Monthly Archives: August 2020

On Naming Categories

Personal Experience and Observations

Sometimes when you are writing a paper or a report, and you have gotten the main things down – the text and the figures – there is this one issue you need to finalize: the naming of categories included in the paper. Although this is an integral part of the paper, you just don’t consider it as a deciding factor for the acceptance of the work for publication, more like an embellishment. As an example, this could involve work on a categorization of a biomedical system, and you have already decided on the number of categories and their definitions. But what to name the categories?

Writing the Draft with Categories Named by Descriptive Terms

When I was writing the draft of a paper on a categorization of drug treatments based on the relationship between the therapeutic effect of a drug and the pathophysiologic process being treated or prevented (i.e., therapeutic specificity) – this was in 1995 and I was on the faculty at an academic medical center – and after I had completed the text and defined six therapeutic specificity categories, the question arose what their names should be. I recall spending quite some time on what to name the categories, and finally I came up with six descriptive names, adjectives like preventative and symptomatic. Feeling pretty good about the draft, I submitted the manuscript to Clinical Pharmacology & Therapeutics, but the manuscript was rejected and not considered for publication.

Finalizing the Paper with Categories Named by Roman Numerals

I told one of my senior medical school administrators about my experience, and he suggested I talk to a professor at the school who had been an editor of a journal in his discipline for a long time. So I sent him the manuscript for comments, and within a day or two he called me back and told me that in his opinion the paper read and looked fine, but I should not use descriptive names for categories of therapeutic specificity, because reviewers – and readers – will all get hung up on descriptive terms. So, that’s what I did, I just changed the names from descriptive names to numbered items, using Roman numerals (plus “0”), based on the relationship between the therapeutic effect of a drug and the pathophysiologic process(es) being treated or prevented, as follows:

  • Category 0: for disease prevention
  • Category I: directed at disease etiology
  • Category II: directed at specific disease processes
  • Category III: directed at specific disease manifestations
  • Category IV: for non-specific disease manifestations
  • Category V: for non-therapeutic drug use

An astute reader will note that if we don’t consider Categories 0 and V, which involve preventative treatments like vaccines and drugs like anesthetics, respectively, then the publication basically involves four categories of drug treatments for specific diseases. I the submitted the manuscript to the Journal of Clinical Pharmacology, which promptly accepted it, and that’s where was published: A Classification of Drug Action Based on Therapeutic Effects. J. Clin. Pharmacol., 36(8):669-673, 1996 (for therapeutic specificity scheme refer to Figure 2).

There are of course other non-descriptive approaches to naming categories that I could have used, such as numbers or letters, but I settled on Roman numerals as that felt less ordered than plain numbers, these categories being more categorical than hierarchical. In more complicated situations where there might be categories and subcategories, other approaches can be used, such as alphanumeric approaches, and those including lower-case and upper-case letters.

Subsequent Evolution of The Therapeutics Categorization Project

What happened next? More than two decades later – this was in 2018 and I had retired from the pharmaceutical industry – this general therapeutics categorization scheme had evolved into a much more comprehensive description of systems therapeutics, addressing how the pharmacologic and pathophysiologic processes interact, to culminate in a therapeutic response. This is based on a systems therapeutics diagram, which consists of two rows of parallel systems components for pharmacologic and pathophysiologic processes, representing four different biologic levels of interactions between these two processes, i.e., at the molecular, cellular, tissue/organ or the clinical levels, as follows:

  • Category I: at the molecular level, involving elements/factors
  • Category II: at the cellular level, involving mechanisms/pathways
  • Category III: at the tissue/organ level, involving responses/processes
  • Category IV: at the clinical level, involving effects/manifestations

This latter work, Systems Therapeutics: Diagram, Definitions and Illustrative Examples, was posted on the Therapeutics Research Institute’s website (, in April, 2018 (for systems therapeutics diagram refer to diagram on page 2).


What are the takeaways from this experience? First, when naming categories avoid using descriptive terms, instead use numbers/numerals or letters. Second, regarding the number of categories, observe the parsimony principle, use only as many as are absolutely needed; three would typically be too few, and often four or five will work fine. Third, conceptual projects may often require many iterations and years to evolve as new experience and knowledge has accumulated to inform the work. Fourth, and not the least, seek help and advice from your colleagues and network; the systems therapeutics project might not have happened if the original therapeutic specificity project had not gotten published.

This post has also been published in Medium in August 2020, under the same title.