Screen Shot 2018-10-04 at 14.09.01.png

Prehospital IM midazolam vs. IV Iorazepam for status epilepticus

  • Included 893 children and adults with status epilepticus (seizures lasting more than 5 minutes) who were still seizing at the time of paramedic arrival
  • Randomized to pre-hospital paramedic administration of either intramuscular (1M) midazolam 10 mg via auto-injector to the outer thigh or intravenous (IV) Iorazepam 4 mg.
    • For children weighing 13 to 40 kg, the dose was reduced by 50% (5 mg or 2 mg was given).
    • Children estimated to weigh less than 13 kg were excluded.
  • If seizures persistent > 10 minutes after the study drug, “rescue” therapy was allowed (based on EMS protocol, but would usually involve a benzodiazepine)
  • Primary endpoint (absence of seizure at the time of emergency department arrival without rescue therapy) was superior with midazolam versus lorazepam (73.4% vs. 63.4%, p<0.001 for superiority, NNT 10)
  • There were no differences regarding the need for endotracheal intubation or recurrent seizure within 12 hours, but those randomized to midazolam were less likely to require hospitalization (RR 0.88, 95% CI 0.65-0.95) and ICU admission (RR 0.79, 95% CI 0.65-0.95)
  • Of those randomized to Iorazepam, 33% did not actually receive Iorazepam either due to seizures stopping before the medication could be given or an inability to obtain IV access.
  • In comparison, only about 1% of midazolam patients did not actually receive active drug (due to auto- injector malfunction)
  • As expected, the time to receipt of the IM injection was much quicker than the time taken to establish an IV line and give the IV medication (1.2 vs. 4.8 minutes).
  • Once given, IV Iorazepam was quicker in stopping seizure activity compared to IM midazolam (1.6 vs. 3.3 minutes).
  • Taken together, the overall time to cessation of convulsions was similar.
  • Although the study was technically designed as a non-inferiority study, it demonstrated both non-inferiority and superiority of midazolam compared to Iorazepam.
  • It is important to recognize that this was a pre-hospital study.
  • For patients in the hospital with established IV access, IV Iorazepam is likely a better choice given that it did demonstrate a quicker onset of seizure cessation versus midazolam once the drug was actually given.
  • Because the study used an auto- injector for IM midazolam, it is possible that the results may not be as externally valid in prehospital settings that use a conventional IM injection (without an auto-injector). A delay in midazolam administration time could negate some of the benefit seen in this study.


In the prehospital setting in patients with status epilepticus, IM midazolam given with an auto-injector was superior to establishing an IV and giving IV Iorazepam in stopping seizures at the time of arrival to the emergency department.

Seer the paper here




Screen Shot 2018-10-07 at 10.27.38

High or low gastric residual threshold for feeding intolerance

  • Included 322 intubated and mechanically ventilated patients receiving enteral nutrition via nasogastric tube from 28 Spanish ICUs
  • Randomized patients to a “high” gastric residual threshold (500 mL) or a control group with a “standard” residual threshold (200 mL).
  • Residuals were checked Q6hr on day 1, Q8hr on day 2, and daily thereafter as long as nutrition was tolerated.
  • A standard management nomogram was used for all patients with an elevated gastric residual (based on their randomized group), including holding enteral nutrition, using prokinetic agents, decreasing the feeding rate, and converting to a nasojejunal tube.
  • Primary endpoint (gastrointestinal complications – abdominal distention, high gastric residual, vomiting, diarrhea) was better with the “high” residual threshold (63.6% vs. 47.8%, p=0.004); however, the composite endpoint was driven specifically by the “high gastric residual” component (42.4% vs. 26.8%, p=0.003).
  • All other components of the composite were not statistically different.
  • In a related trial (NUTRIREA1), vomiting was more common among patients who did not receive gastric residual monitoring (39.6%).
  • In REGANE, vomiting was numerically less common in the high residual group (10.8% vs. 14.5%, p=0.31) with both groups having much less vomiting than in NUTRIREA1.
  • “High” residual patients received a slightly higher percent of prescribed nutrition during the first week (84.5% vs. 88.2%, p<0.01), although this small difference is not clinically meaningful.
  • There were no differences in clinical outcomes between the two groups, including the risk of ventilator- associated pneumonia, duration of mechanical ventilation, length of stay, and mortality.
  • All patients received metoclopramide (Reglan) 10 mg IV Q8hr for three days as a “prokinetic prophylaxis” — this may limit the external validity of the study at centers who do not routinely use metoclopramide



A higher gastric residual threshold of 500 mL during enteral nutrition appears to be safe and does not increase the risk of gastrointestinal complications or ventilator-associated pneumonia.

See the paper here



Screen Shot 2018-10-07 at 10.34.31.png

Higher vs. lower-intensity continuous renal replacement therapy

  • Included 1508 patients with acute kidney injury requiring renal- replacement therapy from 35 ICUs in Australia and New Zealand
  • Randomized patients to receive continuous venovenous hemodiafiltration (CVVHDF) with an effluent rate of 40 mL/kg/hr (“higher- intensity”) or 25 mL/kg/hr (“lower- intensity”)
  • Primary endpoint (90-day mortality) was not different between higher- intensity and lower-intensity arms (44.7% vs. 44.7%, p=0.99)
  • There were no differences demonstrated in 28-day mortality, RRT dependence among survivors, number of RRT days, length of stay, duration of mechanical ventilation, and non-renal SOFA subscores
  • As expected, higher-intensity achieved a higher mean effluent rate (33.4 mL/kg/hr vs. 22 mL/kg/hr), achieving 84-88% of the prescribed dose
  • Higher-intensity patients were more likely to require slightly more CRRT filters per day (0.93 vs. 0.84, p<0.001) and experience hypophosphatemia (65.1% vs. 54%, p<0.001) — both endpoints have questionable clinical relevance
  • There was no difference in mortality in a pre-specified subgroup analysis of patients with severe sepsis, who made up about 50% of the entire study cohort



In patients requiring continuous renal replacement therapy, higher-intensity CVVHDF (40 mL/kg/hr) did not improve any clinical endpofits compared to lower-intensity (25 mL/kg/hr) therapy but was associated with a higher filter replacement rate and hypophosphatemia.

See the paper here




Screen Shot 2018-10-07 at 11.47.20

Decompressive craniectomy in traumatic brain injury

  • Included 408 patients with severe traumatic brain injury and refractory, elevated intracranial pressure (>> 25 mmHg for 1 to 12 hours despite conventional therapies for ICP management) from 20 countries, although the majority (71%) were from the United Kingdom
  • Medical therapy for ICP management included stage 1 therapies (head elevation, hypocapnia, analgesia, paralysis, etc.) and stage 2 therapies (mannitol, hypertonic saline, mop diuretics, hypothermia, etc.)
  • Randomized to either “stage 3” decompressive craniectomy (either hemicraniectomy or bifrontal craniectomy based on the type of injury) or “stage 3” ongoing medical therapy within 4 to 6 hours after randomization (typically the addition of a barbiturate in addition to stage 1 and 2 therapies)
  • Primary endpoint (rating on the Extended Glasgow Outcome Scale [GOS-E] at 6 months) was different based on ordinal analysis (p=0.005).
  • Surgical patients were less likely to die (26.9% vs. 48.9%), but were more likely to be vegetative (8.5% vs- 2.1%) or severely disabled (37.3% vs. 22.3%). “Favorable outcome.” at 6 months (based on GOS-E) was not different between surgical and medical patients (42.8% vs. 34.6%, p=0.12).
  • At 12-months, GOS-E was different based on ordinal analysis (p<0.001). “Favorable outcome” at 12 months was more common with surgical patients than medical patients (45.4% vs. 32.4%, p=0.01).
  • Following randomization, surgical patients has a lower median ICP (14.5 vs. 17.1 mmHg, p<0.001) and a shorter duration of ICP > 25 mmHg (5 vs. 17 hours, p<0.001).
  • Other measures of ICP also indicated that surgical patients had a smaller. proportion of time with an elevated ICP value (P<0-001).
  • Barbiturates (a “stage 3” therapy) were commonly used in medical patients (87%) and were used as a cross-over rescue therapy among 9.4% surgical patients.
  • Craniectomy (primarily bifrontal) was done in 93% of surgical patients and
    was done as a cross-over rescue therapy about 37% medical patients (a relatively high cross-over rate that may have diluted the treatment effect of craniectomy)
  • Given the overall effect of craniectomy being a trade-off exchanging mortality for severe disability, the decision to pursue decompression may not be an obvious choice — the input of family members and possibly documentation like a
    living will may be useful in the decision to pursue surgery
  • The DECRA study appears to conflict with RESCUEicp, but DECRA actually had a very different design. In DECRA, craniectomy was used much earlier (as soon as 15 minutes of elevated ICP) among less severely ill patients (as demonstrated by a 6-month mortality of about 19% vs. 38%). DECRA and RESCUEicp highlight the importance of selecting the appropriate patient for craniectomy.



Among traumatic brain injury patients with a sustained elevation in intracranial pressure despite conventional medical therapy, the use of decompressive craniectomy reduced mortality but increased the risk of long-term severe disability.

See the paper here




Screen Shot 2018-10-07 at 11.52.08.png

Rapid shallow breathing index to predict weaning failure

  • Included 100 patients in a single medical ICU who were considered ready for a weaning trial
  • Physicians were blinded to study design – the decision to extubate was made solely based on the physician’s clinical judgment
  • The study examined many different objective parameters for weaning,
    including the rapid shallow breathing index (RSBT). RSBI is calculated as the respiratory rate (breaths per minute) divided by tidal volume (liters) during a one-minute spontaneous breathing trial (often abbreviated f/Vt)
  • The first 36 study patients (training set) were used to determine the most specific and selective RSBI for extubation success. The next 64 patients were used as a prospective validation set.
  • In the entire study, 60 patients were successfully weaned and 40 patients required reintubation within 24 hours
  • Using a cut-off RSBI of 100 breaths/min/L, patients above this
    value had a 95% likelihood of weaning failure (negative predictive value), whereas patients below this value had an 80% likelihood of weaning success (positive predictive value)
  • RSBI was the most sensitive and specific predictor tested (area under ROC of 0.89i0.05); other indices included CROP index (AUC 0.78i0.06), maximum inspiratory pressure (AUC 0.61+/- 0.07), and minute ventilation (AUC 0.40+/-0.07)
  • Although the decision to extubate is multifactorial, the RSBI is a simple index to help formulate an informed decision, particularly for extubation failure (eg, patients with an RSBI above 100 breaths/min/L are very unlikely to be successfully weaned)


In mechanically ventilated medical ICU Patients, a rapid shallow breathing index (RSBI or f/Vt) cut-off of 100 breaths/min/L was the most sensitive and specific objective measure of extubation success.

See the paper here



Rivers 2001

Screen Shot 2018-10-07 at 12.06.07.png

Early goal-directed therapy for severe sepsis and septic shock

  • Included 263 patients presenting to the ED with severe sepsis or septic shock (SIRS + infection + hypoperfusion).
  • Hypoperfusion defined as SBP < 90 mmHg despite 30 mL/kg fluid bolus or lactate >4 mmol/L
  • Randomized patients to standard of care vs. protocolized early goal- directed therapy (EGDT) prior to ICU admission.
  • ED team was not blinded, but accepting ICU team was blinded to treatment assignment
  • All EGDT patients received central line capable of continuous Cv02 and
    received protocolized care to maximize CVP (goal 8-12 mmHg), MAP (>65 mmHg), Cv02 (>70%), and Hct (>30%)
  • Significant baseline criteria included: lactate 7 mmol/L, anion gap 21 mmol/L, ch02 50%, and 95% receiving appropriate antibiotics
  • Primary endpoint (hospital mortality) was lower with EGDT (30.5% vs. 46.5%, p=0.009, NNT 6). 28-day and 60-day mortality rates were also lower.
  • During the 6-hour resuscitation period, EGDT patients received 5 L of crystalloid, two-thirds received RBC transfusion, more required dobutamine (13.7% vs. 0.8%).
  • Resuscitation metrics (Cv02, lactate clearance, and MAP) were all
    improved with EGDT



Protocolized early goal-directed therapy initiated in the ED in severe sepsis and septic shock patients improved resuscitation parameters and reduced mortality.

See the paper here


Powered by WordPress.com.

Up ↑