FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2916210538> ?p ?o ?g. }

• W2916210538 endingPage "175" @default.
• W2916210538 startingPage "173" @default.
• W2916210538 abstract "There are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns—the ones we do not know we do not know. And if one looks throughout the history of our country and other free countries, it is the latter category that tend to be the difficult ones. Donald Rumsfeld Medicine is often described as a practice of decision-making under uncertainty. Clinicians are called to determine how a patient's care should be managed without knowing, for example, exactly what effect the prescribed therapy will have on that patient's condition. Likewise, patients must accept (or refuse) therapy under the same conditions (and with lesser knowledge of medicine than their clinicians). Uncertainty manifests in several ways in clinical medicine. A task as routine as the measuring of a patient's blood pressure contains various sources of uncertainty related to the size and positioning of the cuff (eg, too big or too small for the arm), the timing of the measurement (Was the patient at rest? Was the patient anxious?), and features of the device used to measure the pressure (Is it valid?). Likewise, in a scientific investigation of an intervention, the estimate of therapeutic effect contains uncertainty due to sampling, confounding variables, reliability and validity of outcome measures, etc. Scientists and clinicians have proposed various methods for reducing the impact of many of the sources of uncertainty in clinical medicine. However, despite our best efforts, uncertainty in clinical medicine is inescapable. The sources of uncertainty are many. As one cannot know what all the sources of uncertainty are, it is no good to attempt to discuss them all here. It may, however, be useful to reflect on the response of the clinical and scientific medicine community to uncertainty. That reflection will reveal that we should be cautious (and humble) about what we profess to know. An important step in dealing with uncertainty is defining what we mean by uncertainty. Djulbegovic et al1 review common definitions of uncertainty and provide a taxonomy of uncertainty in clinical medicine. The gap between what we know and what we can theoretically achieve in our knowledge is conceptualized as “reducible uncertainty.” In their conceptual framework, Djulbegovic et al1 also identify a gap between that theoretical “perfect state of knowledge” and “certainty” (their terms), which they define as “epistemic irreducible uncertainty.” That is, there are some things we cannot know—what they are we cannot know. The authors also recognize an irreducible uncertainty related to statistics. Statistical methods model error in our measurement and approximate observed effects. Those errors can be reduced in part through increasing the size of the sample used to measure the phenomenon, although the infinitely large size sample (theoretically) necessary to achieve certainty in the estimate of effect (assuming all other sources of error are controlled) can never be achieved. However, the impact that statistical uncertainty has on our “perfect” state of knowledge depends on the extent to which we rely on statistical thinking in our investigations of a phenomenon. How much “irreducible” (so-called epistemic or statistical) uncertainty exists relates to the extent to which the nature of the phenomenon is deterministic. It may be that we do not (yet) have the tools to close the gap between what we can know and certainty, in which case, the uncertainty is indeed epistemic (ie, due to limits in our knowledge). We must be careful not to assume that the uncertainty we experience is not ontological; perhaps we live in an indeterministic world. In Miké's view,2 it is an ethical imperative that we “increase the awareness of, and come to terms with, the extent and ultimately irreducible nature of uncertainty” (p. 24). Uncertainty may also be conceptualized using a Bayesian framework. From the Bayesian perspective, one's certainty (or “belief” or confidence in the epistemic position) is determined by both the evidence and the evidentiary weight on his or her belief that individual ascribes to that evidence. An individual will upgrade and downgrade certainty as new information is acquired. An individual may achieve certainty about a phenomenon on the basis of the evidence but that may ultimately prove subjective—weighing the evidence and determining which evidence is relevant requires judgement. Thus, it may be strange to think about our knowledge of a particular healthcare intervention from the extent to which one is certain or uncertain about that knowledge. Rather, it may be more appropriate to think of the threshold to which one is willing to act in some way on the basis of the available evidence. In clinical practice, it may be that what matters is the functional certainty/uncertainty of our knowledge, ie, we can make decisions with reasonable expectation of outcome on the basis of what is known. What is considered “reasonable” under a specific condition or in context is another matter, and one worth considerable discussion and examination. As that is beyond the scope of my current discussion, I will not examine it here. How does one deal with what Rumsfeld described as “known unknowns,” in particular, the epistemically reducible uncertainty? One approach is to do research on the phenomenon of which we are uncertain. Claude Bernard advocated for experimental medicine, so as to reveal the laws of nature that determine health phenomena. The relations between cause and effect described by these laws would provide the practitioner of medicine the information necessary to perform her or his duties in meeting the healthcare needs of the patient. However, the experimental conditions that brought about the phenomenon may not be present in the context of a particular patient, and one cannot be certain that all the important conditions have been met, a priori, when choosing a course of action, thereby limiting the extent to which such laws of nature in fact can eliminate uncertainty for the clinician and the patient. Thus, although the goal was determinism (and in a sense, the annihilation of uncertainty), Bernard recognized that “we shall certainly never reach absolute determinism in everything; man could no longer exist. There will always be some indeterminism then, in all the sciences, and more in medicine that's in any other” (3; p. 140). Much of contemporary clinical research seeks to empirically investigate the relative effect of one intervention compared to another in a study population. Such investigations reveal probabilistic associations between the intervention and the outcome of interest. * The challenge with these population methods is reducing the impact of confounding variables. The presence of a confounding variable can introduce uncertainty into the knowledge derived from a study. Known confounders can be dealt with (to some extent) through study design, tighter specification of the population of interest, and/or statistical adjustments. Unknown confounders are more challenging to the researcher. To deal with the impact of unknown (and known) confounders, many have advocated for the use of randomization (eg, Sackett et al6 and Schunemann et al7). † Randomization, it is thought, will balance the groups being compared on the presence of those confounding variables (known and unknown), thereby “washing out” their impact on the estimate of the relative effect of the intervention on the outcome of interest. To what extent randomization methods do indeed deal with these unknown confounders has been questioned (see arguments by Worrall8 and Thompson9). That is the trouble with things that are unknown—we know little, if anything, about them. An additional uncertainty introduced when applying knowledge derived from these population studies to clinical practice is related to the fidelity between the study population (and the necessary conditions to achieve the observed outcome) and the clinical population where the study results will be applied (and the context of their care). Cartwright10, 11 has provided a detailed discussion on the issue of the generalizability of study results, revealing some of the sources of uncertainty for clinical practice. Thus, contemporary methods of clinical research are not sufficient for eliminating uncertainty in clinical practice. What may be equally concerning is research can only help us reduce (part of the) uncertainty about those things we decide to examine. We may still be ignorant of what we ought to know for any clinical problem. Even if it was the case that the randomization process in a clinical trial did balance the groups with respect to known and unknown confounding variables and there was high fidelity between the trial population and the target population, there is still uncertainty as to what outcome will be achieved for the individual patient the clinician will encounter. Population studies, such as clinical trials, produce average effects. Elsewhere, I (with my colleague Amiram Gafni) have discussed the uncertainty that accompanies the translation of population derived average effects to the care of individual patients.12, ‡ In any trial, there are some patients who benefit from the therapy and some that do not. The trouble is that one cannot know if the individual patient a clinician encounters will be among the group who will benefit from the intervention or in the group who will not. How one is to deal with this uncertainty is not clear. Tonelli and Upshur13 suggest, perhaps mindful of the issues of clinical research evidence I describe, that clinical medicine would be better served if training included philosophical engagement with the nature of uncertainty. Models have been proposed to help researchers and healthcare stakeholders organize their thoughts on uncertainty. For example, Han et al14 offered a conceptual taxonomy “that characterizes uncertainty in health care according to its fundamental sources, issues, and locus” (p. 828). In the current issue of the Journal of Evaluation in Clinical Practice, Pomare et al15 propose a revised model of uncertainty—“the model of uncertainty in complex healthcare settings (MUCH-S)” (see commentary by Han et al16). To what extent these models are helpful in activities that are intended to reduce the uncertainty that plagues clinical practice is unknown. However, such models may constitute a useful starting point. Medicine is ignorant of many things. The “master argument” put forward by Stegenga17 in “Medical Nihilism” would suggest that practitioners of medicine are more ignorant on the usefulness of their interventions than they (or we) are led to believe. On the other hand, with respect to Stegenga's argument, there are in fact some medical interventions (eg, the use of antiretroviral therapies to manage HIV) that are very effective, vastly exceeding expectations. Discriminating between what we know and what we believe we know and do not is a challenge. What we ought to ascribe the status of a “known knowns” is not a trivial task. The threat to medicine is not that we do not know things. The threat is that we create a sense of certainty out of uncertain knowledge—that we profess to know things that we do not. We have a tendency to find solace in scientific findings. Average effects are codified into best practices. Those best practices become benchmarks for quality of care. Failing to meet those benchmarks is conceived as a sign of poor decision-making and ultimately poor patient care. Best practices and quality benchmarks may provide support for clinicians and may even provide a sense that healthcare decisions are grounded in objective science and not arbitrary, but such practices and benchmarks do not eliminate the uncertainty of the patient's outcome. Schwab18 notes that “judgments in medical practice are always accompanied by uncertainty, and this uncertainty is a fickle companion – constant in its presence but inconstant in its expression. This feature of medical judgments gives rise to the moral responsibility of medical practitioners to be epistemically humble” (p. 28). When faced with the uncertainty that abounds in clinical medicine, perhaps epistemic humility is our most reasonable recourse. Some of these thoughts come from conversations with Amiram Gafni, Brian Baigrie, and Ross Upshur who have over the years continued to help me to recognize what I do not know." @default.
• W2916210538 created "2019-03-02" @default.
• W2916210538 creator A5032010271 @default.
• W2916210538 date "2019-02-20" @default.
• W2916210538 modified "2023-10-16" @default.
• W2916210538 title "Humility in the face of uncertainty" @default.
• W2916210538 cites W1247968195 @default.
• W2916210538 cites W156673597 @default.
• W2916210538 cites W1708917685 @default.
• W2916210538 cites W2022856530 @default.
• W2916210538 cites W2033607319 @default.
• W2916210538 cites W2085144863 @default.
• W2916210538 cites W2127142367 @default.
• W2916210538 cites W2130566074 @default.
• W2916210538 cites W2477250644 @default.
• W2916210538 cites W2893246593 @default.
• W2916210538 cites W2901582708 @default.
• W2916210538 cites W2916200923 @default.
• W2916210538 cites W3082867365 @default.
• W2916210538 doi "https://doi.org/10.1111/jep.13116" @default.
• W2916210538 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/30788895" @default.
• W2916210538 hasPublicationYear "2019" @default.
• W2916210538 type Work @default.
• W2916210538 sameAs 2916210538 @default.
• W2916210538 citedByCount "2" @default.
• W2916210538 countsByYear W29162105382020 @default.
• W2916210538 crossrefType "journal-article" @default.
• W2916210538 hasAuthorship W2916210538A5032010271 @default.
• W2916210538 hasBestOaLocation W29162105381 @default.
• W2916210538 hasConcept C144024400 @default.
• W2916210538 hasConcept C15744967 @default.
• W2916210538 hasConcept C17744445 @default.
• W2916210538 hasConcept C199539241 @default.
• W2916210538 hasConcept C2778456462 @default.
• W2916210538 hasConcept C2779304628 @default.
• W2916210538 hasConcept C36289849 @default.
• W2916210538 hasConcept C71924100 @default.
• W2916210538 hasConceptScore W2916210538C144024400 @default.
• W2916210538 hasConceptScore W2916210538C15744967 @default.
• W2916210538 hasConceptScore W2916210538C17744445 @default.
• W2916210538 hasConceptScore W2916210538C199539241 @default.
• W2916210538 hasConceptScore W2916210538C2778456462 @default.
• W2916210538 hasConceptScore W2916210538C2779304628 @default.
• W2916210538 hasConceptScore W2916210538C36289849 @default.
• W2916210538 hasConceptScore W2916210538C71924100 @default.
• W2916210538 hasIssue "2" @default.
• W2916210538 hasLocation W29162105381 @default.
• W2916210538 hasLocation W29162105382 @default.
• W2916210538 hasOpenAccess W2916210538 @default.
• W2916210538 hasPrimaryLocation W29162105381 @default.
• W2916210538 hasRelatedWork W2072674872 @default.
• W2916210538 hasRelatedWork W2748952813 @default.
• W2916210538 hasRelatedWork W2899084033 @default.
• W2916210538 hasRelatedWork W2908834051 @default.
• W2916210538 hasRelatedWork W2972563395 @default.
• W2916210538 hasRelatedWork W3003914079 @default.
• W2916210538 hasRelatedWork W4200578757 @default.
• W2916210538 hasRelatedWork W4223634269 @default.
• W2916210538 hasRelatedWork W4307645252 @default.
• W2916210538 hasRelatedWork W4313463021 @default.
• W2916210538 hasVolume "25" @default.
• W2916210538 isParatext "false" @default.
• W2916210538 isRetracted "false" @default.
• W2916210538 magId "2916210538" @default.
• W2916210538 workType "article" @default.