Featured Post

So You've Decided to Tweet

As the new medical-academic year begins, I'm guessing a bunch of new interns will learn about how great FOAM is, and at the same time,...

March 26, 2014

Real Genius.

The Effect.
Dunning-Kruger has long been a favorite effect of mine, but I never saw the original paper (why would I read an "old" psychology paper when there's wikipedia?). Today I pointed it out to Mark Reid & Chris Wright, there was some banter, some laughs, and Mark linked to the original Dunning-Kruger paper*.

The acknowledgments caught my eye:

No, not the NIH R01 grant that supported this research. Look at the acknowledgments.

Some googling "confirmed" -- that's emergency physician Boris Veysman* back when he was an undergrad psych major at Cornell. He writes a lot of "narrative" pieces -- I particularly like this one.

I guess the lesson is to read the paper, or: "Always... no, no... never... forget to check your references."

I'll end with the epitaph from Dunning & Kruger:
It is one of the essential features of such incompetence that the person so afflicted is incapable of knowing that he is incompetent. To have such knowledge would already be to remedy a good portion of the offense. 
(Miller WI. Humiliation. Ithaca, NY: Cornell University Press)

Great discussion of Dunning-Kruger by Lauren Westafer.

*PMID: 10626367 

March 18, 2014

Trust the ProCESS

ProCESS covered in MedPage Today by Elbert Chu. Also covered by Nick JohnsonSalim Rezaie, Simon Carley, ProCESS author Don Yealy, and of course EMCrit.

It turns out that we're getting pretty good at this. ProCESS is a 3-arm RCT comparing EGDT classic, protocolized "standard care," and usual care, and it showed essentially equal outcomes. The main difference from earlier EGDT studies is that they started collecting in 2008.

I think the main lesson from ProCESS will be that we all took the lessons of the original Rivers study (and the great follow up studies, like the 2010 Jones trial they mention) to heart: we now aggressively look for severe and occult sepsis, and work hard to resuscitate septic patients. Standard therapy before 2001 was not very aggressive, i.e. hang a liter or 2 of fluid and some ceftriaxone, then forget about the patient in the corner.

The main lessons of Rivers were that we should look hard for sepsis, pay close attention to the septic patients, and aggressively resuscitate them. The good news here is that by 2008, we were doing that much better. The study arms not only had similar mortality, they received similar volumes of fluid (much closer than in Rivers) and similar rates of other advanced interventions.

Much like our evolution of trauma care, step 1 was learning how to take these patients seriously and get aggressive protocols in place, and step 2 is to judiciously apply the appropriate interventions to the appropriate patients. That ProCESS was all academic centers might hurt its usefulness in other settings, which don't always have teams of residents and fellows to focus on these patients. Many community hospitals have gotten around this in a number of other creative ways, such as by making a "Code Sepsis" where the MICU team comes to the ED -- treating sepsis like a STEMI or a stroke, which makes a lot of sense, as it's as (or more) serious, and Rivers showed us how much early, aggressive treatment matters.

I still think there is a place for protocols: particularly when we have very sick patients, or we're swamped on a busy shift or exhausted at the end of a shift, it is helpful to have protocols handy to both standardize care and serve as gentle reminders when we need it. I don't think we're completely done innovating in sepsis care, but again and again we're showing the answer is good attention to the basics -- early identification, antibiotics and fluid -- and not technology or wonder-drugs.

...and that vital signs are insufficiently sensitive to identify sepsis, or as markers of adequate resuscitation.

Addendum: A few more thoughts: 
  • A number of people have pointed out that the mortality rate was much higher in the Rivers trial. Rivers had a sicker population, and (I think) we got better at treating severe sepsis. Additionally, lower mortality across the board makes it harder to find a difference between groups.
  • The one potential difference I see in the groups is that the EGDT arm had a somewhat higher rate of intraabdominal infection, and those patients are generally a bit sicker. I doubt this is enough to taint the results.
  • The rates of cardiac and respiratory failure during treatment seemed a little higher in the EGDT and protocolized standard care group, roughly in proportion to the amount of fluid they received. Not statistically significant, but intriguing, and seems to be in line with the MAP study below.

Two other new sepsis studies:

Another nail in the coffin for albumin and other non-crystalloid fluids in sepsis.

BP goals: MAP 65-70 as good as 80-85
No surprise that lower MAP targets as as good or better than higher BPs. Sick patients aren't like healthy people -- our goal should be to resuscitate them just as much as they need and no more, not to make them look like a normal, healthy patients. Over-resuscitation will just expose patients to the potential harms of therapy.