Abstract
Dear Sir,
From time to time even the most influential disease classifications are subjected to fervent criticism calling for a better scientific basis and data supporting the offered diagnostic criteria (1). Such criticism is certainly important as it forces categorical thinking; however, it does not get to the core of the matter. Scientific classifications are undoubtedly of vast importance in scientific research. Their main significance herein lies, however, not in the fact that they offer a most detailed and accurate image of reality, but more in the fact that they provide a scientific agreement about the main features of an object of research and a scientific consensus regarding the name – or, more commonly spoken: the word – we choose to attribute to this very object.
Why is a consensus about the meaning of a word so important in science? Language is our key to the understanding of our world (2). Science, i.e. the aiming at understanding our world, therefore relies on language. Discourse is the basis for scientific progress and a scientific finding that cannot be communicated through language is thus worthless for a society. Language in turn consists of words. How then do words obtain their meaning?
One of the oldest theories concerning this problem is almost as old as philosophy herself: the basic concept of arbitrariness in linguistic philosophy and semiotics (2–4). According to this theory, there is no intrinsic relation between a word and the thing it signifies – a word is arbitrarily assigned to a certain meaning. During language acquisition of a single individual, however, which word is used to designate which thing is not arbitrary, but depends on convention – i.e. a consensus about which word is used for which purpose in a society. This consensus is the basis for understanding each other and the foundation for every possible discourse (4). And this is even more so in artificial languages such as medical terminology. Transferred onto science this means that it does not matter how we call an object of scientific research as long as there is a consensus in the scientific community about the nature of this object, i.e. its basic characteristics. Thus it, for example, does not matter what we call migraine as long as all of us agree on what is called migraine. If, for example, we would elect a big grey animal with a trunk and tusks to carry the name of migraine, then studies investigating migraine would have to focus on animals with these very same characteristics. If we were to define a headache that comes in attacks of 30 minutes and is always accompanied by ipsilateral autonomic symptoms as a migraine, all research concerning migraine would have to focus on syndromes with these very same characteristics. And it is the same with classifications: it does not matter what a certain category in a classification is chosen to signify as long as there is a consensus regarding the essential characteristics of this very thing and as long as everybody using this category refers to the same object with the same characteristics. This, however, i.e. the consensus about a certain object of research, is the very basis for communication and understanding in a (scientific) community (5), which then in turn lays the foundations for scientific progress.
And to raise another problem here: By establishing defining criteria for an object we build a new entity. In the definition of an object thus lies its creation – and demarcation from other objects. If we use the same category of two different classifications, we therefore investigate two different things. This is the very reason why the validation or dis-validation of a classification by systematic studies is simply not possible. Every study trying to validate the present classification using the criteria of this classification as an inclusion criterion would only always verify this very classification again and again and again. But if we were not to use the present classification, there are only two possibilities: one is not to use any classification at all, which is not feasible as every scientific study has to define its object first for consensus. If there is no consensus about the nature of this object in a scientific community, then a scientific discussion is not possible. The second would be to use another classification and other criteria for defining the object of investigation. This, however, is problematic as it would also raise the question as to why this other classification would be better. A study trying to validate these other criteria and at the same time using these criteria would also be doomed to verify this classification over and over and over again. Moreover, the object of the new definition would be a different one than the object of the old. So the matter of discussion here is not two definitions of the same thing but really two different things.
Does this mean that we are never to change an existing classification? Of course not. There are three possible ways of doing so. One would be that in reality no objects can be found to fit a certain category. This category might thus be omitted. The second would be that, in studies strictly applying the defined criteria, additional features of an object may come to light which were not evident when the classification was made and might thus be added to the classification. Third, if a classification differentiates between key features and additional features, it is also conceivable to challenge these additional features with studies strictly relying on the key features of an object. To challenge the key features, however, would not be possible for the named reasons. The question arises whether we can empirically create a classification. Probably not from scratch as every empirical study has to rely on a definition of the object of research. Theoretically, one could investigate an immensely big population of objects with one very general criterion with regard to certain characteristics and then try and statistically identify clusters of characteristics commonly appearing together. Objects presenting these clusters of characteristics would then be classified as belonging to the same category. However, this study would also have to rely on certain predefined criteria for selection of the primary population and thus again also on a classification – however general this might be – with all the problems outlined above. The result of such an attempt would thus also only create other subcategories of an existing classification.
In conclusion:
First and foremost there has to be a consensus about the basic features of an object of research with a certain name. This consensus is arrived at through simple definitions and not through systematic studies. By creating those definitions, the object itself is defined, i.e. created. This definition is positive, as certain characteristics are positively assigned to a certain object, and at the same time negative, as certain other characteristics are assigned to other objects and can therefore not be characteristic of the object in question. So we define what an object is and at the same time – by defining other objects – what an object is not. Classifications – regardless if they are medical or non-medical – provide just such definitions. They define the key – i.e. defining – features of certain entities and thus demarcate them against others. As classifications therefore define and at the same time create entities, it is not possible to disprove classifications for all systematic research aiming at this would also have to predefine its object of research. And in case this very definition is in agreement with that of the classification to be disproved, this research would only always prove the classification in question over and over and over again. In case this new definition is not in agreement with the old classification, the very object of research would not be the same and any studies using this definition could never be used to disprove the old classification as they would investigate different objects. Ergo, the call for changing a classification on the basis of studies using different definitions of a current object is paradoxical.