Use sepsis bundles, or you're breaking the (New York) law
New York state capitol building
In 2013, New York's state government began regulating the care of sepsis. The state has since required its hospitals (and thus its doctors) to adhere to some version of a sepsis protocol that included a "bundle" to be delivered within 3 hours after sepsis recognition:
blood cultures before antibiotics;
lactate measurement;
broad-spectrum antibiotics
The Empire State legislature and health department also mandated a 6-hour sepsis bundle:
30 mL/kg fluid bolus for patients with hypotension or lactate >4 meq/L;
vasopressors for hypotension refractory to fluids;
re-measurement of lactate
These include elements of the core measure for sepsis enacted October 2015 by the U.S. Centers for Medicare and Medicaid. Compliance with each bundle element was recorded in order to comply with New York's new "Rory's Regulations" (more on that in a bit). The data rolled in, and a retrospective observational analysis of 49,000 patients from 2014-2016 at 185 New York hospitals was subsequently published in the New England Journal of Medicine.
Delayed Antibiotics in Sepsis Associated with Mortality
Patients who got all elements of the bundle within 3 hours (=83%) of sepsis identification had lower hospital mortality (the study's primary endpoint).
Each hour of delay in receipt of antibiotics or in completion of the 3-hour bundle as a whole was associated with an increased odds ratio for mortality (+1.04 per hour, p<0.001). Measuring blood cultures quickly was also associated with lower mortality.
These effects were larger in the sickest patients (those on vasopressors, a pre-specified analysis).
An Early IV Fluid Bolus Had No Association with Mortality in Sepsis
Within 6 hours of sepsis identification, another bundle had to be completed for compliance: a 30 ml/kg IV fluid bolus (for hypotension or lactatemia), another lactate level, and vasopressors if hypotension persisted despite fluids.
Delays in administration of IV fluids were not associated with an increase in mortality (odds ratio 1.01, p=0.22).
Sicker (more hypotensive) patients probably got fluids faster (= confounding by indication), rendering any firm conclusion from this data impossible.
That said, intensivists know many patients with sepsis are not intravascularly volume depleted (especially those with normotension), and don't all need 2-3 liter IV fluid boluses. Patients with end-stage renal disease (11% of the cohort) or congestive heart failure (21%) might even be harmed. Most IV crystalloid quickly deposits uselessly in the extravascular space, potentially creating other problems.
Increased lactate production in sepsis may even be adaptive or an epiphenomenon of high aerobic metabolism, not signaling hypoperfusion or anaerobic metabolism (or not always). There is little or no evidence-based rationale for giving large fluid boluses to septic patients with lactatemia who are not hypotensive.
But physicians should be careful saying things like that anywhere near a hospital boardroom these days. The Centers for Medicare and Medicaid's core measure for sepsis also requires ~2-to-3 liter fluid boluses for septic patients with CHF and ESRD with normal or high blood pressure who happen to have elevated lactate levels. No exceptions. If a hospital doesn't comply, they might get "dinged" (i.e., miss out on payments from the federal government, and/or be publicly reported as having "low quality" care).
The government does not seem to be tracking avoidable cases of pulmonary edema, respiratory failure, and mechanical ventilation that might result from the mandated fluid boluses in volume-replete normotensive patients with sepsis and lactatemia.
Sepsis Bundle Compliance Was High
More than half of all hospitals gave antibiotics within one hour after their sepsis diagnosis, and completed the 3-hour bundle in less than one hour and twenty minutes.
Interestingly, the top performers were mostly not New York's large academic centers, but its small community hospitals. Compliance varied between 60% to 90% between hospitals overall.
Time to Bundle Initiation ≅ Time to Sepsis Diagnosis
There is an inherent flaw in the use of bundle completion as a care metric. What matters most in sepsis treatment is not when a bundle gets completed: it's when sepsis gets diagnosed and treatments (primarily antibiotics) are first provided. The two are not necessarily related.
In New York, the compliance clock started when sepsis was identified, not when the patient arrived in the ED. Clever readers might note that you can get a 3-hour bundle done in a few minutes, but if it's after a delayed diagnosis of sepsis, the patient can still lose. For example:
Joe's sepsis is identified 6 hours after ED arrival, and then his bundle was completed "on time" in 2.9 more hours (=8.9 hours to completion of the bundle).
Jane's bundle is completed "late," (4 hours after sepsis identification), but was started after a diagnosis of sepsis 15 minutes after ED arrival (=4:15 total time).
The authors controlled for this problem: when using ED arrival as the time for case identification of sepsis for all patients, the same results held. Thus, the faster completion of 3-hour bundles equated to early identification of sepsis by those hospitals.
The New York laws essentially do not attempt to regulate sepsis diagnosis (case identification) at all, "to facilitate broad adoption" of the rules. Hospitals identified cases of sepsis very differently, from clinical assessment only, to automatic "code sepsis" EMR alerts.
Regulated Sepsis Treatment: a New Era
The New York laws are referred to as "Rory's Regulations", so named after a 12-year old boy tragically died in 2012 from unrecognized sepsis (by his family's account on their foundation website) despite evaluations by a school employee, his pediatrician, and an emergency department physician.
The family has political and financial connections and have channeled their grief into effective political action. The New York regulations were the first victory in the foundation's stated campaign to regulate sepsis care across the U.S. by 2020.
Ironically, it sounds like a missed diagnosis of sepsis -- not the delays in care delivery that bundles target -- resulted in the regulations, which still depend on clinicians' making the sepsis diagnosis to initiate the care bundle.
On a national level, CMS's sepsis core measure SEP-1 became the de facto law of the land in 2015 through recommendation by the National Quality Foundation, after lobbying by the original EGDT advocates with industry trade group representation on NQF's review committee.
Does Regulating Sepsis Care Improve Outcomes?
New York regulators and its governor tout their state's 16% relative reduction in mortality from sepsis (30% to 25%) from 2014 to 2016, combined with a 20% increase in case identification (11,000 to 13,000/year). That math suggests an unchanged absolute number of about 3,300 deaths from sepsis each year at the observed hospitals. Generally speaking, the most efficient way to improve observed survival from a disease is to identify more cases, which tend to be milder in severity, or false positives.
Any such secular trends are not discernible in the aggregated data in the paper, but 45% of the 49,000 patients (2014-2016) were described as having septic shock. However, in a 2015 New York state health department report, 49% of the patients had septic shock. Were the presenting patients less ill as time went on?
Sepsis mortality has been declining nationwide, in states without such regulations. But the study purporting to show this also admits that depending on which competing epidemiologic case definition is used, observed sepsis incidence in the U.S. during a given time period can arbitrarily be tripled, or reduced by 70%. With the uncertainty in case definition, it's hard to have confidence in national numbers.
There's no way to know, but it seems likely that rapid sepsis identification and delivery of antibiotics at high-performing institutions under the New York regulations saved lives. Whether mandating the other elements of the bundles is helpful or necessary is less clear.
Risks Of Regulation of Sepsis Care
The New York legislation represents a new level of governmental control of medical practice, legislating care for a condition affecting more than one million Americans each year.
Editorialists worried that:
Clinical practice guidelines often make recommendations involving proprietary medical devices and pharmaceuticals. Device and pharmaceutical companies could lobby state governments to include these products in future regulations. If these lobbying efforts are not transparent, conflicts of interest may lead to abuse of the regulatory process."
The story behind the CMS sepsis core measure suggests these concerns are valid:
The National Quality Forum (NQF) is the only guideline-issuing entity in the U.S. whose performance measures result in changes to Medicare and Medicaid rules and thus are backed by state power. The NQF's review process is designed to include industry representation.
According to a slides presented by a physician at the National Institutes of Health, in 2013 the National Quality Forum endorsed an earlier version of its sepsis performance measure, proposed by the original advocates of EGDT for sepsis. The NQF's endorsed measure included a requirement for measuring central venous pressure and central venous oxygenation. This decision came despite objections from multiple professional societies, and disregarded the pending results from the Process, Arise, and Promise trials. (In 2014 and 2015, these three trials together definitively refuted any benefit of standardized measurement of CVP or ScvO2 in sepsis.)
It so happened that in 2013, the co-chair of the NQF steering committee reviewing the sepsis core measure was the vice-president of AdvaMed, a trade association for medical device manufacturers. Also on the board of directors of AdvaMed since 2008: the CEO of Edwards Lifesciences, the manufacturer of a proprietary central venous catheter to measure CVP and ScvO2. Edwards Lifesciences funded the original EGDT trial and subsequently provided hundreds of thousands of dollars in funding and speaker's fees to that study's author and institution.
After publication of the Process trial less than a year later, the National Quality Forum revised its performance measure, changing use of CVP and ScvO2 from required to optional.
The New York regulations were created without suggestions of undue industry influence, but it's worth noting they are grounded in similar history, evidence, and politics as the CMS core measure.
At least three states (Pennsylvania, Washington, and Illinois) are considering similar legislation mandating the use of sepsis bundles, based on New York's model.