underfeeding | Standard feeding | 90-day mortality | 27.2% vs 28.9% (NS) | RR 0.94; P=0.58 | — | High | 25896428 | Permissive underfeeding (~40-60% target) is as safe as standard full feeding. |
TARGET | 2018 | Chapman | NEJM | 3957 | Energy-dense enteral | Routine | 90-day mortality | 26.8% vs 25.7% (NS) | RR 1.05; P=0.41 | — | High | 30346225 | Delivering more calories via energy-dense formula does not improve survival — caloric intensity is not the lever that matters. |
REDOXS | 2013 | Heyland | NEJM | 1223 | Glutamine ± antioxidants | Placebo | 28-day mortality | 32.4% vs 27.2% (trend HARM) | Adj OR 1.28; P=0.05 | Harmful in multi-organ failure | High (harm) | 23594003 | Do NOT give glutamine supplementation in critically ill patients with multi-organ failure — signal of harm. |
CALORIES | 2014 | Harvey | NEJM | 2400 | Parenteral route | Enteral route | 30-day mortality | 33.1% vs 34.2% (NS) | RR 0.97; P=0.57 | Route, not dose | High | 25271389 | Route of early nutrition (parenteral vs enteral) does not affect mortality when calories are matched. |
NUTRIREA-2 | 2018 | Reignier | Lancet | 2410 | Early enteral | Early parenteral (shock) | 28-day mortality | 37% vs 35% (NS); more GI ischemia w/ enteral | HR 1.02; P=0.33 | Caution with early enteral in shock | High | 29128300 | In patients on vasopressors for shock, early enteral feeding carries a higher risk of GI ischemic complications — consider delaying enteral feeds until shock is controlled. |
8. Neurocritical Care
Corticosteroids for TBI were routine until MRC CRASH showed increased death. Decompressive craniectomy (DECRA harmful early; RESCUEicp reduces death but increases severe disability survival). Endovascular thrombectomy revolutionized stroke care.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size (95% CI) | P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
HYPERION | 2019 | Lascarrou | NEJM | 581 | Hypothermia 33°C | Normothermia 37°C | 90-day CPC 1–2 (nonshockable arrest) | 10.2% vs 5.7% | Diff 4.5pp (0.1–8.9) | 0.04 | Fragile; open-label | Moderate | 31577396 | Targeted hypothermia (33°C) may improve neurologic outcome after nonshockable-rhythm cardiac arrest — the one subgroup where TTM's benefit persists post-TTM2. |
DECRA | 2011 | Cooper | NEJM | 155 | Early decompressive craniectomy | Standard care | 6-mo GOS-E | Unfavorable 70% vs 51% (worse w/ surgery) | OR 1.84 (1.05–3.24, worse) | 0.03 | Baseline pupil imbalance; early diffuse injury | Moderate | 21434843 | Early prophylactic decompressive craniectomy for diffuse TBI WORSENS functional outcome — do not do this routinely. |
RESCUEicp | 2016 | Hutchinson | NEJM | 408 | Decompressive craniectomy | Medical (barbiturates) | 6-mo GOS-E | Death 26.9% vs 48.9%; more vegetative/severe disability survivors | — | <0.001 | Non-proportional odds; more disability among survivors | Moderate | 27602507 | Last-tier decompressive craniectomy for refractory raised ICP reduces death but trades this for more survivors with severe disability — discuss this tradeoff explicitly with families. |
INTERACT2 | 2013 | Anderson | NEJM | 2839 | Intensive BP (<140) | Guideline (<180) | Death/major disability (90d) | 52.0% vs 55.6% | OR 0.87 (0.75–1.01) | 0.06 (primary NS); ordinal P=0.04 | Primary NS; ordinal secondary sig | Moderate | 23713578 | Early intensive BP lowering (<140) in acute ICH is safe and may modestly improve functional outcome. |
ATACH-2 | 2016 | Qureshi | NEJM | 1000 | Intensive BP (110–139) | Standard (140–179) | Death/disability (90d) | 38.7% vs 37.7% | RR 1.04 (0.85–1.27) | 0.72 | Neutral; more renal AEs intensive | High | 27276234 | Pushing BP even lower (110–139) in ICH offers no added benefit over 140–179 and adds renal risk. |
MISTIE III | 2019 | Hanley | Lancet | 506 | Minimally invasive surgery + alteplase | Standard care | mRS 0–3 at 365d | 45% vs 41% | Adj RD 4% (−4 to 12) | 0.33 | Neutral primary; mortality reduced (secondary) | High | 30739747 | Minimally invasive clot evacuation in ICH is safe and reduces mortality, though it did not meet its primary functional-outcome endpoint. |
ESETT | 2019 | Kapur | NEJM | 384 | Levetiracetam/fosphenytoin/valproate | (3-arm) | Seizure cessation+alertness at 60min | 47% vs 45% vs 46% (equivalent) | Bayesian adaptive | — | No clear "winner" | High | 31774955 | For benzodiazepine-refractory status epilepticus, levetiracetam, fosphenytoin, and valproate are equally effective second-line agents — choose based on side-effect profile/availability. |
MRC CRASH (steroids) | 2004 | Roberts | Lancet | 10008 | Methylprednisolone 48h | Placebo | 2-week death | 21.1% vs 17.9% (HARM) | RR 1.18 (1.09–1.27) | 0.0001 | Stopped early for harm | High (harm) | 15474134 | Corticosteroids INCREASE death after TBI — never give routine steroids for head injury. |
CRASH (6-mo) | 2005 | Edwards | Lancet | 10008 | Methylprednisolone | Placebo | 6-mo death | 25.7% vs 22.3% | RR 1.15 (1.07–1.24) | 0.0001 | Confirms harm | High (harm) | 15936423 | Confirms the harm persists at 6 months — this ended corticosteroid use in TBI worldwide. |
MR CLEAN | 2015 | Berkhemer | NEJM | 500 | Endovascular thrombectomy + usual | Usual care | mRS at 90d (shift) | mRS 0–2: 32.6% vs 19.1% | cOR 1.67 (1.21–2.30) | — | Landmark positive; first of HERMES trials | High | 25517348 | Endovascular thrombectomy for large-vessel occlusion stroke dramatically improves functional outcome — revolutionized acute stroke care. |
ESCAPE-NA1 | 2020 | Hill | Lancet | 1105 | Nerinetide + thrombectomy | Placebo + thrombectomy | mRS 0–2 at 90d | 61.4% vs 59.2% (NS) | Adj RR 1.04 (0.96–1.14) | 0.35 | Neuroprotectant neutral; alteplase interaction | High | 32087818 | Nerinetide neuroprotection adds no benefit on top of thrombectomy — the search for an effective stroke neuroprotectant continues. |
9. Cardiac Critical Care
SHOCK established early revascularization for cardiogenic shock. IABP counterpulsation, long assumed beneficial, was refuted by IABP-SHOCK II. CULPRIT-SHOCK reversed the trend toward complete revascularization.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size (95% CI) | P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
SHOCK | 1999 | Hochman | NEJM | 302 | Early revascularization | Initial medical stabilization | 30-day mortality | 46.7% vs 56.0% (NS at 30d) | RR 0.83 | 0.11 (30d); sig at 6mo/1yr | 30-day NS; benefit emerged at 6mo | Moderate | 10460813 | Early revascularization for MI-associated cardiogenic shock improves longer-term survival — established emergency PCI as standard of care. |
IABP-SHOCK II | 2012 | Thiele | NEJM | 600 | Intra-aortic balloon pump | No IABP | 30-day mortality | 39.7% vs 41.3% (NS) | RR 0.96 (0.79–1.17) | 0.69 | Neutral at 30d, 1yr, 6yr | High | 22920912 | IABP does NOT improve survival in cardiogenic shock — ended its routine use despite decades of prior belief. |
CULPRIT-SHOCK | 2017 | Thiele | NEJM | 706 | Culprit-only PCI | Immediate multivessel PCI | Death/RRT at 30d | 45.9% vs 55.4% | RR 0.83 (0.71–0.96) | 0.01 | Driven by mortality; open-label | High | 29083953 | In multivessel disease with cardiogenic shock, treat the culprit lesion ONLY at the index procedure — immediate complete revascularization is harmful. |
IABP-SHOCK II 6-yr | 2018 | Thiele | Circulation | 591 | IABP | Control | 6-yr mortality | 66.3% vs 67.0% (NS) | RR 0.99 (0.88–1.11) | 0.98 | Confirms no long-term benefit | High | 30586721 | Long-term follow-up confirms IABP has no mortality benefit at any time horizon. |
DOREMI | 2021 | Mathew | NEJM | 192 | Milrinone | Dobutamine | Composite (death, arrest, etc.) | No difference (49% vs 54%) | RR 0.90 (0.69–1.19) | 0.47 | Small; single-center | Moderate | 34379922 | Milrinone and dobutamine are clinically equivalent inotrope choices for cardiogenic shock. |
ECMO-CS | 2023 | Ostadal | Circulation | 122 | Immediate VA-ECMO | Early conservative | Composite death/arrest/other MCS | No difference | — | NS | Small; ECPR evidence still evolving | Moderate | 36335478 | Immediate VA-ECMO vs a conservative early strategy in cardiogenic shock shows no clear advantage yet — mechanical support strategy remains an evolving field. |
10. Infectious Disease / Antibiotics
Activated protein C was approved after PROWESS (2001) but withdrawn after PROWESS-SHOCK (2012)—a landmark reversal. Procalcitonin-guided algorithms and short-course antibiotics reduced exposure without worsening outcomes.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size/P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
PROWESS | 2001 | Bernard | NEJM | 1690 | Drotrecogin alfa (activated) | Placebo | 28-day mortality | 24.7% vs 30.8% | RR 0.81; P=0.005 | Later refuted; withdrawn from market | Low (superseded) | 11236773 | Once hailed as the first "sepsis drug" — now a cautionary tale; see PROWESS-SHOCK. |
PROWESS-SHOCK | 2012 | Ranieri | NEJM | 1697 | Drotrecogin alfa | Placebo | 28-day mortality | 26.4% vs 24.2% (NS) | RR 1.09; P=0.31 | Refutes PROWESS; drug withdrawn | High | 22616830 | Activated protein C does NOT reduce mortality — the drug was withdrawn from the market worldwide. |
PRORATA | 2010 | Bouadma | Lancet | 621 | Procalcitonin-guided antibiotics | Standard | Mortality & antibiotic-free days | Noninferior mortality; more antibiotic-free days (14.3 vs 11.6) | P<0.0001 (abx-free) | Open-label | Moderate | 20096452 | Procalcitonin-guided antibiotic discontinuation safely shortens antibiotic duration in the ICU. |
VAP 8 vs 15 days | 2003 | Chastre | JAMA | 401 | 8-day antibiotics | 15-day (VAP) | Death/recurrence | Noninferior; more relapse w/ non-fermenting GNB | — | Longer duration debated for Pseudomonas | Moderate | 14625335 | 8 days of antibiotics is enough for most VAP — except consider longer courses for non-fermenting gram-negatives (e.g., Pseudomonas). |
SAPS (PCT de-escalation) | 2016 | de Jong | Lancet Infect Dis | 1546 | PCT-guided discontinuation | Standard | Antibiotic duration & 28-d mortality | Shorter duration; lower mortality (20% vs 25%) | P=0.0122 | Secondary mortality finding hypothesis-generating | Moderate | 27333368 | PCT-guided stewardship shortened antibiotics AND reduced mortality — reinforces PRORATA's antibiotic-sparing message. |
DALI | 2014 | Roberts | Clin Infect Dis | 384 | (PK study) | — | β-lactam target attainment | 16% did not achieve targets | — | Observational PK study, not RCT | Low | 24429437 | Standard β-lactam dosing frequently fails to achieve target concentrations in critically ill patients — individualized/extended-infusion dosing may be needed. |
11. Endocrine
Van den Berghe's Leuven trials suggested tight glucose control (80–110) improved outcomes. The larger NICE-SUGAR reversed this, showing tight control increased mortality; current targets are ≤180 mg/dL.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size (95% CI) | P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
NICE-SUGAR | 2009 | NICE-SUGAR Study | NEJM | 6104 | Intensive glucose (81–108) | Conventional (≤180) | 90-day mortality | 27.5% vs 24.9% (HARM) | OR 1.14 (1.02–1.28) | 0.02 | Definitive; severe hypoglycemia 6.8% vs 0.5% | High (harm) | 19318384 | Tight glucose control (81–108) INCREASES mortality via severe hypoglycemia — target ≤180 mg/dL instead. |
Leuven I (surgical) | 2001 | Van den Berghe | NEJM | 1548 | Tight glucose (80–110) | Conventional | ICU mortality | 4.6% vs 8.0% | RR ~0.57 | <0.04 | Single-center; later refuted | Low (superseded) | 11794168 | Originally suggested tight glucose control saves lives in surgical ICU patients — subsequently overturned by NICE-SUGAR. |
Leuven II (medical) | 2006 | Van den Berghe | NEJM | 1200 | Tight glucose | Conventional | In-hospital mortality | 37.3% vs 40.0% (NS overall) | — | 0.33 | Benefit only in ≥3-day stayers; hypoglycemia | Low (superseded) | 16452557 | Benefit was inconsistent (only in longer-stay patients) even before NICE-SUGAR definitively refuted tight control. |
ADRENAL | 2018 | Venkatesh | NEJM | 3800 | Hydrocortisone | Placebo | 90-day mortality | 27.9% vs 28.8% (NS) | OR 0.95 (0.82–1.10) | 0.50 | See Sepsis section | High | 29347874 | See Sepsis table — steroids speed shock reversal but do not change mortality. |
12. GI (Stress Ulcer Prophylaxis)
Routine acid suppression was ubiquitous, but concern grew over pneumonia/C. diff risk. SUP-ICU and PEPTIC showed no mortality difference between agents/placebo; PPIs reduce bleeding modestly.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size/P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
SUP-ICU | 2018 | Krag | NEJM | 3298 | Pantoprazole | Placebo | 90-day mortality | 31.1% vs 30.4% (NS) | RR 1.02; P=0.76 | GI bleed reduced (secondary); high-severity subgroup harm signal | High | 30354950 | PPI stress-ulcer prophylaxis reduces clinically important GI bleeding but does not change mortality — use is reasonable but not mandatory in all patients. |
PEPTIC | 2020 | PEPTIC Investigators | JAMA | 26828 | PPI | H2 receptor blocker | In-hospital mortality (90d) | 18.3% vs 17.5% (NS) | OR 1.05 (1.00–1.10) | — | Cluster crossover; contamination | High | 31950977 |
REVISE | 2024 | Deane/Cook | NEJM | 4821 | Pantoprazole | Placebo | 90-day mortality | No sig diff; less clinically important GI bleeding w/ PPI | — | Confirms bleeding benefit, neutral mortality | High | Verify PMID | Largest, most recent trial confirms: PPI reduces important GI bleeding without a mortality difference. |
13. Trauma
Trauma resuscitation shifted from crystalloid-heavy to hemostatic/balanced blood-product resuscitation. CRASH-2 established early TXA; PROPPR examined transfusion ratios; PAMPer showed prehospital plasma benefit; REBOA remains investigational.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size/P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
CRASH-2 | 2010 | CRASH-2 Collaborators | Lancet | 20211 | Tranexamic acid | Placebo | 28-day death | 14.5% vs 16.0% | RR 0.91; P=0.0035 | Benefit only if given <3h | High | 20554319 | Give TXA within 3 hours of trauma — the earlier the better; delayed TXA may be harmful. |
CRASH-3 | 2019 | CRASH-3 Collaborators | Lancet | 12737 | TXA (TBI) | Placebo | Head-injury death | 18.5% vs 19.8% (overall NS) | RR 0.94 | Mild-moderate subgroup benefit | Moderate | 31623894 | TXA likely benefits mild-moderate TBI when given promptly. |
PROPPR | 2015 | Holcomb | JAMA | 680 | 1:1:1 ratio | 1:1:2 | 24h/30-day mortality | NS (12.7% vs 17.0%; 22.4% vs 26.1%) | P=0.12 (30d) | Primary NS; better hemostasis 1:1:1 | Moderate | 25647203 | 1:1:1 blood product ratio achieves better early hemostasis in massive transfusion for trauma. |
PAMPer | 2018 | Sperry | NEJM | 501 | Prehospital plasma | Standard care | 30-day mortality | 23.2% vs 33.0% | Diff −9.8pp; P=0.03 | Long-transport air-medical setting | Moderate | 30044935 | Giving plasma in the prehospital phase (long-transport air-medical) improves survival in trauma. |
UK-REBOA | 2023 | Jansen | JAMA | 90 | REBOA + standard | Standard care | 90-day mortality | Higher with REBOA (54% vs 42%) | OR 1.58; possible harm | Small; stopped; wide CI | Low | 37824132 | REBOA showed a signal of HARM in this trial — use should remain highly selective, not routine, pending more data. |
14. COVID-19
The platform trials RECOVERY and REMAP-CAP rapidly transformed practice: dexamethasone became the first mortality-reducing therapy; IL-6 inhibition added benefit; remdesivir shortened recovery without clear mortality benefit.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size (95% CI) | P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
RECOVERY (dexamethasone) | 2021 | RECOVERY Group | NEJM | 6425 | Dexamethasone 6mg | Usual care | 28-day mortality | 22.9% vs 25.7% | RR 0.83 (0.75–0.93) | <0.001 | Open-label; benefit only in O2/ventilated | High | 32678530 | Dexamethasone reduces mortality in COVID-19 patients requiring oxygen/ventilation — no benefit (possible harm) if not hypoxic. |
RECOVERY (tocilizumab) | 2021 | RECOVERY Group | Lancet | 4116 | Tocilizumab | Usual care | 28-day mortality | 31% vs 35% | RR 0.85 (0.76–0.94) | 0.0028 | Open-label | High | 33933206 | Adding tocilizumab (IL-6 inhibitor) to steroids further reduces mortality in hypoxic COVID-19. |
REMAP-CAP (IL-6 inhibitors) | 2021 | Gordon | NEJM | 803 | Tocilizumab/sarilumab | Control | Organ support-free days | Median 10 vs 0; improved survival | OR 1.64 (1.25–2.14) | — | Bayesian adaptive; critically ill | High | 33631065 | Confirms IL-6 blockade improves organ-support-free days and survival in critically ill COVID-19. |
ACTT-1 | 2020 | Beigel | NEJM | 1062 | Remdesivir | Placebo | Time to recovery | 10 vs 15 days | RR 1.29 (1.12–1.49) | <0.001 | Mortality benefit NS | High | 32445440 | Remdesivir shortens time to clinical recovery but does not clearly reduce mortality. |
RECOVERY-RS | 2022 | Perkins | JAMA | 1273 | CPAP vs HFNO vs conventional O2 | — | Intubation/death (30d) | CPAP reduced vs conventional (36.3% vs 44.4%); HFNO no benefit | — | 0.03 (CPAP) | Under-recruited; pandemic constraints | Moderate | 34874419 | CPAP reduces intubation/death in COVID-19 respiratory failure better than conventional oxygen; HFNO showed no clear advantage. |
15. ECMO & Extracorporeal Therapies
ECMO for respiratory failure gained traction after CESAR and matured with EOLIA into a conditional therapy for severe ARDS. Blood-purification approaches for sepsis have largely failed to show mortality benefit.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size/P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
EOLIA | 2018 | Combes | NEJM | 249 | Early VV-ECMO | Conventional + rescue ECMO | 60-day mortality | 35% vs 46% | RR 0.76 (0.55–1.04); P=0.09 | 28% control crossover; futility stop | Moderate | 29791822 | See ARDS table — reasonable rescue therapy in very severe ARDS despite a formally "negative" frequentist result. |
CESAR | 2009 | Peek | Lancet | 180 | Referral to ECMO center | Conventional | Death/severe disability (6mo) | 37% vs 53% | RR 0.69 (0.05–0.97); P=0.03 | Only 76% received ECMO; pragmatic | Low–Moderate | 19762075 | Transfer to a specialized ECMO-capable center improves outcomes in severe respiratory failure. |
EUPHRATES | 2018 | Dellinger | JAMA | 450 | Polymyxin B hemoperfusion | Sham | 28-day mortality | 37.7% vs 34.5% (NS) | OR 1.14; P=0.49 | Endotoxin-directed subgroups hypothesis-generating | High | 30326459 | Polymyxin B hemoperfusion does not reduce mortality in septic shock overall — blood purification remains unproven. |
16. ICU Rehabilitation / Long-Term Outcomes
Recognition of "post-intensive care syndrome" reframed ICU care beyond survival. Early mobilization, ABCDEF bundle, and ICU diaries emerged as strategies to improve functional/psychological recovery, though evidence is mixed for more intensive rehab.
Trial | Year | Author | Journal | N | Intervention | Comparator | Primary Outcome | Result | Effect Size/P | Bias/Weaknesses | GRADE | PMID | Take-Home Message |
Schweickert | 2009 | Schweickert | Lancet | 104 | Early PT/OT + SAT | SAT alone | Independent functional status at discharge | 59% vs 35% | P=0.02 | Small, single-center | Moderate | 19446324 | Early mobilization combined with sedation interruption improves functional independence at hospital discharge. |
ICU diary (Jones) | 2010 | Jones | Critical Care | 352 | ICU diary | No diary | New-onset PTSD at 3mo | 5% vs 13% | P=0.02 | Small; blinding difficult | Low–Moderate | 20504360 | An ICU diary kept by staff/family reduces new-onset PTSD symptoms after critical illness — a simple, low-cost intervention. |
TEAM | 2022 | Hodgson | NEJM | 750 | Increased early mobilization | Usual care | Days alive & out of hospital (180d) | No difference; more AEs w/ intervention | — | Neutral; tempers aggressive mobilization | High | 36286256 | More intensive early mobilization beyond usual care does NOT improve outcomes and may increase adverse events — don't over-escalate mobilization protocols. |
CYCLE | 2022 | Kho | (multicenter RCT) | 360 | In-bed cycling + usual PT | Usual PT | Physical function (PFIT-s) at discharge | No significant difference | — | Neutral | Moderate | Verify PMID | Adding in-bed cycling to standard physiotherapy does not further improve physical function at discharge. |
Verification Flags & Superseded Evidence
A. Not fully verified in this research pass (treat as provisional pending direct PMID check):
- REVISE (GI, 2024): conclusions well-established from reporting, exact N/percentages/PMID unconfirmed.
- CYCLE (Rehabilitation): neutral result reported, exact PFIT-s values/PMID unconfirmed.
- Several PMIDs drawn from secondary summaries (ARISE, ProMISe, PLUS, BaSICS, CLASSIC, DOREMI, ECMO-CS, UK-REBOA, TEAM, Jones ICU diary, PRORATA, SAPS, EUPHRATES) were not independently re-fetched — high confidence but verify before publication.
B. Directly verified against primary source or high-fidelity summary quoting primary: ARMA, PROSEVA, EOLIA, ACURASYS, ROSE, ART, ANDROMEDA-SHOCK, ADRENAL, APROCCHSS, SMART, SALT-ED, TRICC, CRASH-2/3, AKIKI, STARRT-AKI, NICE-SUGAR, SUP-ICU, PEPTIC, IABP-SHOCK II, CULPRIT-SHOCK, RECOVERY (dexamethasone), TTM/TTM2, HYPERION, DECRA, RESCUEicp, INTERACT2, ATACH-2, MISTIE III, ESETT, MRC CRASH, MR CLEAN, ESCAPE-NA1, ProCESS.
C. Superseded or refuted landmark results (critical for interpretation):
- TTM2 (2021) superseded TTM (2013): TTM found no difference between 33°C and 36°C; TTM2 (N=1900) then compared 33°C vs targeted normothermia/fever avoidance — 6-month mortality 50% vs 48% (RR 1.04, 95% CI 0.94–1.14, P=0.37), establishing hypothermia confers no benefit over fever control. PMID TTM2: 34133859. HYPERION (nonshockable rhythm) remains the outlier suggesting possible subgroup benefit.
- ROSE (2019) refuted ACURASYS (2010): early routine paralysis is no longer standard — reserved as rescue.
- ProCESS/ARISE/ProMISe (2014–15) refuted Rivers EGDT (2001): protocolized EGDT confers no benefit over good usual care.
- PROWESS-SHOCK (2012) refuted PROWESS (2001): activated protein C withdrawn from market.
- NICE-SUGAR (2009) refuted Leuven tight-glucose trials: targets of ≤180 mg/dL are now standard.
- AKIKI/IDEAL-ICU/STARRT-AKI refuted routine early RRT (ELAIN): indication-driven initiation is standard.
- ART (2017) and OSCILLATE (2013) demonstrated HARM: aggressive recruitment/PEEP titration and HFOV increase mortality.
- TEAM (2022) tempered aggressive early mobilization from the Schweickert (2009) enthusiasm.
- PLUS (2022) and BaSICS (2021) tempered SMART (2018): balanced crystalloids remain a reasonable default but proven mortality benefit is likely small.
Recommendations: Before relying on this for exams/publication, confirm every "verify"-flagged cell and secondary-summary PMID directly against PubMed. Interpretive columns (bias, external validity, GRADE, take-home message) are analytic judgments, not verbatim trial data. This is a snapshot as of July 2026 — the ECPR, mechanical circulatory support, and post-ICU rehabilitation fields are evolving rapidly and warrant periodic refresh.