A belated "Amen, brother." My only constructive criticism would be that having been around before GDT was popularized, it was awful - septic (and other ) shock mortality rates, and permanent renal failure and ischemic limbs were much much higher. A lot of this was due to dawdling on the part of Medicine docs - too much navel gazing, worrying about the effects of fluids (OMG they might have CHF!), etc. I feel pretty strongly that Rivers and GDT sort of shocked people into DOING things (ignoring the issues about the monitoring, data analysis, etc). As Dr. Hassaballa said on another thread of yours, "The keys to sepsis care are simple: (1) high index of suspicion; (2) early, broad-spectrum antibiotics; and (3) aggressive volume resuscitation in those who are volume depleted/responsive, vasopressors for those who need them."
In response to Steven SImpson's comment about sepsis alerts not being part of SEP-1, your "venture into policy" is hardly a non-sequitur. Every hospital is subject to the SEP-1 CMS guidelines, and in turn is using sepsis alert tools in frameworks like EPIC to try to enhance policy compliance. As an ICU Director and unlucky sepsis committee Chair, these two issues are intertwined.
Regarding his criticism of your comments about Abx and fluid timing, while sooner is clearly better than later (as any critical reading of the literature of either shows) we have little idea of what "sooner" means and the more studies are done the more vexing it seems - as it should, as the onset of sepsis has an entirely different nature than the critical break point of an MI or CVA.
I just have to say - sepsis alerts are not a part of SEP-1, so your venture into policy is actually a non-sequitur to this topic.
SEP-1 should be considered along the lines of the Pareto principle in a couple of different ways. First, before SEP-1 more than 90% of US hospitals had no particular approach to sepsis, at all. OK, so that's not 80:20, but the point stands. Secondly, 80% of the benefit may well accrue to 20% of the patients; that's OK, those 20% deserve it, and there has never been indication of harm to patients from getting antibiotics and fluids in a timely fashion. Finally, the principles in SEP-1 apply to at least 80% of patients with sepsis.
In my estimation, the main, underlying beef really is that doctors and hospitals don't like to be held accountable for anything. Not their processes, not their outcomes, not their costs. Nothing. You really stretch the limits of credibility if you say that it doesn't matter how quickly patients who actually have sepsis get their antibiotics, for example. Or how quickly patients with shock get an adequate amount of IV fluid. If you tell me that you're accidentally treating patients who don't have sepsis at all, I can certainly understand that dilemma. So let's fix that with better, more specific diagnostics.
Sepsis alerts are not diagnostic tests, but they do have the potential enrich the population in which diagnostic tests are applied. With application of reasonable diagnostic test, the question would not be "does the alert improve mortality in every patient for whom it fires?", but "does the alert help us find the real sepsis patients and improve their mortality?" Yassim's study doesn't really ask the right questions or have the right tools to answer them. AND it was created well before we even tried to implement AI into alerting algorithms. I think the jury is still out here. When we use poor tools and we don't have a good diagnostic test we can't really expect better results.
BTW, there actually are some reasonable diagnostic aids for sepsis on the market now, but they are not finding much traction, because they are mostly too expensive to use in all comers. However, with enrichment of the population for sepsis by one or another screening test, they can be used rationally.
Love the analysis and the policy proscription in the piece. The whole SEP-1 bundle has been a mess.
A belated "Amen, brother." My only constructive criticism would be that having been around before GDT was popularized, it was awful - septic (and other ) shock mortality rates, and permanent renal failure and ischemic limbs were much much higher. A lot of this was due to dawdling on the part of Medicine docs - too much navel gazing, worrying about the effects of fluids (OMG they might have CHF!), etc. I feel pretty strongly that Rivers and GDT sort of shocked people into DOING things (ignoring the issues about the monitoring, data analysis, etc). As Dr. Hassaballa said on another thread of yours, "The keys to sepsis care are simple: (1) high index of suspicion; (2) early, broad-spectrum antibiotics; and (3) aggressive volume resuscitation in those who are volume depleted/responsive, vasopressors for those who need them."
In response to Steven SImpson's comment about sepsis alerts not being part of SEP-1, your "venture into policy" is hardly a non-sequitur. Every hospital is subject to the SEP-1 CMS guidelines, and in turn is using sepsis alert tools in frameworks like EPIC to try to enhance policy compliance. As an ICU Director and unlucky sepsis committee Chair, these two issues are intertwined.
Regarding his criticism of your comments about Abx and fluid timing, while sooner is clearly better than later (as any critical reading of the literature of either shows) we have little idea of what "sooner" means and the more studies are done the more vexing it seems - as it should, as the onset of sepsis has an entirely different nature than the critical break point of an MI or CVA.
I just have to say - sepsis alerts are not a part of SEP-1, so your venture into policy is actually a non-sequitur to this topic.
SEP-1 should be considered along the lines of the Pareto principle in a couple of different ways. First, before SEP-1 more than 90% of US hospitals had no particular approach to sepsis, at all. OK, so that's not 80:20, but the point stands. Secondly, 80% of the benefit may well accrue to 20% of the patients; that's OK, those 20% deserve it, and there has never been indication of harm to patients from getting antibiotics and fluids in a timely fashion. Finally, the principles in SEP-1 apply to at least 80% of patients with sepsis.
In my estimation, the main, underlying beef really is that doctors and hospitals don't like to be held accountable for anything. Not their processes, not their outcomes, not their costs. Nothing. You really stretch the limits of credibility if you say that it doesn't matter how quickly patients who actually have sepsis get their antibiotics, for example. Or how quickly patients with shock get an adequate amount of IV fluid. If you tell me that you're accidentally treating patients who don't have sepsis at all, I can certainly understand that dilemma. So let's fix that with better, more specific diagnostics.
Sepsis alerts are not diagnostic tests, but they do have the potential enrich the population in which diagnostic tests are applied. With application of reasonable diagnostic test, the question would not be "does the alert improve mortality in every patient for whom it fires?", but "does the alert help us find the real sepsis patients and improve their mortality?" Yassim's study doesn't really ask the right questions or have the right tools to answer them. AND it was created well before we even tried to implement AI into alerting algorithms. I think the jury is still out here. When we use poor tools and we don't have a good diagnostic test we can't really expect better results.
BTW, there actually are some reasonable diagnostic aids for sepsis on the market now, but they are not finding much traction, because they are mostly too expensive to use in all comers. However, with enrichment of the population for sepsis by one or another screening test, they can be used rationally.