When I use a word . . . . Too much healthcare—overdefinitionBMJ 2022; 378 doi: https://doi.org/10.1136/bmj.o2019 (Published 12 August 2022) Cite this as: BMJ 2022;378:o2019
- Jeffrey K Aronson
- Centre for Evidence Based Medicine, Nuffield Department of Primary Care Health Sciences, University of Oxford, Oxford, UK
- Twitter @JKAronson
I have previously discussed medical overdetection1 as one of two major causes of overdiagnosis.2 The other is overdefinition, a term that has not been noticed by major dictionaries, although it has uses in many different disciplines.
Linguistics The earliest English-language instance of “overdefinition” of which I am aware is in a footnote in a 1955 paper by Uriel Weinreich in the journal Language3: “The fallacy of overdefinition has recently been discussed by E. Benveniste ….” In the French-language paper referred to in the footnote,4 Émile Benveniste discussed the difference between polysemy, the possession of several meanings by a single word, and homonymy, the possession of different meanings by words that are spelt the same but have different etymologies. The fallacy is the mistaken attachment of several meanings to a word that has but one meaning, depending on the context in which it is found. For example, as Weinreich explained, “while repasser may appear to the outsider to have three meanings (‘to cross’ ... les monts, ‘to sharpen’ ... un couteau, ‘to iron’ ... du linge), from an internal French point of view one can easily visualize a single meaning (something like “to go over”).” In this case “repasser” is overdefined.
Mathematics Imagine a set of simultaneous equations containing a number of variables. If there are m equations containing n variables, the definition of the set is m/n. If there are more equations than variables (i.e. m >n), the system is said to be overdefined; if the reverse is true (i.e. n>m), the system is said to be underdefined.5 The higher the degree of definition the more easily the variables can be determined. In this case overdefinition is an advantage.
However, confusingly, in some mathematical models, overdefinition occurs when there are more variables than expressions or sources of measurement for them.6 In that case the variables cannot be determined accurately.
Management There are two types of overdefinition in business management.7 First, when a new product line is added and is regarded as a completely new business, and secondly, when excessively precise instructions about how a business should run are laid down, sometimes arbitrarily determined. In the latter case overdefinition introduces too much rigidity in a structure that may be subject to unexpected variation. For example, defining a section of a business by imposing a strict ceiling on the number of individuals in the section, or defining a section according to its major activity, may lead to the creation of new sections, each of which regards itself as a separate entity, introducing the problem of silos and poor communication within the organisation. Journal editors may recognise this problem.
Legislation Overdefinition in legal documents arises from a desire of the legislator to cover every possible eventuality. The complexities that this occasions may cause ambiguity, rather than clarity. Here is a case in point: “I think a good example of our problem with respect to overdefinition is [when you have] about a page and a half defining the words [sic] deception to try to include any conceivable kind of a way that someone may deceive someone else, and it seems to us that this is a good example where a phrase like fraud, false representation, device or artifice to defraud would be better than an attempt to enumerate every conceivable kind of deception.”8
It is clear from this brief, indeed underdefined, account that the dictionary makers have a lot to do in defining “overdefinition.”
Two major mechanisms drive medical overdefinition: first, altering the threshold for a risk factor, typically below some previous value, the treatment of which confers benefit; secondly, by expanding disease definitions to include patients with ambiguous or very mild symptoms.
Altering risk factor thresholds Examples include lowering the blood pressure at which a diagnosis of hypertension is made from, say, a systolic pressure over 150 mmHg to one over 130 mmHg for all adults or lowering the value of glycated haemoglobin or fasting blood glucose concentration at which diabetes is diagnosed, leading to a new disorder called “prediabetes,” the prevalence of which has been described as an epidemic.9 This description was no exaggeration; a contemporary study had suggested that in 2011 about one in three adults in the UK had a glycated haemoglobin concentration (5.7–6.4%) that would have qualified them to be labelled “prediabetic,” up from one in nine in 2003.10 This is also reflected in the numbers of PubMed hits from a search for the word “prediabetes,” over 15 000 hits in all, the annual numbers of which have increased exponentially, from 91 in 2000 to 1517 in 2021, with a doubling time of about 56 months.
Expanding disease definitions As an example, consider the restless legs syndrome, which has also been called Ekbom’s syndrome, since Ekbom described 35 cases in 1945 and a further 70 cases in 1950.11 That he reported that “Mild instances are common, occurring in approximately 5 per cent of a normal series, but lack practical significance” has not prevented others from extending the diagnosis to such cases.
By definition, patients who are labelled with a disorder under a new definition are at lower risk than those diagnosed under earlier definitions. The harmful consequences of overdiagnosis through overdefinition come from labelling and the use of treatments (including lifestyle changes) that offer little, if any, benefit, given the low risks, but that can have important physical, psychological, social, and financial consequences.2
As with overdetection,1 solutions to overdefinition of an existing disorder are hard to find. However, some possible approaches have been suggested. For example, after a thorough literature review, a Delphi discussion of a draft document, and a day-long face-to-face meeting, members of the Preventing Overdiagnosis Working Group of the Guidelines International Network proposed a checklist of actions that might be effective.12 Each item is associated with a pertinent question:
● Definition of the new condition—How do the previous definition and the new definition differ? In some cases this may be problematic, if the condition has not been well defined before; however, in that case one should be even more suspicious about the need for a new definition.
● The precision and accuracy of the definition—What are the repeatability, reproducibility, and accuracy of the new definition? This question, which will often be difficult or impossible to answer, implies that careful studies are necessary before the introduction of any new test or label that may increase the prevalence or incidence of a disorder.
● Number affected—How will the new definition change the incidence and prevalence of the condition? Large changes may have important implications for affordability of healthcare, in finding resources both to treat the new cases and to manage the inevitable adverse outcomes of so doing.
● Trigger—What has suggested that a new definition is needed and is it justified? Possible triggers include the availability of a new treatment, especially a prophylactic one, or a new test, although that itself raises the potential problem of overdetection.
● Benefits—How do the expected incremental benefits for patients defined as having the new condition compare with the benefits under the previous definition? This might include a cost-benefit analysis.
● Harms—How do the expected incremental harms for patients defined as having the new condition compare with the harms under the previous definition? This should take into account the potential harms attributable to treating the condition, the potential harms likely to accrue through not treating it, and other associated harms, such as anxiety engendered by the diagnosis, and the involvement of family members in caregiving and possible investigation in the case of genetic disorders.
● The benefit to harm balance—How does the expected benefit to harm balance for patients defined as having the new condition compare with the benefit to harm balance under the previous definition?
These questions imply a need for careful studies specifically designed to answer them. Finding answers may add perspective to the question of whether an existing entity needs to be redefined.
A final thought
As this discussion may have suggested, in the context of diagnosis “overdefinition” might be better termed “misdefinition,” particularly since in several cases the new definition of a condition involves lowering the criterion for admission to the group (e.g. a lower blood pressure labelled as hypertension, a lower blood glucose labelled as diabetes). However, the word “misdefinition” joins “overdetection” and “overdefinition” in having been ignored by major dictionaries, although the Oxford English Dictionary (OED) traces “mis-define” back to 1867, and contemporary instances of “misdefinition” can be found; here it is, for example, in a text of 1873: “[As a definition of ‘revolt’], Archdeacon Todd has ‘turn’, which as a definition is vague and inexact. His misdefinition shows how unfamiliar revolt, in its modern sense, must have been to him.”13
But in any case, in the world of “too much healthcare” the prefix “over” seems to have become almost compulsory.
Competing interests: none declared.
Provenance and peer review: not commissioned; not externally peer reviewed.