(written in 2004; reviewed/revised 3/2019)
Lab tests for carbamazepine and valproate / divalproex: how often should you do lab tests to monitor safety with these medications?
Carbamazepine (Tegretol, Carbatrol)
Several psychiatrists widely regarded as experts in medication treatments recently were asked about how they use lab tests to monitor for rare but dangerous reactions to medications. Here is the “transcript” (which was edited for release in a “special report” April 2004, which is pretty close to an advertisement, in this case for carbamazepine (Tegretol, Carbatrol). In other words, there is a risk here that their comments were somehow “slanted” by knowing that this was being prepared for one particular product*).
Dr. Blumenfield. What do you advise our psychiatrists to do if they see a dip in the white blood cell count (WBC)?
Dr. Bowden. The white count is not a real issue. Once a patient is stabilized, I don’t routinely check WBCs, not even annually.
Dr. Ketter. I check WBCs if patients get a fever or are sick for a few days.
Dr. Bowden. That’s different. I do it only in relationship to clinical symptoms.
Dr. Blumenfield. How about liver enzymes?
Dr. Bowden. Well, I’ll check those. We’ll see changes, but I don’t do that with great frequency, maybe twice a year.
Dr. Weisler. I tend to do them about twice a year, after patients are stabilized on carbamazepine, to check the LFT’s (liver function tests) and a CBC (complete blood count). It’s not a lot of effort. You’re drawing blood anyway.
Dr. Ketter. I wait for clinical signals to do more of those tests. The best time to test is when patients have a clinical complaint that raises suspicion.
Dr. Gazda. Do you check their sodium as well?
Dr. Ketter. I do it once a year.
Dr. Weisler. We’ll check it periodically, about twice a year or if signs of hyponatremia are present.
Dr. Gupta. Many mental health centers have protocols that suggest every three months. In your own practice, you decide that, but in mental health centers, where the chronically ill are treated, they have set protocols. The state hospitals have set protocols that patients on valproate or carbamazepine will have blood levels and LFTs done every so often, no matter what.
Dr. Skiba. It’s worth adding the clinical point that if you do the center protocol every 3 months and the white count is fine, but then somebody develops a fever, that’s when you really want to check it. You don’t want to say, “We checked it 2 months ago.”
Dr. Weisler. What about blood levels? Once a person is stable, I don’t check them very often at all, unless some medication is changed.
Dr. Bowden. I will still check it periodically, simply because if you have the benefit of an available blood test for any drug in psychiatry that’s metabolized in the liver, it’s important to check because the liver’s capability to metabolize drugs is influenced by so many different factors, such as aging. However, just to make sure that you’re still on the right track, it’s worth doing it for that reason, but it’s not treating the blood level.
Dr. Weisler. I actually feel it’s a plus sometimes with some of these drugs to be able to say “look, we can titrate this drug”. We can find a range where most people will respond. We don’t have perfect blood level data with mania, but we are in the range that we see for epilepsy. Many people feel better when they see themselves in a therapeutic range. They know that you’re not overdosing them or underdosing them, and then they don’t worry about how many milligrams of drug they need to feel better.
Interpreting the experts
Note that Dr. Ketter, of Stanford University, formerly of the National Institutes of Health, one of the really well known names in this group, does not routinely monitor white blood cell count or liver function tests. I don’t understand why he monitors for sodium problems, as these are less dangerous and more likely to manifest themselves with, as he puts it, “clinical complaints” (aka “symptoms”).
Dr. Bowden is similar except that he monitors LFT’s, “maybe twice a year”. He does not monitor white blood cell count (at least for patients on long-term treatment; this article is talking about patients “stabilized” on carbamazepine, not about the first few months when problem reactions are much more common). Similarly, there is no agreement on how often if at all to test carbamazepine blood levels.
This general disagreement and overall lack of commitment to regular monitoring matches the comments I’ve solicited over the years from my local neurology colleagues, several of whom have indicated that they do not routinely monitor white count, liver function tests, or blood levels — though again, amongst them, there is disagreement.
On the basis of all of this, I think we can conclude that there is no clear rationale for testing at any particular interval (at least after 6 months; the bad reactions being discussed here are more common early in treatment. Note that in the discussion above, the emphasis is apparently on long-term lab test frequency, not the first 6 months). However, there is agreement that testing should be used when there are symptoms that suggest a bad reaction may be developing.
(The following statements derive primarily from a Psychiatric Times review in 2004 by Ron Pies, which cites numerous primary sources including an important one from a German consensus conferenceKonig. I trust Dr. Pies as a reliable source, based on the value of many previous, thorough reviews — which is important, because what is said here goes against the usual more fearful recommendations.)
Valproate is associated with liver failure and death — in about one in 40,000 cases, almost all of whom were under 2 years of age or had other medical problems besides epilepsy (for which valproate is a standard medication). However, valproate can much more frequently cause mild increases in liver enzymes. Most experts recommend stopping the medication if routine liver enzyme tests (ALT, AST) are increased above normal — one hears “twice normal” or “three times normal” for such recommendations, though I’ve not seen data on which this might be based (perhaps it is known by liver experts that increases less than these levels are not associated with long-term bad outcomes).
However, Dr. Pies notes that these increases may be temporary, returning to normal on their own, and that repeating the test even several times is a good idea before just stopping the medication.
Finally, Dr. Pies emphasizes that patients feel sick when they have serious problems with their liver reacting to valproate, and that symptoms of nausea, and other flu-like symptoms precede the liver test abnormalities. On this basis, citing the German consensus conference, he concludes:
. . . rather than focusing solely on [ALT/AST], we ought to be asking our patients questions such as, “are you generally feeling well lately?” Are you having any loss of appetite, nausea, malaise [body aches and the overall feeling that goes with them], or flu-like symptoms?”
On the basis of the rarity of the severe liver problems, and the ability to detect such problems through symptoms which probably precede abnormal liver test results, I do not routinely monitor liver tests in people taking divalproex, even though this goes against the usual recommendations. It just makes more sense to me to listen closely to patients’ symptoms than have everyone getting a needle when there aren’t any haystacks in sight.
I am not suggesting that other doctors do it this way, just citing the literature behind my practice (I send my patients here for these references).
Note also that there are other problems with divalproex which also sometimes need lab testing, but for which we don’t routinely do such tests: pancreatitis, and “hyperammonemic encephalopathy“. Any patient with any of these symptoms should be calling her/his doctor to report them:
- abdominal pain, loss of appetite, nausea and vomiting (pancreatitis); or
- lethargy, vomiting, and confusion (HE).
We doctors should remember that for some patients lab testing is not simple, or a minor inconvenience (as opposed to the comment above: “It’s not a lot of effort. You’re drawing blood anyway.”) At some level they also have to pay for these tests. We are looking here at balancing the patient’s need to be as safe as possible, and the doctor’s need to know that she/he is not harming the patient, with the pain and inconvenience and expense of lab testing. There is no “one-size-fits-all” approach to that balancing process.
If a patient really wants to monitor very closely, I have no problem arranging that. But we should not go along with suggested testing intervals unless there is some sort of rationale. The “protocol” testing intervals discussed above may have been strongly influenced by the pharmaceutical company recommendations which, one would think, may be heavily biased toward reducing their own liability (e.g. “we said you should monitor frequently, it’s not our fault, it’s the doctor’s fault, etc”).
By contrast Drs. Ketter and Bowden are precisely the kind of experts I would expect to cite data for a particular approach, if there were reliable data available to go by. These doctors would not go against good research. The approaches they discuss above, what they actually do, speak loudly to me about the lack of any clear standard.
Instead, each patient will need an approach to lab testing that meets her or his need for safety, balanced with his or her need to avoid pain and inconvenience and expense.
*The expert conversation was published by CNS News, April 2004 issue, Special Report insert. I was invited to correct this attribution on 5/28/04 with input from James Prudden of CNS News:
“CNS News was merely the distribution vehicle for this CME (it was prepared by Precept medical communications and accredited by Alabama-Birmingham), and thus had no part in the creation of the program. You also say it was sponsored by CMEZone, again an error. CMEZone is a clearinghouse for all of our CE work and is entirely free to participants. It is not a sponsorer [sic] of CE.”
He also noted:
“your jaundiced view of CE seems to indicate that CNS News creates slanted editorial, and that is an unfair charge, particularly in this case when in fact we did not even create the program.”