Hemodialysis in lithium poisoning: there is no evidence. Full stop.

September 29, 2015, 2:44 pm

0Lithium Carbonate, 300mg, 1,000 Capsules per Bottle McGuffMedical.com★★★½☆

Hemodialysis for lithium poisoning. Lavonas EJ, Buchanan J. Cochrane Database Sys Rev 2015 Sep 16 [Epub ahead of print]


At a session at the ACEP Scientific Assembly last year, Jerry remarked: “Guidelines should regularly conclude that there is no evidence. Full stop.”

I really like this Cochrane review of hemodialysis in lithium poisoning because [SPOILER ALERT] that is exactly what the authors conclude. They note that although hemodialysis undoubtedly increases clearance of lithium, there is no agreement as to whether this enhanced elimination actually improves clinical outcome.

In an attempt to answer that question, they used explicit criteria to review the literature, looking for randomized controlled trials that studied the effect of extracorporeal techniques (hemodialysis, hemofiltration, and continuous renal replacement methods) on outcomes in patients with lithium toxicity.

Their conclusion:

[T]here is no evidence from randomized controlled trials to support nor refute the use of hemodialysis in the management of patients with lithium poisoning.

Full stop.

This seems right on, and is useful counterpoint to the recent ExTRIP review that, based on no logic or evidence that was apparent to this reader, came up with precise lithium levels for which to recommend extracorporeal treatment.

Although this Cochrane review included no studies, the paper is worth reading for the general discussion of lithium toxicity and the logical (if not evidentiary) basis for hemodialysis.

Related post:

Hemodialysis in lithium poisoning: what is the evidence?


  1. steve Sheldon Says:

    Dear Leon.. Thank you for highlighting what is truly a controversial topic. My take on cochrane is that there are no randomized controlled trials to help us and that is accurate. But the entire world of evidence is not made up exclusively of randomized controlled trials and infact there are even bad randomized controlled trials. Should we throw out the baby with the bathwater?

    I would love an RCT, but I can’t get one. Short of that I need to know what to do and it was great of the extrip team to put the existing literature in context.

    Leon, a question to you? Who with lithium poisoning would you dialyze and what is your opinion based on? Help me.. If you dont believ what extrip has to offer what better advice do you have and please tell me how you formulated your opinion.

    Thanks so much for great reviews

  2. Marc Ghannoum Says:


    As a fan of your reviews, you force me to enter the discussion here to settle a few misunderstandings.
    I regret that you once again confuse the absence of evidence with the absence of “good” evidence. They are not the same. We included 166 original studies in our review. Admittedly, most of them consisted of case reports and case series, publications that are considered to be “very low” level of evidence but there were also 5 descriptive cohorts and 3 observational studies.

    As you surely must know, Cochrane reviews concentrate on trials so it is unsurprising that this specific Cochrane review included no studies.

    This being said, we acknowledged and concluded ad-nauseam that there is no good evidence for most of our toxin recommendations. Yet, every day, clinical toxicologists must recommend treatment based on poor or absent evidence. Activated charcoal for caffeine overdose? Alkalinization for methotrexate overdose? Same story. If you’ve recommended dialysis for salicylate overdose before, and I hope you have, (again on poor evidence), then you must admit that there are conditions that justify dialysis. What EXTRIP does is simply that: to identify conditions that justify dialysis. But it does so after doing a systematic review of the evidence, based on discussion, consensus, and review by 30 clinicians and pharmacologists with a process that is reproducible and robust. One may not agree with our recommendations, but I strongly disagree with the cynic’s view that recommendations be withheld until trials are published. If that were the case, very few if any recommendations would exist for antidotes, decontamination, and elimination enhancement techniques.

    Clinicians who may not have the experience and knowledge base to treat poisoned patients should be allowed to benefit from guidelines, as long as limitations are properly disclosed.

  3. James Says:

    I don’t know why people bother writing these toxicology Cochrane reviews. Almost all aspects of the medical or clinical toxicology evidence base are not determined by a randomized controlled trial. A similar question would be: is there any evidence for hemodialysis in any poisoning? Answer: there is no evidence. Full stop. I might waste some time writing a Cochrane review on this… At least the EXTRIP group are giving clinicians something that may be useful for their practice.

  4. Leon Says:

    Thank you all for the comments.

    Steve: By saying that there is no evidence that HD improves outcomes in lithium toxicity, I’m certainly not saying it shouldn’t be used or that I wouldn’t recommend it. Certainly would dialyze a patient with significant signs of toxicity. In contrast to the ExTRIP guidelines, would not dialyze an asymptomatic patient with Li = 4.1 and creatinine = 2.1 mg/dL without trial of volume repletion. Argument for HD in asymptomatic patient is to remove drug before it moves into the CNS. This is a matter of clinical judgment that must take into account amount ingested, time of ingestion, baseline renal function, etc. Too complex to reduce to specific guidelines based on exact levels and test results.

    James: I find all the ExTRIP papers extremely useful, with excellent and comprehensive reviews of the individual topics. The workgroup certainly recognizes the weakness of existing evidence on these topics. My main problem is the use of weak evidence to make specific recommendations based on lab values.

    Marc: In the Cochrane review, the authors note that their conclusion differs from that of the ExTRIP workgroup, explaining that this is a result of the different types of evidence considered, and that Cochrane relies on RCTs. I certainly agree that informed recommendations can — and should — be made even in the absence of RCTs. My concern is that I’m sure many clinicians will just look at the guidelines (e.g., Li > 4 and Cr > 2) without reading the excellent review and discussion in the paper itself, and feel compelled to dialyze in some situations where hydration might be a better option. It is I think more important to understand basic principles and use clinical judgment. Although not all ED practitioners are well-versed in tox principles, expert consultation is available 24/7 through a poison center.

  5. Marc Ghannoum Says:

    Thank you for these clarifications,

    A few more points for the benefit of banter:

    We certainly mention that dialysis is not a first-line therapy and our recommendations assume that full supportive measures including agressive hydration have been initiated.

    Further, although clinicians in the States have the luxury of benefiting on the expertise of poison centers, this is far from being the case in all countries where ER physicians and nephrologists need to make an informed decision quickly.

    Guidelines are nothing more than additional tools to help provide the best care for sick patients. In no case should these precede or replace sound clinical judgement and patient-specific evaluation.



  6. Sophie Gosselin Says:


    I think the choice of words in your original post might reflect a misunderstanding of the work of EXTRIP and thus I invite you to review our initial methodology article. I am puzzled and left wondering whether or not you remember all the detail of our methodology considering these two statements in first post questioning the validity of the EXTRIP lithium review.

    “This seems right on, and is useful counterpoint to the recent ExTRIP review that, based on no logic or evidence that was apparent to this reader, came up with precise lithium levels for which to recommend extracorporeal treatment.

    Although this Cochrane review included no studies, the paper is worth reading for the general discussion of lithium toxicity and the logical (if not evidentiary) basis for hemodialysis.”

    The logic of EXTRIP is explained in detail in our methods papers: a rigorously calculated consensus of experts who appraised the entire body of evidence we could gather. This is surely better than the published opinion of a few authors that keep being cited over and over and over with threshold values we are not entirely sure where they came from.

    Clinical evidence is not simply constituted of RCT’s otherwise what would be the point of our Clintox journal case report section or conference abstracts and of pretty much the entire body of literature in clinical toxicology?

    I disagree with your opinion that an editorialized Cochrane review from one single author that included NO SINGLE PAPER and no guidance as to when to perform HD in lithium poisoning is better than what EXTRIP did with a consensus from large group of interdisciplinary experts reviewing the entire body of literature we could find in every possible languages.

    In your reply you also say this in further criticism to the EXTRIP recommendations:

    “My main problem is the use of weak evidence to make specific recommendations based on lab values.”

    I think this is a bit unfair. Thresholds are needed at some point to make a decision. There is a threshold for ST segment to be considered an STEMI, a threshold value to give vitamin K as per the Chest guidelines, thresholds for NAC with the Rumack Matthew nomogram. What are people to use if they are not given any thresholds? Had EXTRIP not given any specific values, I am quite sure certain might have written that EXTRIP produced fuzzy recommendations that are not clinically useful!

    I agree consultation with a Poison Control Center is always recommended and useful. I am also pretty sure that many PCC have written local policies and protocols incorporating specific threshold values using the same weak evidence that is available to all but perhaps not with the rigorous approach that a systematic review requires.

    I hope this discussion will help clarify the misunderstanding there seems to have been with the work done by EXTRIP and I encourage the readers of this post read both papers and decide for themselves which they find the most useful for their practice.

  7. Leon Says:

    Marc: Well, I think we agree substantially on all of this. Perhaps my main problem was with the possibility of the recommendations themselves being taken on their own out of the context of the entire clinical picture and the limitations of the evidence — which is completely discussed in the ExTRIP papers. Since the whole question of indications for hemodialysis in lithium poisoning is so indications, I certainly appreciate the value of expert opinions and how they may help guide treatment.

    Sophie: Yes, I think my original post on the ExTRIP paper was misunderstood, probably because I didn’t state it precisely enough. I certainly did not intend to imply that the Cochrane Review by Lavonas and Buchanan was “better” than the ExTRIP paper, or that considering only RCTs (especially when there were none) trumps looking at all published “evidence,” even if everyone agrees it is weak. As I stated, I found both papers valuable in their discussions of lithium poisoning, and felt that the Cochrane provided useful “counterpoint.” I also felt (and I could be wrong) that the specific recommendations re: levels were not supported by any good evidence. As I’ve always stressed, given the knowledge and experience of the members of the ExTRIP workgroup, these were valuable even as expert consensus opinions. Nevertheless, I do think that, just based on evidence, Cochrane is correct that the entire question is still unsettled, at least from lack of good evidence.