Wednesday, February 27, 2013

When 1A evidence is not 1A evidence.

Nephrology Merit Badge for digging deep into CPG
A few weeks ago I posted about using ceftriaxone and ampicillin for enterococcal infective endocarditis. There are a few studies which support this aminoglycoside avoidance and to my eyes it seemed like a reasonable therapeutic option, especially in my patients who are often at high risk of aminoglycoside toxicity. I pinged twitter to see if I was fooling myself into believing what I wanted to believe or if this was a viable therapeutic option.
The first responders (of the twitter variety) were the pharmacologists:

Then Med student, Alex Michaels, called me out on how my post relied on observational data:

I replied with increasing desperation:

Then the always insightful but confrontational Jim Smith weighed in with the conservative point of view:

(SOC is standard of care)

This back-and-forth began to crystalize what bothered me most about the ISDA/AHA guidelines, they graded the evidence as 1A but the supporting text did not link to one randomized controlled trial. Up to now I had not received any input from infectious disease experts so I started to fish for them.

Dan Riciuto was the first to get back to me. Here is summary of his 5 tweets (1, 2, 3, 4, 5)

Nice post. Always good to question dogma. I'll try an get back to you later with a bit more detail. I think enterococcus is actually more difficult to treat than say Staph, though two weeks of gentamicin may be fine. I've used ampicillin and ceftriaxone but there is a lot of side effects, fluid and sodium load with the high dose ceftriaxone.
I replied:
The guidelines say amp gent is 1A rec but then they don't give any refs to support "multiple RCT" to satisfy 1A strength. I also searched UpToDate and they also don't cite any RCTs, just observational data. Does the emperor have no clothes? (Tweet 1 and 2)
He continued
There are very few RCTs in ID unless it is with a new antibiotic and hardly any in relatively rare conditions like infective endocarditis. Which is why if your read the definition from the guidelines a "Class I: Conditions for which there is evidence, general agreement, or both that a given procedure or treatment is useful and effective.
I replied, that I am not as concerned about the classification (1, 2, 3) but rather the strength of evidence. Why is this recommendation 1A not 1B
He Concluded
Janine McCready also helped out:

Here is Janine McGready's full 6 tweet reply reassembled and de-abbreviated. (here are the original tweets for verification of my twitter translation 12, 3, 4, 5, 6):
Thanks for question and nice post. Enterococcal IE is harder to treat than other bugs as demonstrated by the high mortality even with bactericidal treatment. Old studies show 60% failure with penicillin alone, prompting the addition of aminoglycosides. If you have an ampicillin sensitive strain with a low MIC it may be ok to use a shorter course of gentamicin or ampicillin with ceftriaxone. I'll take a better look at the new study but I have used high dose ampicillin or ampicillin + ceftriaxone with success. The key is getting a bactericidal combination. The rationale for the addition of ceftriaxone (if I understand it correctly) is that it saturates the penicillin binding proteins (pbps) making the combination bactericidal. In my practice it's always a balancing act and I usually pull the plug on the aminoglycoside after 2 wks and consider substituting ceftriaxone if there is any concern regarding nephrotoxicity or worrisome and irreversible vestibulo/ototoxicity. Sorry the response so is long, hope that helps...
At this point I came to the conclusion that Amp + ceftriaxone is a viable second tier option in patients with aminoglycoside toxicity or high risk for aminoglycoside toxicity. However I felt betrayed by the authors of the AHA/ISDA guidelines. The recommendation for ampicillin and gentamicin appears to be a 1A recommendation.
Here are all of the articles which are referenced in the section on enterococci:

126. streptomycin = gentamicin observation
127. gentamicin dosing observation
128, 129. using duration of symptoms to determine duration of therapy (PubMed 1, PubMed 2) observation
130. Aminoglycoside for only 15 days? observation
131. 5 phenotypes of VRE in vitro
132. linezolid for VRE observation
133. Treating multidrug resistant enerococci, disease model
134. Amp and ceftriaxone, disease model, not human data
135. Amoxicillin and cefotaxime rabbit endocarditis
136. Amp and ceftriaxone observation

That's it. All observational data or experimental data in animals or disease models. Not a single reference to back up the slew A1 grades found in Tables 9 and 10.

Evidence-Based Scoring System from AHA and ISDA
Infective Endocarditis Guidelines

When I first started investigating this I kept expecting to find an RCT buried in some old journal but now I just think the authors broke the rules. I don't know if I should feel foolish for trusting the authors of clinical practice guidelines or self-rightous for smoking these jokers out. Is this kind of deception common in CPGs or is this a particularly sloppy guideline. The nephrology guidelines produced by K/DOQI and KDIGO have all been top notch and transparent with the unfortunate lack of data and reliance on expert opinion. I hope the ISDA/AHA is an exception rather than the rule.

Updates from Twitter (where else?):

Link to a nice post, on a JAMA article looking at the reliability of clinical practice guidelines.

Related Posts Plugin for WordPress, Blogger...