Evidence, Proof & Rationality

 

Definitions of Evidence

How can evidence be defined?

Despite the frequency with which evidence is invoked in the discourse of contemporary health care, it is far from clear that the term is used in an unequivocal manner. Consequently as a first step to clarifying what evidence is it is worthwhile to consider some definitions offered.

The Oxford English Dictionary gives several senses of the term evidence:

  1. The quality or condition of being evident
  2. Manifestation
  3. That which makes evident, mark, trace
  4. Ground for belief; that which tends to prove or disprove any conclusion

Legal definitions of evidence cover similar conceptual ground:

  1. Any species of proof or probative matter, legally presented at the trial of an issue, by the act of the parties and through the medium of witnesses, records, documents, exhibits, concrete objects, for the purpose of inducing beliefs in the minds of the court or jury as to their contention.
  2. All the means by which any alleged matter of fact, the truth of which is submitted to investigation, is established or disproved.

     In health care and medicine, evidence is conceived in a scientific context. Goodman and Royall (1998) have defined scientific evidence as follows:

Evidence is a property of data that makes us alter our beliefs about how the world around us is working. Another way to say this is that evidence is the basis upon which we derive inferences.

A recent editorial in The Canadian Medical Association Journal explicitly seeks to define scientific evidence:

In its simplest form, the available scientific evidence consists of the published report of a single piece of original research. (Miettinen 1998)

     David Sackett characterizes the process of evidence based care as " the conscientious, explicit and judicious use of best evidence in making health care decisions".

     From the definitions cited above, it is clear that evidence in some way is conceptually linked to notions such as proof and belief. The definitions advanced by Goodman and Royall and Miettinen contribute the concepts of research and inference. The definition advanced by Sackett et al uses prudential, moral (conscientious, judicious) and logical (explicit) language.

     Hence, evidence is some observation, fact, or organized body of information, offered to support or justify inferences in the positive demonstration of some proposition or decide a matter at issue. Scientific evidence is information that has been gathered under a systematic approach or protocol.

Back to Top

Characterizing Medical Evidence

Features of Medical Evidence

Evidence derived from scientific studies relating to health care has notable features that are not accurately captured in the definitions considered.

     Evidence from health care research, particularly clinical and epidemiological studies is provisional and defeasible. This means that it never attains absolute certainty and can be revised in light of new evidence. Consequently the occurrence of definitive studies is comparatively rare. This is a common feature encountered in the interpretation of health information.

     An excellent example of the provisional nature of medical evidence is peptic ulcer disease. In three decades the etiology and management of this disease has undergone dramatic and profound changes. In the 1970's textbooks proclaimed the need for white diets and postulated stress as an important element of causation. In the 1980's, with the introduction of cimetidine, hypersecretion and the use of H2 blockers were the cause and treatment. In the late 1990's Helicobacter pylori is the favoured etiologic agent and eradication with triple therapy the preferred treatment. If evidence were not defeasible, that is revisable in light of new information, it could not be accommodated. Hence one salient feature of medical evidence is its inherent provisional nature. Research evidence is rarely an eternal truth or a constant (outside of certain basic biologic laws).

     This contrariness occurs because evidence is emergent and therefore is expected to change with time. There is always the possibility that even the best evidence can be overturned in the course of time. Excellent evidence based therapies can be superseded by newer ones. For example, isoniazid (INH) is used for prophylaxis of tuberculosis. There are several excellent randomized trials that unequivocally establish its effectiveness in the prevention of tuberculosis. It is granted a Grade A recommendation in The Canadian Guide to Clinical Preventive Health Care. However, compliance with this medication is poor, largely for two reasons. The duration of therapy is long (6-12 months) and there are potential side effects. Clinically ill people are poorly compliant so it is no surprise that clinically well people are similarly non-compliant. Diagnostic and screening tests likewise can be superseded. For this reason, despite the existence of good evidence, research will continue, and standards of best evidence will change with time.

     There are also constraints on the amount and type of evidence we can expect to receive as consumers or clinicians. The existence of constraints means that there are some forms of evidence that are unlikely to be forthcoming. This means that evidence will always be incomplete.

In general there are three ways in which evidence can be constrained:

  1. Ethical:

    There are some types of evidence we will not possess so long as current ethical standards prevail. Randomization in certain circumstances is regarded as unethical. This constraint can be overcome only if society markedly changes its views with respect to the nature of experimentation on humans.

  2. Economic resources:

    Some may regard this constraint as capable of being overcome in principle. However, it is well known that sufficient resources do not exist to deliver all health care (evidence based or otherwise) to all persons who need or desire it. Consequently allocation decisions must be made. Evidence may be put forth as a candidate for a criterion of resource allocation.

  3. Computational resources:

    Computational constraint is an epistemological concept. It refers to the extent to which knowledge and certainty is possible. Godelian constraints refer to the limits of proof in open systems. Since closed axiomatic systems cannot demonstrate rigorous proof of their consistency, it is clear that any evidence provided from empirical research will fall well short of certitude. This is inherent to the process of knowing. Following form this is the existence of time constraints in the context of decision making. Spreng's triangle illustrates the interdependence of energy, time and information. One can only achieve maximum information at the cost of increased time.

Evidence is collective. It is found in a community of enquirers. The extent and volume of current scientific evidence is too great for any one person to synthesize or comprehend. Hence there is a collective aspect to the production, dissemination, evaluation and use of evidence.


Historical Asymmetries:

As a consequence of the above considerations, evidence as it exists today is asymmetric. There are differences in evidence access both within and between disciplines. For example the tradition of encouraging research as an important component of professional development has come comparatively late to health care disciplines such as nursing, physiotherapy and occupational therapy. For some elements of the health care system there are no traditions in postgraduate education (for example health inspection, health technology such as radiology technology laboratory technology.) As well, some issues are not sufficiently prevalent to warrant full-scale attention (rare diseases); some lack appeal from the point of view of funders (vitamins, public health interventions e.g. restaurant inspection).

Back to Top

Evidence and Proof

Logical Definition of Proof

In essence, a proof is any finite sequence of well-formed statements such that each statement in the sequence either is an axiom or follows from previous members by a valid rule of inference. A proof of the statement B from the premises any finite sequence of statements (with B the final statement in the sequence) such that each member of the sequence: (a) is one of the premises A or (b) is an axiom, or (c) follows from previous members of the sequence by a rule of inference.

(Stanford Encyclopedia of Philosophy)


Legal Definition of Proof

  1. A body of evidence supporting a contention. Those facts from which a conclusion can be drawn. In this sense, proof can be convincing or unconvincing. But see no. 2.

  2. The result of convincing evidence. The conclusion drawn that the evidence is enough to show that something is true or that an argument about facts is correct. There are various standards of proof including: beyond a reasonable doubt (how convincing evidence must be in a criminal trial); by clear and convincing evidence; and by a preponderance (greater weight) of the evidence. In this sense, proof is always convincing. But see no. 1.

(Oran Legal Dictionary)


Proof in law admits to varying standards according to the context and situation. In other words, the weight of evidence required to meet the standards of proof can also vary. In this sense proof in law is less strict than proof in logic or mathematics. The burden of evidence in law is premised on the context in which it is used. Proof must always be convincing and evidence is that which is used to justify or support the conclusions.

     The question then becomes when does evidence become convincing and what determines or constitutes this. In the standard evidence hierarchy proposed by proponents of EBM, convincing evidence derives from the methodology employed to draw conclusions. Hence the evidence hierarchy can be regarded as setting the standard for proof for a variety of circumstances. This is clearly the sense in which proof is used in law where proof is related to evidence that is compelling. Furthermore,

The greater weight of evidence, not as to quantity (in number of witnesses or facts) but as to quality (believability and greater weight of important facts proved). This is a standard of proof, generally used in civil lawsuits. It is not as high a standard as clear and convincing evidence or beyond a reasonable doubt.

(Oran Legal Dictionary)

Back to Top

Evidence and Rationality

The use of scientific evidence is intended to place decision making on a rational basis. In other words an assumption of evidence-based decision making (EBDM) is that the integration of scientific evidence into health care decision making will result in better decisions by health care providers and better health care outcomes for patients. There is and can be no empirical justification for EBDM, but another underlying assumption is that the use of evidence in practice is more rational.

     The concept of rationality is related closely to that of the employment of reason and reasoning skills. The formal aspects of reasoning are found in logic and mathematics. The applied aspects of reasoning belong to the domain of cognitive science. Recent innovations in cognitive science have emphasized the variability of definitions of rationality in the scientific literature. The relationship between reasoning and decision making has largely been unexplored and been the focus of independent literature. It is often proposed in academic contexts that reasoning and decision making are distinct activities. As Evans notes in the real world: "the distinction between reasoning and decision making is blurred." (Evans et al. 1993)

     Such, however, is not the case in scientific research, where the standards and goals of rationality may differ from that of the clinic, government agency or hospital. The simplest way to understand scientific reasoning is as being concerned for the truth, falsity or degree of probability of propositions. Scientific reasoning and the social process of science attests to the degree of belief in this. However, practical reasoning is concerned with goals and ends and the standard of rationality may be quite distinct from that of a scientific study. One of the barriers to the transfer of evidence is the conflation of the standards of rationality and the mis-specification of the context of the application of evidence. Indeed, given the provisional and defeasible nature of scientific evidence it may be rational (from a practical reasoning perspective) to refrain from introducing scientific evidence into practice until a body of uncontrolled experience is attained (think of post marketing surveillance) The inherent conservatism of some practitioners may not be misplaced.

Evans makes this distinction between practical and scientific rationality clear:

Rationality-1 (rationality of purpose): reasoning in a way which helps one to achieve one's goals
Rationality-2 (rationality of process): reasoning in a way which conforms to a supposedly appropriate normative system such as formal logic.

     According to the idealized model of evidence, rationality in the first sense would predominate in quadrants one (Qualitative/Personal) and two (Qualitative/General) and rationality of the second type would predominate in quadrants three (Quantitative/General) and four (Quantitative/Personal). Similarly the standards by which evidence is adjudicated will vary according to which standard of rationality is being used.

Back to Top

Narrative Contexts

Interpretation

What do we mean by interpretation?

Interpretation is a common daily activity. We interpret a wide variety of verbal, textual, nonverbal and physical data every day. Interpreting information is likely the primary cognitive mode for most people. Some have argued that the aim of science is to eliminate interpretation from the analysis of data.

Interpretation is seen as the contrary to calculation. Calculations follow a set of inexorable rules and yield results consistently. This is the sense in which algorithms are appealed to in health care contexts. An algorithm will specify the steps to take each time a decision is required. There is no room for either subjectivity or arbitrariness.

Interpretation does not follow rules as tightly as calculation. Interpretations are commonly thought to be arbitrary and not capable of either commensuration or adjudication. An oft quoted phrase illustrates this point : "Its just your interpretation."

Yet, some interpretations are arguably better than others. Contextual rules of thumb help here. We can quite clearly recognize self serving or transparently facile interpretations of texts or events, even if we cannot ultimately agree on the finer points.

Back to Top