June 2012 Critical Care Journal Club

Another rapid-fire Journal club this month: We covered 6 articles very concisely.

Stein PD, Matta F. Thrombolytic therapy in unstable patients with acute pulmonary embolism: saves lives but underused.  Am J Med. 2012;125:465-70.  (presented in conference by Henry Leudy). (click here for abstract)

It's interesting to note that there is no definitive clinical trial that proves the benefit of thrombolytic therapy for life-threatening pulmonary embolism.  On first blush, this study looked promising - with a sample size of >72,000 patients, and a huge protective effect for thrombolytic therapy, reducing mortality from 47 to 15% (p<0.0001).

Unfortunately, this study is based on very low quality data.  The clinical diagnoses of pulmonary embolism and hemodynamic instability and the administration of thrombolytic therapy were all based on ICD-9-CM billing codes - notoriously inaccurate for use in clinical research.  The study design was a retrospective cohort - but there was a high risk of selection bias - patients who did not receive thrombolytic therapy were significantly older and sicker to begin with.  The authors made no attempt to adjust for this as they should have, with statistical techniques such as multiple logistic regression.

Although we share a bias towards using thrombolytics in patients with hemodynamically significant pulmonary embolism, the low quality of this study design does not add much to what we already think we know.  Our group routinely performs bedside echocardiography in ICU patients with pulmonary embolism, and feel that right ventricular dilation/dysfunction support the decision to give lytics - as previously suggested by Goldhaber (N Engl J Med 2002; 347:1131-2).  Anecdotally, we have seen several patients with large right atrial clots visible on echo, who immediately died from pulmonary embolism upon receiving lytic therapy - we worried whether lytics might have catastrophically liberated these clots.  The optimal management of life-threatening pulmonary embolism needs to be better defined through high-quality clinical research.

Van Rompaey B, Elseviers MM, Van Drom W, Fromont V, Jorens PG. The effect of earplugs during the night on the onset of delirium and sleep perception: a randomized controlled trial in intensive care patients. Crit Care. 2012;16:R73.  (presented in conference by Jessica Hurley) (click here for full text of article)

This is an intriguing idea - a cheap and simple preventative measure for one of the major iatrogenic complications of critical illness.  Unfortunately, the design of this study was so seriously flawed that this promising hypothesis remains just that. 

The authors randomized 136 patients to nighttime earplug use or control for the first 5 days of ICU stay.  The primary outcome variable was the incidence of delirium based on a validated clinical scoring system.  This all seems pretty good.  Unfortunately no reduction was found in delirium (19% vs 20%), so the authors analyzed a combined endpoint (delirium + confusion) using a 24-hour duration survival analysis.     

Survival analysis (and associated hazard ratios) should only be used when a delay in the outcome of interest - usually death - is of significant duration to be beneficial.  It is often used in studies of cancer chemotherapy.  Most would agree that a therapy that shows significant survival advantage at 5 years is clinically worthwhile.  The shorter the duration of a study, the less applicable survival analysis becomes.  It's simply ridiculous to use it over a 24-hour period of observation as these authors have done.  This particular paper is the worst example I have ever seen of misuse of this statistic.  The author's contention that "a vast improvement" in the incidence of confusion is demonstrated by the calculated hazard ratio shows a lack of understanding of what the statistic means  - in this case, that the onset of confusion is delayed over a 24-hour period in patients wearing earplugs - a clinically meaningless result. Additionally, the external validity of the study was severely limited by the exclusion of all patients requiring sedation - these are the very patients who are at highest risk for ICU delirium. This paper should have been rejected for publication, but the hypothesis has merit.  Hopefully someone will pick up on it and do a better job.  

NHLBI ARDS clinical trials network.  Initial trophic vs full enteral feeding in patients with acute lung injury.  JAMA 2012;307:795-803.  (Tonya Whiting presented this paper in conference). (click here for full text of article)

There are many theoretical reasons why feeding should be better than starving a critically-ill patient, yet definite proof that early nutritional support is beneficial in critical illness is still lacking.  Some observational studies show benefit - others seem to suggest that permissive underfeeding may actually reduce ventilator LOS and mortality.  It is therefore a bit disappointing that the authors set out to compare two nutritional regimens without any non-intervention control group.  This was probably done on the presumption that short-term permissive starvation is harmful.

The authors screened almost 8000 patients with acute lung injury receiving mechanical ventilation but only randomized 1000 - one group received trophic enteral feeding (fixed rate of 20 Kcal/hr), the other group received full enteral feeding (escalating rate to achieve 25-30 Kcal/kg/day).   The intervention was continued for 6 days, then all patients received full enteral feeding.  Initial trophic feeding did not benefit ventilator-free days or reduce 60-day mortality.  Differences in gastrointestinal symptoms were statistically, but not clinically important.     

The study was generally well-designed in terms of internal validity.  However the conclusion that hypocaloric feeding was no better than full enteral feeding is anticlimactic.  The external generalizability of the study is called into question by the large proportion of screened patients who were excluded from participation (87%).  One clinical point we took home from this paper - if you believe early early feeding is beneficial, you needn't fuss too much - trophic feeding through an nasogastric tube is likely to be OK.

Schuetz P, Amin DN, Greenwald JL.  Role of procalcitonin in managing adult patients with respiratory tract infections.  Chest 2012;141:1063-73. (presented in conference by Chris Jivcu).  (click here for full text of article)

We've reviewed procalcitonin several times in the past two years waiting to see whether it will fulfill it's promise.  In this review, we briefly concentrate on sections related to ICU patient care.

Procalcitonin is a serum marker of acute bacterial infection that might be useful in differentiating bacterial from viral respiratory infections.  The test has reasonable operating characteristics in community-acquired pneumonia, with an area-under-the-ROC curve of >80%.  However, it has still not proven to be reliable in the workup of ventilator associated pneumonia.  Studies that have investigated procalcitonin-guided management of sepsis in the ICU, have shown reduction in duration of antibiotics, but mixed effects on ICU length of stay, and no effect on mortality.  The authors attempt a list of recommendations, but these seem premature given the current state of the literature. We will continue to keep an eye on this as more research comes along. 

Schortgen F, Clabault K, Katsahian S, et al.  Fever control using external cooling in septic shock: A randomized controlled trial.  Am J Respir Crit Care Med 2012;185:1088-95. (presented in conference by Emad Wissa). (click here for abstract of article)

Most clinicians treat fever for the patient's comfort, but we've wondered whether doing so might actually be short-circuiting a beneficial host response.  This paper touches on this subject but doesn't provide a clear answer. 

The authors randomized 200 febrile patients with septic shock requiring pressors, mechanical ventilation and sedation to either external cooling for 48 hours to achieve normothermia, or no external cooling.  The primary endpoint was vasopressor dose reduction at 48 hours. 

The conduct of the study seemed reasonable, but the choice of outcome variables, statistics and the conclusions the authors reached are questionable.  There was no statistically significant benefit seen in the primary endpoint, but the authors noted slight improvements in some related secondary outcomes such as an increase in shock reversal during the ICU stay in the cooling group.  They also report a reduction in 14-day mortality from 34 to 19% (p=0.013) with cooling using survival analysis.  It is widely believed that 14-day survival is not a useful clinical outcome because survival over such a short follow-up may simply represent a delay in death rather than a beneficial return to health.  The standard is 28-day survival or longer. 

One take home point is noted in relation to refractory shock - this study suggests that cooling the patient to normothermia in this situation might help support perfusion pressure.  I wonder if this is part of why it sometimes seems that patients in shock do better on continuous renal replacement therapy, which generally cools the patient a degree or two.      

Krikorian A, Limonero JT. An integrated view of suffering in palliative care.  J Palliative Care 2012;28:41-9.  (presented in conference by Roxanne Garciaorr). (no abstract available)

This paper was chosen to help us think about how to respond to the family of a critically-ill patient when they ask: "Is my Dad suffering?"  The authors don't provide an easy answer, but develop several useful concepts related to the question.

Suffering is much more than just pain - and many elements of suffering cannot be ameliorated by analgesics.  Suffering is a form of distress caused by events that threaten the patient's intactness, or even their very existence, over which they feel helpless to save themselves.  It follows that suffering is a very personal experience - the degree to which a patient feels helpless and threatened is the degree to which they will suffer.  Culture, religious beliefs and personality will clearly influence this.  Suffering can be treated by reducing the perception of threat, and by increasing the patient's feeling they can have some control over events.  These are concepts that can be applied at the bedside.  Prayer is one example of an intervention that might address both issues for some patients.

The authors make one point that gave me pause - they say a patient has to be conscious in order to suffer.  It makes me wonder how this applies to a comatose patient in the ICU.  Although we cannot be sure the patient is suffering in this situation, the imminent destruction of the patient causes suffering for the family and medical caregivers.  This suffering is often unfairly discounted when the costs/benefits of potentially futile intensive care unit interventions are weighed.   

Robert A. Raschke, MD

Associate Editor

Critical Care Journal Club

Reference as: Raschke RA. June 2012 critical care journal club. Southwest J Pulm Crit Care 2012;5:20-3. (click here for a PDF version of the journal club)