Wolfgang Gaissmaier

Helping Doctors and Patients Make Sense of Health Statistics

Many doctors, patients, journalists, and politicians alike do not understand what health statistics mean or draw wrong conclusions without noticing. Collective statistical illiteracy refers to the widespread inability to understand the meaning of numbers. For instance, many citizens are unaware that higher survival rates with cancer screening do not imply longer life, or that the statement that mammography screening reduces the risk of dying from breast cancer by 25% in fact means that 1 less woman out of 1,000 will die of the disease. We provide evidence that statistical illiteracy (a) is common to patients, journalists, and physicians; (b) is created by nontransparent framing of information that is sometimes an unintentional result of lack of understanding but can also be a result of intentional efforts to manipulate or persuade people; and (c) can have serious consequences for health. The causes of statistical illiteracy should not be attributed to cognitive biases alone, but to the emotional nature of the doctor–patient relationship and conflicts of interest in the healthcare system. The classic doctor–patient relation is based on (the physicians) paternalism and (the patients) trust in authority, which make statistical literacy seem unnecessary; so does the traditional combination of determinism (physicians who seek causes, not chances) and the illusion of certainty (patients who seek certainty when there is none). We show that information pamphlets, Web sites, leaflets distributed to doctors by the pharmaceutical industry, and even medical journals often report evidence in nontransparent forms that suggest big benefits of featured interventions and small harms. Without understanding the numbers involved, the public is susceptible to political and commercial manipulation of their anxieties and hopes, which undermines the goals of informed consent and shared decision making. What can be done? We discuss the importance of teaching statistical thinking and transparent representations in primary and secondary education as well as in medical school. Yet this requires familiarizing children early on with the concept of probability and teaching statistical literacy as the art of solving real-world problems rather than applying formulas to toy problems about coins and dice. A major precondition for statistical literacy is transparent risk communication. We recommend using frequency statements instead of single-event probabilities, absolute risks instead of relative risks, mortality rates instead of survival rates, and natural frequencies instead of conditional probabilities. Psychological research on transparent visual and numerical forms of risk communication, as well as training of physicians in their use, is called for. Statistical literacy is a necessary precondition for an educated citizenship in a technological democracy. Understanding risks and asking critical questions can also shape the emotional climate in a society so that hopes and anxieties are no longer as easily manipulated from outside and citizens can develop a better-informed and more relaxed attitude toward their health.

Presenting quantitative information about decision outcomes: a risk communication primer for patient decision aid developers

BackgroundMaking evidence-based decisions often requires comparison of two or more options. Research-based evidence may exist which quantifies how likely the outcomes are for each option. Understanding these numeric estimates improves patients’ risk perception and leads to better informed decision making. This paper summarises current “best practices” in communication of evidence-based numeric outcomes for developers of patient decision aids (PtDAs) and other health communication tools.MethodAn expert consensus group of fourteen researchers from North America, Europe, and Australasia identified eleven main issues in risk communication. Two experts for each issue wrote a “state of the art” summary of best evidence, drawing on the PtDA, health, psychological, and broader scientific literature. In addition, commonly used terms were defined and a set of guiding principles and key messages derived from the results.ResultsThe eleven key components of risk communication were: 1) Presenting the chance an event will occur; 2) Presenting changes in numeric outcomes; 3) Outcome estimates for test and screening decisions; 4) Numeric estimates in context and with evaluative labels; 5) Conveying uncertainty; 6) Visual formats; 7) Tailoring estimates; 8) Formats for understanding outcomes over time; 9) Narrative methods for conveying the chance of an event; 10) Important skills for understanding numerical estimates; and 11) Interactive web-based formats. Guiding principles from the evidence summaries advise that risk communication formats should reflect the task required of the user, should always define a relevant reference class (i.e., denominator) over time, should aim to use a consistent format throughout documents, should avoid “1 in x” formats and variable denominators, consider the magnitude of numbers used and the possibility of format bias, and should take into account the numeracy and graph literacy of the audience.ConclusionA substantial and rapidly expanding evidence base exists for risk communication. Developers of tools to facilitate evidence-based decision making should apply these principles to improve the quality of risk communication in practice.

Opportunities and challenges of Web 2.0 for vaccination decisions

A growing number of people use the Internet to obtain health information, including information about vaccines. Websites that allow and promote interaction among users are an increasingly popular source of health information. Users of such so-called Web 2.0 applications (e.g. social media), while still in the minority, represent a growing proportion of online communicators, including vocal and active anti-vaccination groups as well as public health communicators. In this paper, the authors: define Web 2.0 and examine how it may influence vaccination decisions; discuss how anti-vaccination movements use Web 2.0 as well as the challenges Web 2.0 holds for public health communicators; describe the types of information used in these different settings; introduce the theoretical background that can be used to design effective vaccination communication in a Web 2.0 environment; make recommendations for practice and pose open questions for future research. The authors conclude that, as a result of the Internet and Web 2.0, private and public concerns surrounding vaccinations have the potential to virally spread across the globe in a quick, efficient and vivid manner. Web 2.0 may influence vaccination decisions by delivering information that alters the perceived personal risk of vaccine-preventable diseases or vaccination side-effects. It appears useful for public health officials to put effort into increasing the effectiveness of existing communication by implementing interactive, customized communication. A key step to providing successful public health communication is to identify those who are particularly vulnerable to finding and using unreliable and misleading information. Thus, it appears worthwhile that public health websites strive to be easy to find, easy to use, attractive in its presentation and readily provide the information, support and advice that the searcher is looking for. This holds especially when less knowledgeable individuals are in need of reliable information about vaccination risks and benefits.

Good judgments do not require complex cognition

What cognitive capabilities allow Homo sapiens to successfully bet on the stock market, to catch balls in baseball games, to accurately predict the outcomes of political elections, or to correctly decide whether a patient needs to be allocated to the coronary care unit? It is a widespread belief in psychology and beyond that complex judgment tasks require complex solutions. Countering this common intuition, in this article, we argue that in an uncertain world actually the opposite is true: Humans do not need complex cognitive strategies to make good inferences, estimations, and other judgments; rather, it is the very simplicity and robustness of our cognitive repertoire that makes Homo sapiens a capable decision maker.

Sequential processing of cues in Memory-based multiattribute decisions

When probabilistic inferences have to be made from cue values stored in long-term memory, many participants appear to use fast and frugal heuristics, such as “take the best” (TTB), that assume sequential search of cues. A simultaneous global matching process with cue weights that are appropriately chosen would mimic the decision outcomes, albeit assuming different cognitive processes. We present a reanalysis of response times (RTs) from five published experiments (n 5 415) and one new experiment (n 5 82) that support the assumption of sequential search. In all instances in which decision outcomes indicated the use of TTB’s decision rule, decision times increased monotonically with the number of cues that had to be searched in memory. Furthermore, RT patterns fitted the outcome-based strategy classifications, which further validates both measures.

Numbers Can Be Worth a Thousand Pictures: Individual Differences in Understanding Graphical and Numerical Representations of Health-Related Information

OBJECTIVE Informed medical decision making requires comprehending statistical information. We aimed to improve the understanding of conveying health-related statistical information with graphical representations compared with numerical representations. First, we investigated whether the iconicity of representations (i.e., their abstractness vs. concreteness) affected comprehension and recall of statistical information. Second, we investigated whether graph literacy helps to identify individuals who comprehend graphical representations better than numerical representations. METHOD Participants (N = 275) were randomly assigned to receive different representations of health-related statistical information, ranging from very low iconicity (numbers) to very high iconicity (icon arrays including photographs). Comprehension and recall of the information were assessed. Additionally, participants rated the accessibility of the information and the attractiveness of the representation. Graph literacy was assessed by means of a recently developed scale. RESULTS The only difference between representations that affected comprehension and recall was the difference between graphics and numbers; the actual level of iconicity of graphics did not matter. Individuals with high graph literacy had better comprehension and recall when presented with graphics instead of numbers, and they rated graphical information as more accessible than numerical information, whereas the reverse was true for individuals with low graph literacy, F(4, 185) = 2.60, p = .04, η(p)(²) = .05, and F(4, 245) = 2.71, p = .03, η(p)(2) = .04, respectively. Both groups judged graphical representations as more attractive than numerical representations. CONCLUSION An assessment of graph literacy distinguished individuals who are best informed with graphical representations of statistical information from those who are better informed with numerical representations.

From recognition to decisions: Extending and testing recognition-based models for multialternative inference

The recognition heuristic is a noncompensatory strategy for inferring which of two alternatives, one recognized and the other not, scores higher on a criterion. According to it, such inferences are based solely on recognition. We generalize this heuristic to tasks with multiple alternatives, proposing a model of how people identify the consideration sets from which they make their final decisions. In doing so, we address concerns about the heuristic’s adequacy as a model of behavior: Past experiments have led several authors to conclude that there is no evidence for a noncompensatory use of recognition but clear evidence that recognition is integrated with other information. Surprisingly, however, in no study was this competing hypothesis—the compensatory integration of recognition—formally specified as a computational model. In four studies, we specify five competing models, conducting eight model comparisons. In these model comparisons, the recognition heuristic emerges as the best predictor of people’s inferences.

The etiology of diagnostic errors: a controlled trial of system 1 versus system 2 reasoning.

Purpose Diagnostic errors are thought to arise from cognitive biases associated with System 1 reasoning, which is rapid and unconscious. The primary hypothesis of this study was that the instruction to be slow and thorough will have no advantage in diagnostic accuracy over the instruction to proceed rapidly. Method Participants were second-year residents who volunteered after they had taken the Medical Council of Canada (MCC) Qualifying Examination Part II. Participants were tested at three Canadian medical schools (McMaster, Ottawa, and McGill) in 2010 (n = 96) and 2011 (n = 108). The intervention consisted of 20 computer-based internal medicine cases, with instructions either (1) to be as quick as possible but not make mistakes (the Speed cohort, 2010), or (2) to be careful, thorough, and reflective (the Reflect cohort, 2011). The authors examined accuracy scores on the 20 cases, time taken to diagnose cases, and MCC examination performance. Results Overall accuracy in the Speed condition was 44.5%, and in the Reflect condition was 45.0%; this was not significant. The Speed cohort took an average of 69 seconds per case versus 89 seconds for the Reflect cohort (P < .001). In both cohorts, cases diagnosed incorrectly took an average of 17 seconds longer than cases diagnosed correctly. Diagnostic accuracy was moderately correlated with performance on both written and problem-solving components of the MCC licensure examination and inversely correlated with time. Conclusions The study demonstrates that simply encouraging slowing down and increasing attention to analytical thinking is insufficient to increase diagnostic accuracy.

Simple predictions fueled by capacity limitations : When are they successful?

Counterintuitively, Y. Kareev, I. Lieberman, and M. Lev (1997) found that a lower short-term memory capacity benefits performance on a correlation detection task. They assumed that people with low short-term memory capacity (low spans) perceived the correlations as more extreme because they relied on smaller samples, which are known to exaggerate correlations. The authors consider, as an alternative hypothesis, that low spans do not perceive exaggerated correlations but make simpler predictions. Modeling both hypotheses in ACT-R demonstrates that simpler predictions impair performance if the environment changes, whereas a more exaggerated perception of correlation is advantageous to detect a change. Congruent with differences in the way participants make predictions, 2 experiments revealed a low capacity advantage before the environment changes but a high capacity advantage afterward, although this pattern of results surprisingly only existed for men.

Risk perception in natalizumab-treated multiple sclerosis patients and their neurologists

Background: Natalizumab is associated with the potentially life-threatening side-effect progressive multifocal leukoencephalopathy (PML). Little is known about patients’ and physicians’ risk estimates and attitudes towards natalizumab treatment. Methods: Consecutive natalizumab-treated patients (n = 69) and neurologists (n = 66) in two centres and cooperating private practices received an evidence-based three-page information leaflet about natalizumab-associated PML and an evaluation sheet. Results: After reading the information, patients were significantly more likely than physicians to intend continuation of natalizumab treatment and willing to accept higher risks of PML: 49% of physicians would stop treatment at a PML risk of 2 : 10,000 or lower, while only 17% of patients would do so (p < 0.001). This difference could not be explained by risk calculation abilities or lack of understanding. Both groups overestimated natalizumab treatment effects. Conclusion: Patients had a significantly worse perception of multiple sclerosis as a malignant disease. We conclude that patients were willing to accept a higher risk of PML than neurologists. Coherent with their perception of risks and benefits, patients were also more willing to continue treatment. Open information about treatment-related risks is appreciated and might support shared decision making.