- Included 2388 patients requiring nutritional support in England who were admitted to the ICU within the past 36 hours and expected to remain for at least 3 days
- Randomized patients to either 25 kcal/kg of parenteral or enteral nutrition for 5 total days (or until transition to exclusive oral feeding or ICU discharge)
- This was a very pragmatic trial — decisions regarding feeding formulation/composition, threshold for gastric residuals, use of prokinetic agents, and other factors were at the discretion of the clinicians
- Primary endpoint (30-day mortality) was not different between parenteral and enteral nutrition (33.1% vs. 34.2%, p=0.57)
- The study met 90% power to detect a 20% relative risk reduction in 30-day mortality
- Unlike in previous trials, parenteral nutrition was not associated with a greater risk of “infectious complications” per patient (0.22 vs. 0.21, p=0.72), although because the trial was not blinded, there is risk for bias
- There were no differences in SOFA scores, length of stay, or 90-day mortality
- Parenteral nutrition was associated with slightly higher mean calories per day (21.3 vs. 18.5 kcal/kg, no p value); the majority of patients did not meet the 25 kcal/kg goal
- Enteral nutrition was associated with more “clinically significant hypoglycemia (based on clinical judgment, 3.7% vs. 6.2%, p=0.006), although “serious” hypoglycemia was not different (0.4% vs. 0.3%)
- Enteral nutrition was associated with more vomiting (8.4% vs. 16.2%, p<0.001) and prokinetic use (2.2% vs. 35.6%), although there was no difference in the rate of aspiration events or pneumonia
- As mentioned by the authors, there may be a dose-dependent risk of infection with parenteral nutrition. Given that most patients did not achieve the goal number of calories, the risk of infection with parenteral nutrition may have been minimized.
- No cost analysis was done, which has been shown to be a major difference in other trials (eg, EPaNIC)
Among a heterogeneous ICU patient population, there was no difference in mortality between exclusive enteral and parenteral nutrition support.
- Included 21,106 Chinese patients with ischemic stroke within previous 48 hrs
- Randomized to aspirin 160 mg PO daily or placebo x 28 days
- Primary endpoint (28-day mortality) was reduced with aspirin (3.9% vs. 3.3%, p=0.04, NNT 167)
- Recurrent ischemic stroke was reduced with aspirin (1.6% vs. 2.1%, p=0.01, NNT 200)
- Extracranial bleeding was rare but more common with aspirin (0.8% vs. 0.6%, p=0.02, NNH 500)
In patients with acute ischemic stroke, aspirin within 48 hours reduced 28-day mortality and recurrent ischemic stroke, although bleeding events were rare but slightly more common with aspirin.
- Included 4071 patients in China within 48 hours of ischemic stroke with systolic BP between 140-220 mmHg (mean baseline about 165/97 mmHg)
- Excluded patients who received thrombolytic therapy (TPA), diastolic BP > 120 mmHg, atrial fibrillation, and unstable angina
- Randomized patients to antihypertensive treatment (“treatment”) or usual care (“control”). Participants were encouraged to remain hospitalized for 10 days.
- “Treatment” consisted of lowering systolic BP by 10-25% within the first 24 hours (no lower than 126/80) and a goal of < 140/90 mmHg within 7 days. An algorithm using an ACE inhibitor, calcium channel blocker, and diuretic was used to manage antihypertensive regimens.
- “Control” consisted of no antihypertensive medications during hospitalization, but would be able to resume antihypertensives following hospital discharge
- Primary endpoint (14-day mortality or major disability) was not different between “treatment” and “control” (33.6% vs. 33.6%, p=0.98)
- The trial was adequately powered to detect a 5% absolute difference in the primary endpoint
- At 3 months, there was no difference in mortality or major disability (25.2% vs. 25.3%, p=0.93)
- As would be expected, “treatment” patients had lower systolic BP at 24 hours (144.7 vs. 152.9 mmHg), 7 days (137.3 vs. 146.5 mmHg), and 14 days (135.2 vs. 143.7 mmHg) [p<0.001]
- Based on the NIHSS scores (median 4), patient in this trial had less severe symptoms at baseline compared to other stroke trials (eg, NINDS trial median of 14 — higher number indicating worse severity). This may limit the external validity of the trial for patients with more severe stroke.
- Despite a low percentage of patients with embolic stroke (about 5%), use of heparin or low molecular weight heparin was common (33.4%), which is reflective of the typical practice in China (again, a possible risk to external validity for US-based patients)
Among patients with ischemic stroke who do not receive TPA, more aggressive blood pressure reduction during hospitalization does not improve mortality or major disability.
- Included 180 patients with potentially reversible ARDS and a Murray score of 3 or greater
- Excluded those with peak inspiratory pressure > 30 or FiO2 > 80% for more than 7 days or those with contraindications to anticoagulation
- Randomized to conventional ventilator management or referral to Glenfield Hospital with the intent to initiate ECMO (extracorporeal membrane oxygenation)
- Those randomized to “ECMO” were transferred to a single center. On transfer, an ARDS protocol was initiated for 12 hours. If improvement was not observed, percutaneous venovenous ECMO was initiated
- Primary endpoint (survival without severe disability at 6 months) was improved with randomization to ECMO (63% vs. 47%, p=0.03, NNT 6)
- There is a significant lack of safety analysis for ECMO. Five (6%) died prior to or during transport
- Most patients were randomized early (0-48 hrs after intubation), and the degree of ARDS was very severe (mean PaO2/FiO2 ratio of 75)
- 17 patients (19%) were randomized to the ECMO center, but improved with an ARDS protocol and did not require cannulation
- A lack of a standardized protocol in the control is a serious limitation of the study.
- 30% of patients in the control group did not use a lung-protective ventilation strategy (eg, ARDSNet protocol), which has been shown to improve mortality
- Importantly, this is a study of referral to a tertiary care center specializing in ARDS with ECMO capability compared to conventional ventilation at individual centers.
- This study is not a comparison of positive pressure ventilation versus ECMO.
In patients with early ARDS, transfer to a facility specializing in ARDS with the ability to initiate ECMO was associated with an improvement in 6-month survival without severe disability.
- Included 7,000 intensive care unit patients who required fluid resuscitation (a very heterogeneous patient population)
- Excluded patients with dialysis-dependent renal failure or intracranial hemorrhage
Randomized to 6% HES 130/0.4 (Voluven) or 0.9% normal saline for the duration of the patient’s ICU stay. In patients requiring more than 50 mL/kg/day, open-label saline was used
- Primary endpoint (90-day all-cause mortality) was not different between HES and saline (18% vs. 17%, p=0.26)
- This study was powered to detect an absolute difference of 3% in mortality
- Renal replacement therapy was more common with HES (7% vs. 5.8%, NNH 83, p=0.04)
- Surrogate markers of renal function (RIFLE criteria, urine output, serum creatinine) were also worse with HES compared to saline
- HES patients required more blood products (78 vs. 60 mL, p<0.001) and had a higher incidence of rash and pruritus (5.3% vs. 2.8%, p<0.001)
- The overall blinded ratio of HES:saline volume was 1:1.17. Thus, HES was 17% more “potent” to blinded clinicians. This ratio is much lower than was seen with albumin in the SAFE trial (1:1.4 or 40%)
For ICU patients requiring fluid resuscitation, hydroxyethyl starch (HES 130/0.4) was equivalent to normal saline in 90-day mortality, but HES increased the risk of renal failure and the need for blood products.
- Included 1369 pediatric ICU patients (age 36 weeks – 16 years) with an arterial catheter, mechanical ventilation, AND vasoactive drugs after an injury, major surgery, or critical illness. About 60% of patients were less than 1 year old and about 60% of patients were admitted following cardiac surgery.
- Randomized patients to tight glycemic control (72-126 mg/dL) or conventional control (BG 180-216 mg/dL). All blood glucose values were obtained with arterial catheters whenever possible.
- Primary endpoint (number of days alive and free from mechanical ventilation at 30-days) was not different between tight and conventional glycemic control groups (mean -0.36 days [95% CI -0.42 to 1.14])
- There was no difference in 12-month mortality
- Renal replacement therapy was more common with conventional therapy (13.5% vs. 8.9%, OR 0.63 [95% CI 0.45-0.89]); however, this secondary endpoint is at risk for a type-I error due to the number of secondary endpoints analyzed
- Moderate hypoglycemia (BG 36-45 mg/dL) and severe hypoglycemia (BG < 36 mg/dL) were both more common with tight glycemic control (12.5% vs. 3.1% and 7.3% vs. 1.5%, p<0.001)
- Among the entire cohort, hypoglycemia was associated with an increase in mortality (11.1% vs. 4.4%, p=0.001)
- Conventional therapy was associated with a longer hospital length of stay (by 13.5 days) and higher costs (by $13,120) in non-cardiac surgery patients. Cardiac surgery patients had no difference in length of stay or costs.
- Hyperglycemia was uncommon in the conventional group. At 10 days after randomization, mean (SD) blood glucose values were 105 (21) mg/dL vs. 112 (33) mg/dL (p=0.041). Although significantly different, it seems unlikely that a mean difference of about 7 mg/dL would have a profound clinical impact. It is not known whether tight glycemic control would have an effect on the primary endpoint among patients with true hyperglycemia (eg, above 180 mg/dL)
- Only about 7% of those assessed for eligibility met study criteria, which reduces external validity. However, the study criteria selected the “sickest” type of patients in a pediatric ICU, presumably the subgroup of patients that would be most likely to benefit from the study treatment.
Among critically ill pediatric patients, tight glycemic control did not reduce 30-day mortality or duration of mechanical ventilation, but it did increase the risk of hypoglycemia. For unclear reasons, non-cardiac surgery patients had a shorter 12-month hospital length of stay and hospital cost. The study is limited by a lack of hyperglycemia in the control group
- 2×2 factorial, open-label trial including 509 patients receiving hydrocortisone 50 mg IV Q6h x 7 days for septic shock
- Randomized patients to receive conventional (<150 mg/dL) vs. intensive (80-110 mg/dL) glucose control and fludrocortisone 50 mcg PO daily vs. placebo x 7 days
- Primary endpoint (hospital mortality) was not different between conventional and intensive glucose control (45.9% vs. 42.9%, p=0.50) or fludrocortisone and placebo (42.9% vs. 45.8%, p=0.50)
- Hypoglycemia (< 40 mg/dL) was more common with intensive glucose control (16.4% vs. 7.8%, p=0.003, NNH 12)
- New infections (primarily UTI) were more common with fludrocortisone than placebo (21.6% vs. 14%, p=0.02, NNH 13)
- This study helped clarify the role of intensive glucose control in septic shock (van den Berghe and NICE-SUGAR included all types of ICU patients), and the role of fludrocortisone as an adjunct to hydrocortisone in septic shock (Annane 2002 vs. CORTICUS)
Among patients with septic shock receiving hydrocortisone, neither intensive glucose control nor fludrocortisone improved mortality. Intensive glucose control was associated with a higher incidence of hypoglycemia and fludrocortisone with a higher incidence of new infections.
- Included 45,852 patients with acute MI (87% STEMI) in Chinese hospital
- Excluded those with systolic BP < 100 mmHg, heart rate < 50 bpm, heart block, cardiogenic shock, or primary PCI (study was 2×2 factorial with clopidogrel)
- Randomized to placebo vs. metoprolol IV 5 mg x 3 doses Q3min (unless HR < 50 or SBP < 90), then metoprolol 50 mg PO Q6hr (days 0-1), then metoprolol succinate XL 200 mg daily (days 2-28)
- Neither co-primary endpoint death/reinfarction/cardiac arrest nor 28-day mortality was different between metoprolol and placebo
- Death due to arrhythmias was more common with placebo (1.7% vs. 2.2%, p=0.0002, NNT 200)
- Reinfarction (2.0% vs. 2.5%, p=0.001, NNT 200) and ventricular fibrillation (2.5% vs. 3.0%, p=0.001, NNT 200) was more common with placebo than metoprolol
- Death due to shock was more common with metoprolol (2.2% vs. 1.7%, p=0.0002, NNH 200)
- Metoprolol was more associated with heart failure, hypotension, shock, and bradycardia compared to placebo
- Risk factors for cardiogenic shock included: age > 70 years, presenting systolic BP < 120 mmHg, presenting HR > 110 bpm, and Killip class III
- The authors conclude that early metoprolol should not routinely be used in acute MI, particularly in certain high-risk groups, but should be considered once hemodynamic stability occurs
The use of early, aggressive-dose metoprolol in acute MI decreased arrhythmias and reinfarction, but increased cardiogenic shock.
- Included 499 patients with septic shock within the past 72 hours. (Annane 2002 enrolled patients within 8 hours of shock)
- Randomized patients to placebo or hydrocortisone (without fludrocortisone)
- All patients received a stim test. Non-responders were those with < 9 mcg/dL increase in serum cortisol at 60 minutes
- Hydrocortisone regimen: 50 mg IV Q6h x 5 days, then 50 mg IV Q12h (days 6-8), then 50 mg IV Q24h (days 9-11), then stop
- Primary endpoint (28-day mortality among non-responders) was not different between hydrocortisone or placebo (39.2% vs. 36.1%, p=0.69)
- Mortality among responders was not different (28.8% vs. 28.7%, p=1.00)
Time to reversal of shock (SBP > 90 mmHg without vasopressors for at least 24 hrs) was quicker with hydrocortisone (3.3 vs. 5.8 days) for both responders and non-responders
Patients receiving hydrocortisone had a higher incidence of new infections occurring 48+ hrs after study drug (OR 1.37, 95% CI 1.05 to 1.79), hyperglycemia, and hypernatremia
Study was underpowered for the primary endpoint (only enrolling 499 of a projected 800 for adequate power). Given the lack of effect seen, study is at high risk for type-II error
Hydrocortisone therapy did not improve outcomes among patients with septic shock (onset within 72 hours), although it did shorten the duration of vasopressor dependence.
- Included 20,211 trauma patients with significant hemorrhage (or at risk) who were within 8 hours of initial injury and at least 16 years old
- Randomized to receive tranexamic acid (1 gm loading dose over 10 minutes, then 1 gm infusion over 8 hrs) or placebo
- Tranexamic acid is a lysine derivative that blocks the lysine site on plasminogen (thus inhibiting fibrinolysis)
- Primary endpoint (hospital mortality within 4 weeks of injury) was reduced with tranexamic acid (14.5% vs. 16.0%, p=0.0035, NNT 66)
- Death due to hemorrhage was also reduced with tranexamic acid (4.9% vs. 5.7%, p=0.0077, NNT 125)
- Interestingly, there was no difference in blood transfusions (50.4% vs. 51.3%, p=0.21) or the median number of blood products transfused (3 vs. 3 units, p=0.59).
- This lack of difference in transfusion brings the beneficial mechanism of tranexamic acid into question
- Safety analysis did not reveal an increase in vascular occlusion (MI, CVA, PE) between tranexamic acid and placebo (1.7% vs. 2.0%, p=0.084)
Tranexamic acid reduced all-cause mortality in a broad population of trauma patients without an increase in vascular occlusion complications.
- Included 2,857 patients requiring fluid resuscitation for hypovolemia (hypotension with low cardiac filling pressures and signs of hypoxia/hypoperfusion). Patients could NOT have received any form of fluid resuscitation in the ICU prior to the study enrollment.
- Patients were primarily medical (about 70%) with sepsis (55%) or non-traumatic hypovolemic shock (40%)
- Randomized patients to crystalloids or colloids. Investigators were non-blinded and allowed to select from a variety of fluid options. In the crystalloid group, normal saline was used in 85% of cases and Ringers lactate in 18%. In the colloid group, hydroxyethyl starch was used in 69% of patients, gelatins in 35%, albumin-4% in 6%, and albumin-20% in 14%. The dose of fluid was also at the discretion of the unblinded investigator.
- Primary endpoint (28-day mortality) was not different between crystalloids and colloids (27.0% vs. 25.4%, p=0.26). The study was halted early due to futility — an additional 153 patients were planned based on the initial power calculation.
- 90-day mortality was higher in patients receiving crystalloids (34.2% vs. 30.7%, p=0.03, NNT 29). ICU and hospital mortality trended towards benefit with colloids but were not statistically significant.
- Subgroup analysis indicates that 90-day mortality was driven by HES (hydroxyethyl starch) having a greater mortality benefit compared to isotonic saline (HR 0.79, 95% CI 0.66-0.95). Previous studies (CHEST, 6S), which were not secondary subgroup analyses, refute this finding
- Appropriately, the authors conclude that the benefit in 90-day mortality is exploratory and requires further study
- Colloids were associated with a modest reduction in days alive without mechanical ventilation in the first 7 days (2.1 vs. 1.8 days, p=0.01) and 28 days (14.6 vs. 13.5, p=0.01)
- Colloids were associated with a modest reduction in days alive without vasopressors in the first 7 days (5 vs. 4.7 days, p=0.04) and 28 days (16.2 vs. 15.2 days, p=0.03)
- Within the first 7 days, patients receiving crystalloids demonstrated a higher median cumulative fluid intake (3000 mL vs. 2000 mL, p<0.001). Because the trial was not blinded, the potency ratio (1:1.5) should be considered with caution.
- Given the fact that about half of the patients had severe sepsis or septic shock, the median total fluid received in the first 7 days (3000 mL crystalloid) appears to be quite low, which questions the severity of hypovolemia in the patients
- Unlike previous colloids or hetastarch trials, there was no difference between the two groups regarding blood transfusions or renal replacement therapy
- Given the heterogeneity of the type of fluids used in this trial and the conflicting data with previous trials (SAFE, CHEST, 6S, and more), the topic of crystalloids versus colloids remains controversial
In a heterogeneous ICU population with hypovolemia, there was no difference in 28-day mortality between crystalloids and colloids. Colloids did demonstrate benefit in duration of mechanical ventilation, vasopressor use, and 90-day mortality, although these were secondary endpoints that are conflicting with previous literature and deserve further study.
- Included 197 non-immunocompromised, culture-positive patients with late-onset (>5 days on vent) ventilator-associated pneumonia (VAP) or early onset (>48 hrs on vent) with recent antibiotic exposure
- Randomized patients to receive 8 days or 15 days of appropriate antibiotic therapy
- Primary outcome (28-day all-cause mortality) was non-inferior between 8 and 15 days (18.8% vs. 17.2%, 90% CI -3.7% to 6.9%)
- 28-day pneumonia recurrence was non-inferior between 8 and 15 days (28.9% vs. 26.0%, 90% CI -3.2% to 9.1%)
- On day 1, about 90% of patients received double coverage with a B-lactam plus either quinolone or aminoglycoside
- In subgroup analysis, there was no difference in recurrence for MRSA, but there was a higher incidence of recurrence (40.6% vs. 25.4%) for nonfermenting gram negative rods (Pseudomonas, Acinetobacter, Stenotrophomonas)
- Study may have been underpowered (power estimate of 40% mortality, actual rate closer to 20%)
Among patients with late-onset VAP or early-onset VAP with recent antibiotic exposure, 8-days was non-inferior to 15-days of appropriate antibiotic therapy for mortality and pneumonia recurrence. In patients with certain non-fermenting gram negatives, a 15-day course may be more appropriate.