Conversations in Cardiology: Is a hematocrit drop after uncomplicated PCI really a bleeding complication?

Compiled by Morton Kern from contributions from Richard Bach, MD, St. Louis, Missouri; John Bittl, MD, Ocala, Florida; Barry Borlaug, MD, Rochester, Minnesota; Greg Dehmer, MD, Temple, Texas; Kirk Garrett, MD, New York, New York; Lloyd Klein, MD, Chicago, Illinois; Mitch Krucoff, MD, Durham, North Carolina; Gus Pichard, MD, Washington, D.C.; Habib Samady, MD, Atlanta, Georgia; and Bonnie Weiner, MD, Worcester, Massachusetts.

Compiled by Morton Kern from contributions from Richard Bach, MD, St. Louis, Missouri; John Bittl, MD, Ocala, Florida; Barry Borlaug, MD, Rochester, Minnesota; Greg Dehmer, MD, Temple, Texas; Kirk Garrett, MD, New York, New York; Lloyd Klein, MD, Chicago, Illinois; Mitch Krucoff, MD, Durham, North Carolina; Gus Pichard, MD, Washington, D.C.; Habib Samady, MD, Atlanta, Georgia; and Bonnie Weiner, MD, Worcester, Massachusetts.

Our colleague Kirk Garratt in NYC asks an interesting question about hemoglobin (Hgb) drop after PCI:

Hey Mort, 

We’ve recently learned that our percutaneous coronary intervention (PCI) bleeding rates are up, even though we’ve added a number of additional radial operators. In reviewing the data, it seems that many of the “bleeds” were patients without any apparent bleeding event but who had a fall in hemoglobin (Hgb) of 3gm or more. I don’t think our processes involve unusual procedural blood losses, but we do use a fairly aggressive hydration policy: routine patients get at least a liter, and renal patients may get quite a bit more. It seems sensible that prolonged NPO [nothing by mouth], insensible losses associated with anxiety, and vasodilatation from medications would warrant aggressive hydration, but in comparing our practice with the other PCI-capable hospitals in our system, we give quite a bit more fluid than the others. In asking my colleagues, I think we’ll find an incredible non-uniformity in protocols for yet another practice that ought to be well defined by now. 

Question: Do people think that aggressive hydration combined with typical procedural blood loss could push the Hgb down by 3gms, or am I fooling myself?

Mort Kern, VA Long Beach, California: I believe the drop in hematocrit (HCT) by itself is not bleeding per se, but a function of hospital-related blood draws and hydration. Three grams seems like a lot, but we all have noted post procedure drop in HCT in patients in the hospital for several days. According to the BARC (Bleeding Academic Research Consortium) categories1, this drop in HCT seems to be in type 1 category and of no clinical significance. I think it’s possible to drop the HCT 3gms by fluids, but I’d double-check for occult bleeding, too. Of course, accurate reporting from operators about bleeding during and after the procedure is required to be secure about this conjecture. 

Regarding aggressive hydration, we don’t give a liter of saline to each patient before the procedure since many come as outpatients. We do give a liter to patients with impaired renal function whenever possible before the procedure, whether in or outpatient. In-patients are hydrated, but rarely more than a liter extra. We probably should hydrate patients based on increasing urine output rather than estimating state of hydration. And I agree that, unfortunately, there is no standard for this policy on aggressive pre procedure hydration.

Mitch Krucoff, Duke University, Durham, North Carolina: I routinely hydrate patients who have reasonable ventricular function for immediate issues such as nitrate/sedation tolerance, to blunt potential vagal events, and for what I believe is still, to date, the only really proven measure to reduce kidney injury, e.g. salt water. A drop of 3gm hemoglobin in this setting (NPO prior to procedure, blood draws and arteriotomies, IV hydration) can be achieved without any “bleeding” per se. More importantly, such drops rarely have clinical symptoms or a need for transfusion (depending on what the initial Hgb is, of course), and hence ended up in BARC 1 designation. For quality metrics, I would not classify these as “bleeds” at all, unless a bleeding site was specifically identified.

Greg Dehmer, Baylor Scott & White Health, Temple, Texas: At the risk of being laughed at for sharing such an old article, this issue has been formally examined in a study2 from the early days of angioplasty. Of course, times were different then, with larger access catheters, etc., but this level of decline in HCT is not unheard of. I am a firm believer in aggressive hydration post PCI if you believe the patient can handle it. Even if they have marginal left ventricular (LV) function, I tell the fellows “It’s easier to treat heart failure than renal failure.” You are not fooling yourself; this does happen in some patients without a bleeding event.

Barry Borlaug, Mayo Clinic, Rochester, Minnesota: We did a study where we compared hemodynamic responses to saline infusion and exercise (Andersen et al, Circ Heart Failure, e-pub ahead of press, 2015). We measured hemoglobin immediately preceding and immediately following infusion of 10 ml/kg NS at 150 ml/min (e.g. 1 liter on average over 5 minutes). The saline was pre-warmed so we wouldn’t drop core temperature. The mean hemoglobin dropped from 12.5 to 11.0 gm/dl with this infusion (n=26). The 95% CI for the reduction with saline was -1.3 to -1.6 gm/dl. The largest drop observed was -2.3 gm/dl.

These patients got 4000-5000 U heparin, but no other anticoagulants, and there were no PCIs performed. But this saline infusion is obviously much more rapid than what is given in practice, and the slower you give it, the more the fluid redistributes from the vascular space to the tissues.

Gus Pichard, Washington Hospital Center, Washington, D.C.: A drop of 3 grams Hgb is unusual but possible after hydration, blood draws, etc. I always give at least one liter of saline before diagnostic or interventional procedures (including transcatheter aortic valve replacements [TAVRs]). For those with severe congestive heart failure (CHF), I monitor the pulmonary artery (PA) pressure during rapid infusion, and at times use 5% DW [distilled water] (the dextrose molecule keeps the fluid intravascular, until the liver slowly metabolizes it). Recently our cath lab established a clinical pathway that requires 1 liter of saline for all cath lab procedures, with modification for profound LV dysfunction. 

Habib Samady, Emory University, Atlanta, Georgia: We have physician-specific “report cards” that include bleeding, and have also noted higher rates that we expected. We hydrate aggressively as well. Another issue is whether you discard or give back the 20cc or so of blood drawn prior to ACT [activated clotting time] draws from the manifold (I personally give that back immediately, but I know other colleagues don’t). During a long case, you may discard >50cc of blood which in concert with the hydration could drop the Hgb. Although this is not bleeding, I think we should reduce blood loss. 

Lloyd Klein, Rush University, Chicago, Illinois: Although I do agree with what has been said, a 3 gram drop in Hgb is not due to hydration alone in most cases. If the patient was adequately volume repleted at the time of the baseline Hgb, additional volume isn’t going to create such a substantial dilutional effect. On the other hand, if there was dehydration at the time of the baseline draw, then it could easily be related to re-hydration. I have spoken with an excellent hematologist here who concurs with this: maybe a 1 or 1.5 gram drop, not 3. A drop of that magnitude is most often due to a blood loss. But whether that loss is due to blood drawing, catheter flushing, or other losses versus a bleeding complication is another matter. Definitely it can be technical in origin, but a concerted effort to find the blood is also called for. Before radial access, this was often ascribed to loss of blood in the femoral canal and thigh, and I can understand that possibility. In the arm though, there is no place for 3 grams of blood to hide. Kirk, how hard are your guys looking for a retroperitoneal hematoma?

Bonnie Weiner, University of Massachusetts, Worchester, Mass.: Lloyd, you are probably correct, but I suspect a significant percentage of our patients fall into the “dehydrated” category. Etiologies will vary, but I suspect it is more frequent than we appreciate.

Lloyd Klein, Rush University, Chicago, Illinois: I think that is right, Bonnie. Especially when there is cardiomyopathy, valvular disease, or hypertension treated with diuretics.

Richard Bach, Washington University, St. Louis, Missouri: We have been vigorously tracking and trying to reduce our NCDR [National Cardiovascular Data Registry] bleeding and AKI [acute kidney injury] rates among our PCI patients for the last few years. Our impression has definitely been that for many cases, aggressive hydration alone (along with minor if not trivial amounts of blood loss during the procedure, but without any signs of bleeding and minimal likelihood of retroperitoneal [RP] bleed) can be responsible for an observed 3gm Hg drop (and admittedly, more likely among patients who seem “dehydrated” at the baseline blood draw). In response to this, we have attempted to delay the “baseline” blood draw until after a standardized pre-procedural hydration protocol, but this is challenging from a process perspective and ironically, we have observed on the flip side that sometimes the serum creatinine appears diluted to a lower value than the patient’s previous creatinine on that sample, and then after the procedure, if it rises ≥0.3 mg/dl, even back to the patient’s true baseline, that patient reaches criteria for an AKI event by NCDR definition. [Sounds like the spot between the rock and the hard place – MK]

John Bittl, Ocala, Florida: Canning the NPO order maintains homeostasis better than any other approach and would likely eliminate the possibility of dilutional anemia. Please see this editorial comment3 if you have any remaining reservations after earlier collegial “debate” on the topic.

The bottom line

A drop in Hgb after PCI should prompt a close look for the site of bleeding, especially for femoral access cases. Beyond sites of obvious bleeding (gastrointestinal [GI], retroperitoneum, access site), other causes of lower Hgb include dilution, blood draws from lab tests, blood loss during catheter exchanges (both for flushing and leaking), asymptomatic retroperitoneal or other location accumulation (unlikely), and unexpected destruction of red cells (e.g. hemolysis). Before calling the asymptomatic drop in Hgb with no source a complication, it would be good to review the alternate possibilities and the BARC categories first.

My thanks to my cath lab experts for helping us understand this problem. I always learn new things through these conversations. I hope you do, too. 


  1. Mehran R, Rao SV, Bhatt DL, Gibson CM, Caixeta A, Eikelboom J, et al. Standardized bleeding definitions for cardiovascular clinical trials: a consensus report from the Bleeding Academic Research Consortium. Circulation. 2011 Jun 14;123(23):2736-47. doi: 10.1161/CIRCULATIONAHA.110.009449.
  2. Phillips SJ, Spector M, Zeff RH, Skinner JR, Toon RS, Grignon A, Kongt ahworn C. Hematocrit changes after uncomplicated percutaneous transluminal coronary angioplasty. Am J Cardiol. 1989; Oct 15; 64(14): 940.
  3. Bittl JA. A proposal to reduce contrast nephropathy: eliminate the NPO order. Catheter Cardiovasc Interv. 2014 May 1; 83(6): 913-914. doi: 10.1002/ccd.25482.

Disclosure: Dr. Kern reports he is a consultant and speaker for St. Jude Medical and Volcano Therapeutics, and a consultant for Boston Scientific, Opsens, ACIST Medical, and Merit Medical.