How Do Family Physicians Communicate About Cardiovascular Risk?

Frequencies and Determinants of Different Communication Formats


Stefan Neuner-Jehle; Oliver Senn; Odette Wegwarth; Thomas Rosemann; Johann Steurer
Abstract


Background: Patients understand
information about risk better if it is communicated in numerical or
visual formats (e.g. graphs) compared to verbal qualifiers only. How
frequently different communication formats are used in clinical primary
care settings is unknown.
Methods: We collected socioeconomic and patient understanding
data using questionnaires and audio-recorded consultations about
cardiovascular disease risk. The frequencies of the communication
formats were calculated and multivariate regression analysis of
associations between communication formats, patient and general
practitioner characteristics, and patient subjective understanding was
performed.
Results: In 73% of 70 consultations, verbal qualifiers were used
exclusively to communicate cardiovascular risk, compared to numerical
(11%) and visual (16%) formats. Female GPs and female patient's gender
were significantly associated with a higher use of verbal formats
compared to visual formats (p = 0.001 and p = 0.039,
respectively). Patient subjective understanding was significantly higher
in visual counseling compared to verbal counseling (p = 0.001).
Conclusions: Verbal qualifiers are the most often used
communication format, though recommendations favor numerical and visual
formats, with visual formats resulting in better understanding than
others. Also, gender is associated with the choice of communication
format. Barriers against numerical and visual communication formats
among GPs and patients should be studied, including gender aspects.
Adequate risk communication should be integrated into physicians'
education.Background




A major challenge in preventive medicine is to effectively inform
patients about the risks and benefits of possible treatment options.
Risk can be communicated by words (verbal qualifiers, e.g., "your risk
is high" or "this is not good for your health"), by numerical formats
(absolute percentages, relative percentages, natural frequencies), by
visual formats, or by a combination of these methods.^[1,2]
Although there is no clear evidence as to which format is the best for
communicating risk most effectively, recent research allows a ranking
(hierarchy) of the different risk communication formats in terms of
effectiveness and patient understanding: the use of natural frequencies,
graphical formats (e.g., bar charts), and combinations of these are
more comprehensive than percentages or a purely verbal translation of
risk,^[3–6] and patients prefer these formats as opposed to percentages.^[7]
Patient characteristics as age, education, cultural background,
psychosocial aspects as well as literacy influence a patient's
understanding of information about risk and communication preferences.^[8]
The numeracy of patients refers not only to the capacity of
understanding numerical formats accurately, but also to graphical
formats. Numeracy has an impact on risk perception, adherence to
interventions and even health outcomes.^[9]
Also physician characteristics are important determinants of which
formats are used, thus leading to a large variation in the use of risk
communication formats among primary care providers.^[10]
Growing evidence suggests that involving patients in decision-making
has positive effects in terms of patient satisfaction, adherence, and
even health outcomes.^[11,12] Patients increasingly seek more active participation in healthcare decisions, though not all of them to the same degree,^[13] and there has been a call for a shift towards a meaningful dialogue between patients and physicians and shared decision making.^[14]

Cardiovascular disease (CVD) is a major issue in public health,
contributing excessively to the overall morbidity and mortality of
populations in industrialized societies.^[15]
Only a minority of people with cardiovascular risk factors (CVRF) who
qualify for prophylactic interventions to lower their risk is treated
adequately.^[16,17]
Possible explanations are low adherence of doctors to guidelines and
ineffective risk communication between doctors and patients.^[18]
Furthermore, cardiovascular risk is often perceived inappropriately by
primary care patients, leading to over- or underestimation of the risk.
Communicating risk by understandable formats has the potential to
correct inappropriate risk perception.^[19]
Systematic reviews suggest that providing adults at moderate to high
CVD risk with information about their global CVD risk (using different
communication formats) improves their accuracy of risk perception and
probably increases their intent to initiate CVD prevention.^[12,20]
Gain-framed messages (e.g. presenting a survival benefit), as apposed
to lost-framed messages (e.g. presenting a potential damage), shorter
timeframes (5 or 10 years until realisation of the risk, versus 15 or 20
years) and visual formats seem to enhance understanding of risk and to
increase the self-efficacy to prevent CVD.^[21]

Thus, improving comprehension and the effectiveness of risk
communication in the field of CVD is an important and challenging task
in medicine with public health consequences. How often primary care
physicians use the various risk communication formats while counseling
patients with CVRF is yet unknown.
The aim of this study was to investigate how frequent verbal,
numerical, and visual formats, or a combination of formats, are used in
the CVRF communication process between primary care physicians and
patients. An additional aim was to identify patient and GP
characteristics associated with different counseling formats. In
contrast to previous research studying the effectiveness of different
risk communication formats in healthy volunteers,^[2,4,6]
we chose a real physician-patient encounter and assessed the primary
outcome (counseling format) by audio-recordings rather than
self-reporting.Methods



Procedure


To investigate how CVD risk is communicated, primary care physicians
audio-taped counseling sessions with patients at CVD risk. The sessions
were part of the usual care performance of the general practitioners
(GPs; specialists for general and internal medicine). In three northern
and central Swiss cantons (Zurich, Lucerne, and Zug), all GPs approved
by the Swiss Medical Doctors Federation (Foederatio Medicorum
Helveticorum, FMH) who were either running a medical office
independently or employed in a medical office were invited to
participate by information leaflet sent to postal addresses provided by
the FMH. The recruitment and instruction of doctors started in December
2008, and data were collected until the end of February 2010. The study
was received and approved by the ethical committees of the cantons of
Zurich, Lucerne and Zug.
The doctors who were willing to participate were visited and provided
with information about the aims of the study in so far that we told
them that the study is about investigating risk communication in daily
practice. The participants were further introduced to the details of how
to collect the requested data and provided with the required materials
(digital audio-recorder, several copies of a GP questionnaire, several
copies of a patient questionnaire), as they had to collect all data by
themselves. No suggestions were given to physicians on the method or
program of risk calculation. No information was given neither to
physicians nor patients as to what we exactly planned to analyze in the
interviews, in order to avoid desirability bias. The doctors were
informed about the planned analysis only after data collection,
including the right to withdraw the data (which none of the GPs finally
did).
During the next three months, GPs identified eligible patients among
those in their practice and asked them to participate based on the
following inclusion criteria: 35 to 65 years of age (in order to adapt
to the age range of risk calculators; exceptionally younger or older
participants could be included if the GP calculated their risk by other
methods), and with at least one risk factor of the following - lipid
disorder; high blood pressure (using internationally accepted cut-off
points), and ongoing tobacco smoking. Exclusion criteria were: cognitive
deficiency (dementia, stroke); terminal disease with poor prognosis in
terms of survival time; acute somatic or psychiatric disorder; and
clinical manifestation of any CVD. Patients willing to participate
signed an informed consent form before starting the counseling session.
These sessions were audio-taped, and patients and GPs completed a
questionnaire immediately after the session.
Patient questionnaires referred to their age, gender, level of
education, ethnicity, and the presence of CVD in first degree relatives.
The self-rated understanding of the received information, the awareness
(estimate) of the risk of developing CVD, and the anxiety regards
developing CVD were measured on a visual analogue scale (VAS) ranging
from 0 (low level of understanding, awareness, and anxiety) to 100
(highest level of comprehension, awareness, and anxiety) points. In
addition, patients documented a translated 8-item-set of a validated
index of the need and expectation for medical information, the autonomy
preference index (API).^[22,23]
The GP questionnaire asked for an estimate of the patient's risk of
developing CVD and an estimate of the patient's anxiety towards
developing CVD, again measured on a VAS, ranging from 0 (low level of
risk and anxiety) to 100 (highest level of risk and anxiety) points.
All audio tapes and questionnaires were collected by the principal
investigator three months after the first visit. After the recorded
files were transcribed, data from each counseling session were
classified independently by two researchers into different format
categories: verbal, numerical, visual, or combined. In case of
disagreement, discussion of the interview would follow in order to reach
consensus; if this would not be possible, data would be excluded. In
fact, there was no disagreement among raters in the communication format
classification of interviews.
A consultation was classified as "visual format" if the information
from the audio-taped interview indicated the use of a table or graph.
The total counseling time and the ratio of patient talking time to total
consultation time was calculated from the audio record.


Participants


Out of 1188 primary care physicians invited to participate in the
study, 35 (2.9%) were willing to participate. Finally, 22 primary care
physicians (1.9% of invited GPs) enrolled 77 patients (1–9 patients
each). For seven patients, audio records were incomplete due to
technical reasons and, therefore, excluded from the main analysis. No
values were missing in the questionnaires. The baseline characteristics
of the physicians and patients are summarized in Table 1.
The median age of the participating primary care physicians was 46.5
years (IQR 38 - 57 years), and 64% were male. The GPs had a median 11
years (IQR 5 - 20 years) of experience as a primary care physician. The
median age of patients was 51 years (IQR 45.5 - 59 years), and 58% were
male.


Data Analysis


For our main outcome, the prevalence of different risk communication
formats, we calculated the frequency of each format and the
corresponding 95% confidence intervals. We tested bivariate associations
between the different counseling formats and patient and physician
characteristics using the Kruskal-Wallis and Fisher's exact tests for
continuous and categorical variables, respectively. The intraclass
correlation coefficient was calculated to assess the agreement between
different counseling formats in GPs who recorded at least two
encounters. To further investigate independent determinants of the
counseling format used (verbal-numerical-visual) we performed a
multinomial logistic regression analysis including all physician and
patient characteristics showing at least a borderline significant (i.e. p<0.1)
bivariate association with the counseling format. In addition, the
model was controlled for the clustering effect by the GP. In a secondary
analysis, we investigated whether the counseling format was an
independent determinant of the level of understanding by performing a
multiple linear regression. Potential confounding factors, such as
patient age and gender, education, API, total counseling time, patient's
self-assessed level of CVD risk, and anxiety, were included as
covariates. For the quantitative analysis, we used the SPSS Statistical
Software Package (SPSS version 14, SPSS Inc, Chicago, Illinois).Results




In 51 of 70 consultations (73%; 95% CI 62–84%), GPs communicated
cardiovascular risk to their patients using verbal qualifiers only. In
eight consultations (11%; 95% CI 4–19%), the GPs combined such verbal
qualifiers with numerical information. Graphical formats were
exclusively used in one consultation (1.4%; 95% CI 0.3–8%) and together
with numerical information in 10 consultations (14%; 95% CI 8–24%).
Among the 18 consultations in which GPs presented numerical
information to their patients, they exclusively used absolute
percentages in 10 sessions, combined absolute risk information and
natural frequencies in seven sessions, and used relative risk
information to communicate the patient's risk in one session.
The majority of visual formats (91%) were tables with a
color-coding system analogous to traffic light colors, indicating low,
medium, and high risk values. The underlying risk calculation was always
based on a 10-years timeframe. Only one doctor used another graphical
format, a survival curve. Bar charts or population diagrams were not
used at all in our study sample.
Ten of 16 GPs (62.5%) who provided more than one consultation used
the same risk communication format throughout the sessions without
variation, whereas six GPs switched between different formats. The
intraclass correlation for the counseling formats used was 0.63 (95% CI
0.41–0.85; p < 0.01) in GPs who performed more than one consultation on risk, indicating a significant clustering effect.
The mean consultation time for communicating risk was 9 min 46 sec
(SD, 5 min 35 sec), ranging from 1 min 26 sec to 32 min 04 sec. During
the consultations, patients were talking 24% of the time. On average,
GPs' estimate of patients' CVD risk was 29 (range: 0–81) on the VAS
scale (0–100), whereas patients' mean estimate of their own CVD risk was
28 (range: 3–100). In the direct comparison between GPs' and patients'
CVD risk estimates (estimates on the same patient), they differed by
17.5 points (SD 17.3). For the estimates of the anxiety of developing
CVD, the mean estimate of GPs was 31 (range: 1–98), the mean estimate of
patients 26 (range: 0–100). On average, the estimates of GPs and
patients differed by 16.0 points (SD 16.4) in the direct comparison.
Results of the bivariate analysis between risk counseling formats and patient and physician characteristics are listed in Table 2. Gender
was the only GP characteristic strongly associated with the
communication format; only one female physician used numerical formats
and none used visual formats (Figure 1). Patient gender and degree of
subjective understanding the given information were patient determinants
associated with the communication format. Female patient gender
remained significantly associated with a higher use of purely verbal
qualifiers compared to visual formats when controlled for practice type
(i.e. a single workplace medical office vs. a primary care center with
several physicians), duration of GP-patient relationship, patient
subjective understanding, the ratio of patient talking time to total
consultation time, and the clustering effect of the GP (OR 1.4, p
= 0.039). The use of a visual format resulted in significantly higher
subjective perceived patient understanding compared to pure verbal
counseling (adjusted mean (SE) difference of 10.3 (2.7) points on the
0–100 VAS scale), which remained independently associated when
controlled for patient age and gender, education, API, total counseling
time, patient's self-assessed level of CVD risk and anxiety, and GP
clustering effect (Figure 2). None of the other potential covariates
analyzed showed any association with the communication format.

(Enlarge Image)

Figure 1. Applied risk communication format depending on the gender of the GPs (bars show the frequencies of consultations conducted by female GPs and male GPs).

(Enlarge Image)

Figure 2. Association between the subjectively perceived level of understanding and the applied risk communication format. Difference between the communication formats regarding to the mean estimates of subjectively perceived unterstanding: p = 0.001 for visual vs. verbal format and p = 0.12 for visual vs. numerical format

Discussion




The majority (73%) of primary care physicians participating in this
study used exclusively verbal formats to communicate cardiovascular risk
to their patients. The combination of numerical and visual formats was
used to a minor degree, as was the combination of verbal and numerical
formats. Our findings demonstrate a gap between the recommendations of
medical associations, which favor numerical and visual formats for
communicating risk, and the reality in clinical practice.
In our study, the frequency of the use of verbal formats in risk
communication appeared to be associated with gender; female physicians
communicated risk more often in verbal formats than male physicians. We
found the same association, though to a lesser degree, in female
patients, independent of the GP's gender. There are only few data
existing on this gender issue in doctor-patient communication; a recent
systematic review showed that female doctor-patient dyads are talking
longer and combine different communication styles in one consultation in
comparison to mixed or male dyads.^[24]
To our knowledge, the gender association in risk communication formats
has not been described previously and merits further research. This
tendency of female physicians and female patients to communicate
exclusively in a verbal format should also be addressed in the
development of medical education programs and tools.
The statistically quantifiable association between the use of visual
formats and understanding rated by patients in our study is not
necessarily translating into a clinically relevant benefit of
understanding among patients. Though, it is consistent with data from
the literature, which show that visual formats are easier to understand
than other, especially verbal, formats.^[5–7] Only a few studies address the visual formats used by the physicians in our study.^[25] However, compared to verbal and numerical formats alone, findings for other visual formats such as "risk ladders",^[26] population charts,^[27] pie charts,^[28] and histograms^[29–31]
suggest better understanding, increased risk perception, higher risk
aversion, and better acceptance of interventions in patients.
Restrictions on the use of visual formats should not be forgotten:
depending on the skills of a person to understand graphics^[32] and the danger to manipulate patients by emphasizing parts of the graphic.^[33–36]
The same is true for numerical formats. Moreover, some visual formats
(e.g., survival curves and scales) need additional verbal or numerical
explanations to be sufficiently clear,^[37,38] and others, such as population figures, are not easier to understand than numerical formats.^[39]
The use of colors as additional information in visual formats, which
was also used by participating physicians, seems to be powerful and
familiar to patients, providing them with an important reference point
about the severity of risk (e.g., the red color indicating urgency and
necessity to stop proceeding in the same way as until now).^[40]

Nevertheless, a high proportion of patients in our sample indicated
that they subjectively understood the communicated risk fairly well when
confronted with verbal qualifiers or numerical formats alone.
Sixty-three percent of GPs used their preferred format without
variation, which indicates that, in addition to the associations with
doctor and patient gender, the choice of the communication format
depends more on physician characteristics than those of the patient.
The difference between GPs' and patients' estimates of CVD risk and
the anxiety of developing CVD is significant. Previous research already
reported a high frequency of underestimation of CVD risk (and, to a
minor degree, overestimation) among patients with CVD risk.^[19]
In the absence of biometric data from patients in our study population
in order to calculate their CVD risk, it is not possible to tell whether
rather patients' or GPs' estimate of risk is closer to the real CVD
risks. The result is nevertheless noteworthy as it makes obvious that to
a certain extent physicians are not able to fully transfer their risk
estimation to their patients.


Strengths


This study is the first to provide data from a real clinical setting
targeting the use of different formats in communication of CVRF between
physicians and patients. We chose a study design that optimally handles
desirability or reporting bias, in contrast to studies using
self-reported data from questionnaires, as our main covariate (i.e. risk
communication format) is not self-reported but recorded and objectively
analyzed.


Limitations


Due to the small sample size, especially the number of GPs, the
generalizability of our results is limited. The participation rate of
doctors (1.9%) was low. We think that GPs are not used to making audio
recordings of their own consultations, and this may be a major
psychological barrier for participation. The drop-out rate of 37% was
explained by the physicians who were willing to participate but unable
to enroll patients because of their high work load and lack of time.
Physician age, gender, and experience in ambulatory primary care in our
sample matches well with the statistics reported by the FMH, but the
sample size is far too small to be representative of the entire
population of Swiss primary care doctors. Moreover, the recruitment
strategy might have led to selection bias towards a group of doctors
with more motivation and skills in communication techniques. However,
this bias may support our findings as one would expect that less
motivated colleagues might also be less likely to catch up with the
recent recommendations of their medical associations.


Possible Explanations of the Findings


One reason for the high extent of verbal qualifier use compared to
other formats might be that primary care physicians are not sure about
the numerical facts regarding risk. Several studies suggest that,
regardless of specialization, physicians often face problems
understanding medical statistics.^[41–43]
Also, many of the physicians may not know the effects of different
formats of risk communication on patient understanding and their
emotional perception of risk. Many physicians might judge risk
presentation as verbal "value" to be good enough, not being aware of the
low correlation between objective risks and perceived risk in patients.
Similarly, the need and desire of patients to obtain maximum
information, which was also shown in our study by the API, might be
underestimated by physicians.
A second reason is that teaching activities regarding communication
skills are still at a low level in the professional education of Swiss
physicians. Consequently, many doctors do not feel comfortable with
different communication techniques. Furthermore, analogous to findings
in the use of calculation and communication tools to predict
cardiovascular risk in primary care, doctors face many barriers, such as
distrust in their validity or their contribution to encourage decision
making, lack of time, and low reimbursement.^[44]

Another issue contributing to our observations is the trust and confidence
of patients in their doctors. As we know from research in the field of
shared decision making, many patients prefer not to decide themselves,
but to leave the decision to the doctor,^[13] or prefer to know his opinion rather than being informed of facts.^[11] Asked for a personal opinion,
the physician may prefer verbal communication formats to others, taking
the risk of inducing different meanings and confusion in patients.
Similarly, by purposes like reassuring or persuading, GPs may prefer
verbal formats, as was shown e.g. for clinical geneticists.^[45]
The asymmetry in mean talking time indicates a more one-sided sort of
communication (with the physician as a source of information), rather
than an interactive discussion among equal partners. Moreover, the
straightforward use of verbal qualifiers by doctors is a time-saving
approach not requiring any preparation time, unlike choosing adequate
numbers or drawing graphs.
At present, several programs are in development in Switzerland,
yielding to patient-centered health-promoting activities in primary care
settings.^[46]
Within these programs, risk communication is a major issue. Visual
formats like colour-coded graphics are emphasized, in order to transport
information about risk to patients in the most understandable way and
to facilitate discussion about risk between physician and patient.^[47]Conclusion




In summary, our data demonstrate a gap between the recommendations
from medical associations and clinical reality in communicating CVD
risk. The verbal formats that are mainly used are rated lowest in
recommendations regarding understanding and effectiveness. Similarly,
the highly recommended formats, such as natural frequencies and visual
formats like bar charts, were rarely or never used by the primary care
physicians in our study. Visual formats resulted in significantly higher
subjective understanding of the information given. Also, gender is
significantly associated with the choice of communication formats, which
was unknown thus far for CVD risk communication.
As implication for clinical practice, relevant barriers towards the
use of "highly ranked" communication formats among doctors and patients
are to be identified and addressed. Furthermore, strategies for
improving communication skills among doctors should be developed, such
as practicable tools and medical education programs. The gender aspect
should be addressed, especially in regard to target groups of
interventions. The results of our study stress the need for developing
health-promoting programs in primary care that would more clearly focus
on a transparent communication on risk. Our final intention is to close
the gap between theory and daily clinical practice in the field of CVRF
communication, which does not necessarily mean that doctors and patients
have to change their communication style - it might lead to input for
the further development of theoretical models adapted to reality.


References

1. Lipkus IM: Numeric, verbal, and visual formats of conveying health risks: suggested best practices and future recommendations. Med Decis Making 2007, 27:696–713.
   
2. Edwards A, Elwyn G, Mullay A: Explaining risks: turning numerical data into meaningful pictures. Brit Med J 2002, 324:827–30.
   
3. Alaszewski A, Horlick-Jones T: How can doctors communicate information about risk more effectively? Brit Med J 2003, 326:728–731.
   
4. Price M, Cameron R, Butow P: Communicating risk
   information: The influence of graphical display format on quantitative
   information perception - accuracy, comprehension and preferences. Pat Educ Couns 2007, 69:121–28.
   
5. Timmermans DRM, Ockhuysen-Vermey CF, Henneman L:
   Presenting health information in different formats: The effect on
   participants' cognitive and emotional evaluation and decisions. Pat Educ Couns 2008, 73:443–47.
   
6. Fagerlin A, Zikmund-Fisher BJ, Ubel PA: "If I'm
   better than average, then I'm ok?": Comparative information influences
   beliefs about risk and benefits. Pat Educ Couns 2007, 69:140–44.
   
7. Goodyear-Smith F, Arroll B, Chan L, Jackson R, Wells
   S, Kenealy T: Patients prefer pictures to numbers to express
   cardiovascular benefit from treatment. Ann Fam Med 2008, 6:213–7.
   
8. Fuller R, Dudley N, Blacktop J: Risk communication
   and older people - understanding of probability and risk information by
   medical inpatients aged 75 years and older. Age Ageing 2001, 30:473–6.
   
9. Reyna VF, Nelson WL, Han PK, Dieckham NF: How numeracy influences risk comprehension and medical decision making. Psychol Bull 2009, 135(6):943–73.
   
10. Edwards AGK, Matthews E, Pill R, Bloor M: Communication about risk: diversity among primary care professionals. Fam Pract 1998, 15:296–300.
    
11. Van Steenkiste B, van der Weijden T, Stoffers HEJH:
    Improving cardiovascular risk managment: a randomized controlled trial
    on the effect of a decision support tool for patients and physicians. Eur J Cardiovasc Prev Rehabil 2007, 14:44–50.
    
12. O'Connor AM, Stacey D, Entwistle V: Decision aids for people facing health treatment or screening decisions. Cochrane Database Syst Rev 2003, CD001431.
    
13. Levinson W, Kao A, Kuby A, Thisted R: Not all
    patients want to participate in decision making. A National Study of
    Public Preferences. J Gen Intern Med 2005, 20:531–35.
    
14. Politi MC, Han PKJ, Col NF: Communicating the uncertainty of harms and benefits of medical interventions. Med Decis Making 2007, 27:681–695.
    
15. American Heart Association. Heart and Stroke Statistics: 2008 Update. Dallax, TX: American Heart Association; 2008.
    
16. Deepak L, Bhatt et al for the REACH Register
    Investigators: International Prevalence, Recognition and Treatment of
    Cardiovascular Risk Factors in Outpatients with Atherothrombosis. J Amer Med Assoc 2006, 295:180–89.
    
17. Boersma E, Keil U, De Bacquer D, De Backer G,
    Pyörälä K, for the EUROASPIRE I and II Study Groups: Blood pressure is
    insufficiently controlled in European patients with established coronary
    heart disease. J Hypertens 2003, 21:1831–1840.
    
18. Turnball F: Managing cardiovascular risk factors: The gap between evidence and practice. PloS Med 2005, 2(5):e131.
    
19. Van der Weijden T, Bos LBJ, Koelewijn-van Loon MS:
    Primary care patients' recognition of their own risk for cardiovascular
    disease: implications for risk communication in practice. Curr Opinion in Cardiology 2008, 23:471–76.
    
20. Sheridan SL, Viera AJ, Krantz MJ, Ice CL, Steinmann LE, et al.: The effect of giving global coronary risk information to adults. Arch Intern Med 2010, 170(3):230–39.
    
21. Waldron CA, van der Weijden T, Ludt S, Gallacher J,
    Elwyn G: What are effective strategies to communicate cardiovascular
    risk information to patients? A systematic review. Patient Educ Couns 2010.
    
22. Ende J, Kazis L, Ash A, Moskowitz MA: Measuring
    patients' desire for autonomy: decision making an information-seeking
    preferences among medical patients. J Gen Intern Med 1989, 4(1):23–30.
    
23. Giersdorf N, Loh A, Bieber C: Development and validation of assessment instruments for shared decision making. Bundesgesundheitsbl - Gesundheitsforsch - Gesundheitsschutz. Springer Verlag 2004, 47:969–976.
    
24. Sandhu H, Adams A, Singleton L, Clark-Carter D, Kidd
    J: The impact of gender dyads on doctor-patient communication: a
    systematic review. Pat Educ Couns 2009, 76(3):348–55.
    
25. Weinstein ND, Sandman PM, Roberts NE: Communicating effectively about risk magnitudes. New Brunswick (NJ): Environmental Communication Research Program, Cook College, Rutgers University 1989.
    
26. Sandman PM, Weinstein ND, Miller P: High risk or low: How location on a risk ladder affects percieved risk. Risk Anal 1984, 14:35–45.
    
27. Stone ER, Yates JF, Parker AM: Effects of numerical and graphical display on a professed risk-taking behaviour. J Exp Psychol Appl 1997, 3:243–56.
    
28. Hollands JG, Spence I: Judging proportions with graphs: the summation model. Appl Cogn Psychol 1998, 12:173–90.
    
29. National Cancer Institute, Officer of Cancer Communications: How the public percieves, processes and interprets risk information: findings from focus group research with the general public. Bethesda (MD): National Cancer Institute; 1998. Report-No POS-T 086
    
30. Lipkus IM, Hollands JG: The visual communication of risk. J Nat Cancer Inst Monogr 1999, 25:149–63.
    
31. Aucker J, Senathirajah Y, Kukafka R, Starren J: Design feature of graphs in health risk communication: a systematic review. J Am Med Inform Assoc 2006, 13:608–18.
    
32. Han P, Klein W, Lehman T, Massett H, Lee S, Freedman
    A: Laypersons' responses to the communication of uncertainty regarding
    cancer risk estimates. Med Decis Making 2009, 29:391–403.
    
33. Jarvenpaa S: Graphic displays in decision making - the visual salience effect. J Behav Decis Making 1990, 3:247–62.
    
34. Schirillo J, Stone E: The greater ability of
    graphical versus numerical displays to increase risk avoidance involves a
    common mechanism. Risk Anal 2005, 25:555–66.
    
35. Chua HF, Yates JF, Shah P: Risk avoidance: graphs versus members. Mem cogn 2006, 34:399–410.
    
36. Stone E, Sieck W, Bull B, Yates J, Parsh S, Rush C:
    Foreground to background salience: explaining the effects of graphical
    displays on risk avoidance. Organ Behav Hum Decis Proc 2003, 90:19–36.
    
37. Armstrong K, Fitzgerald G, Schwartz JS, Ubel PA:
    Using survival curve comparisons to informed patient decision making:
    can a practice exercise improve understanding? J Gen Intern Med 2001, 16:482–5.
    
38. Parrott R, Silk K, Dorgan K, Condit C, Harris T:
    Risk comprehension and judgements of statistical evidentiary appeals:
    when a picture is not worth thousand words. Health Comm Res 2005, 31:423–52.
    
39. Timmermans DRM, Ockhuysen-Vermey CF, Henneman L:
    Presenting health risk information in different formats: The effect on
    participants' cognitive and emotional evaluation and decisions. Pat Educ Couns 2008, 73:443–47.
    
40. Hill S, Spink J, Cadilhac D, Edwards A, Kaufman C,
    Rogers S, Ryan R, Tonkin A: Absolute risk representation in
    cardiovascular disease prevention: comprehension and preferences of
    health consumers and general practitioners involved in a focus group
    study. BMC Public Health 2010, 10:108.
    
41. Casscells W, Schoenberger A, Grayboys T: Interpretation by physicians of clinical laboratory results. N Engl J Med 1978, 299:999–1000.
    
42. Gigerenzer G, Gaissmaier W, Kurz-Milcke E, Schwartz
    LM, Woloshin S: Helping doctors and patients to make sense of health
    statistics. Psychological Science in the Public Interest 2007, 8:53–96.
    
43. Kalet A, Roberts JC, Fletcher R: How do physicians talk with their patients about risks? J Gen Intern Med 1994, 9:402–404.
    
44. Eichler K, Zoller M, Tschudi P, Steurer J: Barriers to apply cardiovascular prediction rules in primary care: a postal survey. BMC Fam Pract 2007, 8:1.
    
45. Henneman L, Marteau TM, Timmermanns TRM: Clinical
    geneticists' and genetic counselors' views on the communication of
    genetic risks: a qualitative study. Pat Educ and Couns 2008, 73:42–49.
    
46. Neuner-Jehle S, Grueninger U, Hoesli R, Schmid M,
    Somaini B: A multidimensional health promoting counseling program in
    primary care setting: Coaching Your Health. Eur J Public Health 2006, 16(Suppl 1):80–81.
    
47. Hill Smith I, Mathie E, Little P: Involving patients in decisions about preventive medication: a focus group study. Prim Care Cardiovasc J 2010, 3:51–56.