Secondary Logo

Journal Logo

Viewpoint

CMS & SEPSIS

How the Past Should Change Us

Mosley, Mark MD

doi: 10.1097/01.EEM.0000488836.23573.55
Viewpoint

Dr. Mosleyis the director of quality improvement, the medical director for residency education, and the director of operations management at Wesley Emergency Center in Wichita, KS.

Figure

Figure

The Centers for Medicare & Medicaid Services required IV beta-blockers for all acute MI patients when they first began setting quality measures for the ED. Hospitals that did not comply were fined. By the time their recommendations became widespread, however, we learned from the COMMIT trial that IV beta-blockers in the first 24 hours increase mortality. (Lancet 2005;366[9497]:1622.) It took CMS several years to catch up with the literature and rescind the IV beta-blocker requirement. What harm was caused because the hospital wanted to maximize financial reimbursements?

And then there was door-to-balloon in less than 90 minutes. We now know from good scientific data that door-to-balloon time in less than 90 minutes does not decrease mortality. (New Engl J Med 2013;369[10];901.) Even worse, an aggressive approach in decreasing door-to-balloon times increases mortality by increasing false-positive STEMIs that delay diagnosis of dangerous conditions such as aortic dissection. (Mayo Clin Proc 2015;90(12):1614.) It is likely that these CMS quality metrics of IV beta blockers and door-to-balloon time less than 90 minutes have physically harmed patients and dramatically increased costs for unnecessary cath lab initiations because the hospital protocol ran quickly before us, through us, and over us.

Then CMS recommended blood cultures and IV antibiotics within the first four hours in the ED for community-acquired pneumonia based upon no good scientific data. Later, we collected data that it was not beneficial. (EMN 2015;37[4]:2; http://emn.online/ECASSIII.) Still, we all ordered blood cultures and slammed antibiotics in as fast as possible to be good stewards of the hospital and secure its CMS reimbursement.

The amount of gamesmanship and over-treatment to make this metric was egregious. It took CMS several years to get rid of the four-hour rule as well as the required blood cultures — metrics they told us were essential for quality medical care. How many unnecessary cases of anaphylaxis, Clostridium difficile colitis, increased antibiotic resistance, and other harms were incurred for no reason other than getting the hospital its financial reimbursement — and perhaps to obtain our own financial bonus for meeting our metrics? That's not to mention the cost to the patient.

Next came thrombolytics for stroke. The three-hour window of NINDS (1995), which to this day stands alone against a mountain of data that proves otherwise, became a 4.5-hour window based on data that showed no benefit if you use NINDS criteria for outcome. (New Engl J Med 2015;372[14]:1301.) But strong science pulling the rope against the arms of capitalism proved no match, especially with CMS as the referee.

Now we are being told that the sepsis bundle based upon SIRS criteria and early goal-directed therapy is good quality medicine. This recommendation by CMS is simply not supported by any good scientific literature over the past two years. (New Engl J Med 2015;372[14]:1301; New Engl J Med 2015;372[17]:1629.) And an expert panel recently published Sepsis-3, an international consensus that SIRS and EGDT are no longer accepted. (JAMA 2016;315[8]:801.)

Any suggestion that a sepsis protocol decreases mortality should be met with extreme suspicion. The initial data touted to prove this was from retrospective chart reviews with poor methods. We have recent systematic reviews that show no difference in mortality for earlier antibiotics in sepsis. (Crit Care Med 2015;43[9]:1907.) Using SIRS criteria to start sepsis protocols on as many people as you can will have you including a lot of patients with the flu or overdoses who are tachycardic with mental status changes. Including a lot of low-acuity patients into your sepsis numbers will markedly improve your overall mortality, without really changing anything. This is why the range of sepsis mortality is 16 to 49 percent. It all depends on whom you include in your numbers. We are joining this CMS charade again for no good scientific reason. We just don't want to get in trouble while we make our metric bonus and the hospital gets paid.

We do not, of course, have data that running everyone who meets SIRS criteria through a sepsis protocol actually does harm. Some physicians argue, “What can it hurt to give two liters of fluid, draw cultures, and start antibiotics?” But plenty of harm can be done. Here are four examples:

False positive blood cultures: Will we call back everyone who has a blood culture that grows Staphylococcus aureus? What about Staphylococcus epidermis? What will we do if a patient has no primary care doctor with whom to follow up? If he is admitted, will you keep him on vancomycin until the cultures come back? Will you put him on home infusion therapy just to be safe?

Valuable time and nurse resources: Our volumes, like everyone else's, seem to be exploding without additional rooms or staff. The worst thing that you can do to affect patient outcomes, not to mention patient satisfaction, is having your nurse draw blood cultures on a population of patients from whom it is increasingly difficult to find IV access while other patients are not attended to in a timely manner. There is an opportunity cost to increasing the number of sepsis protocols per shift.

Destruction of gains made in palliative care and the Choosing Wisely initiatives: A patient who has end-stage pancreatic cancer with hypotension and mental status change is brought in. Every hospital employee will be asking you to initiate the sepsis bundle. Or maybe they won't ask you. They may have a preapproved, nurse-initiated protocol that starts it on every SIRS-positive patient. Unfortunately, while we are slapping each other on the back for improving our quarterly metrics and our monthly check, we will have missed what may be best for this particular patient. Has anyone had the difficult and sometimes long discussion about what we should or should not do for this patient at this time — regardless of the timed metric!? Doing fewer labs and more talking is often the best choice.

Automatic hospital-driven protocols: These erode our willingness to think critically. Reducing extremely complex pathophysiology into a marketing sound bite like “Time is muscle,” “Time is brain,” and “Time is tissue” is simplified to the point of being deceptive and just flat wrong. Dead tissue is not recoverable by going faster. And the door of the ED does not represent the onset of ischemia. We know this. We have simply quit acting upon it because we are too tired to fight the protocol.

Some believe this cry for scientific integrity in the face of corporate control of medicine is naive and too late. Some argue that we have already drunk from the well with STEMI and stroke, and sepsis is just the same thing. But STEMI and stroke protocols are unlike sepsis protocols. Sepsis is even more difficult to define and to recognize, and enormous harm can be done with unnecessary antibiotics, which has already been demonstrated with CMS pneumonia measures. Sepsis protocols greatly expand the target of harm. STEMI and stroke are relatively easy to recognize quickly and harm is hardly ever done by giving someone an aspirin (IV tPA is a different story). But sepsis is difficult to define and recognize, and enormous harm can be done with unnecessary antibiotics, which we learned from CMS pneumonia measures.

What do we do? If we lose our scientific integrity, if we do things that potentially harm our patients for no reason other than making money, we are no longer physicians. We become another provider who wants to stay employed by doing what the hospital wants. Certainly we do not have great data on some areas of medicine, and one can fall on one side or the other of a therapy. I don't believe the SIRS criteria and early goal-directed sepsis bundle is one of them. The current literature is solidly against their benefit. It is time for us to be physicians and say no to therapies that are not supported by good literature and that have the potential to do great harm.

Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.