The British Committee for Standards in Haematology (all be it back in 2007) tells us that it is ‘generally accepted’ that the platelet count should be >50x109/l before insertion of a central line, and that the correction of coagulopathy should be considered ‘on an individual patient basis’. They are consistent, with this document produced in 2012 by NHS Blood and Transplant quoting the same value. The AAGBI use the BCSH guidance as the sole reference for their recommendation that platelets should be above 50x109/l for a central line as well.
In the UK at least, thrombocytopenia seems to be getting special attention….
The general acceptance referred to by the BCSH is, I’m assuming, their own 2003 guideline on the use of platelets, which says:
“For lumbar puncture, epidural anaesthesia, gastroscopy and biopsy, insertion of indwelling lines, transbronchial biopsy, liver biopsy, laparotomy or similar procedures, the platelet count should be raised to at least 50x109/l”
I am somewhat sceptical of 50 being a very round number, so thought it was worth having a look to see whether it has any basis beyond accepted wisdom, not really expecting to find any. The key reference for the BCSH guidance is another guideline which is always suspicious, written in 2001 by the American Society of Clinical Oncology, who give a level C recommendation of 40-50x109/l for invasive procedures. As suspected, 50 seems to be a round number rather than an evidence based threshold, but one that has been accepted for many years.
But what is the evidence that a platelet count of less than this will cause significant bleeding? And even if the risk is increased, at what level does that risk outweigh the risk and cost of giving a platelet transfusion?
The only clues to help answer these questions (accepting that the answer will be different for each patient) will be from case series – what have people been ‘getting away with’?
One such example is this paper by Doerfler et al, reporting a series of 104 central line insertions in patients who all had disorders of coagulation. 41 had platelets of <50x109/l, 24 had a raised PT with INR>1.5 and 13 had both raised PT and thrombocytopenia (the others had a platelet count of <100 but >50 or INR>1.2, so really there were only 78 with what I would consider to be significant coagulopathy). Seven patients had bleeding, and these all came from the thrombocytopenia group (<50x109/l). 5/7 had bleeding from the puncture or sutures, and the other 2 developed 2-3cm haematomas. One of these patients had Kaposi's sarcoma of the skin and a platelet count of 6x109/l. This patient received a transfusion of 5 bags of donor platelets (which seems a lot), requiring 1 h of direct pressure to stop the bleeding from the skin. The others just required 10-20mins of pressure. So from this series, thrombocytopenia seems to be the only significant risk, with bleeding that required platelet transfusion in one patient.
Whether transfusion to >50x109/l in that patient would have prevented the bleeding is not certain (or whether the platelets were what stopped the bleeding), neither is whether any harm came of transfusing once the bleeding had become an issue rather than prophylactically. How many more patients would have had lines inserted before another significant bleeding episode is also unknown, so it wouldn’t be fair to say that the rate is 1/78. There are plenty of other series out there, and it’s not worth trailing through them all now but they all come down on the side of showing safety with thrombocytopenia (there will of course be publication bias).
But does any of this matter? Well it turns out it does. This paper looked at why patients meeting criteria for EGDT for sepsis (in 2013, when it was still in fashion!) didn’t get central access. The OR for not attempting central access was just shy of 4 for those patients with a platelet count of <50x109/l. It’s too big a leap to say that harm resulted, but if this threshold is stopping active management then that can’t be a good thing. The paper also reports local bleeding rates, adding support to 50x109/l being conservative.
As for the costs of platelet transfusion, there’s a financial one (about £250) as well as the risks of TRALI, allergy and infection, all of which are more common with platelets than other blood products. There’s also the issue of platelets not being a limitless resource. According to this document the demand on platelets in the UK is increasing, with 15% of platelets being given pre-procedure. The same document quotes the magical 50x109/l threshold however, so they’re maybe not helping their cause!
I think it’s fair to say at this point that with coagulopathy not caused by thrombocytopenia my stance is definitely unchanged. With regards thrombocytopenia, a threshold of 50x109/l is not evidence based and there are numerous examples showing a lower threshold would be fine. I don’t think that the platelet count can be ignored, but I also don’t buy into the concept that the risk of bleeding at laparotomy is the same as inserting a 16G needle under USS guidance.
It is difficult to know for the individual patient what my threshold should be (especially when thinking about function as well as number). Maybe <10x109/l would be better, but that’s just another round number………