“Effect of postextubation high-flow nasal cannula vs conventional oxygen therapy on reintubation in low-risk patients,” JAMA, 2016, Spain

“Effect of postextubation high-flow nasal cannula vs conventional oxygen therapy on reintubation in low-risk patients,” JAMA, 2016, Spain

Question: Does HFNC reduce the need for reintubation in patients at low risk of post-extubation respiratory failure?

Study Type: Multicenter, randomized clinical trial in 7 ICUs in Spain

Study Population: Patients who passed an SBT after at least 12 hrs of mechanical ventilation were eligible if they meet the following inclusion criteria: Age <65, not initially intubated for CHF, absence of mod-severe COPD, APACHE II <12, BMI <30, no known airway problems and low risk of developing laryngeal edema, adequate cough and requiring suctioning <2xs Q8hrs, not difficult to wean, mechanical ventilation <7 days, <2 co-morbidities.  Patients were excluded if they had a tracheostomy or had evidence of hypercapnia during an SBT.

Study Groups: Patients in the intervention arm were placed on HFNC with flow set at 10L/min which was titrated up at 5L/min intervals until pts experienced discomfort.  FIO2 was titrated to keep SpO2 > 92%.  Patients in the control arm had conventional oxygen applied through a facemask or nasal cannula titrated to keep SpO2>92% for 24 hrs.

Primary Outcome: Need for reintubation at 72 hrs.

 Results: 527 patients randomized. Notable patient characteristics: primary neurologic diagnosis (29%), scheduled or urgent surgery at admission (47%), primary respiratory failure (17%).  HFNC significantly reduced the need for reintubation at 72 hours (4.9% vs 12.2%, p=0.004) with a number needed to treat to prevent one reintubation of 14.  Patients treated with HFNC also had lower rates of reintubation secondary to respiratory causes (8.3% vs 14.4%, p=0.03). There was no difference in ICU length of stay or mortality.

Caveats: Primarily neurology or surgery patients (not a typical MICU population), most common cause for reintubation in the control arm was inability to clear secretions which is perhaps a function of having so many pts with neurologic injuries.

 Take-home Point: In a cohort of mostly surgical and neurologic patients at low risk for reintubation, HFNC reduced the need for reintubation compared to conventional oxygen therapy.  It is unclear if these results are generalizable to a more typical MICU population.

 

“Effect of postextubation high-flow nasal cannula vs noninvasive ventilation on reintubation and postextubation respiratory failure in high-risk patients,” JAMA, 2016, Spain

“Effect of postextubation high-flow nasal cannula vs noninvasive ventilation on reintubation and postextubation respiratory failure in high-risk patients,” JAMA, 2016, Spain

 Question: Is high-flow nasal cannula non-inferior to non-invasive ventilation (NIV) in reducing rates of reintubation and postextubation respiratory failure in high risk patients?

Study Type: Multicenter, randomized, non-inferiority trial in 3 ICUs in Spain

 Study Population: Patients receiving mechanical ventilation > 12 hours were eligible for the trial if they met at least one of the following criteria: age >65, CHF as primary indication for MV, mod-severe COPD, APACHE II >12 on day of extubation, BMI >30, airway patency problems, inability to deal with respiratory secretions, difficult weaning, 2 or more co-morbidities, or MV > 7 days.  Exclusions included DNR, tracheostomies, and self-extubations.

 Study Groups: Patients in the HFNC arm were placed on HFNC with flow set at 10L/min which was titrated up at 5L/min intervals until pts experienced discomfort.  FIO2 was titrated to keep SpO2 > 92%.  Patients in the NIV arm had PEEP and IPAP adjusted to targets of RR 25, SaO2 92, and pH 7.35.  FiO2 was titrated to an SpO2 of at least 92%.  Both groups were switched to conventional O2 after 24 hrs.

 Primary Outcomes: Reintubation and postextubation respiratory failure within 72 hrs.  Non-inferiority margin set at 10%

 Results: 604 patients were randomized. Notable patient characteristics: respiratory failure as primary diagnosis (56%), surgical patients (38%), neurology patients (3%), median # of high risk factors (3).  HNFC was non-inferior to NIV (19.1% of pts required reintubation in NIV group vs 22.8% in HFNC group).  More patients in the NIV group experienced postextubation respiratory failure (26.9% HFNC vs 39.8% NIV).  Close to 50% of NIV patients were unable to tolerate therapy for the full 24 hours (median duration 14 hrs).  Other than a small reduction in ICU LOS with HFNC, secondary outcomes were similar between the two groups.

Caveats: Both HFNC and NIV were given for only 24hrs (in contrast to some other studies with longer periods of therapy), sedatives to facilitate NIV tolerance were not allowed, trial was not blinded.

 Take-home Point: In high risk patients, HFNC appears non inferior to NIV in reducing rates of reintubation and postextubation respiratory failure.

 

“Effect of lung recruitment and titrated positive end-expiratory pressure (PEEP) vs low PEEP on mortality in patients with acute respiratory distress syndrome,” JAMA, 2017

“Effect of lung recruitment and titrated positive end-expiratory pressure (PEEP) vs low PEEP on mortality in patients with acute respiratory distress syndrome,” JAMA, 2017, Acute Respiratory Distress Syndrome Trial (ART) Investigators

 Question: Do lung recruitment maneuvers paired with PEEP titrated to best respiratory system compliance improve outcomes for patients with moderate-severe ARDS?

Study Type: Multicenter randomized trial (120 ICUs in 9 countries)

Study Population: Mechanically ventilated patients with moderate-severe ARDS for <72 hours were eligible. Notable exclusion criteria included escalating doses of vasopressors, contraindications to hypercapnia (intracranial hypertension, ACS), and pneumothorax.

Study Groups: Patients in the control arm were managed per the ARMA/ARDSnet protocol (NEJM, 2000). Patients in the intervention arm received a bolus of a neuromuscular blocking agent followed by a recruitment maneuver with subsequent decremental PEEP titration (PEEP decreased by 3 cmH20 every 4 minutes until the highest respiratory system compliance was achieved). If P/F levels were stable or increasing for 24 hrs, PEEP was decreased by 2 cmH20 every 8 hrs. The initial recruitment maneuver utilized the following protocol: pressure control ventilation, Pi 15 cmH20, PEEP 25 cmH20 x 1 min, PEEP 35 cmH20 x 1 min, PEEP 45 cmH20 x 2 min (so breaths on the final step started at a PEEP of 45 cmH20 and increased to a Ppeak of 60 cmH20). After PEEP titration, the patients received a 2nd recruitment maneuver of PEEP 45 cmH20 x 2 minutes. After 556 pts, this protocol was modified after 3 patients in the intervention arm had a cardiac arrest possibly associated with the recruitment maneuver (!!). The revised protocol started at a PEEP of 25 cmH20 x 1 min then 30 cmH20 x 1 min then 35 cmH20 x 1 min with PEEP subsequently decreased in 3 minute intervals to find the highest compliance followed by a final recruitment maneuver of PEEP 35 cmH20 for 2 mins.

https://www.thebottomline.org.uk/summaries/icm/art/

 

Primary Outcome: 28-day mortality

Results: 1,010 patients were included in the final analysis. There were no notable differences between the two groups. 65% of patients had septic shock and 55% had pneumonia. The mean P/F was 119, both groups were receiving mean TVs of 5.8 mL/Kg IBW at enrollment, and 10% received prone positioning. At 28 days, 55.3% of pts in the intervention arm had died compared to 49.3% of pts in the control arm (HR 1.2 95% CI, 1.01-1.42; p=0.41). Patients in the intervention arm had significantly higher rates of death within 6 months, death within 7 days, pneumothorax requiring drainage within 7 days, and need for commencement or increase of vasopressors or hypotension within 1 hr. There were also 3 cases of cardiac arrest in the intervention arm.

Caveats: Study protocol changed in the middle of the study, multiple interventions used in experimental arm (more below)

Take-home Point: For patients with moderate-severe ARDS, a strategy that paired recruitment maneuvers with PEEP titrated to best respiratory system compliance was associated with harm compared to usual care.

Comments:

  • Just so everyone is clear, for patients in the intervention arm, following the recruitment maneuver, the bedside clinician would switch the ventilator mode from AC-PC to AC-VC, drop PEEP from 45 cmH20 to 23 cmH20, wait 4 minutes, perform an inspiratory hold, and calculate compliance (Tv/Plt-PEEP). PEEP would then be decreased by 3 cmH20 and compliance again calculated after 4 minutes. This process was repeated until the PEEP with the highest compliance was identified. At this point, 2 cmH20 was added to that PEEP. The idea here is to “open up” as much of the lung as possible with the recruitment maneuver and then “keep it open” with some high level of PEEP that balances overdistension with de-recruitment.
  • A major issue with interpreting this study is that the intervention arm actually includes multiple interventions (paralytics often paired with IVF, recruitment maneuvers, and a decremental PEEP trial). It is impossible to determine if one of those in particular contributed to the signal of harm in the trial (perhaps obscuring a signal of benefit from one of the other interventions) or if they acted synergistically.
  • As in the ALVEOLI trial, higher levels of PEEP produced a physiologic benefit. Patients in the intervention arm had lower driving pressures, lower Pplts, and higher respiratory system compliance. So why didn’t this improve outcomes? The recruitment maneuvers were very aggressive, up to a PEEP of 45 cmH20 for 2 minutes until the protocol was changed after 3 cardiac arrests. Perhaps the signal of increased pneumothorax and barotrauma was because pts in the intervention arm were being subjected (often repeatedly) to very high airway pressures often combined with IVF which can worsen edema.
  • As with the ALVEOLI trial pts in the control received exceptional care (far better than we do in practice) with a mean Tv <6 cc/kg IBW and Pplt <30. I think these trials show that when the ARMA (ARDSnet) protocol is followed, the additive survival benefit of any single additional vent maneuver is marginal. There is a lot of enthusiasm in the critical care community re: the use of driving pressure (Pplt – PEEP) to set PEEP and guide vent management. I think this trial should temper expectations that such a strategy will dramatically alter outcomes (despite a decrease in driving pressure, more patients in the intervention arm here died).
  • The major PEEP trials are ALVEOLI (reviewed last time), ART (this trial), EXPRESS (JAMA, 2008), and LOVS (JAMA, 2008). All of them failed to show a signal of benefit in their primary outcome and one of the (ART) showed harm. Much has been written about benefits of higher PEEP in subgroups, PEEP responders (defined a bunch of different ways), and when data is pooled in meta-analyses. Those are all interesting but I think it’s really important to understand the major clinical trials, their limitations, and form your own conclusions to inform your practice.

See also:

The Bottom Line

 

Liberation from mechanical ventilation in critically ill adults

“Liberation from mechanical ventilation in critically ill adults: an official American College of Chest Physicians/American Thoracic Society clinical practice guideline,” CHEST, 2017

Question: In acutely hospitalized patients ventilated more than 24 hours, should the spontaneous breathing trial (SBT) be conducted with or without inspiratory pressure augmentation? (1 of 6 questions addressed in the guidelines).

Study Type: Systematic review, meta-analysis, and multidisciplinary expert panel discussion

Methods: Panel co-chairs were selected by CHEST and ATS leadership.  14 panelists were then selected by co-chairs (7 pulm/CCM MDs, 4 CC MDs, 1 CC RN, 1 physical therapist, 1 pharmacist).   Co-chairs drafted 6 key questions which were reviewed by the panelists.  After finalization of search terms, inclusion/exclusion criteria, and databases to be searched, a systematic review was performed by a methodologist.  Studies were screened, reviewed, and extracted by panelists.  Final recommendations required 75% panel participation and 80% consensus.

Studies Included: Systematic reviews, RCTs, or observational studies comparing SBT with inspiratory pressure augmentation (PS or automatic tube compensation) to SBT without pressure augmentation in acutely hospitalized patients ventilated for >24 hrs.  This review excluded patients who did not pass an initial SBT (i.e. the 1994 Brochard and 1995 Esteban RCTs we reviewed were excluded).

Outcomes: Duration of ventilation, ventilator-free days, successful SBT, extubation success (not requiring NIV or reintubation for 48 hrs), duration of ICU stay, short-term mortality (<60 days), and long-term mortality.

Results: 560 articles screened, 14 included in full text review, 5 included in qualitative synthesis, 4 included in quantitative synthesis.  When the 4 trials were pooled through meta-analysis, conducting an SBT with pressure augmentation was more likely to be successful (84.6% vs 76.7%; RR 1.11; 95% CI, 1.02-1.18), produced a higher rate of extubation success (75.4% vs 68.9%; RR, 1.09; 95% CI, 1.02-1.18), and was associated with a trend toward lower ICU mortality (8.6% vs 11.6%; RR, 0.74; 95% CI, 0.45-1.24).

Caveats: Only 4 trials included in analysis (including studies from the Croatian Medical Journal and the American Journal of Medical Sciences), 3 of the 4 trials were single-center studies, all trials non-blinded.

ACCP/ATS Recommendation: “We suggest that the initial SBT be conducted with inspiratory pressure augmentation (5-8 cmH2O) rather than without (T-piece or CPAP). Conditional recommendation, moderate quality evidence.”

Comments:

– The 1997 study by Esteban is by far the most rigorous.

– To summarize, in parts 1 and 2 of this review we looked at landmark RCTs by Brochard and Esteban.  I chose these trials because they are frequently mentioned on rounds and in textbooks when discussing liberation from MV.  It’s important to be familiar with these trials and also to understand how their unique methodologies (only randomized patients who failed a 2-hr T-piece trial, step-wise decrease in PS, multiple SBTs per day) do not apply to usual practice today.

– In my opinion, the ACCP/ATS recommendations are sensible but it’s important to acknowledge that they are based on limited data (only 4 trials of varied quality) and that the confidence intervals for both SBT success and extubation success are underwhelming and very nearly include 1

– I think for the vast majority of patients, a PS trial is a reasonable strategy for SBTs.  However, at NMH we see a lot of patients with COPD and CHF.  In these patients, the 5 cmH20 of PEEP used with traditional PS SBTs may be physiologically relevant by easing work of breathing in the setting of dynamic hyperinflation (COPD) or improving LV performance by decreasing LV afterload (CHF).  In these settings, passing a PS SBT may falsely reassure providers that a patient is ready to be liberated from MV.  When caring for these specific patients, I still find T-piece trials very helpful.  In the end, probably what matters much than the weaning strategy used is making sure you are performing daily SBTs in patients who are eligible, extubating those who pass, and aggressively treating reversible problems in those who fail.  I would be interested in the opinions of attendings on this list.

“Effect of aspirin on development of ARDS in at-risk patients presenting to the emergency department, the LIPS-A randomized clinical trial” JAMA, 2016, USA

“Effect of aspirin on development of ARDS in at-risk patients presenting to the emergency department, the LIPS-A randomized clinical trial” JAMA, 2016, USA

Question: Does aspirin prevent the development of ARDS in at-risk patients?

Study Type:  Multicenter, double-blind, placebo-controlled, parallel-group, randomized clinical trial at 16 medical centers across the United States

Study Population: Adults presenting to the emergency department with a Lung Injury Prediction score ≥ 4 were eligible (see commentary for more regarding LIPS).  Numerous exclusions included ARDS on presentation, chronic use of antiplatelet therapy, admission for elective surgery, chronic tracheostomy, presentation due to CHF, active peptic ulcer disease, bleeding disorders, and severe chronic liver disease.  Randomization had to occur within 12 hours of presentation.

Study Groups: Patients were randomized to either placebo or daily ASA given as a 325mg loading dose within 24 hours of presentation followed by 81mg daily up to hospital day 7, discharge, or death.  Clinicians were encouraged to use the Checklist for Lung Injury Prevention to guide clinical care.

Primary Outcome: Development of ARDS requiring mechanical ventilation within 7 days of hospital admission

Results: 390 patients included in intention-to-treat analysis. Notable patient characteristics: median age (67), Caucasian (72%), suspected sepsis (78%), non-cardiogenic shock (21%), possible PNA (61%), ventilated on day of randomization (17%), median LIPS (≈5.75).  ASA did not decrease the development of ARDS (10.3% in ASA group, 8.7% in placebo group, P=0.53). ASA did not significantly impact any secondary outcome measures.  There was no difference in adverse events between groups

Caveats: Generalizability a concern as 95% of the 7,673 patients screened were excluded (with prior antiplatelet use, inability to consent with 12 hours, and suspected active bleeding being the most common reasons for exclusions), delay between consent and medication administration (median delay 12.7 hours), development of ARDS occurred almost half as frequently as anticipated.

Take-home Point: Among at-risk patients presenting to the emergency department, daily ASA use did not decrease the development of ARDS at 7 days.

Commentary

–       I chose to review this study because it is important for residents to be aware of the increasing focus on ARDS prevention.  Indeed, the NHLBI-funded ARDS clinical trial network is named PETAL (Prevention and Early Treatment of Acute Lung Injury).  This refocusing of ARDS clinical trials is based on the observation that ARDS incidence has decreased (suggesting perhaps that through early abx, thoughtful IVF, less injurious ventilator strategies, and restrictive transfusion thresholds we have shown that ARDS prevention is possible) and the hypothesis that so many therapeutic trials in ARDS have been negative because therapies have been initiated AFTER significant alveolar-capillary injury has already occurred.

–       The Lung Injury Prediction Score has been found to have an AUC for the prediction for ARDS of 0.8 (see attached AJRCCM article).  LIPS variables include shock, aspiration, sepsis, PNA, high-risk surgery, high-risk trauma, EtOH abuse, obesity, low albumin, chemotherapy, FIO2>0.35, RR>30, SpO2<95%, pH <7.35, DM.  At a cut-off of 4, LIPS has a sensitivity of 69% and specificity of 78% for the development of ARDS.  Based on the AJRCCM data, with a LIPS of 6 (roughly the average of pts in the JAMA trial), you would expect roughly 15% to develop ARDS.

–       ASA was chosen given the known pathogenic role of platelet activation and aggregation in the development of ARDS, the anti-inflammatory effects of ASA, and encouraging pre-clinical data.

“Incidence, predictors, and outcomes of new-onset atrial fibrillation in critically ill patients with sepsis,” AJRCCM, 2017, Netherlands

“Incidence, predictors, and outcomes of new-onset atrial fibrillation in critically ill patients with sepsis,” AJRCCM, 2017, Netherlands

Question: What is the incidence of new-onset atrial fibrillation/atrial flutter (AF) in ICU patients with sepsis and is the development of AF associated with mortality?

Study Type: Sub-study of a prospective cohort study at two hospitals in the Netherlands

Study Population: Adults admitted to the ICU with sepsis over a two-year period were included.  Exclusion criteria included a history of AF, recent cardiotomy, or cardiac arrest.

Study Groups: AF detection based on hourly recordings of the rhythm observed by the bedside nurse.  AF needed to be sustained for 1 hour or require pharmacologic treatment or cardioversion to be included.

Primary Outcome: Incidence of new-onset AF

Results: 1,782 patients included in analysis.  Notable patient characteristics: Age (»62), immunosuppression (26%), malignancy (10%), lung as source of infection (62%), mechanical ventilation (83%), surgical admission (23%), 90-day mortality (34%).  The overall incidence of AF was 23%.  For patients with septic shock, the risk of AF was 40%.  The most common treatments administered were amiodarone (71%), magnesium (69%), electrical cardioversion (11%), and sotalol (8%).  After multivariable regression, AF remained associated with ICU mortality (HR 2.10; 95% CI, 1.61-2.73).  An AF risk score was developed (the “SAFE” score) and validated in an independent cohort.  The score can be accessed here: https://safescore.shinyapps.io/safe/.

Caveats: Unmeasured confounding a concern in observational studies, unable to prove AF has causal role in poor outcomes, relied on identification and classification of arrhythmia by bedside nurse so possible missed or misclassified episodes, data collected from only two centers.

Take-home Point: New-onset AF is common in ICU patients with sepsis and may lead to increased mortality.

Commentary 

–        Part of a growing body of literature helping to describe the incidence and significance of new-onset AF in critical illness.  Certainly notable that the incidence of AF in septic shock approaches 50%

–        The SAFE score still needs to be independently validated, but it is an interesting attempt by the authors to come up with a way to predict the development of sepsis-induced AF (and, in doing so, potentially identify a group of septic patients to enroll in trials of AF prophylaxis).​

 

 

“The timing of early antibiotics and hospital mortality in sepsis” AJRCCM, 2017, USA

“The timing of early antibiotics and hospital mortality in sepsis” AJRCCM, 2017, USA

Question: Does the timing of antibiotic administration in patients presenting with sepsis impact mortality?

Study Type: Retrospective study of a large healthcare database

Study Population: Adult patients presenting with sepsis (defined by ICD 9 codes + the receipt of abx within 6 hours of ED registration) to any of the 21 hospitals in the Kaiser Permanente Northern California system between 7/2010 and 12/2013.

Study Groups: Septic shock defined as need for vasopressors or initial serum lactate ³4; severe sepsis as a lactate ³2, ³1 instance of hypotension, need for invasive or non-invasive ventilation, or laboratory-determined organ dysfunction; all other patients classified as having sepsis. All vital signs obtained during the first hour were recorded and averaged. Abx administration calculated from the time of ED registration to the administration of the first abx in hours. Logistic regression was used to estimate the odds of hospital mortality based on abx timing and patient factors.

Primary Outcome: Hospital mortality

Results: 35,000 patients randomly selected. Notable patient characteristics: sepsis (35%), severe sepsis (52%), septic shock (13%), median age (73), mean first lactate (1.8), mechanical ventilation (1.4%), mortality of entire group (9.4%), mortality of patients with sepsis (3.9%), mortality of patients with severe sepsis (8.8%), mortality of patients with septic shock (26%). The fully adjusted odds ratio for hospital mortality based on abx timing was 1.09 (95% CI, 1.05-1.13) per elapsed hour after ED presentation. The absolute increase in mortality associated with a 1-hr delay in abx administration was 0.3% (95% CI, 0.01%-0.6%) for sepsis, 0.4% (95% CI, 0.1%-0.8%) for severe sepsis, and 1.8% (95% CI, 0.8%-3%) for shock.

Caveats: Retrospective, only able to describe an association not prove causation, no data provided on other aspects of early resuscitation which may impact outcome (IVF, did abx given cover causative pathogen, etc), only included patients presenting to the ED not patients who develop sepsis while hospitalized, relied upon ICD 9 codes to identify patients with sepsis.

Take-home Point: Among patients presenting to the ED with sepsis, severe sepsis, and septic shock, delays in abx administration are associated with increased mortality.

Commentary 

–        The most frequently cited paper on this subject is by Kumar (CCM, 2006) which retrospectively analyzed 2,731 patients with septic shock and found that for each hour delay in appropriate abx therapy after the onset of hypotension, mortality increased by 7.6%.

–        The result is less dramatic here (1.8% increase for each hour delay in patients with shock) and the authors highlight several methodological differences which may account for their findings.

–        The study has flaws, but it perhaps provides a more accurate look at the true mortality benefit of early abx in the modern era of sepsis management. 

 

High-impact PCCM Article Summaries: Restriction of Intravenous Fluid in Patients with Septic Shock

Meyhoff et al. Restriction of Intravenous Fluid in Patients with Septic Shock: The CLASSIC Trial. NEJM 2022

Question: Does a restrictive fluid strategy improve 90-day mortality in ICU patients with septic shock?

 

Why ask it: When to administer intravenous fluids (IVF) to patients with sepsis and septic shock remains a source of debate. Restrictive resuscitation may improve outcomes by avoiding the harms associated with unnecessary fluid loading. However, the safety and efficacy of this approach has not been studied in a large, randomized trial.

Intervention: 1,554 ICU patients in 8 European countries who developed septic shock within 12 hours of screening randomized to a restrictive or standard fluid strategy (see comment for details) for the duration of their ICU stay. Septic shock was defined as suspected or confirmed infection + lactate  ≥ 2 + ongoing vasopressor or inotrope use + ≥ 1L IVF administered in the 24 hours prior to screening.

Results:

  • Death within 90 days (primary outcome)
    • 42.3% in the restrictive fluid group vs 42.1% in the standard-fluid group (adjusted absolute difference, 0.1 percentage points [95% CI, -4.7 to 4.9], p=0.96)
  • No difference in serious adverse events, serious adverse reactions, number of days alive without life support, or number of days alive and out of the hospital
  • Results were robust to a sensitivity analysis
  • There was no heterogeneity in the effect of the intervention across a range of pre-specified subgroups

Conclusion: In ICU patients with septic shock, a restrictive fluid strategy did not improve 90-day mortality compared to a standard fluid strategy

Comment:

  • Worth being aware of some the arguments in favor of restrictive resuscitation (an incomplete list):
  • CLASSIC was an international, stratified, parallel-group, open-label RCT conducted in both university and community ICUs
  • Notable patient characteristics
    • Median age: ~ 70 years
    • Male: 59%
    • Comorbidities
      • Ischemic heart disease or heart failure: 17%
      • Long-term dialysis: 1%
    • Median time from ICU admission to randomization: ~3 hours
    • Source of admission
      • ED or prehospital: 39%
      • Hospital ward: 37%
    • Source of infection
      • GI: 38%
      • Pulmonary: 27%
    • Median volume of IVF in the 24 hours prior to randomization: ~3L
    • Use of respiratory support: 51%
  • Details of interventions
    • Restrictive-fluid group: Indications for 250-500 mL crystalloid bolus:
      • Severe hypoperfusion
        • Lactate  ≥ 4
        • MAP < 50 mmHg despite vasopressors
        • Mottling of skin beyond the edge of the kneecap
        • Urine output < 0.1 mL/kg/hr during the first 2 hours after randomization
      • Replace documented fluid loss
      • Correct dehydration or electrolyte deficiencies because of limited enteral access
      • Ensure a total daily fluid intake of 1L
    • Standard-fluid group: No upper limit on the amount of IVF that could be given for any of the following:
      • Improvement in hemodynamic factors in response to IVF
      • Replace fluid losses or correct electrolyte deficiencies
      • Need for maintenance fluids
  • Results of intervention (all written as restrictive-fluid vs standard-fluid group)
    • Median IVF volume (Ls)
      • Day 1: 0.5 vs 1.3
      • Day 5: 1.5 vs 3.1
      • Day 90: 1.8 vs 3.8
    • Median total fluid volume (Ls)
      • Day 1: 1.8 vs 2.7
      • Day 5: 8.9 vs 10.8
      • Day 90: 10.4 vs 12.8
    • Median cumulative fluid balance (Ls)
      • Day 1: 0.7 vs 1.3
      • Day 5: 1.7 vs 2.4
      • Day 90: 1.7 vs 2.4
  • My take
    • CLASSIC is a rigorous, large RCT with very few patients lost to follow-up, a robust primary outcome, and a diverse range of study sites that increase its external validity. The intervention produced a durable difference in management in that patients in the standard-fluid group received more fluid at every time point. Yes, the trial was open label, there were some protocol violations, and the trial was powered for an eyebrow-raising 7% difference in mortality, but I do not think these limitations detract from the study’s major findings.
    • Some, including the editorial, have raised concerns about the standard-fluid arm of this trial – noting that this approach was too “restrictive” to be viewed as usual practice and thus biased the trial towards a negative result. As just one comparison, in the PROCESS trial of early goal-directed therapy for sepsis, the protocol-based standard therapy arm received a mean of 8L of IVF between 0 and 72 hours and the usual care arm received 7L over that time frame. These differences are worth noting and comparing to your practice, but I think most would agree that the CLASSIC investigators should not have mandated extra IVF beyond what they felt was reflective of standard care.
    • Depending on one’s enthusiasm for restrictive resuscitation, you will hear the results of this trial framed as “look, restrictive resuscitation is safe” or “look, restrictive resuscitation isn’t helpful.”
    • Coupled with the preliminary results of the CLOVERS trial, CLASSIC pours cold water on the idea that restrictive resuscitation is the next innovation in sepsis care and a necessary response to the fluid-happy practice that followed the Rivers trial. We should give IVF when we think it will benefit our patients, with that decision informed by all of the (imperfect) clinical markers at our disposal. Early antibiotics, prompt source control, and frequent bedside reassessment remain the critical components of early sepsis care.

High-Impact PCCM article summaries: Ultrasound Assessment of JVP to Predict Elevated RAP

Wang et al. Accuracy of Ultrasound Jugular Venous Pressure Height in Predicting Central Venous Congestion. Annals of Internal Medicine. 2022

Question: Does ultrasound assessment of jugular venous pressure height (uJVP) in the semi-upright and upright positions predict elevated right atrial pressure (RAP) measured during right heart catheterization (RHC)?

Why ask it: Visual inspection of the height of the JVP is often used to estimate RAP. Patient anatomy and variation in the distance between the sternal angle and right atrium may limit the accuracy of this measurement. While ultrasound assessment of inferior vena cava (IVC) diameter and collapsibility is commonly used to estimate RAP, there are many patient and operator-specific limitations to this technique.

Intervention: Convenience sample of 100 patients undergoing RHC at the University of Utah. Two POCUS-trained cardiology fellows and 1 attending physician obtained the following values using a handheld ultrasound (Butterfly Network):

1. Quantitative uJVP: measured with HOB at 30-45° and defined as the point at which the internal jugular (IJ) vein tapered to smaller than the adjacent carotid artery throughout the respiratory cycle (vertical height of this taper above sternal angle + 5 cm = uJVP).

2. Qualitative uJVP based on neck zone where the IJ collapse point was identified: zone 1 (below clavicle), zone 2 (lower 3rd of neck), zone 3 (middle 3rd of neck), zone 4 (upper 3rd of neck), and zone 5 (above mandible).

3. Qualitative uJVP with HOB 90°: If IJ distended to at least the size of the adjacent carotid artery, this test was considered positive

Results:

  • ROC of uJVP as a continuous variable to predict RAP > 10 mmHg:
    • AUC 0.84 (95% CI, 0.76 – 0.92)
  • Test characteristics of uJVP > 8 cm to predict RAP > 10 mmHg:
    • Sensitivity: 72.7% (95% CI, 57.2% – 85.0%), Specificity: 78.6% (95% CI, 65.5% – 88.4%), likelihood ratio: 3.4 (95% CI, 2.0 – 5.8)
  • Test characteristics of qualitative uJVP with HOB 90° to predict RAP > 10 mmHg:
    • Sensitivity 54.5% (95% CI, 38.8% – 69.6%), Specificity: 94.6% (85.1% – 98.9%)
  • Evaluating correlation between 5 neck zones and uJVP showed a linear relationship with increasing RAP (figure pasted below)

Conclusion: Point-of-care ultrasound assessment of JVP can identify patients with an elevated RAP on RHC with a reasonable degree of accuracy

Comment:

  • Important to emphasize this study looked at the relationship between JVP and RAP, not JVP and fluid responsiveness (i.e., will this patient increase their stroke volume in response to a fluid challenge). Two very different questions.
  • Study cohort:
    • LVEF < 50%: 44%
    • Male: 64%
    • White: 76%
    • Presenting for evaluation of heart failure: 67%
    • Outpatients: 40%
    • Mean RAP: 10.3 mmHg
    • Mean PCWP: 17 mmHh
  • Limitations: Single-center study, convenience sample of 100 patients out of a potential 4,436 patients presenting for RHC during study period (possible selection bias), largely white and male population, not necessarily representative of MICU population, unclear generalizability of study findings to less experienced ultrasonographers
  • Visual assessment of JVP had similar test characteristics to uJVP (supporting use of traditional method of JVP assessment for those who like this technique). However, uJVP was possible in 100% of patients while visual estimation was possible in only 60% of assessed patients.
  • I like this study because it asks a simple question and gives helpful test characteristics to inform an individual clinician’s decision on whether to incorporate this technique into their clinical practice.
  • All measured values in the ICU (e.g., Pplt, Pes, P0.1, CVP, PCWP, IVC collapsibility index, PPV, etc) are imperfect surrogates for complex physiology. In my view, the utility of any of these tests depends on the clinical question being asked, a clinician’s awareness of a specific test’s characteristics and limitations, and how thoughtfully a given value is combined with other data. Based on this study, if your clinical question is,“does my patient have an elevated RAP?,” ultrasound assessment of the IJ may be helpful.
  • My take-home: I have always liked scanning the IJ (seems like a logical extension of scanning the heart and IVC + our MICU population seems enriched for patients in whom visual assessment of JVP is challenging). I prefer qualitative “big picture” patterns over quantitative bedside assessments as I worry about accuracy and over-interpretation with the latter. What I will consider using in practice:
    • Seeing a zone 4 or 5 uJVP seems to be a reasonable surrogate for a RAP > 10 mmHg
    • Visualization of the uJVP with the patient at 90° (not feasible in some MICU pts) has a high specificity for a RAP > 10 mmHg

High-Impact PCCM article summaries: A Three-step Communication Strategy at the End of Life

Kentish-Barnes et al. A Three-step Strategy for Relatives of Patients Dying in the Intensive Care Unit. The COSMIC-EOL Cluster Randomized Trial. Lancet 2022

Question: Does a proactive intervention involving repeated meetings with relatives of ICU patients dying after a decision to withdraw or withhold life-sustaining therapies decrease the presence of prolonged grief disorder at 6 months?

Why ask it: Effective and empathic communication with family members is an essential component of high-quality end-of-life care in the ICU. Poor communication has been linked to increased rates of PTSD and complicated grief for family members in the months that follow a patient’s death.

Intervention: 875 patient relatives in 34 French ICUs randomized to standard end-of-life communication or a three-step, physician-driven, nurse-aided support strategy after the decision was made to withdraw or withhold life-sustaining therapies. For relatives in the intervention arm, meetings with the treating physician and nurse were scheduled at 3 distinct time points: An initial end-of-life preparatory conference to plan end-of-life care and discuss the needs of both the patient and the family, a meeting during the dying process to show non-abandonment and detect unmet needs, and finally a meeting after death to answer questions about the ICU stay and acknowledge emotions.

Results:

  • Median prolonged-grief 13 questionnaire score during a telephone interview at 6 months (primary endpoint): 19 (IQR 14 – 26) in the intervention group vs 21 (15 – 29) in the control group [mean difference 2.5 (95% CI, 1.04 – 3.95)
  • Proportion of relatives with a score > 30 indicating complicated grief: 15% intervention arm vs 21% control arm, p=0.035
  • At 3 months, relatives in the intervention arm had statistically lower hospital anxiety and depression scale assessment scores as well as less PTSD-related symptoms.

Conclusion: A three-step proactive communication intervention involving the treating clinician and nurse reduced the burden of prolonged grief in relatives of dying patients following the decision to withhold or withdrawal life-sustaining therapies.

Comment:

  • A well-done trial with a simple intervention on an important topic. 90% of relatives in the intervention arm had all 3 steps completed and 79% completed 6-month follow-up.
  • The majority of relatives were either a spouse/partner or a child of the patient. Most had very strong social support
  • 68% of relatives in control arm received meetings at the first 2 time points suggesting perhaps a unique impact of the scheduled meeting after a patient’s death
  • The authors rightly point out that symptom scores are imperfect surrogates for the complex emotional and physical burden family members shoulder after the death of a loved one. Additionally, there is no clear minimally important difference in the PG-13 score used in the primary outcome
  • really like the emphasis on shared clinician and nurse communication. Aligned clinician + RN communication and clear shared support for families I think an important model to emulate
  • This trial will make me more deliberately consider empathic clinician + RN communication at these 3 distinct phases in the care of our dying patients. I will make a more concerted effort to partner with nursing during these time points. The communication model outlined in the intervention may be worth trying to replicate in our ICU.