different measures: relative risk (RR) or absolute (some- times called ..... of absolute risk reduction should give patients a greater ability to be informed and to ...
The Framing Effect of Relative and Absolute Risk DAVID J. MALENKA, MD, JOHN A. BARON, MD, MS, MSc, SARAH JOHANSEN, MD, JON W. WAHRENBERGER, MD, JONATHAN M. ROSS, MD Objective: To test w h e t h e r a p a t i e n t ' s p e r c e p t i o n o f benefit is influenced by w h e t h e r t h e benefit is p r e s e n t e d in relative or absolute terms. Design: Questionnaire-based study. Setting: A general m e d i c i n e o u t p a t i e n t clinic at a rural tertiary care center associated w i t h a medical school. Patients: 4 7 0 of 511 c o n s e c u t i v e patients w h o agreed to a n s w e r a q u e s t i o n n a i r e w h i l e waiting for their clinic visit. Mean age was 49.1 years, 62.1% were female, a n d 51.9% had at least o n e year o f education b e y o n d h i g h school. Main o u t c o m e measures: Patient response to t h e c h o i c e of t w o e q u a l l y efficacious m e d i c a t i o n s for t h e m a n a g e m e n t of a hypothetical serious disease. T h e benefit of o n e m e d i c a t i o n w a s stated in relative terms, t h e o t h e r in absolute terms. Patients c o u l d c h o o s e either medication alone, indicate indifference to t h e c h o i c e of medication, or choose n o t to answer. Main results: 56.8% of t h e patients c h o s e the m e d i c a t i o n w h o s e benefit was in relative terms. 14.7% c h o s e t h e m e d i c a t i o n w h o s e benefit was in absolute terms. O n l y 15.5% were indifferent to the c h o i c e o f medication. The patients preferred the m e d i c a t i o n w h o s e benefit was in relative t e r m s across a w i d e range of ages a n d educational levels. Further q u e s t i o n i n g s u g g e s t e d that t h e patients t h o u g h t benefit was greater w h e n e x p r e s s e d in relative terms b e c a u s e t h e y ignored t h e u n d e r l y i n g risk o f disease a n d a s s u m e d it was one. Conclusions: The " f r a m i n g " o f benefit (or risk) in relative versus absolute t e r m s m a y have a major influence o n patient preference. Key words: framing; risk; patient preferences; benefit; decision making.
d GEN INTERN MED 1993;8:543-548.
IT IS WELLKNOWN that c o m m o n expressions of probability have different meanings to different p e o p l e , t and that e v e n quantitative expressions can be i n t e r p r e t e d differently. 2 This difficulty in c o m m u n i c a t i n g is comp o u n d e d b y " f r a m i n g , " the influence on decision making of the descriptions of acts, contingencies, and outcomes. 3 For e x a m p l e , McNeil et al. f o u n d that patients, students, and physicians w e r e m o r e likely to choose surgery o v e r radiation t h e r a p y for treating patients w h o have lung cancer w h e n the o u t c o m e o f treatment was framed as the p r o b a b i l i t y of surviving rather than as the p r o b a b i l i t y o f dying. 4 In m e d i c a l decision making, c o n c e r n s w i t h framing have b e e n largely confined to the d e s c r i p t i o n of outcomes. 4-7 O n e aspect o f framing that has not b e e n w e l l investigated is the i m p a c t of different measures of risk.
Received from t h e D e p a r t m e n t s of Medicine (DJM, JAB, SJ, JWW, JMR) and C o m m u n i t y and Family Medicine (DJM, JAB) a n d t h e Center for the Evaluative Clinical Sciences (DJM, JAB), D a r t m o u t h Medical School, Hanover, N e w Hampshire. Address c o r r e s p o n d e n c e reprint r e q u e s t s to Dr. Malenka: Section of Cardiology, D a r t m o u t h - H i t c h c o c k Medical Center, Lebanon, NH 0 3 7 5 6 .
In clinical situations, risk m a y b e s u m m a r i z e d using two different measures: relative risk (RR) or absolute (sometimes called attributable) risk (AR). W h e n describing the effect of a risk factor, such as a high level of s e r u m cholesterol in relation to c o r o n a r y heart disease (CHD), RR is the ratio of the risk o f disease in o n e g r o u p (i.e., CHD a m o n g those w i t h high s e r u m cholesterol) to that in another (i.e., CHD in those w i t h l o w s e r u m cholesterol); AR is the difference b e t w e e n the risks o f disease in the two p o p u l a t i o n s and defines the absolute magnitude of the risk. We 8 and others 9, ~o have argued that in clinical decision making AR is a m u c h m o r e meaningful measure of risk. However, RR is p r e f e r r e d in etiologic research, is m o r e f r e q u e n t l y r e p o r t e d in m e d i c a l literature as w e l l as in the p o p u l a r press and is c o m m o n l y used in clinical encounters. If the underlying risk o f disease is known, RR can be c o n v e r t e d to AR. 8, 9, ~t T h e two measures are logically equivalent and, in this sense, presentation o f either should lead to the same decision (if information a b o u t the u n d e r l y i n g risk of disease is p r e s e n t e d ) . H o w e v e r , the literature on p e r c e p t i o n of risks suggests this m i g h t not b e true. t2, t3 W h e n t r e a t m e n t efficacy is e x p r e s s e d in relative terms, larger percentages result than w h e n the same t r e a t m e n t is discussed in absolute terms. For example, s u p p o s e a m e d i c a t i o n r e d u c e s the risk of an adverse o u t c o m e f r o m 0.05 to 0.025. In relative t e r m s it reduces the risk b y 50%, w h i l e in absolute terms it r e d u c e s the risk b y 2.5%. Thus, the presentation of RR m a y magnify the p e r c e p t i o n of efficacy. To evaluate this question w e c o n d u c t e d a questionnaire-based study to establish w h e t h e r patients p e r c e p tions of risk w e r e influenced b y the presentation of risk in relative versus absolute terms. We h y p o t h e s i z e d that patients w o u l d m i s i n t e r p r e t RR and not a c c o u n t for underlying risk o f disease w h e n c o m p a r i n g it w i t h a measure of AR.
METHODS Study Population Patients w e r e r e c r u i t e d f r o m the o u t p a t i e n t practice of an a c a d e m i c general internal m e d i c i n e g r o u p in rural N e w H a m p s h i r e . The g r o u p is part of a large multispecialty p r a c t i c e and at the t i m e o f this study consisted of 12 internists and four nurse practitioners or physician's assistants seeing a m i x of s c h e d u l e d and unschedu l e d (walk-in-clinic) patients. DuringAugust-. S e p t e m b e r 1989 o n e of the investigators spent eight half-days in the designated general 543
Malenka etal., FRAMINGEFFECTOF RISK
5,44
m e d i c i n e waiting r o o m recruiting subjects. All schedu l e d and u n s c h e d u l e d patients w h o registered w e r e asked to c o m p l e t e a three-page questionnaire. O n l y patients w h o c o u l d not read English, w e r e visually impaired, or had physical disability that limited their ability to read or write w e r e e x c l u d e d . A written i n f o r m e d consent was not required, t h o u g h a c o v e r letter app r o v e d b y the institutional r e v i e w b o a r d for the protection of h u m a n subjects was attached to each questionnaire stating the intent of the study, that participation was voluntary, and that no u n i q u e patient identifier w o u l d b e obtained.
Questionnaire The self-administered questionnaire consisted of eight items requesting d e m o g r a p h i c information and t w o hypothetical situations, e a c h w i t h an a c c o m p a n y ing m u l t i p l e - c h o i c e question. The d e m o g r a p h i c inform a t i o n i n c l u d e d data describing gender, age, level of education, a measure of self-reported overall health, and w h e t h e r subjects w e r e u n d e r medical t r e a t m e n t for h y p e r t e n s i o n a n d / o r h y p e r c h o l e s t e r o l e m i a . We collected the latter information b e c a u s e w e h y p o t h e s i z e d that patients b e i n g treated for a cardiac risk factor may have had an explicit discussion a b o u t risks and benefits and might b e m o r e attuned to different measures of risk than w o u l d patients not b e i n g so treated. In the hypothetical situations, the patients w e r e asked to imagine they had a disease and to answer a question a b o u t their p r e f e r e n c e s for treatment. The disease was described as serious, w i t h one-year mortality that was clearly specified numerically. O n e year was chosen as the t i m e frame to m i n i m i z e the effect of patients' attitudes toward length of l i f e ) 4 The first situation (Appendix A) was designed to test the effect o f risk presentation. The patients w e r e asked to make a c h o i c e b e t w e e n taking one of two medi-
TABLE 2 Percentageof Respondents(n = 470) ChoosingEach PossibleAnswer for Situation One
Answer
Percentage
MedicationA (relative benefit)
56.8
MedicationB (absolute benefit)
14.7
Medication A or B
15.5
Can't decide
13.0
Z2 = 254.3 with 3 df.
cations for the treatment of their life-threatening disease. "Medication A" and "Medication B" w e r e described as costing the same and having almost no side effect. The efficacies o f the drugs w e r e actually equivalent. However, the benefit of Medication A was presented in relative terms w h i l e the benefit of Medication B was presented in absolute terms. We asked the patients to choose to take Medication A, Medication B, either Medication A or B, or not to answer. The second situation (Appendix A) was designed to test patients' understanding of the relationship b e t w e e n RR and AR. The patients w e r e told there was o n l y one medication to treat a disease. T h e y w e r e told the benefit of this medication in relative terms and then asked to indicate this benefit in absolute terms. The one-year mortality associated with the disease was again clearly specified. T h e y w e r e given a c h o i c e of five possible answers, including one that was the p r o d u c t of the RR and the p o p u l a t i o n size (the " e x p e c t e d m i s t a k e " ) , one that was the p r o d u c t o f the RR and the baseline risk (the c o r r e c t answer), and one indication they c o u l d not decide. Four versions of the questionnaire w e r e administered to provide in the first situation all combinations of an underlying risk o f death (10 or 80%) and a treatment benefit (10 or 80%). The questionnaires w e r e ordered to ensure that equal n u m b e r s of patients w o u l d receive e a c h version t h r o u g h o u t the p e r i o d of study.
TABLE 1 Patient Characteristics(n = 470) Age--median (years)
49.1 (range 15-89)
Gender (%) Female Male
62.1 37.9
Education (%) -< 12 grade - 13 grade
48.1 51.9
Medicalhistory (%) Hypertension Medicallytreated Hypercholesterolemia Medicallytreated
22.8 14,2 20.5 2.6
Overall health (%) Poor- fair Good- very good Excellent
17.7 65.1 17.2
Data Entry and Analysis All data w e r e entered using SAS/FSP with error trapping. Standard chi-squared contingency-table m e t h o d s w e r e used to test for associations b e t w e e n categorical variables. Odds ratios (ORs) w i t h 95% confidence intervals (95% CIs) w e r e calculated as described b y H o s m e r and L e m e s h o w ~s using p a r a m e t e r estimates and standard errors f r o m the a p p r o p r i a t e logistic regression analysis (SAS Proc Logistt6).
RESULTS During the study p e r i o d 511 patients registered to b e seen as outpatients in the Section of General Medicine; 96.3% ( 4 9 2 ) r e t u r n e d a questionnaire. O f these, 4.5% ( 2 2 ) p r o v i d e d no answer for the first hypothetical situation. These patients w e r e m o r e likely to be over the
JOURNALOFGENERALINTERNALMEDICINE.Volume 8 (October), 1993
age of 70 years and less likely to have any high school education than w e r e those w h o provided an answer. Because we w e r e primarily interested in patients' choices in this first hypothetical situation, those patients not choosing an answer were eliminated from further analysis. The characteristics of the remaining 470 patients w h o constituted our study p o p u l a t i o n are shown in Table 1. There was no difference in patient characteristics among those w h o r e s p o n d e d to the different versions of the form (data not shown). When asked in the first hypothetical situation to choose a medication to treat their disease, 56.8% of the subjects chose the medication with benefit presented in relative terms, 14.7% chose the medication with benefit presented in absolute terms, 15.5% w e r e willing to be treated with either medication, and 13.0% c o u l d not d e c i d e (Table 2). This result was not e x p e c t e d by chance alone (Z 2 = 254.3 with 3 df). Subjects w e r e most likely to choose the medication whose benefit was presented in relative terms regardless of age, gender, level of education, self-reported health, e x p e r i e n c e with medical treatment for hypertension or high serum cholesterol (Table 3), or questionnaire version (data not shown). In a logistic regression m o d e l that included all patient characteristics as i n d e p e n d e n t variables, subjects with at least some college e d u c a t i o n and those being medically treated for hypertension or hypercholestrolemia were significantly more likely to select the medication with benefit presented in relative terms than
54S
w e r e those with less education or not taking medications. There was no statistical evidence o f interaction b e t w e e n these variables in predicting w h o w o u l d choose the medication w i t h benefit expressed in relative terms. The subjects w e r e asked to specify in free-text w h y they chose their answer. Two h u n d r e d thirty-one (86.5%) patients w h o chose the medication with benefit in relative terms responded. Twenty-eight (12.1%) c o m p a r e d the benefit of Medication A (expressed as a relative benefit) with that of Medication B (expressed as an absolute benefit) w i t h o u t accounting for the underlying risk of dying. O t h e r respondents thought Medication A offered more benefit but w e r e not specific as to why. Sixty-five (89.0%) of the patients w h o were indifferent to the choice of medication responded. Thirtyfour (52.3%) clearly indicated that they k n e w the benefits of the medications w e r e equivalent. In the second hypothetical situation the subjects w e r e given the underlying risk o f dying and the benefit of medication in relative terms and asked to specify the benefit of this medication in absolute terms. Most subjects (47.7%, Table 4) apparently m u l t i p l i e d the relative benefit by the size of the p o p u l a t i o n to p r o d u c e the treatment effect that w o u l d be e x p e c t e d if the probability of dying from the disease w e r e 1.0 (the e x p e c t e d mistake); 28.2% correctly identified the e x p e c t e d benefit and apparently w e r e able to convert relative into absolute benefit. Some patient characteristics w e r e as-
TABLE 3 Percentage of Respondents(n = 470) Preferring the Medication Expressedas a Relative Benefit (Medication A) or the Medications Expressedas Relative or Absolute Benefits ( MedicationsA and B, respectively) and Odds Ratio for Selectingthe Medication Expressedas a Relative Benefit, by Patient Characteristic, for Situation One
B
Percentage Preferring Medication Expressed as: Relative Relative or Absolute Benefit (A) Benefit (A or B)
Odds Ratio* of Preferring Medication Expressed as Relative Benefit (A)
Age -- 39 years 4 0 - 59 years - - 6 0 years
171 152 147
49.1 63.2 59.2
22.2 12.5 10.9
1.00t 1.44 (0.90, 2.30)* 1.21 (0.73, 1.99)
Gender Female Male
287 175
54.7 61.1
17.1 12.6
1.00 1.20 (0.80, 1.78)
Education -< 12 grade >-- 13 grade
226 244
49.1 63.9
14.2 16.8
1.00 1.68 (1.12, 2.51)
Health -- Fair Good Excellent
82 302 80
51.2 57.6 60.0
12.2 15.6 20.0
1.00 1.27 (0.75, 2.15) 1.37 (0.69, 2.70)
Treated for high blood pressure and/or cholesterol No Yes
392 68
54.9 72.1
17.6 4.4
1.00 2.21 (1.19, 4.12)
*As calculatedfrom the coefficients of a logistic regression model that included all listed patient characteristics as independent variables. tReference category. *95% confidenceintervals.
546
Malenka et aL, FRAMINGEFFECTOF RISK TABLE 4
Percentage of Respondents(n = 470) Choosing Each PossibleAnswer for Situation Two Answer
Percentage
25 deaths can be prevented (correct answer) 50 deaths can be prevented (expected mistake) Other choices Don't know
28.2 47.7 5.3 18.8
X2 = 329.4 with 4 df.
sociated w i t h choosing the c o r r e c t answer. Men, patients w i t h at least one year of college education (--> 13 grade), and patients in " e x c e l l e n t " health w e r e all m o r e likely than their c o m p a r i s o n groups to successfully convert relative into absolute benefit (Table 5). Despite this, these subjects w e r e almost as likely as others to make the e x p e c t e d mistake. Interestingly, they w e r e less likely than those in their c o m p a r i s o n g r o u p s to indicate they c o u l d not answer the question. Patients w h o w e r e able to equate relative benefit w i t h absolute benefit in the second hypothetical situation w e r e m o r e likely to be indifferent to the choice of m e d i c a t i o n in the first situation (OR = 1.84, 95% CI 1 . 0 1 , 3 . 3 5 ) . However, like o t h e r subjects, m o r e than half o f t h e m (58.1%) still p r e f e r r e d the m e d i c a t i o n w i t h benefit e x p r e s s e d in relative t e r m s in the first hypothetical situation.
patients u n d e r g o i n g medical t r e a t m e n t for hypertension or a high serum cholesterol level. Thus, most patients, e v e n w e l l - e d u c a t e d ones, may b e influenced b y the framing of risk. O u r results may be a special case o f the " p s e u d o certainty" effect described b y Tversky and Kahneman ~2 in the framing of contingencies. This p h e n o m e n o n can o c c u r w h e n a decision p r o b l e m necessitates conditional evaluation. In o u r e x a m p l e , sequential processing was r e q u i r e d to equate relative benefit w i t h absolute benefit. It a p p e a r s that a " c o n d i t i o n a l f r a m e " was generated in w h i c h the underlying risk of death was eliminated from consideration. The majority of patients may simply have c o m p a r e d the relative benefit (i.e., it will decrease y o u r risk of dying b y 80%) w i t h the absolute benefit (i.e., 8 deaths p r e v e n t e d p e r 100 treated, or 8%) in choosing their answers. T h e sense o f certainty associated w i t h this choice is illusory, however, because the benefit of the m e d i c a t i o n p r e s e n t e d in relative terms is conditional on the underlying risk o f dying. O t h e r possibilities exist to explain w h y most patients chose the m e d i c a t i o n w h o s e benefit was presented in relative terms. O u r descriptions o f relative and absolute benefits might have p r e s e n t e d patients with different framings of o u t c o m e s , not just of probabilities. The description of relative benefit read " I f you take this
TABLE 5
CONCLUSIONS The central finding o f this study is that the majority of patients faced w i t h choosing b e t w e e n two treatments, one e x p r e s s e d in terms of its relative benefit and the other in terms of its absolute benefit, chose the o n e f r a m e d in relative terms. This was true even w h e n adequate information (about the underlying risk of death) was p r o v i d e d so that the relative benefit c o u l d be converted to absolute benefit. O f course, this r e q u i r e d multiplying the risk of death b y the relative benefit of treatm e n t to d e t e r m i n e the absolute benefit. Therefore, it m i g h t b e e x p e c t e d that patient characteristics, especially level of education, w o u l d b e associated w i t h the ability to equate relative benefit w i t h absolute benefit. To a degree, this was the case. In the second hypothetical scenario, w h e n the subjects w e r e specifically asked to convert relative to absolute benefit, those w i t h a -> 13 grade e d u c a t i o n w e r e m o r e than t w i c e as likely as those w i t h a -< 12 grade e d u c a t i o n to c o r r e c t l y p e r f o r m this task. Men and patients in " e x c e l l e n t " health w e r e also m o r e likely than their c o m p a r i s o n groups to answer correctly, w h i l e those aged -> 60 years w e r e less likely. However, even in the first h y p o t h e t i c a l situation w h e n all the information necessary to c o n v e r t relative to absolute benefit was available, subjects w i t h at least s o m e college e d u c a t i o n w e r e m o r e l i k e l y to prefer the medication w i t h the benefit e x p r e s s e d in relative terms than w e r e subjects w i t h less education. The same was true for
Percentage of Respondents(n = 470) Answering " 5 0 Deaths Can Be Prevented" (the Expected Mistake) or " 2 5 Deaths Can Be Prevented" (the Correct Answer) and Odds Ratio of Selecting " 2 5 DeathsCan Be Prevented," by Patient Characteristic, for Situation Two Percentage Answering: 50 25
Odds Ratio* of Answering "25 Deaths Can Be Prevented"
Age < 39 years 4 0 - 5 9 years > 60 years
165 149 143
37.6 53.7 53.2
33.9 30.9 18.9
1.00t 0.77 (0.46, 1.29)* 0.40 (0.21,0.74)
Gender Female Male
281 168
51.6 42.3
20.3 42.3
1.00 2.74 (1.76, 4.26)
Education - 12 grade -> 13 grade
216 241
47.7 47.7
19.4 36.1
1.00 2.21 (1.38. 3.54)
Health -< Fair Good Excellent
79 292 80
44.3 50.0 42.5
15.2 28.4 42.5
1.00 1.96 (0.97, 3.95) 2.56 (1.13, 5.82)
Treated for high blood pressure and/or cholesterol No Yes
381 66
48.0 47.0
29.9 21.2
1.00 1.07 (0.53, 2.19)
*As calculated from the coefficients of a logistic regression model that included all listed patient characteristics as independent variables. TReference category. *95% confidenceintervals.
JOURNALOFGENERAl.INTERNALMEDICINE, Volume 8 (October), 1993
m e d i c a t i o n it w i l l d e c r e a s e y o u r r i s k o f d y i n g . . . . " It is c l e a r t h a t t h e b e n e f i t o f m e d i c a t i o n is d i r e c t l y a p p l i c a b l e to t h e p a t i e n t . T h e d e s c r i p t i o n o f a b s o l u t e b e n e f i t r e a d " I f 1 0 0 p e o p l e w i t h t h e d i s e a s e , l i k e y o u , t a k e this medication .... " H e r e , it m a y h a v e b e e n less c l e a r t h a t t h e b e n e f i t o f t h e m e d i c a t i o n w o u l d a p p l y t o t h e pat i e n t . T h e c o m m e n t s o f at l e a s t o n e r e s p o n d e n t sugg e s t e d this i n t e r p r e t a t i o n . A n o t h e r p o s s i b i l i t y is t h a t patients could have known that both medications were e q u a l l y e f f e c t i v e a n d s i m p l y c h o s e n t h e first o n e p r e s e n t e d as b e i n g as r e a s o n a b l e a c h o i c e as a n y o t h e r . T h e o r d e r o f a n s w e r s w a s n o t v a r i e d t o t e s t this h y p o t h e s i s . H o w e v e r , this p o s s i b i l i t y w a s n o t r e f l e c t e d in t h e c o m ments of any respondent. The high response rate and consistency of findings across subgroups in our study indicate that our results a r e v e r y l i k e l y v a l i d for t h e p o p u l a t i o n w e s t u d i e d : a s o c i o e c o n o m i c a l l y m i x e d p a t i e n t g r o u p o f ( a l m o s t ent i r e l y w h i t e ) p a t i e n t s in a s e m i r u r a l p a r t o f t h e N o r t h east. O f c o u r s e , g e n e r a l i z a t i o n t o o t h e r p a t i e n t g r o u p s may be hazardous, but we know of no reason to suspect t h a t t h e r e s u l t s d o n o t a p p l y for o t h e r g r o u p s o f c u l t u r a l l y s i m i l a r p a t i e n t s . S i m i l a r l y , w h i l e it c o u l d b e a r g u e d t h a t o u r s c e n a r i o s a r e artificial, w e m i m i c k e d t h e lang u a g e c l i n i c i a n s u s e in e x p l a i n i n g r i s k t o p a t i e n t s . Thus, w e s u c c e s s f u l l y e x p l o r e d r i s k c o m m u n i c a t i o n as c o m m o n l y p r a c t i c e d in t h i s m e d i c a l c e n t e r . O f c o u r s e , it is true that language distinct from ours might be perceived d i f f e r e n t l y b y p a t i e n t s a n d l e a d to d i f f e r e n t c o n c l u s i o n s . T h e effect o f f r a m i n g o n t h e p e r c e p t i o n o f risks a n d b e n e f i t s is n o t l i m i t e d t o p a t i e n t p o p u l a t i o n s . F o r r o w e t al. t7 p r e s e n t e d d a t a to p h y s i c i a n s f r o m p u b l i s h e d s t u d ies o f t h e b e n e f i t s o f t r e a t i n g p a t i e n t s w h o h a v e h y p e r tension and hypercholesterolemia. Each study was pres e n t e d in t w o d i f f e r e n t ways, o n c e in t e r m s o f r e l a t i v e a n d o n c e in t e r m s o f a b s o l u t e b e n e f i t . W h e n a s k e d h o w t h e i n f o r m a t i o n w o u l d i n f l u e n c e t h e i r d e c i s i o n s to t r e a t , 46% o f t h e p h y s i c i a n s g a v e d i f f e r e n t r e s p o n s e s t o t h e s a m e r e s u l t s p r e s e n t e d in d i f f e r e n t w a y s a n d m o s t indicated a stronger preference to treat after reading of the relative benefit. Though the authors did not include t h e u n d e r l y i n g r i s k o f d i s e a s e in t h e i r d e s c r i p t i o n o f relative benefit, their findings are clearly consistent w i t h o u r s . P h y s i c i a n s , as w e l l as p a t i e n t s , a r e i n f l u e n c e d b y t h e f r a m i n g o f risks a n d b e n e f i t s , a n d . t h e f r a m e w i l l have an influence on clinical decision making. Our study demonstrates the difficulty introduced by attempts to express measures of risk for patients. Compelling information published in terms of relative r i s k a l o n e a r e c o m m o n p l a c e a n d m a y h a v e t h e effect o f blinding the clinician and patient to the absolute benefit. F o r e x a m p l e , at a b a s e l i n e r i s k o f 10%, a n d a r e l a t i v e r i s k r e d u c t i o n o f a t r e a t m e n t o f 50%, t h e a b s o l u t e r i s k
$47
r e d u c t i o n w o u l d b e 5%, w h i l e w i t h a b a s e l i n e r i s k o f 1.0%, it w o u l d b e 0 . 0 5 % . I n t h e f o r m e r c a s e t h e r a p y m i g h t b e a t t r a c t i v e , w h i l e in t h e l a t t e r it m i g h t b e ignored, even though the relative risk reductions are equal. The implications of these findings are critical to t h e w a y in w h i c h c l i n i c i a n s d i s c u s s p o t e n t i a l risks a n d benefits with their patients. Expressions that magnify t h e b e n e f i t b e c a u s e t h e y a r e in r e l a t i v e t e r m s m i g h t t e n d t o b e m o r e c o m p e l l i n g b o t h t o c l i n i c i a n a n d to p a t i e n t , a n d m i g h t e v e n b e c o e r c i v e in t h e i r effect. E x p r e s s i o n s of absolute risk reduction should give patients a greater ability to be informed and to choose treatment preferences rationally.
The authors thank the nurses and receptionists in General Internal Medicine for their help with this study.
REFERENCES 1. Nisbett R, Ross L. Human Inference: Strategies and Shortcomings of Social Judgment. Englewood Cliffs, NJ: Prentice-Hall, 1980. 2. Kong A, Barnett GO, Mosteller F, Youtz C. How medical professionals evaluate expressions of probability. N EngIJ Med. 1986; 315:740-4. 3. Tversky A, Kahneman D. The framing of decisions and the psychology of choice. Science. 1981 ;211:453-8. 4. McNeil BJ, Pauker SG, Sox HCJr, Tversky A. On the elicitation of preferences for alternative therapies. N Engl J Med. 1982; 306:1259-62. 5. Eraker SA,SOxHCJr. Assessment of patients' preferences for therapeutic outcomes. Med Decis Making 1981; 1:29-39. 6. O'Connor AMC, Boyd NF, Warde P, Stolbach WL, Till JE. Eliciting preferences for alternative drug therapies in oncology: influence of treatment outcome description, elicitation technique and treatment experience on preferences. J Chron Dis. 1987; 40:811-8. 7. Siminoff LA, Fetting JH. Effects of outcome framing on treatment decisions in the real world: impact of framing on adjuvant breast cancer decisions. Med Decis Making. 1989;9:262-71. 8. Malenka DJ, BaronJA. Cholesterol and coronary heart disease; the importance of patient-specificattributable risk. Arch Intern Med. 1989;148:2247-52. 9. Laupacis A, Sackett DL, Roberts RS. An assessment of clinically useful measures of the consequences of treatment. N Engl J Med. 1988;318:1728-33. 10. Browner WS, Hulley SB. Effect of risk status on treatment criteria; implications of hypertension trials. Hypertension. 1988; 13(suppl I):I51-I56. 11. Benichou J, Gail MH. Estimates of absolute cause-specific risk in cohort studies. Biometrics. 1990;46:813-26. 12. TverskyA, Kahneman D.Judgement under uncertainty; heuristics and biases. Science. 1974;185:1124-31. 13. Elstein AS. Clinical judgment: psychological research and medical practice. Science. 1976; 194:696-700. 14. Sox HC Jr, Blatt MA, Higgins MC, Marton KI. Medical Decision Making. 1st ed. Stoneham, MA: Butterworths, 1988;213-6. 15. Hosmer DW Jr, Lemeshow S. Applied Logistic Regression. New York: John Wiley & Sons, 1989;38-63. 16. SASInstitute Inc. SUGI Supplemental Library User's Guide, Version 5 Edition. Cary, NC: SASInstitute Inc. 1986;269-94. 17. Forrow L, Taylor WC, Arnold RM. Absolutely relative: how research results are summarized can effect treatment decisions. Am J Med. 1992;92:121-4.
(The appendix follows)
548
Malenka eta/., FRAMINGEFFECTOF RISK
APPENDIX A
S i t u a t i o n One
S i t u a t i o n Two
Suppose you have a serious disease that needs to be treated with medication. Your risk of dying over the next year is 10% if you d o n ' t receive treatment.
The situation is now a little different. Suppose you have a serious disease that needs to be treated with medication. Your risk of dying over the next year is 50% if you don't receive treatment.
There are only 2 possible medications for the disease: Medication A and Medication B: They cost about the same and have almost no side effects.
There is only one medication for the disease. Your doctor provides you with the following information:
Your doctor provides you with the following information about these medications: Medication A:
If you take this medication it will decrease your risk of dying by 8096 (four fifths) over the next year.
Medication B:
If 100 people with the disease, like you, take this medication 8 deaths can be prevented over the next year.
Question:
Which medication do you want? (Circle your answer.) (1) (2) (3) (4)
Medication A Medication B Either Medication A or B Can't decide
Medication A:
If you take this medication it will decrease your risk of dying by 50% (by half) over the next year.
Question:
If 1O0 people with the disease, like you, were treated with Medication A, how many deaths could be prevented over the next year? (Circle your answer.) (1) (2) (3) (4) (5)
5 deaths can be prevented 10 deaths can be prevented 25 deaths can be prevented 50 deaths can be prevented Don't know
