In the Game of Thrones, a great wall surrounds civilization. There are rumors of terrible creatures beyond the wall, deadly White Walkers, but few believe such rumors. There is no solid evidence that the White Walkers exist. It’s more comfortable to believe that the White Walkers are imaginary. Unfortunately, the White Walkers turn out to be very real. Eventually they breech the wall, placing civilization at risk.
As scientists, we live within our own wall, the scientific method. It will only allow in ideas which are capable of being reproducibly demonstrated to be true. Beliefs which cannot be subjected to testing are thrown out, beyond the wall: religion, mysticism, and the like. Many ideas exiled beyond the wall are junk. However, some untestable ideas lurking beyond of the wall are quite powerful.
There are a number of reasons that an idea may be impossible to test scientifically. Perhaps the disease state that the idea applies to is extremely rare. Perhaps it is logistically impossible to test the idea, due to issues with blinding or time constraints. This post will focus on an insidious reason that ideas are thrown beyond the wall: the timeframe and study size required to test them are unachievable.
Example #1: Antibiotics for skin abscess
Recently a meta-analysis was published in the Annals of Emergency Medicine demonstrating that antibiotic therapy for skin abscesses improves treatment success:1
Antibiotics for that abscess? A #FOAMed meta-analysis https://t.co/WayiUA9nW3 #VisualAbstract @MGottliebMD pic.twitter.com/2kMJPSYE1R
— AnnalsofEM (@AnnalsofEM) December 27, 2018
For patients with skin abscess, treatment with clindamycin or trimethoprim-sulfamethoxazole decreased treatment failure by 8%. However, it remains debatable whether this treatment is truly justified. Most treatment “failures” weren’t particularly dire, but instead represented patients who needed repeat drainage or initiation of antibiotics. Studies included abscesses with surrounding cellulitis, leaving it unclear whether antibiotics are necessary for an isolated abscess without cellulitis.
Currently, antibiotics for uncomplicated skin abscess is not usually the standard of care. This seems to be working out OK. I’m not aware of patients dropping dead due to lack of antibiotics following drainage of skin abscess.
One risk of broadly applying antibiotics to skin abscesses is increasing the rate of antibiotic resistance. This isn’t a merely theoretical problem. Widespread use of trimethoprim-sulfamethoxazole for urinary tract infection has fostered resistance, rendering the antibiotic increasingly ineffective for urinary tract infection. It’s entirely plausible that a similar phenomenon could occur for trimethoprim-sulfamethoxazole and staph aureus. What would happen then? Well, by that point, the standard of care will be to provide antibiotics, so the only logical next step would be to use more powerful antibiotics for abscess (e.g. linezolid). Once you enter the antibiotic rat race, it’s hard to get out.
Unfortunately, the long-term harms of unleashing millions of prescriptions for trimethoprim-sulfamethoxazole to patients with skin abscess is unknown. The emergence of drug resistance usually takes years and thousands of exposures. It’s impossible to scientifically test this. Ultimately, this is left beyond the wall of science, in the abyss of unsubstantiated fears and scary rumors. I’m sure that someone in the comments section will excoriate me for fear-mongering, perhaps rightfully so.
This leaves practitioners to make an impossible choice between something which is proven (the benefit of antibiotics for a single patient with a skin abscess) versus something which is fundamentally unknowable (the long-term population-based risks of antibiotic over-use). And we know how this plays out. We will inevitably be swayed by the meta-analyses and infographics showing benefit for our patients. We ignore the long-term risks of antibiotic resistance. Because people aren’t afraid of white walkers until they’re knocking at the front door.
It’s possible that reserving antibiotics for more serious infections would be wiser in the long run. For skin abscess, this would require accepting a low re-treatment rate and the need to drain a few more abscesses (could we please stop causing it “failure” if the patient needs a minimal amounts of follow-up care?). However, it might be worth it to preserve our antibiotic armamentarium for future generations.
Example #2: Antibiotic decontamination in the ICU
The gut is a reservoir of bacteria, which may lead to infection at other sites in the body (e.g. aspiration with subsequent pneumonia). This has led to the concept of decontamination of the digestive tract – if we could eliminate all pathogenic organisms from the gut, perhaps this could avoid nosocomial infections.
Currently there is a large study underway (the SuDDICU trial) to test decontamination of the digestive tract among ICU patients using a combination of three antibiotics (oral colistin, oral tobramycin, and intravenous ceftriaxone or ciprofloxacin). This trial suffers from the same limitations as the use of antibiotics for skin abscesses. Even if patients treated with broad-spectrum antibiotics do better, carpet-bombing ICU patients with antibiotics will inevitably lead to problems with CDiff and antibiotic resistance.
I like to think of myself as a fairly scientific person. However, I must admit that regardless of what the SuDDICU trial shows I will still be opposed to decontamination of the GI tract. The risks of drug resistance will inevitably take longer to emerge than any immediate benefits, so it’s impossible for the study to accurately measure the long-term, real-life consequences. Even if gut decontamination works wonders for the span of the trial, it’s still likely to break down over time. It’s like test-driving a car: even if it drives beautifully, that doesn’t mean it will hold up for ten years.
This is a patently un-scientific perspective, one which isn’t supported by any high-quality evidence. In fact, I am prepared to leverage my unsubstantiated fears against the findings of a scientific multi-center trial. Mea culpa. I’ll admit it, I’m afraid of the White Walkers.
Example #3: Nocturnal in-house attending coverage for ICUs
Some studies have suggested that on-site overnight attending staff might decrease mortality in some ICUs.2,3 That created quite a stir, leading to a push for in-house attending staff in ICUs.
As I’ve explored on the blog in the past, demonstrating a mortality benefit for any intervention in the ICU is damnably difficult. So it should come as little surprise that a subsequent study wasn't able to show that nocturnal attendings improved mortality.4 This doesn’t prove anything one way or the other.
My bias is that in-house nocturnal attendings probably do improve care. Even as a fully trained intensivist with years of experience, I still often struggle with tough patients in the ICU. It’s unrealistic to expect trainees to manage these patients perfectly at night. Trainees aren’t werewolves who become stronger and smarter at night; if anything, they are more fatigued and more likely to make a mistake.
So, for the sake of argument, let’s say that patient care is better if senior attending staff is present. Unfortunately, the tradeoff is that reduced trainee autonomy may impair education. For the attendings, in-house call may promote burnout.
Once again, we are forced to make a choice between a short-term, more palpable benefit (improved nocturnal patient care) versus long-term, theoretical risks (impaired trainee education, burnout of attending staff). Yet again, this decision falls beyond the wall of science. We could attempt to have a debate, but there is no data on the long-term consequences of nocturnal attendings. As such, the debate would be more of a philosophic one than a scientific one.
- Science only pertains to hypotheses which are testable. Unfortunately, this excludes many important concepts which are logistically impossible to test.
- Clinical trials usually last 1-2 years at most. This often makes it impossible to test long-term consequences of an intervention (especially at the larger population level).
- By its nature, science has a tendency towards near-sightedness: we can study a phenomenon in front of our eyes, be we often can’t predict the long-term consequences of our actions.
- Just because we can’t prove something doesn’t mean that it’s not real. Looming issues such as growing antibiotic resistance and global warming are difficult to prove, but that doesn’t mean that they don't exist.
- White walkers are real. Be afraid. Be very afraid.
More in-depth exploration of antibiotics for subcutaneous abscess:
- Utility of antibiotics in abscess management: systematic review and meta-analysis (Anand Swaminathan, RebelEM)
- The case of the pragmatic wound (Rory Spiegel, EMNerd)
- Trimethoprim-sulfamethoxazole for uncomplicated skin abscesses? (Gillian Schmitz and Salim Rezaie, RebelEM)
- Cuts like a knife (Chip Lange and Ken Milne, SGEM, covering Talan 2016)
References
- PulmCrit Blogitorial – Use of ECGs for management of (sub)massive PE - March 24, 2024
- PulmCrit Wee: Propofol induced eyelid opening apraxia – the struggle is real - March 20, 2024
- PulmCrit wee: Why I like central lines for GI bleed resuscitation - March 13, 2024
“Treatment failure” just doesn’t seem like a particularly useful outcome compared to, say, a composite of need for subsequent IV antibiotics, surgical debridement under anesthesia or hospitalization. Saying that antibiotics decrease treatment failure in these trials to me seems more akin to saying that not extubating ventilated patients leads to fewer reintubations.
Agree: basically antibiotics given now prevents the need to give antibiotics later. This is is the challenge of trial design. If you choose a more meaningful endpoint (e.g. IV antibiotics, hospitalization, or surgical debridement under anesthesia), then the rate of that endpoint is low making it difficult/impossible to power the trial appropriately. Alternatively, if you choose a less meaningful endpoint (treatment failure of any sort) then it’s easier to power the trial and to generate a “positive” trial – but the clinical significance of the trial remains hazy. The underlying issue is that clinically *meaningful* treatment failure is uncommon, so… Read more »
Being afraid of White Walkers may not be scientific, but it is rational. Nassim Taleb defines rationality as avoiding ruin, even if the probability of ruin is small and/or incalculable. We know antibiotic resistance results from antibiotic use, thus use needs to be limited. Additionally, colistin is a rapidly deteriorating last line of defense against MDR gram negative bacteria. Unless the results of the trial are spectacular, avoiding rapidly accelerating antibiotic resistance is the rational choice. Also, if the results are spectacular, then they are probably too good to be true.
Rationality vs. science – neat.
It’s nearly impossible to imagine a scenario in which SUDDICU trial would change my management. With modern infection-prevention techniques, rates of most nosocomial infections have fallen. For example, ventilator-associated pneumonia really isn’t much of an issue at Genius General Hospital. Meanwhile, CDiff does continue to be a problem. In a clinical context where non-CDiff nosocomial infections are increasingly rare but CDiff remains a problem, gut decontamination will only hurt you.
I wouldn’t call this unscientific. There is a difference between strictly evidence-based medicine (with the strict hierarchy of evidence) and science-based medicine which takes into account prevailing scientific ideas of pathophysiology, epidemiology etc. Ultimately these ideas are still (though perhaps not easily) falsifiable. One could imagine a prospective study where hundreds of patients with drained abscesses develop endocarditis, or another where a population is shown to require fewer antibiotics at all when they are given up front as SDD. I suspect in both cases you might change your mind about antibiotic use. RCT’s are not the only form of science.… Read more »
Everyone in medicine (and science at large) should read Kuhn’s “The Structure of Scientific Revolutions”. It is irrational and we see it everyday in practice.