The Latest in Critical Care, 7/31/23 (Issue #10)

New Practice Guidelines for Diagnosis and Management of Acute Respiratory Distress Syndrome

As mentioned here previously, a group of U.S. and European experts arrived at a consensus of new recommendations for the management of acute respiratory distress syndrome (ARDS), now publishing their guidance in JAMA. The highlights include:

• A trial of high-flow nasal cannula oxygen was advised before intubating patients.

• Noninvasive ventilation was neither endorsed nor discouraged based on lack of evidence.

• Prone positioning for moderate or severe hypoxemia was strongly encouraged, largely based on PROSEVA.

• Use of higher PEEP was neither advised nor discouraged, based on lack of evidence.

• Recruitment maneuvers were discouraged, based on evidence of harm greater than benefit.

• Extracorporeal membrane oxygenation (ECMO) was encouraged, based largely on the EOLIA trial. Referral to ECMO centers was endorsed for patients who “meet criteria” (i.e., severe hypoxemia not doing well or on a bad trajectory while receiving conventional care).

• Continuous neuromuscular blockade to reduce ventilator dyssynchrony was discouraged, based on inconsistent findings between two randomized trials.

There’s rarely anything surprising in these guideline updates, and that’s to be expected. Clinical practice naturally runs out ahead of the slow, consensus-seeking review process. Its value lies in its serious analysis of recently accumulated evidence, which can give practicing clinicians a sense of how strongly supported our practices are (or aren’t). Read in JAMA

Should we feed critically ill patients less?

Caloric targets in critical illness have been set somewhat arbitrarily to approximate intake during periods of health, without strong evidence. But when we’re sick, don’t we lose our appetites and eat less—and isn’t there maybe a reason for that? Multiple randomized trials have shown that providing critically ill patients fewer calories than the advised targets is not harmful. A recent study even suggested that higher protein targets may harm critically ill patients with renal failure. What if providing fewer calories to critically ill patients is actually beneficial?

In the NUTRIREA-3 trial conducted in France, 3044 patients receiving invasive mechanical ventilation and vasopressors were randomized to receive either standard calorie and protein feedings (1750 kcal / day with 84 g protein for a 70 kg person) or a restricted diet (~420 kcal / day with 21 g protein per 70 kg) for the first 7 ICU days.

Patients were receiving high-dose norepinephrine (median 0.5 μg/kg/min, or 35 mcg/min for a 70 kg person) and enteral feedings were started and advanced without regard to the vasopressor dose.

There was no observed difference in 90 day mortality (41% vs 43%), but the restricted calorie group were ready for ICU discharge one day sooner and had fewer adverse effects attributable to enteral feedings (vomiting, diarrhea, bowel ischemia and liver dysfunction).

Usual practice in the U.S. is to restrict caloric delivery in patients with this degree of shock. The complications observed in the full-caloric feeding arm validates that practice, for me.

Because of the decision to feed full-calorie diets to patients on high dose vasopressors, it’s hard to conclude that feeding less was actually superior. To make that leap, I for one would need to see improved outcomes with caloric restriction in patients with less-severe shock. Read in Lancet Respiratory Diseases

In The News

A tornado destroyed Pfizer’s North Carolina factory, but mostly hit the storage facilities, rather than production areas. Drug shortages aren’t expected to worsen as a result.

The NYT went hard after vascular interventionalists’ hugely profitable $2 billion business of providing allegedly unnecessary, off-guideline atherectomies and other procedures to “treat” often-asymptomatic stenoses in patients with peripheral artery disease. It’s reminiscent of all the unnecessary stents that went into coronary arteries for stable coronary disease in the 90s and 00s (and still do, just less often).

The OpenAI folks tried to woo the medical community with an artfully constructed demo for the NEJM showing ChatGPT’s potential to write progress notes for us. But as of today, one of the few companies actually providing this service (DeepScribe, which uses an alternate AI model) relies heavily on hundreds of behind-the-scenes humans to train the algorithm, correct the error-filled notes, and eliminate the made-up medicines the AI hallucinates/confabulates. After all that, doctors still often need to extensively edit before signing. Still, about 1,000 clinicians are using DeepScribe, which at $6.5 million in annual revenue, today depends on VC funding. Computers writing progress notes holds huge potential for reducing the 50% of working hours time doctors spend documenting. But it also will transform daily workflows and revenue capture in unpredictable ways, mostly not controlled by physicians. This is a space worth watching.