Extracorporeal Therapy: is it indicated?
Dialysis, hemoperfusion and plasma exchange for dogs and cats. When to refer, how urgent it is, and who to call. Written for clinicians who do not perform these treatments themselves.
Which question is closest to yours?
Not sure what these treatments actually are?
Patient
Species
Body weight kg
Used for the circuit-size feasibility check, the likely platform, and the printed record.
Travel time to the nearest center hours
Do not know? Leave blank - or use the center directories at the bottom of this page.
Kidney - is dialytic intervention indicated?
!First, rule out prerenal and postrenal causes
Azotemia is not the same thing as kidney failure. The IRIS criteria require the problem to be refractory to medical or surgical management, so a patient who is simply volume-depleted, or who cannot pass urine, is treated for that first. Both are reversible, and neither needs renal replacement therapy.
1Azotemia
The threshold below is not a number on its own. It only counts in a patient that is already rehydrated and still climbing.
Creatinine
BUN optional
Which way is it moving?
Needs at least two values to answer.
Hydration status right now
A high creatinine in a dehydrated patient is a fluid problem until proven otherwise.
2Urine output
Output
How long has that been the case? hours
The IRIS bullet needs more than 6 hours, despite appropriate treatment.
Once this patient is rehydrated, stop the maintenance rate. A patient who is not
passing urine cannot excrete a standard fluid rate, so continuing one simply accumulates it, and
severe overhydration is itself a reason to dialyze. The full instruction is in the action list
with your answer below.
3What has already been tried, and what is refractory
These are the qualifiers that turn a lab value into an indication. Check only what is genuinely true.
Potassium mmol/L, optional
Recorded for the printout. The indication above is driven by whether it is refractory and cardiotoxic, not by the number alone - the guidelines deliberately do not set a threshold.
Poisoning - when can a toxin be removed from the blood?
This section does not assess your patient. It gives you the published
reasoning and the numbers so you can work through your own case.
Call poison control for advice - ASPCA APCC 888-426-4435 or Pet Poison Helpline 855-764-7661 in the United States, your own national service elsewhere. Ask about extracorporeal removal directly; it is not raised routinely.
Call poison control for advice - ASPCA APCC 888-426-4435 or Pet Poison Helpline 855-764-7661 in the United States, your own national service elsewhere. Ask about extracorporeal removal directly; it is not raised routinely.
What you need before any of this decides anything
Whether extracorporeal treatment can work depends on five things. Collect all five before you work through the parts below, because a decision made without one of them is a guess.
- Volume of distribution - how much of the drug is still in blood, where a machine can reach it.
- Protein binding - which half of the modality algorithm in Part 2 applies.
- Half-life - how fast the patient is clearing it without help.
- Time elapsed since exposure - read against the half-life, this is what is still on board.
- The largest possible dose - not the most likely one.
The first three are published for every agent in the table below, along with molecular weight, which Part 2 also needs: . If your drug is not in it, the lookup resources are at the foot of this page: .
The last two come from the history rather than from a reference, and they are the two most often wrong. Owners rarely know how many tablets were in the bottle, and the time of exposure is frequently a range rather than a moment.
1Should anything be removed at all?
Four questions in sequence, as Fig. 1 prints them.
1
Is the toxin likely to result in significant organ dysfunction, injury or death?
Judge from the largest possible dose, not the most likely one. Owners rarely know how many tablets were in the bottle, and the guidance is explicit that the recommendation should be based on the maximum possible quantity.
Yes → go onNo → removal is not indicated
2
Will removal by machine clear more than the patient is clearing alone?
This is the step the pharmacokinetics answer. Two numbers decide it.
- Volume of distribution - above or below 2 L/kg? A machine only cleans blood. Below 2 L/kg enough of the drug is still there to reach. Above it, most of the drug has left the blood for the tissues and removal is unlikely to be worth it - call poison control or a center anyway, because there are a few exceptions (Vet Clin North Am 2020).
- Half-life against the dose - what is still on board? Work out what is left, not what fraction has gone.
Worked example A dog eats 100 mg/kg of carprofen. Half-life about 6 hours. Renal toxicity starts around 40 mg/kg.at 6 h50 mg/kgwhat is left after one half-life - still above the renal threshold, so removal is worth itat 12 h25 mg/kgwhat is left after two half-lives - below the renal threshold, so probably not worth extracorporeal treatmentPublished half-lives come from therapeutic dosing. In a large overdose the enzymes clearing the drug saturate and the real half-life is longer, so treat this arithmetic as the best case rather than the expected one.
Yes → go onNo → removal is not indicated
3
Is another antidote or life-saving treatment available?
If there is an antidote, give it first. In the published figure that ends the pathway, because removal is aimed at the poisons with no pharmacological alternative.
In practice it is softer than that
An antidote that blocks further damage does not always clear what is already circulating, and the two can be worth doing together. Ethylene glycol is the example. Fomepizole stops more alcohol being converted, but it does not clear the alcohol already in the blood or the glycolic acid already made, and hemodialysis removes both. Timing counts as much as availability: one given too late does not undo the exposure.
With an antidote in hand, this is a call to poison control rather than a decision to make on your own.
Yes → removal is not indicatedNo → go on
4
Do the benefits of removal exceed the risks to this patient?
Anemia, thrombocytopenia, cardiovascular instability and small body size all raise the risk. None of them rules it out on its own - a center can usually work around a specific one - but each belongs in the conversation. The circuit should hold less than 20% of the patient's blood volume; hemoperfusion drops the platelet count by 20 to 50% depending on the device.
Yes → go on to Part 2No → removal is not indicated
2Which method?
Only use this after Part 1. It picks a method to suit the drug's chemistry. It does not ask whether removal is worth it, and it will name a method even for a drug Part 1 ruled out. Protein binding picks the side, then molecular weight, then volume of distribution.
Start hereIs the drug more or less than 80% protein bound? That single number decides which of the two lists below you read. Only then molecular weight, and only then volume of distribution.
Ifprotein binding is under 80%
under 500 Da→Conventional hemodialysis ± hemoperfusionSmall, unbound molecules cross an ordinary dialyzer membrane by diffusion.The hemoperfusion half does not apply to alcohols, glycols or inorganics - carbon does not adsorb them at any size. Hemodialysis still does, which is why ethylene glycol is a dialysis case and not a charcoal one.
500 to 1,500 Da→High-efficiency hemodialysis ± hemoperfusionStill small enough to cross, and free rather than bound, so diffusion does the work.
1,500 to 40,000 Da→Hemodiafiltration ± hemoperfusionToo large for diffusion alone. Convection - plasma water pushed across the membrane - carries molecules this size with it.
over 40,000 Da, Vd under 1 L/kg→Plasma exchange ± hemoperfusion up to 70,000 DaPlasma exchange removes it with the plasma it sits in, so size does not limit it - but only if it stays concentrated in plasma.Hemoperfusion stops at about 70,000 Da, so it is an option at the lower end of this band only.
Ifprotein binding is 80% or over
over 40,000 Da→Plasma exchange ± hemoperfusion up to 70,000 DaPlasma exchange is not limited by molecular weight at all - solutes over 1,000,000 Da can be removed with the plasma they sit in.Hemoperfusion is limited, and stops at about 70,000 Da. The million-dalton figure is plasma exchange only.
under 40,000 Da, Vd under 0.5 L/kg→Plasma exchange ± hemoperfusionHeavily bound and confined to plasma - the ideal profile, and the one many NSAIDs fit.
under 40,000 Da, Vd 0.5 L/kg or over→HemoperfusionA sorbent column binds the drug directly out of blood, including the fraction carried on albumin.
The 2020 figure understates hemoperfusion. It treats size and protein binding as limits. The 2026 consensus does not: carbon hemoperfusion is listed for 100 Da to 70,000 Da and is independent of protein binding, which puts it alongside almost every row above. What limits it is volume of distribution - under 1 L/kg preferred, 1 to 2 acceptable, over 2 suboptimal - and chemistry: it does not adsorb alcohols, glycols, inorganics or minerals at any size.Vet J 2026;315:106519, Tables 3 and 7
Redrawn from Fig. 1 of Foster JD, Extracorporeal therapies in the emergency room and intensive care unit, Vet Clin North Am Small Anim Pract 2020;50:1215-1236. The boxes and thresholds are as the figure prints them; the explanatory lines under each are ours.
3The published numbers
Every agent the veterinary guidelines carry, with what is published for it and what is not. A blank is information - it means the veterinary literature does not have that value, which is itself worth saying on the phone. Check anything you act on against your own reference: figures differ between sources and may not hold in overdose.
4If your drug is not in that table
Look the four numbers up and carry them through Parts 1 and 2 yourself.
Plumb's Veterinary Drugs - veterinary pharmacokinetics where they exist ·
Merck Veterinary Manual, Toxicology ·
DrugBank - human volume of distribution, protein binding and half-life ·
PubChem - molecular weight for any compound ·
EXTRIP - human consensus on removal, poison by poison.
Human values may not hold in dogs and cats, and a value published only for the other species can differ severalfold. Say which source you used when you call.
Immune-mediated disease - would plasma exchange help?
Unlike the kidney and poison arms, this list is expert opinion rather than a consensus guideline - there are no veterinary randomized trials of plasma exchange (Vet Clin North Am Small Anim Pract 2020;50:1215-1236).
1Does any of this describe your patient?
Check everything that applies. Any single one of these is worth a phone call.
2Weaker evidence
Extrapolated from human apheresis guidelines and shared pathophysiology, where published veterinary experience is sparse. Treat these as a conversation, not a recommendation.
3How is the patient doing?
Providers are unanimous that plasma exchange should not be the last thing tried before euthanasia. Earlier referral is what makes it work.
Find a center
These treatments are not widely available, and most hospitals that offer them have only one machine. Find your nearest provider before you need one, and save the number.
ASVNU - facilities list →
American Society of Veterinary Nephrology and Urology. Around 50 hospitals worldwide, grouped by region, with the treatments each one offers. This is the directory cited in Vet Clin North Am Small Anim Pract 2020;50:1215-1236. It is a plain list, not a search box.
Queen of the Nephron - unit map →
A map of veterinary dialysis and extracorporeal units, with a page for submitting a clinic that is missing. If it does not open, use the ASVNU list above.
Not sure whether to call? Call. Every provider quoted in these guidelines would rather take an early telephone call than a late patient.
Before you refer - can this patient be treated safely?
Does it actually work? The honest numbers
Pooled mortality in acute kidney injury is around 45% in dogs and 53% in cats. A meta-analysis found higher mortality in dialyzed patients (53%) than in those managed conservatively (37%) - which reflects that dialyzed patients are considerably sicker, with more comorbidities and higher illness scores, rather than that dialysis causes harm. The same pattern is seen in human medicine. About 19% of dogs and 25% of cats were discharged with a creatinine back inside the reference range - a proportion of the whole cohort, not of the survivors.
All figures from Foster JD. Extracorporeal therapies in the emergency room and intensive care unit. Vet Clin North Am Small Anim Pract 2020;50:1215-1236.
The counterweight: among dogs and cats that survived acute kidney injury with dialysis, one-year mortality was not much different from 30-day mortality. Patients who get through it can do well for a long time afterwards. Animals with infectious causes did better than non-infectious ones.
For toxin removal the picture is different and more favorable, because the therapy is aimed at a known molecule with a known window rather than at an established organ injury. How much comes out in one session varies widely with the toxin, the modality, the device and the timing, so there is no single figure to quote. For a very large exposure one treatment may still leave a toxic concentration, and a second the next day is sometimes needed.
References
Show the four sources this tool is built on
- Foster JD. Extracorporeal therapies in the emergency room and intensive care unit. Vet Clin North Am Small Anim Pract 2020;50:1215-1236. doi:10.1016/j.cvsm.2020.07.014Source of the four decision conditions and the modality algorithm (Fig. 1), the modality table for common intoxications (Table 1), and the plasma exchange indication list.
- Segev G, Foster JD, Francey T, Langston C, Schweighauser A, Cowgill LD. International Renal Interest Society best practice consensus guidelines for intermittent hemodialysis in dogs and cats. Vet J 2024;305:106092. doi:10.1016/j.tvjl.2024.106092Source of the four indication bullets for acute kidney injury (100% consensus), the toxin-removal statement, and the circuit and hematocrit safety limits.
- Chen H, Cowgill LD, Francey T, Jepson RE, Langston C, Schweighauser A, Segev G. International Renal Interest Society best practice consensus guidelines on the use of continuous renal replacement therapy in dogs and cats. Vet J 2026;315:106548. doi:10.1016/j.tvjl.2026.106548Same indication bullets (91% consensus), plus the platform-choice guidance between continuous and intermittent therapy.
- Butty EM, Barnes J, Reineke E, Dufayet C, Her J, Cowgill LD. Hemoperfusion for veterinary toxicities: International Renal Interest Society best practices consensus guidelines. Vet J 2026;315:106519. doi:10.1016/j.tvjl.2025.106519Source of the hemoperfusion indications and safety recommendations, and of the pharmacokinetic values in the agent library (Table 4).
This tool is clinical decision support for qualified veterinary professionals. It does not replace clinical judgment, a poison control center, or a conversation with the center that would take the case. Pharmacokinetic values are published figures for the species stated and may not hold in overdose, where enzyme and transporter saturation can prolong half-life and increase total exposure.
