What Counts as a PCI Complication? That Can Be Hard to Pin Down

Morton Kern, MD, often engages his colleagues via email in brief, informal dialogue on clinically relevant topics in cardiology.

Morton Kern, MD, of VA Long Beach Healthcare System and University of California, Irvine, often engages his colleagues via email in brief, informal dialogue on clinically relevant topics in interventional cardiology. With permission from the participants, TCTMD presents their conversations for the benefit of the cardiology community. Your feedback is welcome—feel free to comment at the bottom of the page.

 

William Fearon, MD (Stanford University, CA), asks:

At our monthly QA meeting, we had several cases which our coders had flagged and submitted to the American College of Cardiology’s National Cardiovascular Data Registry (NCDR) as “complications.”

One case is of an NSTEMI patient who had a subtotal proximal LAD, which was successfully stented; however, on post-PCI echo, this person had a low EF, without a record of EF before admission. The resident wrote in their note “new heart failure,” and the coders flagged it as a complication of the PCI. The patient, of course, did fine. 

A second case was of a patient undergoing thoracic aortic aneurysm surgical repair who had intraoperative ventricular fibrillation. He was brought to the hybrid cath lab and found to have a clot in the LAD, which was successfully aspirated. Patient then had completion of surgery but had a > 3 mg/dL drop in hemoglobin, and so the case was flagged and reported as a post-PCI bleeding complication. Patient did fine and had no obvious source of bleeding. 

We have had many others like these, as I am sure others on this email chain have as well.

Questions for the group:

1. Our coders tell us that they are audited frequently and are concerned that we might lose accreditation if we do not include these events. Are other sites finding that their coders are so meticulous?
2. Do other sites have suggestions regarding how to handle these types of cases?
3. Anything the group can do to change the criteria for complications?

Kern:

As I send this note around to our list, I’ll give you my short version. How and why the coders do this before asking an involved physician is mysterious. I have received queries from the coders about certain language that I needed, which they wanted to code at a higher/lower level. Regarding them confusing normal clinical events for complications, I don’t think there is a good system other than that each complication they code undergoes a second look.

---

Bonnie Weiner, MD (Saint Vincent Hospital, Worcester, MA):

I think your NCDR coders are being overly cautious but before they submit, someone should review the complications to ensure they are clinically appropriate.

I don’t think this is a criteria issue. If someone asks NCDR, they are likely to get inconsistent answers.

---

James Blankenship, MD (University of New Mexico, Albuquerque):

About 15 years ago our cath lab noted a surge in post-PCI “strokes” with a rate of about 10%. When we investigated it, we found that the nurses who were completing the NCDR CathPCI Registry data forms were characterizing any complaints such as “dizziness” as neurologic events. This anecdote leads to several conclusions. First, education of coders about the definitions of complications is critical. Second, cath lab QI personnel need to keep a close watch on their NCDR reports. Third, it may be wise to review all coded “complications” before the data is submitted to the NCDR. We implemented that practice in our cath lab and found that the post-PCI stroke rate rapidly returned to normal (rare) levels. Cases like those described above and by Dr. Fearon and others led to this paper describing the NCDR Data Quality Program. I think the biggest opportunity for improvement is, as Dr. Kern suggests, to screen your “complications” before they are submitted to the NCDR.

---

Matthew Cavender, MD (University of North Carolina at Chapel Hill):

Great question—I am currently the chair of the NCDR Metrics, Reporting, and Methodology subcommittee, which reviews all the definitions and data fields in the NCDR. We are always happy to hear from users regarding suggestions as to what makes the most sense from a scientific/quality-improvement perspective. We do have robust risk-adjustment strategies that are designed to help mitigate the effects of these types of events on a program’s overall results. We also review all of our NCDR data/outcomes prior to submitting to make sure we are in agreement with the facts of the case and that the case meets the definitions.

For scenario #1 - In my opinion, this would not meet the definition for post-PCI heart failure. Here is the definition:

• Indicate if the patient has been diagnosed with heart failure
• Target value: any occurrence between birth and current procedure
• Supporting definition: heart failure

Heart failure is a complex clinical syndrome that results from any structural or functional impairment of ventricular filling or ejection of blood. The cardinal manifestations of HF are dyspnea and fatigue, which may limit exercise tolerance, and fluid retention, which may lead to pulmonary and/or splanchnic congestion and/or peripheral edema. Some patients have exercise intolerance but little evidence of fluid retention, whereas others complain primarily of edema, dyspnea, or fatigue. Because some patients present without signs or symptoms of volume overload, the term “heart failure” is preferred over “congestive heart failure.” There is no single diagnostic test for HF because it is largely a clinical diagnosis based on a careful history and physical examination.

For scenario #2 - As you know, having standardized definitions is important to make sure that you get an “apples to apples” comparison across sites/providers. The sites have access to NCDR staff to help with coding complex/unique situations. Any time you apply standard definitions, there can be situations that result in a complication that doesn’t seem “fair.” We try to handle that with risk adjustment, which isn’t perfect but does tend to make these types of situations less impactful to the program metrics.

Here are the current bleeding definitions and risk-adjustment strategy. 

---

Michael Ragosta, MD (University of Virginia Health System, Charlottesville):

At University of Virginia, we have two full-time FTEs (former cath lab nurses) that review all procedures and look for complications using the NCDR criteria. They are experienced and know what to look for and are well versed in the data definitions.

Each quarter, a committee including two physicians reviews the complications with the nurse reviewers and agrees (or disagrees) on whether they meet criteria for complications. If there are any questions, there is a resource with NCDR that can tell us if they meet criteria.

The “heart failure” case you describe would not meet criteria.

---

Nauman Siddiqi, MD (WellSpan Health, York, PA):

At MedStar Baltimore, the core group of three interventional cardiologists would meet monthly in person with our coders, a cath viewer, and the NCDR bible to adjudicate every negative outcome and then meet quarterly for final clean up prior to submitting. This would help us get the cleanest data possible, visualize trends rapidly, and educate the coders. Case 1 would have been knocked out easily. Case 2 is unique as it was in the context of open-heart surgery.

But seeing these clinically insignificant negative outcomes based purely on lab values was not infrequent. Creatinine goes up by 0.3 or greater, and it is acute renal failure. It is what it is. For the hemoglobin one I’ve even seen places not check a hemoglobin the next day for electives, just a hematocrit, but this was found to be tedious and then same-day discharge came around anyway. And, if I’m not mistaken, post-PCI the clock only resets after another cath-based procedure. Meaning even if a post-PCI patient undergoes open-heart surgery during the same admit and has a bleed, we get dinged. I also heard hemoglobin criteria may have changed to 4.

---

David Cox, MD (Sanger Heart & Vascular Institute, Charlotte, NC):

Sadly, this is a common issue: as Mort points out, you’d think your coders would review with your docs before submitting a PCI complication, but they often don’t. I suspect some believe they should act independently and not ‘be influenced’ by a ‘rogue doc not allowing them to appropriately code.’ I’ve heard that before in many places I’ve worked.

A process does exist within NCDR to submit a query for clarification, but frankly it’s often not a timely response and often just involves reading the print, not thinking about the situation.

That said, NCDR staff I don’t think view their role as a collaborative force to advance quality in PCI—they’re more often oddly rigid in their interpretation of complications. Not sure I see a clear path out of this one, and their audits can be chilling, but let’s see if we can find a ‘good faith’ review process that works for all. Maybe NCDR needs a physician ombudsman to adjudicate some of these.

---

Ajay Kirtane, MD (NewYork-Presbyterian/Columbia University Irving Medical Center, New York, NY):

This may be a controversial take, but I do think that for a minimally audited database like this there is possibly the same (or greater) variability in ascertainment protocols for measuring some of these “complications” than there is variability in meaningful patient outcomes for some of these measures. 

The “Hgb-sensitized” bleeding outcome is a great example. How many patients have reliable hemoglobin levels that reflect their volume status immediately prior to a procedure and that is then (apples to apples) compared to a hemoglobin drawn post procedure? I personally have advised hospitals with bleeding rates that were high (but no identifiable quality issue when reviewing every single bleed) to make sure they check a point-of-care or an in-lab assay so that they don’t get unfairly penalized for prehydrating patients. 

How many labs don’t even check a postprocedure hemoglobin (especially with same-day discharges these days)? How many hospitals do/don’t document oozes or swelling at access sites or have protocols in place to minimize the impact of these documentation issues on the outcome?

Yes, I agree that at some point a definition needs to be agreed upon and made. That’s not my point.

My point is that I would bet that there is the same amount of variability in protocols for ascertainment, gaming the system, etc., as there is in actual, clinically meaningful outcomes when it comes to some of these endpoints like bleeding. 

---

Weiner:

Ajay, although you are correct in many respects, I would suggest that the largest issue is not in the reporting of complications but of indications. I think we have all seen the gaming of the system with patients labeled as ACS just because they are admitted with “chest pain.” It is hard to not be “appropriate” when labeled as such even where many of these cases clearly have noncardiac symptoms and these patients have neither ECG changes nor biomarker abnormalities.

---

The Bottom Line From Mort Kern

Some insignificant events during and after PCI can be confused for a complication. Coders may not distinguish between normal clinical events for the procedure and true complications. Without input from an oversight review process by a knowledgeable clinician/nurse, these reports can damage the record of the lab. There is no foolproof system to eliminate coding errors completely. A consensus from the experts would agree that each event entered as a complication should undergo a second look.