This blog post is in reference to my article in the May 2023 edition of the Lancet Psychiatry – Lithium Story: Eight Guidelines, Eight Recommendations. It is adapted from notes I sent to one of the editors when constructing the article.

The blog post is a criticism of the clarity and precision of the language used in lithium pharmacokinetic literature. The post focuses on a specific paragraph of the Grandjean paper that contains two examples of imprecise language. Alex explains the implications of the authoritative and inaccurate language used, which can lead to a false sense of understanding and clarity in a reader not familiar with the subject, and may cause a slow deterioration of the original research finding citation to citation, decade to decade. Alex also argues that imprecise language can cause the loss of scientific knowledge, using as an example the case of Amdisen and colleagues, who proposed taking patient serum concentrations at 12 hours in the 1970s, and soon after became standard practice worldwide, but with time the original research finding has been lost.


It might seem strange to say from someone writing a blog post criticising clarity and precision of language, but I personally find it very difficult to write clear and precise language. Goodness, if you could see my first-year PhD report!

But, because I had to work very hard to clarify the muddled thoughts in my head, I recognised mistakes in lithium pharmacokinetic literature similar to those I used to make in my writing.

In the Lancet Psychiatry article, I focus on the paper “Lithium: updated human knowledge using an evidence-based approach” by Etienne Marc Grandjean and Jean-Michael Aubry. It is an extensive collation of knowledge on lithium treatment. However, producing a paper with such breadth of knowledge, can in turn, lead to unclear and imprecise language given how much the authors are required to understand.

I will focus on the following paragraph of the Grandjean paper. Quoting directly

It is universally accepted that lithium doses should be adjusted on the basis of the concentration in serum drawn (optimally) 12 hours (security interval 10–14 hours)[48] after the last dose. Serum concentrations at this time are in the ‘flat part’ of the pharmacokinetic curve.[81]”

There are two significant examples of unclear and imprecise language

Example 1: “Universally Accepted”

Firstly, I think it is important to note that this phrase gives a false presentation of authority to the claim following it. As if the claim is unquestionable. In my opinion, this type of language should not be used in scientific articles. I think it is telling that the author only used one citation. I would expect at least five, preferably ten or twenty, to make such a strong claim.

The word “Universally” is vague. It also suggests every single guideline, government body and scientific paper, advocates for 12-hour serum concentrations with a security interval of 10–14 hours.

The word “accepted” is a strange one. I don’t really know who is accepting this claim? In scientific literature (across all fields) there is no official body that “accepts” whether a statement is true or not. That isn’t how the scientific method works. Instead, a consensus is reached after experimental results from many different sources agree with a statement/hypothesis. But there is no such thing as universal acceptance. If the author is referring to official bodies like NICE or the EMA, why not say so?

Example 2: “’flat part’ of the pharmacokinetic curve”

This statement is as confusing as it is unclear. Curves, by definition, cannot be flat. Even if the author meant “function” instead of “curve”, they are referencing the lithium pharmacokinetic function which, in theory, should obey exponential decay. The derivative of an exponential function is always an exponential function. This means the derivative is never constant i.e. the curve is never flat (or better put – never parallel with the x-axis).

In practice, the only situation where the lithium serum concentration will be “flat” is if the patient is coming off lithium. The serum concentration will eventually drop to zero and stop obeying exponential decay. Which, I think it is safe to say, is not what the author intended!

The author’s use of language in this example has further disquieting implications. The lithium pharmacokinetic curve after peak concentration, according to Chapter 17 of the book “Applied Clinical Pharmacokinetics” and references therein, can be described through a two-compartmental model. An alpha or distribution phase, where lithium is distributed into the tissues (first compartment) and a beta or elimination phase, where lithium is excreted through the kidneys (second compartment).

The “flat part” description seems to show a lack of knowledge, or possibly understanding of the lithium two-compartment pharmacokinetic model. “The elimination phase of the pharmacokinetic curve” is by far a clearer and more accurate description.

The combination of authoritative language in example 1 and the inaccurate language in example 2 is liable to create a false sense of understanding and clarity in a reader not familiar with the two-compartment model displayed by the lithium pharmacokinetic curve after peak concentration.

Like the children’s game “Telephone“, imprecise language can cause a slow deterioration of the original research finding, citation to citation, decade to decade.

The consequences of imprecise language

In fact, I think this is exactly what has happened.

In the 1970s, Amdisen and colleagues, through a series of high-quality papers that included clear and precise language , proposed the time of 12 hours (±0.5 hours) to take patient serum concentrations. This, soon after, became standard practice.

Part of the reason for taking 12-hour levels was to ensure lithium serum concentrations were taken in the elimination phase (to accurately reflect kidney function), avoiding the long distribution phase that came before it. All data I have seen (at least in immediate release tablets) shows a long distribution phase .

In 1992, Gitlin published a paper about lithium pharmacokinetics. In the section titled “The twelve-hour lithium level” the author referenced the work by Amdisen as the reason for the 12-hour level. They continue with a very familiar looking couple of sentences

It is now virtually universally accepted that lithium doses should be adjusted on the basis of the serum level drawn (optimally) 11½ – 12½ hours after the last lithium dose. Serum levels at this time are in the ‘flat part’ of the pharmacokinetic curve.

In trying to summarise the results by Amdisen, Gitlin introduced vague and unclear language. In 2009, Grandjean and Aubry published the aforementioned review paper “Lithium: Updated Human Knowledge Using an Evidence Based Approach”, where they write

It is universally accepted that lithium doses should be adjusted on the basis of the concentration in serum drawn (optimally) 12 hours (security interval 10–14 hours)[48] after the last dose. Serum concentrations at this time are in the ‘flat part’ of the pharmacokinetic curve.[81]”

Different readers will have different interpretations of the aforementioned paragraph by Grandjean and Aubry and all will believe they have a clear understanding.

For instance, an individual responsible for creating guidelines might interpret the “Universal acceptance” as meaning government bodies, while an academic might interpret it as scientific consensus (as I originally did).

In all those cases, the meaning behind the 12-hour level, outlined by Amdisen and colleagues, has been lost.


References