Decision-Making Lecture 5 Applications In Healthcare PDF

Decision-Making Lecture 5 APPLICATIONS IN HEALTHCARE  Lecture outline Focus questions Lecture sections How might confirmation bias (defined by Confirmation bias and misdiagnosis research on heuristics and biases) contri...

Decision-Making Lecture 5 APPLICATIONS IN HEALTHCARE  Lecture outline Focus questions Lecture sections How might confirmation bias (defined by Confirmation bias and misdiagnosis research on heuristics and biases) contribute to misdiagnosis, and how can confirmation bias be reduced? How might Prospect Theory help to motivate Prospect Theory and health promotion health-related behaviour? How has the fast-and-frugal heuristics Presentation of health-related information the model shaped the way in which health- fast-and-frugal way related information is presented to practitioners and the public? How has the Bayesian approach improved A Bayesian model of impulsive decision-making our understanding of impulsive decision- making in people with addictions? The University of Adelaide Slide 2  Readings for this lecture The University of Adelaide Slide 3 Confirmation bias and misdiagnosis  The confirmation bias The confirmation bias is the tendency to test your hunches (or hypotheses) by asking questions that will yield an affirmative response if the hypothesis is true. The bias arises from the application of the positive test heuristic: “When testing a hypothesis, ask questions that will yield a ‘yes’ response if the hypothesis under consideration is true”. The University of Adelaide Slide 5  Confirmation bias in a medical setting In medical settings, confirmation bias occurs when: A health professional attends to symptoms that are present without considering symptoms that are absent or A health professional ignores evidence that disagrees with a leading diagnosis These patterns of decision-making can result in misdiagnosis, overtreatment, and unnecessary use of diagnostic tests. The University of Adelaide Slide 6  Considering present rather than absent symptoms In a field setting (a US medical clinic), Christensen-Szalanski and Bushyhead (1983) examined the extent to which physicians’ predictions about the probability that a patient has pneumonia were sensitive to both the presence and absence of symptoms. Nine physicians considered symptoms from 1,531 patients. Consistent with a confirmation bias, physicians’ predictions reflected the patient’s status on 12 out of 14 present symptoms but only 1 out of 5 absent symptoms. This occurred even though status on all 19 symptoms was predictive of patients’ actual pneumonia status based on an X-Ray. All 19 cues were predictive in that a patient presenting with a relevant symptom (say, an acute cough) was more likely to have pneumonia according to an X-Ray if a number of other symptoms were absent (as they would not be if the patient had the flu rather than pneumonia). Slide 7 https://www.eurekalert.org/news-releases/647789 Ignoring evidence that disagrees with a leading diagnosis Patient presented with severe headache, followed by a collapse and loss of consciousness for a few minutes. CT scan was conducted immediately, and was the basis for a preliminary diagnosis of a brain hemorrhage. The patient was immediately transferred to another hospital, where she could begin invasive treatment and further testing. Mohmed et al. (2021)  Ignoring evidence that disagrees with a leading diagnosis At the second hospital, it was noticed that the patient had undergone a lumbar puncture to test cerebrospinal fluid even before being admitted by Mohmed’s team at the first hospital. The results of this test almost fully ruled out a hemorrhage, but had been ignored. An MRI at the second hospital confirmed that the patient had a benign tumour that had the appearance of a hemorrhage in the CT scan. That MRI could have been conducted at the first hospital to save the patient stress and discomfort. Image source Mohmed et al. (2021)  What can be done? Systems promoting awareness of the bias, tool adoption, and a team-based approach Tools supporting reflection Individuals embracing processes and tools individuals Tools Systems/ processes/ policies Slide 11  What can be done? Systems promoting awareness of the bias, tool adoption, and a team-based approach “A cerebral aneurysm or vascular anomaly continued to be sought, despite a negative lumbar puncture result. In retrospect, the negative LP result was either overlooked, misattributed or forgotten. An unbiased “fresh pair of eyes” was instrumental in revising individuals the radiological diagnosis.” Mohmed et al. (2021) p. 1 Tools The ‘fresh pair of eyes’ was part of a routine re- Systems/ processes/ policies review process at the second hospital. Slide 12  What can be done? Tools supporting reflection Individuals and/or teams could be provided with a case summary and a specially designed form, in which there is space to describe: 1. clinical findings in the case that favour the initial diagnosis 2. clinical findings in the case that contradict the initial diagnosis individuals 3. clinical findings absent from the case but that would be expected to be present under the initial diagnosis Tools 4. two alternative diagnoses and points 1-3 relating to them Systems/ processes/ policies Chaves et al. (2022) Slide 13  What can be done? Tools supporting reflection Given that health professionals Individuals and/or teams could be provided with a case have been found to attend to summary and a specially designed form, in which there is symptoms that are present space to describe: 1. clinical findings in the case that favour the initial without considering symptoms diagnosis that are absent, what additional 2. clinical findings in the case that contradict the initial field might it be useful to add to diagnosis this form? 3. clinical findings absent from the case but that would be expected to be present under the initial diagnosis 4. _________ 5. two alternative diagnoses and points 1-4 relating to them Chaves et al. (2022)  What can be done? Tools supporting reflection Given that health professionals Individuals and/or teams could be provided with a case have been found to attend to summary and a specially designed form, in which there is symptoms that are present space to describe: 1. clinical findings in the case that favour the initial without considering symptoms diagnosis that are absent, what additional 2. clinical findings in the case that contradict the initial field might it be useful to add to diagnosis this form? 3. clinical findings absent from the case but that would be expected to be present under the initial diagnosis 4. clinical findings absent from the case that you would expect to be absent under the initial diagnosis 5. two alternative diagnoses and points 1-4 relating to them Chaves et al. (2022) Team + tool was also recommended for intelligence analysts (see Lecture 2) Slide 16 Prospect theory and health promotion We have seen an example already. Which phenomenon proposed by Prospect Theory causes this ad to be effective? Image source: https://order.hpa.org.nz/products/what-s- your-look-don-t-make-it-dumbburn-poster- female  Another postulated phenomenon: The framing effect In addition to whatever you own, In addition to whatever you own, you have been given $1,000. you have been given $2,000. Which would you prefer next? Which would you prefer? A. A 50% chance of gaining A. A 50% chance of losing $1,000 $1,000 B. A sure loss of $500 B. A sure gain of $500 People are risk-averse in the People are risk-seeking in the gain frame – i.e., they choose loss frame – i.e., they choose the more certain outcome. the more uncertain outcome.  The framing effect Reverses for small probabilities Which would you prefer? Which would you prefer? A. A 1% chance of gaining $5,000 A. A 1% chance of losing $5,000 B. A sure gain of $5 B. A sure loss of $5 Risk seeking with a gain frame. Risk averse with a loss frame.  Framing of health messages In health messaging, uncertainty has traditionally been considered roughly equivalent to outcome severity – the extent to which a health behavior could lead to undesirable outcomes. Preventative behaviours that preserve good health through potentially preventing disease – behaviours such as applying sunscreen and exercising – are considered unlikely to lead to undesirable/severe consequences, and are thus considered low risk (~ low in uncertainty). Detection behaviours that can confirm potential disease – behaviours such as taking a mammogram and testing for diabetes – are considered more likely to lead to undesirable consequences (i.e., finding out one has a disease); they are considered high risk (~ high in uncertainty). By this (questionable) logic, it follows that preventative (low severity/low uncertainty) behaviours should be more appealing in a gain frame – that is, in messages emphasising attaining something desirable or avoiding something undesirable. On the other hand, detection behaviours (high severity/high uncertainty) should be more appealing in a loss frame.  Framing of health messages This week’s reading on Prospect Theory tests this logic, which is known to some health promoters. In the video below, which promotes a detection behaviour, the loss frame is presented in terms of: a small probability of losing life to by bowel cancer a sure probability of losing a sense of comfort during testing or associated conversations. The bowel cancer is the more uncertain and severe of the two losses, and it is expected that the viewer will choose this option, given the reverse framing effectfor small probabilities and the assumption of equality between uncertainty and outcome severity. This is just my application of Prospect Theory to this advertisement. I’m not sure that the advertisement was designed based on Prospect Theory principles. https://www.youtube.com/watch?v=WG18pPbvSpE Does the logic of equality between outcome uncertainty and outcome severity work for you? Which of the following messages makes you want to exercise more? A. Pat and Chris didn't like to work out; that is, until B. Pat and Chris didn't like to work out; that is, until recently. The two friends were going to Florida for recently. The two friends were going to Florida for spring break. Pat invited friends to an aunt's home in spring break. Pat invited friends to an aunt's home in Daytona Beach. A few months before the trip, Pat and Daytona Beach. A few months before the trip, Pat and Chris decided to try to get in shape by working out Chris decided to try to get in shape by working out together regularly. They started running for at least 20 together regularly. They started running for at least 20 minutes, after their classes on Mondays, Wednesdays, minutes, after their classes on Mondays, Wednesdays, and Fridays. They were worried that they didn't have and Fridays. They were worried that they didn't have time to work out and wouldn't stay motivated. Pat has time to work out and wouldn't stay motivated. Pat has a a full-time job in addition to taking 12 hours each full-time job in addition to taking 12 hours each semester, and Chris takes 15 hours and has an semester, and Chris takes 15 hours and has an internship. Pat saved time by studying on the internship. Pat could've saved time by studying on the treadmill, and Chris stayed motivated by thinking treadmill, and Chris could've stayed motivated by the about how being in shape would help during surfing prospect of being out of shape during surfing lessons in lessons in Florida. They started to look and feel better Florida, but both became overwhelmed with school and and stronger. They were also surprised that they felt a work and quit exercising. They started to look and feel sense of accomplishment in doing something good worse and get out of shape. And, they lost the sense of for their bodies. They felt great when they went to accomplishment they'd felt in doing something good for Daytona and had a wonderful time. Pat felt a new their bodies. Pat and Chris did not have such a great confidence and met some great friends, while Chris, time in Daytona. Pat lacked the confidence to meet new reenergized, learned to surf. They continued to friends, and Chris didn't have the energy to learn to exercise long after spring break. Chris is beginning to surf. They also couldn't get motivated to work out when think that the small amount of time they take to work they got back to school. Chris is beginning to wonder if out is benefiting them in a very big way. the small amount of time they're not taking to work out will hurt them in a very big way. A. Pat and Chris didn't like to work out; that is, until recently. The two friends were going to Florida for spring break. Pat invited friends to an aunt's home in Daytona Beach. A few months before the trip, Pat and Chris decided to try to get in shape by working out together regularly. They started running for at least 20 minutes, after their classes on Mondays, Wednesdays, For me, Passage A and Fridays. They were worried that they didn't have provides more time to work out and wouldn't stay motivated. Pat has a full-time job in addition to taking 12 hours each encouragement. semester, and Chris takes 15 hours and has an internship. Pat saved time by studying on the treadmill, and Chris stayed motivated by thinking about how being in shape would help during surfing lessons in Florida. They started to look and feel better and stronger. They were also surprised that they felt a sense of accomplishment in doing something good for their bodies. They felt great when they went to Daytona and had a wonderful time. Pat felt a new confidence and met some great friends, while Chris, reenergized, learned to surf. They continued to exercise long after spring break. Chris is beginning to think that the small amount of time they take to work out is benefiting them in a very big way.  Disentangling severity and uncertainty Confirming the logic, Gray and Harrington (2010) found that participants rated Message A to be more effective, and reported greater intentions to exercise after reading it – i.e., greater intentions of a preventative behavior. Message A is presented in a gain frame (in, which, arguably, people choose the more certain outcome, such as gaining improved health). In your second reading, Harrington and Kerr (2017) seek to disentangle the effects of outcome uncertainty and outcome severity. They manipulate outcome severity, and determine whether – regardless of severity – participants select the less uncertain outcome in a gain frame and the more uncertain outcome in a loss frame.  More severe Selected by majority of people Selected by minority of people Selected by approx. 50% of people The University of Adelaide 27  Less severe Selected by majority of people Selected by minority of people Selected by approx. 50% of people The University of Adelaide 28  Disentangling severity and uncertainty Critically, Harrington and Kerr (2017) found that outcome severity – the factor supposedly distinguishing preventative and detection-directed health behaviors – does not affect the direction of the framing effect. Overall, also, only half of the framing effect was observed – risk-aversion in the gain frame.  Conclusion Despite large amounts of research on appropriate framing for messages depending on the kind of health behaviour they are aimed at promoting, there is no scientific consensus on the best frame (gain vs. loss) for preventative and detection-oriented behaviours. In light of ongoing debates, Prospect Theory has not become an overarching theoretical framework for health advertising. Many detection-oriented behaviours are promoted in a gain frame. https://www.youtube.com/watch?v=7jCCjoVexpM Presentation of health-related information the fast-and-frugal way  Fast and frugal heuristics According to Gigerenzer, people are highly attuned to their environments, and can, thus, select cues and heuristics that are most appropriate in various environments. Heuristics identified by Gigerenzer and his colleagues are listed on this slide. The ability to select appropriate cues The Take-the-Best heuristic and heuristics is likely to be shaped The Recognition heuristic by a number of processes: evolutionary hard-wiring The Fluency heuristic individual learning (within that Tallying environment and similar ones) social processes – imitation and explicit teaching The University of Adelaide Slide 32  Evidence for the ‘fast-and-frugal’ model Gigerenzer provided evidence for his claim that people are highly attuned to their environments. Gigerenzer developed studies that showed 'reversals' of the biases demonstrated by Kahneman and Tversky when questions were asked in a way that was sensitive to the differences people might see across different environments: in a natural frequency format. The University of Adelaide Slide 33  Widespread impact on information presentation Research by Hoffrage and Gigerenzer (1998) in your third reading showed that the natural frequency format is more effective for communicating with doctors about the predictive value of diagnostic tests. Gigerenzer (2007) has since written a more comprehensive guide for teaching and presenting healthcare-related statistics, and we can see the influence of this work in how infographics relating to healthcare and many other topics are designed these days. The University of Adelaide Slide 34 Recall also that applying Take-the-Best, one of the heuristics defined by the fast-and-frugal approach, involves following a decision tree.  Widespread impact on information presentation (cont.) Green and Mehr (1997) developed as a decision aid for emergency room doctors at a Michigan hospital when allocating patients to a specialist heart—or ’coronary’—unit, as opposed to a regular bed. Application of the tree by doctors resulted in fewer patients at risk of heart attack being wrongly sent to a regular bed. Application of the tree also reduced load on the coronary unit. Prior to the tree’s introduction, doctors had been sending 90% of patients with any coronary symptoms directly to the coronary unit, preferring to err on the side of caution. The University of Adelaide Slide 37 A Bayesian model of impulsive decision-making Recap: Bayesian models enable complex processes to be ‘run through’ by the model to check whether the observed data (the decisions of participants) are well-approximated by assuming that different kinds of decision-makers adopt some reasonable complex strategies.  Bayesian modelling was used to clearly demonstrate that: the degree to which an individual is reluctant to accumulate evidence prior to making a decision, also known as reflection impulsivity, is heightened in people who binge drink, particularly in the sense that they have difficulty thinking ahead Banca et al. (2015) The University of Adelaide Slide 41 Participants: 30 people who binge drink 30 controls Banca et al. (2015) Banca et al. (2015) The findings provide an indirect evidence base for ads like this that support thinking ahead for people at risk of binge drinking. https://www.youtube.com/watch?v=uZtwRqRyEN0 Summary  Returning to the focus questions Focus questions In terms of an evidence base How might confirmation bias (defined by for health promotion strategies, research on heuristics and biases) the theories of decision-making contribute to misdiagnosis, and how can confirmation bias be reduced? in each focus question provide: a body of research (in terms How might Prospect Theory help to motivate of number of studies) health-related behaviour? indicated by the size of the How has the fast-and-frugal heuristics traffic light model shaped the way in which health- study findings in the related information is presented to direction predicted by the practitioners and the public? theory to a degree indicated How has the Bayesian approach improved by the colour of the traffic our understanding of impulsive decision- light (green = as predicted; making in people with addictions? yellow = partially as predicted) The University of Adelaide Slide 46 47  References Banca, P., Lange, I., Worbe, Y., Howell, N. A., Irvine, M., Harrison, N. A., Moutoussis, M., & Voon, V. (2016). Reflection impulsivity in binge drinking: behavioural and volumetric correlates. Addiction Biology, 21(2), 504–515. Chaves, A. B., Moura, A. S., de Faria, R. M. D., & Ribeiro, L. C. (2022). The use of deliberate reflection to reduce confirmation bias among orthopedic surgery residents. Scientia Medica, 32(1), e42216. Gigerenzer, G., Gaissmaier, W., Kurz-Milcke, E., Schwartz, L. M., & Woloshin, S. (2007). Helping Doctors and Patients Make Sense of Health Statistics. Psychological Science in the Public Interest, 8(2), 53–96. Gray, J. B., & Harrington, N. G. (2011). Narrative and framing: a test of an integrated message strategy in the exercise context. Journal of Health Communication, 16(3), 264–281. Mohmed, A. I. H., Kumaria, A., & White, B. (2021). Risk of confirmatory bias: Parafalcine meningioma mimicking acute subarachnoid haemorrhage. Interdisciplinary Neurosurgery, 25, 101155. The University of Adelaide Slide 48