Family Practice Vol. 19, No. 1, 7-11
© Oxford University Press 2002
Original Paper |
Interpretation of and preference for probability expressions among Japanese patients and physicians
Department of General Medicine and Clinical Epidemiology, Kyoto University Graduate School of Medicine,
a Yomise-dori Clinic,
b Nagasaki-chuo National Hospital,
c Primary Care Unit, Hokkaido University Medical Hospital,
d Aso Iizuka Hospital and
e Ishinkai Yao Hospital, Japan.
Dr Fukui, Professor of Medicine, Department of General Medicine and Clinical Epidemiology, Kyoto University Hospital, Kyoto University Graduate School of Medicine, 54 Kawahara-cho, Shogoin, Sakyo-ku, Kyoto 606-8507 Japan.
Ohnishi M, Fukui T, Matsui K, Hira K, Shinozuka M, Ezaki H, Otaki J, Kurokawa W, Imura H, Koyama H and Shimbo T. Interpretation of and preference for probability expressions among Japanese patients and physicians. Family Practice 2002; 19: 7–11.
Received 19 January 2001; Accepted 9 July 2001.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Background. Patients and physicians often communicate using qualitative probability expressions that describe expected outcomes or risks of a medical intervention without knowing whether or not they share the same understanding.
Objective. The aim of the present study was to determine interpretations of qualitative probability expressions in clinical settings by Japanese patients and physicians as well as their assessments and preferences about physicians' use of such expressions.
Methods. One hundred and sixty-eight consecutive patients aged 16 years or older, who attended a university hospital during a 2-week period in 1999, and 156 physicians recruited through the Japanese General Medicine Research Network participated in this cross-sectional survey, using a self-administered questionnaire. Participants were asked to assign numerical interpretations as a percentage to 10 qualitative expressions of probability in two clinical situations related to prescribing a medicine for a cold and an anti-cancer drug. They were also asked which type of expression, qualitative or quantitative, they usually use when communicating probabilistic clinical information and which they prefer.
Results. The estimates of probability expressions showed wide variations, especially among patients. Patients tended to assign lower and higher values to highly positive and negative probability expressions, respectively, than physicians. Clinical context also influenced the estimation: both groups tended to assign higher estimates in the anti-cancer drug situation than in the cold treatment situation. Factor analysis revealed three psychologically meaningful factors in each situation. More patients than physicians (64.6% versus 50.3%) thought that physicians do not use numbers in routine practice. More than 20% of both patients and physicians considered that the actual use of qualitative terms by physicians is undesirable. Nevertheless, a sizable number of patients (41.4%) and physicians (15.2%) considered it preferable that physicians do not use numbers.
Conclusion. Since interpretation of qualitative expressions of probability in Japanese is subject to large interpersonal variability and differences between patients and physicians, as well as context dependence, the use of qualitative expressions alone might cause misunderstanding among the parties involved. However, the majority of patients prefer words to numbers at present. Therefore, physicians, at least in Japan, ought to provide patients with both numbers and words when critical decisions need to be communicated.
Keywords. Communication, decision making, physician–patient relations, probability, qualitative expressions.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Patients and physicians communicate in a clinical environment under the implicit assumption that they share expressions with the same meaning. However, expressions are often vague, so that the same expression may be interpreted in several ways. In such a situation, the ‘message sent’ may differ from the ‘message received’. Ambiguous expressions include qualitative descriptions of probability. Physicians extensively use probability or likelihood expressions such as ‘often’ or ‘rare’ without quantification about disease prevalence, expected outcomes of a recommended medical intervention or the risk of adverse outcomes, among others.
Previous studies on interpretation of qualitative expressions have focused on numerical values assigned to each expression so as to compare the expressions among patients/laymen,1,2 medical professional workers3–5 or both.6 The results showed that interpersonal differences in the interpretations of such expressions were sizable and that the interpretations were influenced by the experience of the decision maker2,4 and by the context of a situation.2 Some researchers have found consistency in the usage of qualitative expressions,5 but others have not.2,5,6 Therefore, the relevance of the use of qualitative expressions of probability by physicians in clinical situations remains controversial.2
People in Japan are familiar with many qualitative expressions of probability in everyday as well as clinical situations. However, numerical interpretations of qualitative expressions have never been examined from the viewpoint of medical cognitive psychology. Meanwhile, ‘ambiguity’ is said to be characteristic of communication in Japanese.7 If so, it is the very question of whether or not there is implicit consistency in qualitative expressions. It is important to understand how ambiguously, if at all, such expressions are interpreted in clinical situations.
The type of preferred expression is no less important than how consistently it is interpreted. Some researchers have explored actual patients' preferences for probability expressions.2,8 The results revealed that a substantial proportion of patients preferred qualitative expressions. Although previous studies focused only on patient preferences, it would be interesting to know whether or not physicians differ from patients in preference for probability expression.
The present study was thus intended to assess the meanings ascribed by both patients and physicians to Japanese probability expressions frequently used in clinical situations. We also examined which expressions of probability, i.e. qualitative or quantitative, patients and physicians actually used and preferred in clinical decision making.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Study participants were Japanese patients aged 16 years or older and physicians. The patients were those who consecutively visited the General Medicine Clinic at Kyoto University Hospital during a 2-week period in 1999. Physicians were recruited through the Japanese General Medicine Research Network.
A three-page questionnaire was given to each participant after informed written consent was obtained. Patients in moderate or severe distress and those with cognitive problems were excluded from the study. Each patient received the questionnaire before being seen by a physician. They were informed as part of the consent process that their decision to participate would not affect their medical care. They completed and returned the questionnaire before leaving the clinic.
The main questionnaire consisted of two components. In the first component, participants were asked to express numerical meanings as percentages (represented by ‘of 100’) for qualitative expressions of probability. We focused on 10 probability expressions in Japanese that are used frequently in clinical situations: "kakuzitsuni (certainly)", "osoraku (probably)", "tabun (perhaps)", "shibashiba (often)", "tokini (sometimes)", "mareni (infrequently)", "tabun ... nai (unlikely)", "osoraku ... nai (improbably)", "mettani ... nai (rarely)" and "zettai ... nai (never)". Respondents were asked to translate each term into a number on a scale from 0 to 100 in two situations with regard to a side effect of a medicine for a cold (1 or 3) and of an anti-cancer drug (3 or 1). For example (question given to patients),
Your physician prescribes a cold medicine and explains the possibility of its side effect such as a rash. Please write in parentheses the number (decimal figure permitted) you think appropriate so that left and right sentences appear equivalent.It perhaps occurs. It occurs in ( ) of 100 ...
The order of both a set of expressions and that of a medicine were presented randomly, avoiding any effects of sequential or anchoring influences.
In the next component (2), sex and age were elicited from all participants and domains of expertise were determined from the physicians. Participants were also asked whether they thought physicians routinely use ‘numbers’ (numerical or quantitative expressions) or ‘words’ (qualitative ones) when trying to impart information about probabilities such as adverse effects of medicine. They were then asked which type of expressions, numbers or words, they liked physicians to use.
The probability estimates were summarized by calculating mean values and standard deviations for each term. F-test, t-test and paired t-tests were used to determine whether or not the reported probability estimates differed between patient and physician groups, or between the medicines to be prescribed for each term. A P-value of <0.05 was considered significant. Principal component factor analysis was performed on the 10 probability terms, followed by a calculation of Cronbach's 
to determine the internal consistency of the resulting scales. After varimax rotation, a term was included in the factor if it was loaded with a weight of at least 0.50. The strength of correlations among the items of a factor was estimated by the eigenvalue. Eigenvalues above 1 were considered significant. Differences between patient and physician groups regarding assessment of actual use of qualitative or quantitative expressions and their preferences were determined by chi-square tests.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
The questionnaire was completed by 168 (88%) of the patients and all of the recruited physicians (156). The mean ages of the patients and physicians were 51 years (range 17–83, SD 18.1) and 36 years (range 24–76, SD 8.2), respectively. The proportions of males among the patients and physicians were 43.5 and 86.5%, respectively. The domains of expertise among the physicians were internal medicine (n = 112) and others (n = 44).
Figure 1
summarizes the estimates for the 10 expressions of probability and shows wide variations, especially for those estimated as intermediate values. For example, the interquartile range for numerical estimates assigned to "tabun (perhaps)" by patients and physicians was 55 and 20, respectively. The variability was prominent within the patient group. Variances with respect to all terms except "shibashiba (often)" were significantly larger within patients than physicians. Mean values of interpretations by patients were significantly different from those by physicians for five of the 10 expressions in each prescription situation (P < 0.05). More patients than physicians gave lower values to the highly estimated probability term "kakuzitsuni (certainly)" and higher values to lower estimated terms such as "mareni (infrequently)", "osoraku ... nai (improbably)" and "zettai ... nai (never)".
|
Context in terms of disease and its related medicine influenced the probability estimation. In the anti-cancer medicine situation, both groups tended to give higher estimates to expressions of intermediate probability such as "osoraku (perhaps)", "shibashiba (often)" than in the situation of a medicine for a cold. Differences of estimated values in the same situation were significant in four or five terms in each group (P < 0.05). Both groups tended to assign higher estimates in the anti-cancer, than in the medicine for a cold situation.
Principal components factor analysis revealed that three psychologically meaningful factors (Table 1) were associated with the 10 probability terms in each medicine situation (cold medicine, anti-cancer medicine). These factors centred on positive (four items), moderate (two items) and negative (four items) interpretations of the qualitative expressions in both situations. The four probability terms that centred on negative interpretations were all composed of grammatically negative terms bearing the "nai (not)" form. Cronbach's for ‘positive’ estimates in each situation was 0.869 and 0.841, respectively, and for ‘negative’ estimates 0.834 and 0.866, respectively.
|
Table 2
shows whether patients and physicians prefer numbers or words. Half or more of both groups (65% of patients, 50% of physicians) thought that physicians actually used more words than numbers. With respect to the preference for probability explanation, the proportion of those who preferred words was lower in both groups compared with the rate stated above. However, patients who preferred words (41%) outnumbered those preferring numbers (36%), in contrast to 58% of physicians who preferred numbers. Thus, the ratio of patients who thought that physicians routinely used only words was larger than that of physicians, and >20% of each group considered the actual words used by physicians to be undesirable. Nevertheless, a sizable number of individuals, especially among the patients, considered that physicians should use words rather than numbers.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Improving communication between patients and physicians is of paramount importance worldwide. As a prerequisite for effective communication, mutually understandable information must be exchanged using words or numbers. However, little is understood about how people in Japan communicate in terms of probability. We therefore examined how qualitative expressions of probability are interpreted quantitatively by patients and physicians and which expressions, qualitative or quantitative, are mainly used and preferred.
Descriptive analysis of interpretations of words
Previous studies on the interpretation of qualitative expressions have shown that interpersonal differences are large and that interpretations are influenced by various factors such as situational context and respondents' attributes. Kong et al. have found some consistency in the usage of qualitative expressions.4 Many researchers, however, have found little or limited consistency in interpretation of qualitative expressions.2,5,6 We focused on 10 qualitative probability expressions in Japanese and found large interpersonal variability in numeric interpretations of these expressions, especially those of intermediate probability, with less variability among physicians than among patients. Moreover, patients tended to avoid assigning extreme values, and both patients and physicians were influenced by the type of medicines being prescribed. According to ‘prospect theory’ (Tversky and Kahneman9), outcomes that are merely probable are underweighted in comparison with outcomes that are obtained with certainty. Patients might have such a tendency more than physicians because of unfamiliarity of probability of medical events. Overall, these interpretations of qualitative expressions of probability came under the influence of various factors, and we found little consistency among them.
Comparing the mean values and SDs for specific probability terms in our results with those in previous studies is difficult, since strict (one-to-one) translation between Japanese and English words in many cases is neither feasible nor established. However, when rough comparison is made, the mean value and SD for "shibashiba (often)" by patients in relation to medicine for a cold in our study, for example, were 52.9 and 23.3, respectively, whereas those for "possible" were 49.7 and 27.2 according to Mazur et al.,1 and 36.9 and 22.5 according to Woloshin.2 Since these SDs are quite close, interpersonal variability among probability interpretations might have similarity regardless of linguistic differences.
Factor analysis of words
Matur1 and Woloshin2 found using factor analysis of probability estimates that three psychologically meaningful factors were associated with qualitative expressions. The present study also identified three factors. Individuals may have common categorical recognition of probability expressions regardless of cultural and linguistic differences. Moreover, low estimates of probability expressions bearing the negative modifiers in the present study were psychologically classified under the same, exclusive factor. Our results suggest that negatively modified terms, in general, constitute a key factor abstracted from enriched probability expressions in the Japanese language. Although probability expressions are abundant, we might have a bigeminal or two-phased recognition of probability, that is a rough classification and a finer modification.
Attitudes towards numbers and words
Although patient preference for probability expressions is a matter of concern, how content both patients and physicians are with the status quo and which probability expressions both groups prefer should be considered. Previous studies on patient preference for numbers or words have not been conclusive. Woloshin et al. found that patients preferred quantitative to qualitative information.2 Mazur and Hickam, however, reported the opposite results.8 The present study found that Japanese patients prefer qualitative to quantitative terms, as did Mazur and Hickam, whereas the preference of physicians was opposite.
Our study revealed a discrepancy between patients and physicians in the assessment of physicians' actual expressions of probability. Patients did not think that physicians used numbers in clinical situations as frequently as physicians themselves thought. Patients might give weight to and listen attentively to qualitative expressions, ignoring physicians' use of numbers, or physicians tend to place more value on numbers and overestimate using quantitative expressions. At any rate, physicians should recognize this discrepancy when referring to probability.
Pertinent usage of words
The best way to communicate with patients regarding probability should be carefully thought out. Because of the inherent vagueness of qualitative probability terms, three choices are available: quantify probabilities whenever possible,5,6 use qualitative and quantitative estimates together1,2 or codify interpretations of qualitative expressions.4,10 Codifying vague expressions does not appear to be a realistic solution according to the inconsistency described above, lack of consensus and difficulties of the task, among others. Considering that most patients preferred words to numbers, we believe that both numbers and words should be provided when faced with critical decision making. Confirmation of mutual understanding is of course mandatory in most of the clinical situations.
Limitations
This study has several limitations. First, the subjects are limited samples of convenience. Therefore, the results might not be applicable to other parts of the population in Japan. Specifically, the patients in our study all lived in Kyoto, whereas the physicians were recruited from all over Japan. This could confer some limitation to the comparison between patients and physicians because of a possible difference in local cultural and linguistic background. Secondly, estimating one point of probability for a vague expression is difficult in nature. It would have been easier for participants to assign a numerical range, rather than one point of probability. Variability might become larger in such a situation. Thirdly, the responses were sought based solely on a scale of 0–100. Participants would have difficulty in giving estimates <1 even though they were allowed to respond using a decimal figure if necessary. Actually, only two patients assigned figures <1.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Interpretations of qualitative expressions of probability in Japanese vary widely between the groups of patients and physicians as well as among each group, and the clinical situation significantly influences the interpretations. Thus the use of qualitative expressions could cause misunderstandings in clinical settings. Since more patients than physicians wanted to have information in terms of words rather than numbers, physicians ought to provide patients with both numbers and words in critical decision-making settings
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
1 Mazur DJ, Hickam DH. Patients' interpretations of probability terms. J Gen Intern Med 1991; 6:237–240.[Web of Science][Medline]
2
Woloshin KK, Ruffin MT 4th, Gorenflo DW. Patients' interpretation of qualitative probability statements. Arch Fam Med 1994; 3:961–966.
3 Bryant GD, Norman GR. Expressions of probability: words and numbers [letter]. N Engl J Med 1980; 302:411.[Web of Science][Medline]
4 Kong A, Barnett GO, Mosteller F, Youtz C. How medical professionals evaluate expressions of probability. N Engl J Med 1986; 315:740–744.[Abstract]
5
Robertson WO. Quantifying the meanings of words. J Am Med Assoc 1983; 249:2631–2632.
6 Nakao MA, Axelrod S. Numbers are better than words. Verbal specifications of frequency have no place in medicine. Am J Med 1983; 74:1061–1065.[Web of Science][Medline]
7 Takai J, Ota H. Assessing Japanese interpersonal communication competence. Jap J Exp Soc Psychol. 1983; 33:224–236.
8 Mazur DJ, Hickam DH. Patients' preferences for risk disclosure and role in decision making for invasive medical procedures. J Gen Intern Med 1997; 12:114–117.[Web of Science][Medline]
9 Kahneman D, Tversky A. Prospect theory: an analysis of decisions under risk. Econometrica 1979; 47:263–291.[Web of Science]
10 O'Brien BJ. Words or numbers? The evaluation of probability expressions in general practice. J R Coll Gen Pract 1989; 39:98–100.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  P. Webster and J. Austoker Women's knowledge about breast cancer risk and their views of the purpose and implications of breast screening--a questionnaire survey J. Public Health Med., September 1, 2006; 28(3): 197 - 202. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Thorne, T. G. Hislop, M. Kuo, and E.-A. Armstrong Hope and Probability: Patient Perspectives of the Meaning of Numerical Information in Cancer Communication Qual Health Res, March 1, 2006; 16(3): 318 - 336. [Abstract] [PDF]
|
|
|
|
 |
|
 D. Misselbrook and D. Armstrong Thinking about risk. Can doctors and patients talk the same language? Fam. Pract., February 1, 2002; 19(1): 1 - 2. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||