Blog | Friday, December 23, 2016

Sepsis bundles and why sensitivity and specificity matter


Graham Walker(@grahamwalker) tweeted this in response to a blog post:

“Agree w @medrants on Abx usage. Sepsis guidelines mandating Abx for anything that COULD be sepsis is the problem”

I responded that his example is brilliant. Let's dissect the problem.

Sepsis is a severe problem that responds better to early aggressive treatment.

Those invested in diagnosing sepsis desire bundles that have a high sensitivity. In case you forgot the definition of sensitivity, it is the true positive rate. Sensitivity here represents the percentage of sepsis patients that you treat promptly. Sounds good; we do not want to miss any patients with sepsis.

But wait! All tests or bundles have both false negatives and false positives. We want to minimize our false negatives, but we cannot do that without increasing the false positives. Since specificity equals 1, the false positive rate, we have a tautology. Increasing sensitivity means decreasing specificity.

Anyone who spends some time considering this problem will understand that a sepsis bundle that errs on the side of diagnosing sepsis will have the expected consequence of giving antibiotics to a significant number of patients who do not have sepsis, but rather other reasons for matching the criteria in the sepsis bundle.

We love antibiotics when appropriately used, but antibiotics are not benign. Patients who received broad spectrum antibiotics suffer the risk of antibiotic associated diarrhea, allergic reactions and other side effects. Broad spectrum antibiotics increase the emergence of antibiotic resistance.

The problem that Dr. Walker describes (and again I quote one of his tweets):

“Lact 2-3.9 and admitting them for their GI bleed? You'd better give them abx, and quickly!”

As he implies, we need physicians to understand the patient's context. Simple guidelines that include a check box will treat many patients appropriately, but still too many patients inappropriately.

Any premortem examination of these bundles would quickly identify this problem, yet most hospitalists have had these bundles enacted without considering the patient's context.

Now the good news:

The Centers for Medicare & Medicaid Services (CMS) updated the Severe Sepsis and Septic Shock: Management Bundle (SEP-1) measure specifications several times in response to newly published evidence. As a result, CMS will not score the SEP-1 measure validation for Hospital Inpatient Quality Reporting (IQR) Fiscal Year (FY) 2018. CMS is also postponing the public reporting of the SEP-1 measure on Hospital Compare until it is confident that it has valid data that reflects hospitals' performance.

But I doubt that hospitals will reconsider these bundles. CMS discontinued the 4 and then 6 hour pneumonia rule, but still too many patients get antibiotics and are labeled community acquired pneumonia in most emergency departments. While CMS will not be reporting, the sepsis boulder will keep rolling downhill and gaining speed.

And then we will wonder about overuse of antibiotics and emerging antibiotic resistance.

db is the nickname for Robert M. Centor, MD, FACP. db stands both for Dr. Bob and da boss. He is an academic general internist at the University of Alabama School of Medicine, and is the Regional Associate Dean for the Huntsville Regional Medical Campus of UASOM. He still makes inpatient rounds over 100 days each year. This post originally appeared at his blog, db's Medical Rants.