In the 1979 paper in which he outlined a number of biases that can affect research studies, David Sackett included “compliance bias”, which he described in the following terms: “In experiments requiring patient adherence to therapy, issues of efficacy become confounded with those of compliance, e.g. it is the high risk coronary patients who quit exercise programs.”
The underlying problem to which he was alluding was that participants who are compliant with an intervention may differ from those who are not, and in ways that might affect assessment of the outcome being studied. This means that compliance, in this sense, is a confounding factor in assessing the effect of the intervention.
However, it is notable that in his original description of compliance bias Sackett referred to “adherence”. Modern terminology prefers the term “adherence bias” to “compliance bias”.
Compliance, concordance, adherence–the history of the terminology
The word “compliance” comes from the Latin verb complire, meaning to fill up and hence to complete an action, transaction, or process and to fulfil a promise. In the Oxford English Dictionary the relevant definition is “The acting in accordance with, or the yielding to a desire, request, condition, direction, etc.; a consenting to act in conformity with; an acceding to; practical assent.”[“compliance, n.” OED Online, Oxford University Press, March 2023, www.oed.com/view/Entry/37698.]. In using the word “direction”, this definition also implies acting in accordance with advice, and specifically advice given by a healthcare professional discussing the use of a medicine or some other form of therapeutic intervention, such as a surgical operation, accompanied by directions on how the medicine is to be used or information about how the intervention will be implemented.
However, the modern attitude to the word “compliance” is that it betrays a paternalistic attitude towards the patient on the part of the healthcare practitioner, and that it should therefore not be used.
For this reason, the idea of concordance was introduced. The relevant meaning of concordance in the OED is “The fact of agreeing or being concordant; agreement, harmony” [“concordance, n.” OED Online, Oxford University Press, March 2023, www.oed.com/view/Entry/38351.]. This implies that, for example, in the case of a pharmacological intervention the prescriber and patient should come to an agreement about the regimen that the patient will take. It is not clear, however, whether this seemingly laudable aim is actually beneficial to the patient, and although it is generally assumed to be, the true benefit to harm balance of this strategy has not been determined. Nor is it clear how one can identify patients who desire participation of this type and for whom it would be beneficial and those for whom it would not be beneficial or might even be harmful. Indeed, it can be argued that negotiation between the patient and the prescriber, which is what concordance implies, is often inappropriate to effective healthcare.
Use of the term “concordance” also carries the implication that patients should take greater responsibility for their management, even though not all are willing to do that. Furthermore, there are philosophical reasons supporting imbalance between the responsibilities of the prescriber and those of the patient.
Thus, the terms “compliance” and “concordance” have been replaced by the term “adherence” The term comes from the Latin verb adhaerere, which means to cling to, keep close, or remain constant. In the OED, the relevant definition of “adherence” is “Steadfast commitment to a belief, practice, etc.; fidelity, devotion; (later also) strict or faithful observance of (a rule, promise, etc.)” [“adherence, n.” OED Online, Oxford University Press, March 2023, www.oed.com/view/Entry/2328.], a definition that appropriately conjures up the tenacity that patients need to achieve in sticking to a therapeutic regimen or fulfilling the requirements of other forms of intervention.
For this reason, the original entry titled “Compliance bias” in the Catalogue of Bias has been replaced by a similar entry titled “Adherence bias”.