| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH |
|
|
|
|||||||
|
|||||||||
TRACK to:
|
|
Electronic letters published:
![[Read Comment]](/icons/misc/sectionDOWN.gif){kind=icon}
A preference does not equate with understanding
We agree that very little is ever equal
Theraputic agnosticism, or, when all other things are not equal
Is relative risk really the way to go for numerical presenation?
|
|
|||
|
Odette Wegwarth, Berlin, Germany Max Planck Institute for Human Development, Gerd Gigerenzer
Send response to journal:
|
Goodyear-Smith et al.’s study sought to determine which method of expressing benefits of treatment (statins) a) would most encourage patients with known cardiovascular disease to decide to take preventive medicine, and b) is preferred in helping to understand the risk information. With respect to b), there are several issues we would like to challenge. We question the authors’ presumption that a patient’s preference for a given format reflects ease of understanding the respective information. Although the authors explicitly say that they sought to differentiate between the persuasiveness of the method in which data are presented and the method that patients found best in helping them understand their risks and benefits, they end up measuring persuasiveness by asking for preferences instead of investigating real understanding of the information delivered by each method. Our concern that preference does not equate with understanding particularly relates to the second most preferred method of study participants, the relative risk format. A patient may prefer, for instance, to participate in mammography screening after being given information about the reduction of breast cancer mortality by the method of relative risk. However, this does not mean that the patient actually understood the risk information. In a recent representative study of 1,000 German citizens, people were given information in the form of relative risks about the risk reduction of dying from breast cancer when participating in mammography screening.(1) Subsequently, they were asked to indicate how many fewer women out of 1,000 would die of breast cancer. Only about 1% of the participants arrived at the correct conclusion and hence understood what the information meant. The majority (> 75%) highly overestimated the risk reduction by a magnitude of 100 times and more. A similar overestimation of benefits has been reported for PSA screening and for citizens of several other European countries (2) as well as for physicians.(3) The format of relative risk information, known for its persuasiveness (making effects appear bigger) and its lack of absolute reference (no information about the respective base rates), is least likely to foster understanding and informed decision making. Given that the sample encompassed only patients who have already been taking an angina or heart attack preventing drug (statins) and that 53% of them were innumerate, respective results suggest that authors have measured affection to bigger numbers and the best justification of a kind of drug choice already made, not a real understanding of the risk information. Interest in fostering informed consent and shared decision making requires presenting the risks and benefits by methods that allow patients to truly understand the numbers. There is evidence that methods like absolute risk and natural frequencies do so, while relative risk does not.(4) Of course, treatment effects no longer appear as favourable in the former methods, where numbers are usually smaller, and patients may thus opt for a particular therapy less often. For instance, individuals who received information about a treatment only in form of relative risk reduction compared to absolute risk or NNT were significantly more likely to endorse therapy, least confident in their decision, and highly likely to change their decision when presented with measures of absolute risk and NNT afterwards. (5) However, in this respect, we would like to make the same point as Smith(6) did when asking why it is necessarily bad when some patients decline potentially helpful interventions once they fully understand the risks and benefits. After all, health decisions are personal decisions. Thus, the utmost goal of health authorities and researchers in this field should instead be to provide patients with objective information than with persuasions justified by the idea that medical experts are better at knowing what is good or bad for the individual patient. Finally, it is a shortcoming of the article that neither the actual measurement nor the final results of the outcome “perceived risk of a heart attack” are reported and thus made traceable for the reader. 1 Gigerenzer, G., Gaissmaier, W., Kurz-Milcke, E., Schwartz, L. M., & Woloshin, S. (in press). Helping doctors and patients make sense of health statistics. Psychological Science in the Public Interest. 2 Gigerenzer, G., Mata, J., & Frank, R.: A survey of health knowledge in nine European countries. Unpublished manuscript 2008. 3 Schüssler, B.: Im Dialog: Ist Risiko überhaupt kommunizierbar, Herr Prof. Gigerenzer? [Interview with Gerd Gigerenzer: Can risk be communicated at all?]. Frauenheilkunde Aktuell 2005; 14, 25–31. 4 Gigerenzer, G.: Reckoning with risk: Learning to live with uncertainty. London: Penguin Books 2002. 5 Cao, C, Studts, J.L., Abell, T et al.: Adjuvant chemotherapy for breast cancer: How representation risk influences decision making. J Clin Oncol 2003; 21: 4299-4305. 6 Smith, P.C.: Therapeutic agnosticism, or, when all other things are not equal. Online Comment on „Patients prefer pictures to numbers to express cardiovascular benefit from treatment“. Ann Fam Med 2008; 6: 213-217. Competing interests: None declared |
|||
|
|
|||
|
Felicity A Goodyear-Smith, Auckland, New Zealand Associate professor, University of Auckland
Send response to journal:
|
We agree that while patients might prefer RR, that information presented in this manner may be misleading and that AR is preferable. In other words, while patients might prefer it, we do not see RR as an acceptable means of expressing these types of information. We are currently conducting a much larger study looking at how patients with a range of CVD risk (from 5% to 30%) rank various ways of numerically and pictorially expressing risk, both as to which would most encourage them to action and which is the most helpful to them to understand their risk and the possible benefits of making changes. This study also addresses decision-making. We are asking patients both what format of information most encourages them to take a medication, and what format helps them best understand. These are often quite different. It is not best practice to merely present patients with information that might overly encourage them to embark on life-long medication (which unlike our hypothetical drug, will have harms as well as risks). Our study is attempting to understand what formats might be unduly coercive to patients. We agree that not all patients will prefer or understand information presented in a certain way and ideally physicians should have available a range of options to explain risks and benefits. We also agree that patients may decline to take potentially helpful medications or to make lifestyle changes that might reduce their risk of illness. This is their choice, not ours. Our role is to provide them with information in a form that they can understand, that makes sense to them, in order for them to make these decisions. Competing interests: None declared |
|||
|
|
|||
|
Peter C. Smith, Denver, CO USA Associate Professor of Family Medicine, University of Colorado Denver School of Medicine
Send response to journal:
|
Goodyear-Smith and colleagues offered 100 patients with cardiovascular disease a theoretical choice: whether or not to take a medication for secondary prevention, depending on how the benefits are presented.1 Patients overwhelmingly chose the medication regardless of how the information was portrayed. One exception was the use of NNTs, which caused patients to feel less certain and less “encouraged” to take it. I agree with Dr. Krones that selection bias may be at play here; some of the study’s language hints at a pro-treatment bias on the part of the authors. Rather than ask about patients’ willingness “to take daily medication for the rest of their lives to protect their hearts,” would more treatment-neutral phrasing, such as “to potentially reduce the risk of heart attack”, have altered the results? The authors’ choice of the 4S Study for the theoretical medication requires some scrutiny. Some data suggest that 4S was skewed towards higher risk patients, given the more conservative results from later studies such as CARE2 and LIPID3, whose 5 year NNTs for major coronary events were at least twice as high. The use of the optimistic 4S results in the absence of risk and cost data presented the patients in this study with a proverbial “no brainer”– take a pill and reduce your risk without potential hazard. Were the less sanguine patients (the 11-26% who would not be encouraged to take it) the same individuals as the 19% who were unwilling “to take daily medication for the rest of their lives to protect their hearts”? Prescriptions are free or highly subsidized in New Zealand. How would these results have differed among our 47 million uninsured, especially if presented with only half the 4S effect size, qualified by out-of-pocket expenses for a statin and liver function monitoring for 5 years, and the caveat that every 7 coronary events prevented would be balanced by an adverse event?4 Most research tends to raise more questions than it answers. How can we study preferences given the uncertain nature of the enterprise and the scant evidence base for our discipline? How do we balance increasing emphasis on shared decision making with pressures to meet measurable quality metrics? Is it necessarily bad that some patients decline potentially helpful interventions once they fully understand the risks and benefits? Why do doctors tend to fall into one of three categories: therapeutic nihilists, positivists, and agnostics? Do our patients fall into similar categories? Why or why not? Finally, how can we assess the true impact of interventions in the messy complexity of real life? After all, we do not practice in a frictionless universe, and all other things are never equal. 1. Goodyear-Smith F. Arroll B, Chan L, Jackson R, Wells S, Kenealy T. Patients Prefer Pictures to Numbers to Express Cardiovascular Benefit From Treatment. Annals of Family Medicine 2008; 6:213-217. 2. Plehn JF, Davis BR, Sacks FM, Rouleau JL, Pfeffer MA, Bernstein V, et al. Reduction of stroke incidence after myocardial infarction with pravastatin: the cholesterol and recurrent events (CARE) study. Circulation 1999;99:216-23. 3. The long-term intervention with pravastatin in ischaemic disease (LIPID) study group. Prevention of cardiovascular events and death with pravastatin in patients with coronary heart disease and a broad range of initial cholesterol levels. N Engl J Med 1998;339:1349-57. 4. Silva MA, Swanson AC, Gandhi PJ, Tataronis GR. Statin-related adverse events: a meta-analysis. Clinical Therapeutics 2006;28:26-35. Competing interests: None declared |
|||
|
|
|||
|
Norbert Donner-Banzhoff, Marburg, Germany Academic Family Physician, Dept. of Family Medicine, University of Marburg
Send response to journal:
|
I agree with the authors’ conclusion that patients prefer graphical presentations, and these should be available as part of decision aids if ever possible. However, I would like to challenge their conclusion with regard to relative risk (RR) as the preferred numerical format. Remember who was included: patients who had a history of heart disease, had been taking statins, and were willing to take part. I think this was a sample very much biased towards people who were convinced that treatment (statin, or the hypothetical drug that was almost identical to statins) was good for them. It is not surprising, that these patients preferred RR, since the bias that comes with this kind of presentation, i.e. effects look bigger, corresponds well with their attitudes towards their treatment. Moreover, the wording related to RR is much easier to understand than with the other choices they had. I think that even these patients would think differently about the issue if they had learnt about possible distortions by different numerical formats. The RR-format certainly looks simpler than the other presentations. However, this does not mean that it is closer to the truth. I still think that absolute risk is to be preferred, see our paper in this issue (Krones et al.). Kind regards Norbert Donner-Banzhoff Competing interests: I am one of the authors of the ARRIBA-decision-aid (see article by Krones et al. in this issue) which emphasizes absolute risk reduction |
|||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH |
TRACK: