Physicians' Perceptions of Patients' Knowledge and Opinions Regarding Breast Cancer: Associations with Patient Education and Physician Numeracy

Abstract

Summary

Background: The aim of this study was to assess physicians' perception of their patients' knowledge and opinions regarding regular screening, and the association of their perceptions with physician numeracy and patient education level. Methods: We carried out a survey study of 240 obstetrician-gynecologists. Results: Overall, 99.6% physicians perceive that their patients know that breast cancer is hereditary, 86.5% predicted that there is a gene mutation related to breast cancer, and 79.4% predicted that most breast cancer cases occur in women aged 50 years or greater. Physicians with less educated patients thought that their patients would not know about genetic screening, and physicians with more educated patients thought that their patients would know that mammography does not reduce the risk of getting breast cancer. A total of 66.0% of obstetrician-gynecologists answered all 3 numeracy questions correctly. Less numerate physicians were more likely to indicate that their typical patient would agree with the statement about regular mammography screens than the more numerate physicians. Conclusions: Obstetrician-gynecologists expect that their patients know some things about breast cancer and not others. Some of the physicians' perceptions about patients differ based on numeracy.

Introduction

Educating patients about breast cancer screening, testing, and treatment is an important aspect of many gynecologists' practice. Research has shown that women's knowledge about breast cancer is lacking [1, 2, 3], and previous studies suggest that patients' and physicians' perceptions of their communication with each other are not always in line [4, 5]. For example, in a national sample of women, only 29% considered age and 28% considered high-fat diet as risk factors for breast cancer [6]. Having an accurate perception of patients' knowledge and opinions can enhance physician-patient communication and education. For example, if physicians overestimate their patients' knowledge, physicians may not communicate enough information to their patient. Little research has been done on whether physicians have accurate perceptions of their patients' knowledge and opinions about breast cancer.

The ability to use and understand numbers (referred to as numeracy) has been shown to be associated with medical decision-making and health outcomes [7, 8, 9]. Those who are less numerate tend be more vulnerable to decision biases (such as framing and loss aversion). Ancker and Kaufman [10] suggest that 3 variables interact to affect health numeracy in the communication between patients and health care providers: i) the individual numeracy levels of the patient and provider; ii) the oral communication skills of the patient and provider (including accurate perception of other's knowledge and opinions); and iii) the quality and ease of use of materials that provide health information. Physician numeracy is often overlooked as it is thought that health literacy deficits are a problem among patients, not physicians. However, research shows that physicians have problems with numeracy at times; for example, many obstetrician-gynecologists could not calculate the positive predictive value after a positive mammography [7]. While lots of research has looked at patient numeracy and patient decision-making, few studies have measured physician numeracy and how it affects physician decision-making.

This study looks at physicians' perception of their patients' knowledge and opinion regarding regular screening, and physician numeracy. We look at patient education as a possible factor in physicians' perceptions of their patient's knowledge and screening preferences. We hypothesize that physicians will report that their patients know that age and high-fat diet are risk factors for breast cancer. We hypothesize that almost all physicians will answer the numeracy questions correctly, and finally, we hypothesize that there will be no difference in the perception of patient knowledge and opinions regarding breast cancer among high and low numerate physicians.

Methods

Sample

A survey on breast cancer screening was administered to 400 Fellows of the American College of Obstetricians and Gynecologists in the fall of 2009. Fellows were members of the Collaborative Ambulatory Research Network (CARN) which is a group of practicing obstetrician-gynecologists who volunteer to participate in survey research.

Survey

The survey was based on a mixed-method design (both an electronic and a paper version of the survey were used). Information about the study and the link to the online survey were emailed to the sample. After 4 email reminders, a paper version of the survey was mailed to all non-responders (and those who did not have email addresses). One paper reminder mailing was sent. Survey questions included demographics, 6 questions about their patients' education level (physicians were asked to estimate the percentages of their patient population that were in the following educational ranges: some high school, high school or GED (General Educational Development), some college, associate degree, college degree, graduate degree (PhD), professional degree (Juris Doctor, JD or Medical Doctor, MD)), 9 questions to which physicians responded as if they were their average patient (8 true/false questions about patients' knowledge (e.g. ‘Breast cancer is hereditary’), and 1 question about their patients' agreement with: ‘All women greater than or equal to 40 years old, even if they are at low risk, should get a mammogram every 1-2 years’). The Schwartz numeracy scale [11] was used to assess numeracy as it is the measure used in previous research with medical providers [12, 13]. It comprises 3 questions: i) a conversion from 1% to 10 in 1,000; ii) a conversion from 1 in 1,000 to 0.1%; and iii) an estimation of how many heads there will be in 1,000 coins flips.

Data Analysis

Data were analyzed using a personal computer-based version of SPSS 16.0 (SPSS Inc. Chicago, IL, USA). Descriptive and frequency data was computed for primary analysis for demographic questions, score on numeracy scales, true/false questions, and the question about regular mammography. ANOVA was use to assess whether the education level of physicians' patients was associated with their predictions of the total number of questions about breast cancer. Chi-square was used to assess whether physician numeracy was associated with individual items. Mann-Whitney U was used to assess whether predictions about regular mammography screening differed among physicians with high and low educated patients. Physicians estimated the proportion of their patient populations with various education achievements (table 1). A cluster analysis was conducted to form 2 groups of physicians based on their patients' education: one group with a more educated patient population (n = 122) and another group with a less educated patient population (n = 125). Significance was evaluated at alpha < 0.05 and confidence intervals of 95%.

Table 1.

Cluster groups for education: the number of participants in each cluster and the average percent of patients physicians reported for each category

	Cluster	n	mean %	SD
Some High School	more educated	122	3.57	4.32
	less educated	125	17.13	17.74
High School or GED	more educated	122	9.92	8.78
	less educated	125	40.79	15.29
Some college	more educated	122	12.68	13.44
	less educated	125	13.52	10.31
Associate degree	more educated	122	7.41	8.72
	less educated	125	7.74	6.88
College degree	more educated	122	34.67	23.62
	less educated	125	14.32	8.56
Graduate degree (PhD)	more educated	122	7.64	8.77
	less educated	125	4.13	4.22
Professional degree (JD or MD)	more educated	122	5.73	5.82
	less educated	125	2.72	2.89

SD = Standard deviation; GED = General Educational Development; JD = Juris Doctor; MD = Medical Doctor.

Results

Of the 240 responders (60% response rate), only those who answered the numeracy questions were included in analyses (n = 191). The mean age was 50.9 years (standard deviation (SD) = 10.4), and 50.0% were female.

Numeracy

Overall, each numeracy item was answered correctly by 88% or more of the sample: 174 (90.6%) correctly converted a percent to frequency, 170 (88.5%) correctly converted a frequency to a percentage, and 169 (88.0%) correctly answered 500 rolls out of 1,000. Three-fourths of the sample answered all 3 questions correctly (n = 144, 75.3%), and of the remaining, 4 (1.6%) answered 0 correctly, 7 (3.7%) answered 1 correctly, and 37 (19.4%) answered 2 correctly. Given the non-normal distribution of the scores on the numeracy scale, data were combined into 2 groups for analysis; the more numerate group (answered all 3 correct) and the less numerate group (answered < 3 correct). Physician numeracy was not associated with age, gender, or patient education cluster.

Prediction of Patient Knowledge and Preferences

For the prediction of patient knowledge questions (true/ false), physicians were asked, ‘How would your typical patient complete the following’; physicians predicted that patients would know an average of 4.8 (SD = 2.0) of the 8 questions regarding breast cancer (table 2). When asked how their typical patient would respond to, ‘All women greater than or equal to 40 years old, even if they are at low risk, should get a mammogram every 1-2 years’ (answer on a scale from strongly disagree (-5) to agree strongly (5), recoded 1-11), the mean response was 9.04 (SD = 2.0).

Table 2.

Descriptive statistics on the study questions

There is an increased risk of breast cancer associated with early age of first menstrual period, n (%)
True	86 (36.6)
False	149 (63.4)
Percent correct to each numeracy questions n (%)
Question 1	174 (90.6)
Question 2	170 (88.5)
Question 3	169 (88.0)
Numeracy scores, n (%)
0 correct	4 (1.6)
1 correct	7 (3.7)
2 correct	37 (19.4)
3 correct	144 (70.3)
Response about regular mammography screening on a scale from 0 (strongly disagree) to 10 (strongly agree), mean (SD)	9.04 (2.0)

SD = Standard deviation.

Inferential Statistics

Physicians with more educated patients correspondingly predicted that their patients would answer significantly more questions correctly than physicians with less educated patients (M = 5.1 vs. M = 4.5, F(1,224) = 4.1, p = 0.04)). A greater number of physicians who tended to have low educated patients indicated that their patients would incorrectly respond false to ‘Women can be tested for a gene mutation that is related to her risk of breast cancer’ (χ² = 4.1, p = 0.04). A greater number of physicians who tended to have low educated patients also indicated that their patients would incorrectly respond false to ‘There are multiple genes that are related to breast cancer’ (χ² = 10.1, p = 0.002). A greater number of physicians who tended to have high educated patients indicated that their patients would correctly respond false to ‘Mammography reduces the risk of developing breast cancer’ (χ² = 5.4, p = 0.020). Prediction of patients' responses on these knowledge questions did not differ between high and low numerate physicians. Numeracy level was associated with physicians' predictions of patients' preferences about regular mammography screening. The less numerate group was more likely to indicate that their typical patient would agree with getting regular mammography screens. The mean response was 9.5 for the moderate/low numerate group and 8.8 for the high numerate group (z = 2.52, p = 0.012).

Discussion

The results of this study suggest that, on average, physicians perceive that their patients know that breast cancer is hereditary and that there is a gene mutation related to breast cancer. These findings are in line with what surveys of patients have found about women's breast cancer knowledge [14, 15]. Some of our results of physicians' perceptions are not in line with reports of women's aggregated knowledge. For example, nearly 80% of physicians thought that patients would know that most breast cancer cases occur in women 50 years or older, however, studies with nationally representative samples of women have shown that 29% [14] and 30% [15] of women consider age as a risk factor.

Physicians with lower educated patient populations predicted that their patients would not know that there are multiple genes associated with breast cancer and that patients can be tested for them. This finding is supported by previous studies which suggested that lower educated patients do in fact tend to know less about genetic testing [16, 17]. A direct comparison of physicians' predictions with their actual patients' knowledge would be interesting future research.

Similar to previous studies where 72% [12] and 60% [13] of the physician samples answered all 3 of the Schwartz numeracy questions correctly, 75% of the sample in this study answered all 3 correctly. Studies in the general population and educated samples have found that the hardest question on the Schwartz measure is the one that requires converting a frequency to a percentage [18], about 20% complete the task correctly. In our physician sample, 88.5% of the sample correctly converted a frequency to a percentage. Though the Schwartz measure allowed us to compare high and low numerate individuals in this study by dividing the sample based on whether they answered all 3 correct or answered < 3 correctly, perhaps future studies could use more difficult problems that are relevant to physicians' practice in order to distinguish moderate numerate level and low numerate level.

This is one of the first studies to our knowledge that assesses whether physician numeracy is associated with any aspect of their practice. Our results suggest that physician numeracy is associated with physicians' perception of their patients' opinions about regular screening. Given that numeracy has been found to be associated with one's own risk assessments [19], it makes sense that numeracy would also be associated with perceptions of others' health practices and risk assessments. Less numerate physicians were more likely to think that their patients agreed with regular mammography.

Further studies will be needed to determine whether these findings are replicable and examine whether high or low numerate physicians are demonstrating any perceptional biases, as low numerate individuals tend to be more susceptible to biases and heuristics under conditions of uncertainty and risk [19]. As already mentioned, a direct comparison of physicians' predictions with their actual patients' knowledge would be a useful direction for future research on this topic. Along with physicians educating patients, decision tools are also available to provide education and counseling to patients; for example cancer-related decision aids have been found to be an effective way to increase patient knowledge [20].

Disclosure Statement

The authors have no conflicts of interest.