People often talk about the time critical nature of sepsis. But it may surprise you to hear that the evidence to support “early” antibiotics actually isn't a settled debate. Why is that? The key question is “early in relation to what”? 🧵1/n
Early usually means "early from diagnosis of sepsis" or "early from hospital admission". Both of these are *administrative* time points and have little to no basis in the biology of sepsis. 2/n
The point of admission is arbitrary. Some arrive via ambulance, others make their own way. People respond to prompts from their own physiology in different ways. Some family members are more cautious than others. 3/n
The point of diagnosis is arbitrary. When did the clinician review the patient? When were the obs taken? Neither time point reflects the time that an individual “became septic”. Is there such a time? Perhaps not. Why does this matter? 4/n
In experimental animal models of sepsis we *know* the time of inoculation with a standardised infection, and so the variability in the onset of sepsis is tightly controlled. 5/n
Timing of interventions in these models are very important and the experimental design offers maximum statistical power to find a treatment effect should it exist. 6/n
But humans are unstandardised, messy and fool us into defining a "time zero" for the onset of their sepsis that really doesn't exist, at least not collectively. 7/n
Imagine an animal model where the infective inoculation was randomised and hidden from the analysis. Are you treating sepsis at hour 1 or hour 8? It would make identification of therapeutics extremely challenging. 8/n
I like the waterfall analogy. Imagine two people moving in a flowing river that ends in a waterfall. Person A (blue) is still upstream of the waterfall, we can intervene “early” and save them from falling. Person B (red) however has already gone over the cliff edge. 9/n
We can intervene in the exact same amount of time as we did for Person A; we can still intervene “early”. But our actions will no longer affect the outcome. 10/n
Their position in the river reflects an anchoring in the biological timings of sepsis. Person B has already reached a critical level of inflammation. Their physiology has transitioned to a self perpetuating and maladaptive response to infection. 11/n
Arguably their infection is no longer the issue; the host response is. Antibiotics should still be given, but in many scenarios they are likely less useful than when this was a simple infection. 12/n
This is why, IMO, research in this area has produced weak results. Isn’t this heresy? Doesn’t every hours delay in antibiotics result in a 7.6% reduction in survival? Doubtful. 13/n
The original Kumar paper from 2006 was observational. What often isn't discussed is that patients who received antibiotics *before* they became hypotensive had *worse* outcomes. I’ve adjusted their headline figure to show this group. 14/n
There is no plausible biological mechanism for their findings which likely represent confounding. Antibiotics were probably a surrogate marker for the rapidity of care in general. As methodological rigour improves, this signal tends to get smaller 15/n
Generally, we have been unable to replicate these findings under experimental conditions. Moving antibiotic administration into the pre-hospital setting hasn’t been shown to improve outcomes in sepsis, and it would be hard to get them in any faster! thelancet.com 16/n
This thinking is replicated in the latest surviving sepsis guidelines that suggest a welcomed and more nuanced approach to antibiotic administration in these settings. 17/n
SSC splits out septic shock from sepsis, as that’s where the strongest evidence sits. These guidelines have changed markedly over the years, which shows the dynamic nature of this field and a willingness to be able to change your practice based on new evidence. 18/n
The evidence landscape is of course more complex than just these few papers, but they are illustrative of a broader issue in the field. 19/n
Early observational research that highlights potential treatment avenues for hard to treat diseases often fail to replicate when tested experimentally. 20/n
Forcing through large scale policy decisions on low quality evidence without considering negative consequences isn’t the best idea. Antibiotics aren't a free lunch, overuse comes with risks of allergic reactions, c. diff infections, and the promotion of resistant organisms. 21/n
Timely, senior led care is the absolute cornerstone of sepsis management. Critical thought is also vital and ongoing research is required in this area, particularly to understand the nature of biological time zero and disease trajectory in sepsis. 22/n
To close, I’ll say that of course I give antibiotics in sepsis and naturally I don’t dawdle (meropenem is a close personal friend and twitter doesn’t have room for ambiguity!). But I think there is likely more equipoise here than popular doctrine would suggest. n/n
Loading suggestions...