Antipsychotic Plasma Levels and Adherence

Jonathan Meyer, MD

Clinical Professor of Psychiatry
University of California, San Diego

Last updated: January 4, 2019


  • Plasma level is the best proxy for antipsychotic CNS action
    • Detects nonadherence
  • For non-responders, levels can help determine:
    • Point of futility
    • Kinetic or adherence issues
    • Need for dose increase




As I alluded to before, occasionally, there are patients who don’t take medicines in the way you prescribe them. Of course, this doesn’t happen to my patients. It’s always someone else’s patients. But that being said, we’re often very poor judges of who is actually taking their medications even when you put people in clinical trials.

So this is a schizophrenia patient who knows that here she is in a study. And often, they do these studies with what’s called a MEMS cap. MEMS cap is a cap which has an electronic chip embedded in it. Every time the person opens the pill bottle, it registers the date and the time. It’s a good proxy for adherence. The assumption is if they’re going through the trouble of opening the pill bottle, maybe they’ll actually be taking their pill.

Even when you have people on a study and you say we’re going to consider you adherent if you take your medicines five days out of seven which is about 70%, the actual level of adherence was about 43%. Meaning 57% of people were nonadherent even though they’re on a study and even though they’re being watched.

But of course, if you ask the patient well what proportion of you are actually taking your medicines five days out of seven, 95% said they were adherent. And if you ask the psychiatrist, their estimate of adherence was 93%. And so we often have no idea what actually is going on with the patient. We often assume that the person is taking their medication simply because they show up to their appointments and they refill their prescriptions And the fact that they remained stable just has to do with biological features of their illness. With whatever level of adherence or whatever level of drug exposure they are experiencing, it keeps them together.


The other important concept is that a lot of times people are deemed treatment-resistant. Like wow, I prescribed this guy 40 mg of olanzapine and he refills his prescriptions and I’m pretty sure he’s taking them because all my patients are adherent. Well, guess what. We have great data here from McCutcheon out of the UK which looked at a group of 99 outpatients who were thought to be treatment-resistant by their clinicians. And surprise, surprise. More than a third had subtherapeutic plasma levels and a good chunk of those actually had undetectable plasma levels whatsoever.


So you say okay, that’s fine. I like to maybe see what’s going on with my patient. How do I have some gross sense of whether they’re taking their medication? Well, the way to have an understanding of this for people on oral medications is at least to have some numbers which correlate an oral dose with the plasma level. Now, the assumption in this table that I’m showing you for all these drugs is that the person is an extensive metabolizer, that they’re not on inducers with the one exception of clozapine where I show the effect of smoking.

So for example, you have somebody on 20 mg at bedtime of aripiprazole. Typically, we get these as 12-hour troughs and you order it. So what result would you expect? Well, I would expect something in the middle range of 200 ng/mL. And so, here we see a mean level of around 230 with a large standard deviation. Again, you have to account for variations there. But certainly if somebody was on 20 mg a day of aripiprazole and their level was 20, you know they’re not taking their medication.


For clozapine, we have to take into account the impact of smoking. So I give you some basic information depending upon whether the person is a smoker or not and whether they are male or female.

For haloperidol, a good relationship to remember is that 10 mg per day, again given at bedtime, so 10 mg qhs should generate a trough plasma level of around 7.8 ng/mL. We have a few other ones as well here.


For fluphenazine, the data are kind of spotty and this is the best I could come up with.

For olanzapine, the data are much more solid. You can see for a nonsmoker, your plasma level is about two times your oral dose. So if you had somebody on 10 mg qhs, you would expect a trough plasma level of 20. On the other hand, if they are a smoker, you could see the relationship is diminished somewhat.


And you know, we have a lot of medicines regardless of adherence and all of that which just have multiple possibilities for kinetic issues and lurasidone is one of them.
So lurasidone as many of you know goes through P450 3A4.

And so there are a number of drugs which both inhibit and induce this. You can see it actually has a significant interaction with strong 3A4 inhibitors which increase the area under the curve ninefold. And you cannot use it with strong inhibitors.

You can use it with moderate inhibitors where it increases it only twofold.

Conversely, strong inducers reduce the plasma levels by 80% and you’re just wasting your time if you try to combine lurasidone with let’s say carbamazepine or phenytoin. There’s also a food effect as well.


And so given all these complexities and there are also some variations in P450 3A4 genetics, the conclusion from the imaging studies is that D2 occupancy correlates very poorly with lurasidone dose but very well with the plasma level of the active moiety. And here’s a nice graph from Steven Potkin to illustrate this.


So here are just a couple comments about getting plasma levels. In general the rule is give the bulk of your antipsychotic at bedtime ideally all of it if possible and get your levels as a 12-hour trough after the bedtime dose in the morning and before they take any morning medication. Even among adherent patients, you may often see levels fluctuating up to 30%. Changes beyond this probably especially if you see this repeatedly where they bounce around quite a bit are probably due to nonadherence or maybe some new kinetic issue. But assuming that they haven’t changed their other medications or habits, almost certainly it’s due to nonadherence. So if you see levels bouncing around by 50% or more, probably there’s an adherence issue, again assuming that they’re getting the levels obtained at the same time in the morning. Of course, you should always be mindful that there are ultrarapid metabolizers or people who are inducing let’s say 1A2 by smoking and be mindful of that. And again, a lot of people are just nonadherent. And so you just have to understand that this is the nature of the game when you’re treating people with chronic severe mental illness.


The other point I want to make here which I’ll mention is that laboratory reference ranges are idiosyncratic at best. And so simply because your lab might say 20 ng/mL is the highest haloperidol level you’re supposed to prescribe, this is not borne out by the literature.

Then I say we’ve published data on people who tolerate levels up to 30 and there are people who tolerate even higher levels. So the best way to use laboratory ranges for antipsychotics is to look at the lowest number, meaning the threshold for response. That is probably pretty good data. The upper numbers are usually based on just some idiosyncratic reading of the literature.


And the most important concept is let’s say the lab says the upper limit of clozapine is 700 but your patient has been on a stable dose for years and their plasma level is 850. Do not reflexively reduce doses simply because the patient’s plasma level exceeds the laboratory range. Antipsychotics have a much broader safety margin than do drugs like lithium or anticonvulsants. Often, we will get consultations in the state hospital for patients. So they say, hey, this person is stable but their clozapine level is 1000.

What should I do?
The most important thing is to document that the patient is tolerating this particular level and then recheck it. Occasionally, there are lab errors. But if it comes back 1000 again on a repeated basis, you can say, all right, this is what this person needs to be stable and they’re showing no adverse effects.
But rapid reduction of doses is most likely going to destabilize your patient and more likely than not the patient may need that level for response.


Now, if the level is excessively high, you got a plasma haloperidol level of 54, generally what we tell people is they may not need a level much beyond 30 but do not cut their dose in half.
There is a concept of what’s called super sensitivity psychosis that people exposed to very high levels of D2 blockade don’t look so good when you take away a lot of that D2 blockade very quickly.
So often in those patients, we will say very slowly reduce the haloperidol by no more than perhaps 5% per month. The idea is that you don’t want to unmask their psychosis in a way you might unmask tardive dyskinesia if you take away an antipsychotic too quickly.


So the key points, once again, plasma level is the best proxy for antipsychotic CNS action.
Getting levels during periods of stability can often help detect nonadherence. And for nonresponders, levels can be helpful for three reasons: To see if you’ve reached the point of futility, to see if there are kinetic or adherence issues leading to subtherapeutic levels, and most importantly to tell you whether you have room for further dose increases.


  1. Byerly M, et al. Psychiatric Services 2007; 58:844–847.
  2. McCutcheon R, et al. Antipsychotic plasma levels in the assessment of poor treatment response in schizophrenia. Acta Psychiatrica Scandinavica 2018; 137:39-46.
  3. Meyer JM, et al. Psych Clin N Am 2016; 39(4):541-556.
  4. Potkin SG, et al. CNS Spectrums 2013
  5. Potkin SG, et al. D2 receptor occupancy following lurasidone treatment in patients with schizophrenia or schizoaffective disorder.  CNS Spectrums 2014; 19(2):176-81.
  6. Meyer JM. Current Psychiatry. 2015; 14(11): 16:19-20 2.
  7. Eap CB, et al. J Clin Psychopharmacol 2004; 24(2):214-9.
  8. Remington G, et al. J Clin Psychopharmacol 2013; 33(2): 161-162
  9. Midha KK, et al. J Psychiatr Neurosci 1994; 19(4): 254-264.

Published in: ,

Earn CME/SA credits

  • Up to 36 CME credits/year
  • Up to 16 SA credits/year
  • Download more than 70 videos
  • Exclusive updates

Free Course: “SSRIs: The Essentials”

Learn the essentials of SSRIs with our free online course.

One module per week:

  • The Mechanism of Action of SSRIs
  • Fluoxetine
  • Fluvoxamine
  • Paroxetine
  • Sertraline
  • Citalopram and Escitalopram

2019-20 Psychopharmacology Online Update

Earn Up to 36 CME Credits / Year

  • Access exclusive updates: new videos every month
  • Download our entire library: more than 70 videos
  • Earn up to 36 CME credits /year