Influence of Cardiovascular Risk Communication Tools and Presentation Formats on Patient Perceptions and Preferences | Cardiology | JAMA Cardiology | JAMA Network
[Skip to Content]
[Skip to Content Landing]
Figure 1.  Patient-Perceived Risk and Willingness for Drug Therapy
Patient-Perceived Risk and Willingness for Drug Therapy

Patient-perceived risk severity (A) and willingness for drug therapy (B) when shown a 10-year atherosclerotic cardiovascular disease (ASCVD) death risk of 4%, a 10-year ASCVD risk of 15%, or a lifetime ASCVD risk of 50%.

aP <.001.

Figure 2.  Perceived Risk and Willingness for Drug Therapy by Risk Estimate
Perceived Risk and Willingness for Drug Therapy by Risk Estimate

Perceived risk (A) and willingness for therapy (B) by risk estimate when shown with text only, bar graphs, or a pictogram.

aP <.001.

Table 1.  Characteristics of Patients Participating in PALM Registry Survey Risk Questions
Characteristics of Patients Participating in PALM Registry Survey Risk Questions
Table 2.  Association of Education, Numeracy, Age, Prior ASCVD, and Statin Use With Perceived Risk and Therapy Willingnessa
Association of Education, Numeracy, Age, Prior ASCVD, and Statin Use With Perceived Risk and Therapy Willingnessa
1.
The Expert Panel.  Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults.  Arch Intern Med. 1988;148(1):36-69. doi:10.1001/archinte.1988.00380010040006PubMedGoogle ScholarCrossref
2.
Stone  NJ, Robinson  JG, Lichtenstein  AH,  et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  J Am Coll Cardiol. 2014;63(25 Pt B):2889-2934. doi:10.1016/j.jacc.2013.11.002PubMedGoogle ScholarCrossref
3.
Whelton  PK, Carey  RM, Aronow  WS,  et al.  2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines.  Hypertension. 2018;71(6):1269-1324. doi:10.1161/HYP.0000000000000066PubMedGoogle ScholarCrossref
4.
Catapano  AL, Graham  I, De Backer  G,  et al; Authors/Task Force Members.  2016 ESC/EAS guidelines for the management of dyslipidaemias: the Task Force for the Management of Dyslipidaemias of the European Society of Cardiology (ESC) and European Atherosclerosis Society (EAS) developed with the special contribution of the European Association for Cardiovascular Prevention & Rehabilitation (EACPR).  Atherosclerosis. 2016;253:281-344. doi:10.1016/j.atherosclerosis.2016.08.018PubMedGoogle ScholarCrossref
5.
Zipkin  DA, Umscheid  CA, Keating  NL,  et al.  Evidence-based risk communication: a systematic review.  Ann Intern Med. 2014;161(4):270-280. doi:10.7326/M14-0295PubMedGoogle ScholarCrossref
6.
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(3):213-217. doi:10.1370/afm.795PubMedGoogle ScholarCrossref
7.
Edwards  A, Thomas  R, Williams  R, Ellner  AL, Brown  P, Elwyn  G.  Presenting risk information to people with diabetes: evaluating effects and preferences for different formats by a web-based randomised controlled trial.  Patient Educ Couns. 2006;63(3):336-349. doi:10.1016/j.pec.2005.12.016PubMedGoogle ScholarCrossref
8.
Stacey  D, Légaré  F, Lewis  K,  et al.  Decision aids for people facing health treatment or screening decisions.  Cochrane Database Syst Rev. 2017;4:CD001431.PubMedGoogle Scholar
9.
Koelewijn-van Loon  MS, van der Weijden  T, Ronda  G,  et al.  Improving lifestyle and risk perception through patient involvement in nurse-led cardiovascular risk management: a cluster-randomized controlled trial in primary care.  Prev Med. 2010;50(1-2):35-44. doi:10.1016/j.ypmed.2009.11.007PubMedGoogle ScholarCrossref
10.
Krones  T, Keller  H, Sönnichsen  A,  et al.  Absolute cardiovascular disease risk and shared decision making in primary care: a randomized controlled trial.  Ann Fam Med. 2008;6(3):218-227. doi:10.1370/afm.854PubMedGoogle ScholarCrossref
11.
Karmali  KN, Persell  SD, Perel  P, Lloyd-Jones  DM, Berendsen  MA, Huffman  MD.  Risk scoring for the primary prevention of cardiovascular disease.  Cochrane Database Syst Rev. 2017;3:CD006887.PubMedGoogle Scholar
12.
Navar  AM, Wang  TY, Li  S,  et al.  Lipid management in contemporary community practice: results from the Provider Assessment of Lipid Management (PALM) Registry.  Am Heart J. 2017;193:84-92. doi:10.1016/j.ahj.2017.08.005PubMedGoogle ScholarCrossref
13.
Navar  AM, Wang  TY, Goldberg  AC,  et al.  Design and rationale for the Patient and Provider Assessment of Lipid Management (PALM) registry.  Am Heart J. 2015;170(5):865-871. doi:10.1016/j.ahj.2015.08.002PubMedGoogle ScholarCrossref
14.
Fagerlin  A, Zikmund-Fisher  BJ, Ubel  PA, Jankovic  A, Derry  HA, Smith  DM.  Measuring numeracy without a math test: development of the Subjective Numeracy Scale.  Med Decis Making. 2007;27(5):672-680. doi:10.1177/0272989X07304449PubMedGoogle ScholarCrossref
15.
Navar  AM, Pencina  MJ, Mulder  H, Elias  P, Peterson  ED.  Improving patient risk communication: translating cardiovascular risk into standardized risk percentiles.  Am Heart J. 2018;198:18-24. doi:10.1016/j.ahj.2017.12.005PubMedGoogle ScholarCrossref
16.
Mann  DM, Ponieman  D, Montori  VM, Arciniega  J, McGinn  T.  The statin choice decision aid in primary care: a randomized trial.  Patient Educ Couns. 2010;80(1):138-140. doi:10.1016/j.pec.2009.10.008PubMedGoogle ScholarCrossref
Original Investigation
December 2018

Influence of Cardiovascular Risk Communication Tools and Presentation Formats on Patient Perceptions and Preferences

Author Affiliations
  • 1Duke Clinical Research Institute, Duke University School of Medicine, Durham, North Carolina
  • 2University of Iowa, Iowa City
  • 3Baylor College of Medicine, Houston, Texas
  • 4Mayo Clinic, Rochester, Minnesota
  • 5Emory University, Atlanta, Georgia
  • 6Washington University in St Louis, St Louis, Missouri
JAMA Cardiol. 2018;3(12):1192-1199. doi:10.1001/jamacardio.2018.3680
Key Points

Question  How do different presentation formats for atherosclerotic cardiovascular disease (ASCVD) risk influence patient perceptions and treatment preferences?

Findings  In this randomized survey study, risk perception and therapy willingness were highest when patients were shown estimates for lifetime risk, followed by 10-year ASCVD risk and then 10-year cardiovascular disease death risk. Using a pictogram led to lower risk perception and therapy willingness than a bar graph or no graphic.

Meaning  Patient perceptions of ASCVD disease severity and treatment preferences are influenced by the tool used for risk communication, and risk perception influences a patient’s willingness to consider therapy.

Abstract

Importance  Practice guidelines recommend that clinicians engage patients in treatment decisions and explain atherosclerotic cardiovascular disease (ASCVD) risk but do not describe how to communicate this risk most effectively.

Objective  To determine how the ASCVD risk time horizon, outcome, and presentation format influence risk perceptions and treatment preferences.

Design, Setting, and Participants  From May 27, 2015, through November 12, 2015, participants from the Patient and Provider Assessment of Lipid Management Registry at 140 US cardiology, primary care, and endocrinology practices were presented 3 independent scenarios (representing the same hypothetical patient) and asked to rate their perceived risk and willingness to take medication to lower risk in light of (1) a 15% 10-year ASCVD event risk, (2) a 4% 10-year cardiovascular disease (CVD) death risk, and (3) a 50% lifetime ASCVD event risk.

Exposures  Participants were randomized to receive risk estimates using numbers only, a bar graph, or a face pictogram.

Results  Of 3566 eligible participants, 2708 (76.9%) responded (median age, 67 years [interquartile range, 61-76 years]; 280 [10.3%] African American; 1491 men [55.1%]). When shown the lifetime ASCVD risk, respondents were more likely to consider the risk “high to very high” than when presented the 10-year ASCVD risk or the CVD death risk (70.1% vs 31.4% vs 25.7%, respectively; both P < .001). Treatment willingness was also the highest for lifetime ASCVD risk (77.9% very willing) followed by 10-year ASCVD risk (68.1%) and 10-year CVD death risk (63.1%; both P < .001). Compared with participants who were shown a bar graph or no graphic, those who were shown the risk information with a pictogram had the lowest perception of disease severity and the lowest willingness to consider therapy. These findings were robust across demographic and socioeconomic subgroups.

Conclusions and Relevance  The format, time horizon, and outcome used for risk estimation influence patient perceptions and should be considered when designing risk communication tools. When shown lifetime risk estimates, patients had higher risk perception and willingness for therapy than when shown 10-year estimates. Pictogram risk displays may decrease risk perception and consideration for treatment.

Introduction

Patient cardiovascular risk estimates are increasingly recommended to guide treatment decisions for atherosclerotic cardiovascular disease (ASCVD) prevention, including statin initiation and blood pressure treatment.1-3 Current practice guidelines recommend that clinicians engage patients as active participants in treatment decisions. Such shared decision making requires that patients understand their individual risk and the potential benefit of prevention treatments.2 While current guidelines recommend discussing individual ASCVD risks with patients, less guidance is provided on the optimal risk estimate to provide and how to display it.

Many ASCVD risk calculators are available. The current US cholesterol and hypertension guidelines emphasize a 10-year risk of myocardial infarction or stroke based on the Pooled Cohorts Equations but also recommend considering treatment for those with a high lifetime risk of disease.2 Current European cholesterol guidelines emphasize the Systematic Coronary Risk Evaluation risk, which estimates a patient’s risk for cardiovascular disease (CVD) death in the next 10 years.4 Regardless of the tool used, the data presentation format may influence patient perceptions. Many individuals have a limited understanding of numeracy and probability; therefore, studies have previously suggested that incorporating visual aids to portray risks (bar charts or face pictograms) may improve patient understanding and satisfaction.5-7

To date, research on risk communication has focused on the influence of risk communication approaches with patients’ overall risk perception, knowledge, and satisfaction.8-10 A recent systematic review suggests that communicating ASCVD risk may improve the treatment of ASCVD risk factors, but the authors noted that research remains limited.11

Given that risk estimates are increasingly recommended to guide therapy, we sought to evaluate how different risk estimation and presentation formats influenced patients’ perception of risk and patients’ willingness to initiate medication therapy. Specifically, we determined (1) whether an individual’s perception of disease risk severity varied depending on whether he or she was presented with an estimate of 10-year ASCVD risk, lifetime ASCVD risk, or 10-year risk of CVD death for a given patient risk factor scenario; (2) whether changing these risk estimate types influenced an individual’s reported willingness to take medication therapy to reduce the risk of ASCVD; (3) whether the format of visual aids that accompanied the risks estimate would alter either disease perception and/or treatment preferences; and (4) the degree to which these conclusions were robust across various patient age, education, or numeracy levels.

Methods

The Patient and Provider Assessment of Lipid Management (PALM) registry was conducted across 140 US cardiology, primary care, and endocrinology practices from May 27, 2015, through November 12, 2015.12,13 Participants in the registry provided signed informed consent at the time of their physician visit, after which they completed an electronic survey that was administered via an iPad (Apple). The survey was designed to collect information about participants’ prior statin use and beliefs about ASCVD (eFigure 1 in the Supplement). Patient characteristics, including education, income, and race, were obtained by self-report. Numeracy was assessed using the Subjective Numeracy ability subscale, a 4-item scale that assesses an individual’s perceived numerical ability.14 Medical record abstractions for clinical characteristics were performed by trained study coordinators at each site. Sites obtained local institutional review board approval. The PALM Registry was registered in clinicaltrials.gov (NCT02341664).

Overall, 3566 of the PALM participants were randomly selected to answer a series of questions about ASCVD risk based on a hypothetical patient with an elevated 10-year and lifetime risk. First, each patient was asked to imagine that their physician told them that they had a 15% chance of a heart attack or stroke in the next 10 years. They were then asked to rate the severity of the risk estimate using a sliding scale (very low, low, medium, high, and very high), and to indicate their willingness to take a medication to reduce their risk of disease by approximately one-third (very unwilling, slightly unwilling, possibly, somewhat willing, or very willing). Next, the patient was asked the same questions, but instead of a 15% 10-year risk of stroke or myocardial infarction (MI), he or she was asked to consider a 4% chance of CV death in 10 years. Finally, the questions were repeated for a lifetime risk of stroke or MI of 50%. These scenarios were presented independently. However, the scenarios were constructed based on a hypothetical individual with high short- and long-term risk. The thresholds were chosen to represent an individual at high short- and long-term risk and for whom current US guidelines would recommend statin therapy. Among those presented with these survey questions, participants were further randomized to be presented with the risk scenarios in 1 of 3 different formats: (1) questions presented as text only, (2) questions presented with an accompanying bar graph, or (3) questions presented with an accompanying face pictogram (eFigure 2 in the Supplement). All PALM surveys were conducted on an application that was designed for this study, which took patients through the informed consent process and the survey; the randomization was conducted by the application itself. The sample size for this analysis was determined by the sample size of the overall PALM registry. We prespecified that half of the participants would receive these risk questions, with equal distribution into the 3 arms (bar graph, pictogram, or text).

Participants who skipped the risk questions or who reported “I don’t know” or “I don’t understand” were excluded from analyses that compared risk perception and therapy willingness. The proportion of participants who reported their perceived risk as high or very high was compared for 10-year ASCVD risk, Systematic Coronary Risk Evaluation CVD death risk, and lifetime ASCVD risk information. Similarly, the proportions of adults who reported being “willing or very willing” to take therapy based on that risk were compared across risk information groups using McNemar tests. To determine the association between perceived risk and the willingness to take therapy, the odds that a participant who reported a willingness for therapy given a high to very high perceived risk were calculated within each scenario using a conditional logistic regression to account for a within-participant correlation.

The potential effect of risk presentation format was evaluated by comparing the proportions of participants who reported high to very high perceived risk and willingness for therapy when shown information using a bar graph, a face pictogram, or text alone for each question. A logistic regression was used to estimate the differences in the odds of reporting high perceived risk or high therapy willingness by type of visual support used, stratified by age, education, numeracy, ASCVD status, and statin use. The association of age (< or ≥65 years), numeracy (top 50% vs bottom 50%), education (at least some college vs no college), ASCVD status, and statin treatment status with risk perception and therapy willingness was evaluated using a logistic regression. Interaction terms were created to assess whether differences in risk perception or therapy willingness as seen by risk horizons varied by age, education, numeracy, ASCVD, or statin use.

To account for multiple comparisons, a 2-tailed test with an α of .01 was used to establish statistical significance. For this analysis, a prespecified statistical analysis plan was followed, with primary analyses conducted between November 2016 and January 2017.

Results

Of the 3566 PALM participants who were randomized for this analysis, 1022 (28.7%) were shown questions with text alone, 1489 (41.8%) were shown a bar graph, and 1046 (29.3%) were shown face pictograms. Of these, 858 participants (24.1%) either skipped the risk questions or marked “I don’t know” or “I don’t understand” for a final sample size of 2708. The characteristics of those participants are presented in Table 1. Among responders, the median age was 67.0 years (interquartile range [IQR], 61.0-76.0 years), 1491 (55.1%) were men, 1249 (46.1%) had prior ASCVD, 2366 (87.4%) self-reported as being white, and 1715 (63.3%) had private insurance.

Information about those who did and did not answer the questions about risk (defined as complete responses to all the questions regarding risk) is also shown in Table 1. Those who did not respond to the risk questions were older (median age, 69 years [IQR, 61-76 years] vs 67 years [IQR, 59-74 years]; P < .001), had lower subjective numeracy scores (median, 15 [IQR, 9.3-20] vs 17 [IQR, 12-21]; P < .001), were more often African American (16.1% vs 10.3%; P < .001), completed less education (59.2% had at least some college vs 65.0%; P < .001), and were less likely to have private insurance (58.0 vs 63.3%; P = .007). There was no difference in aspirin (52.0 vs 48.0%; P = .04) or statin (73.9 vs 69.9; P = .02) use between the 2 groups. There was no difference in survey response rates by which format patients were presented (no graphic [74.1%], bar graph [77.7%], or pictogram [75.2%]; P = .09), nor were there statistically significant differences in the characteristics of patients who were randomized to the pictogram, bar graph, or text-only version of the survey (eTable 1 in the Supplement). Many patients had partial responses to the survey, therefore the response rates for individual questions were higher than the response rate for the complete set of risk questions. eTable 2 in the Supplement shows question-specific response rates. The characteristics of subgroups of participants who were presented with face pictograms, bar graphs, and text-only versions of the survey, as well as the characteristics of nonresponders by version type, are shown in eTables 3 to 5 in the Supplement.

Figure 1 shows the participant-perceived risk severity and willingness for therapy for each risk estimate (15% 10-year ASCVD risk, 50% lifetime ASCVD risk, and 4% 10-year CVD death risk). Participants reported the lowest perceived risk when shown risk of CVD death and the highest perceived risk when shown lifetime ASCVD risk; 2553 of 3356 respondents (70.1%) perceived a lifetime risk of 50% to be “high to very high” compared with 982 of 3132 respondents (31.4%) when shown a 10-year ASCVD risk of 15% and 844 of 3280 respondents (25.7%) when shown a 10-year CVD death risk of 4%. Similarly, participants were most likely to report being willing to take medication therapy to lower their risk of heart attack or stroke when presented with a 50% lifetime ASCVD risk. However, the differences by time horizon were attenuated for therapy willingness compared with what was observed for risk perception. When shown a 50% lifetime risk, 2621 of 3363 respondents (77.9%) reported willingness to take medication therapy compared with 2214 of 3252 respondents (68.1%; P < .001) when shown a 15% 10-year risk and 2097 of 3324 respondents (63.1%) when shown a 4% 10-year CVD death risk (P < .001).

Participants who reported a higher perceived risk also reported an increased willingness to take therapy to reduce risk. When shown a 10-year ASCVD risk of 15%, participants who reported their perceived risk to be high to very high had a 2.1-fold increased odds of also reporting being very willing to take therapy to lower that risk (odds ratio [OR], 2.1; 95% CI, 1.86-2.44). Similar associations were seen between perceived risk and willingness for therapy when shown a 4% 10-year CVD death risk (OR, 3.45; 95% CI, 3.00-3.98) and when shown a 50% ASCVD lifetime risk (OR, 3.27; 95% CI, 2.89-3.70).

Subgroup Analysis by Age, Education, Numeracy, ASCVD Status, and Statin Use

The differences in risk perception and therapy willingness were compared by education, numeracy, age, ASCVD status, and statin use (Table 2). Statin use (OR, 1.27; 95% CI, 1.14-1.41), prior ASCVD (OR, 1.28; 95% CI, 1.16-1.39), younger age (OR, 1.20; 95% CI, 1.09-1.32), and higher educational levels (OR, 1.18; 95% CI, 1.06-1.30) were all associated with an increased perceived risk. Statin use (OR, 2.13; 95% CI, 1.85-2.44), prior ASCVD (OR, 1.61; 95% CI 1.43-1.82), higher educational levels (OR, 1.33; 95% CI, 1.16-1.52), and higher numeracy scores (OR, 1.45; 95% CI, 1.28-1.64) were associated with an increased willingness for therapy. The association between these factors and the perceived risk and willingness for therapy was generally consistent by the risk estimate used (ie, younger adults were more likely to report risk as high compared with older adults across all 3 scenarios; 10-year ASCVD, 33.8% vs 29.7%; 10-year CVD death, 27.4% vs 24.6%; or lifetime ASCVD risk, 74.3% vs 67.4%).

In formal interaction testing, the magnitude of the differences in risk perception by scenario, but not the directionality of those differences, varied by education and numeracy but not by age, prior ASCVD, or statin use. Specifically, across educational groups perceived risk was higher for 10-year ASCVD risk (<college, 318 [29.9%]; at least some college, 655 [32.3%]) compared with 10-year CVD death risk (<college, 315 [27.8%]; at least some college, 518 [24.6%]), but that difference was greater in those with at least some college (interaction P = .002). Similarly, the perceived risk was higher with lifetime risk compared with 10-year ASCVD risk across both educational groups (<college: lifetime risk, 771 [62.5]; at least some college: lifetime risk, 1616 [74.8%]) and numeracy levels (bottom 50%: lifetime risk, 1110 [66.7%]; top 50%: lifetime risk, 1236 [74.4%]). However, this difference was amplified in those with higher numeracy (interaction P < .001) and higher education (interaction P < .001). There was no interaction between the effect of the risk estimate provided and the willingness for therapy by education, numeracy, age, prior ASCVD, or statin use.

Format of Risk Presentation

Figure 2 shows the proportions of participants who reported a high perceived risk or high willingness for drug therapy when shown a face pictogram, bar graph, or no graphic, stratified by each risk estimate horizon used. When risk estimates were shown with a pictogram, the perceived risk was consistently lower than when presented as a bar graph or without a graphic for all 3 risk estimate horizons. For example, when asked about a 10-year ASCVD event risk of 15%, 204 of 924 (22.1%) who were shown a pictogram reported a high perceived risk compared with 319 of 892 (35.8%) who were shown no graphic and 459 of 1315 (34.9%) who were shown a bar graph (P < .001 for both). Similarly, approximately 10% fewer adults who were shown a pictogram reported a high perceived risk than those who were shown a bar graph or no graphic for lifetime ASCVD risk and 10-year CVD death risk. There was no significant difference in perceived risk between groups that were shown a bar graph or no graphic. Similarly, the willingness for drug therapy was somewhat lower when presented with a face pictogram compared with a bar graph, although the magnitude of the difference was lower. Across all 3 risk scenarios, 5% to 6% more adults reported a high therapy willingness when shown a bar graph compared with a pictogram.

Discussion

There are multiple means to provide risk estimates to adults to estimate their risk for CVD events, including a 10-year risk of ASCVD event, lifetime risk of ASCVD events, and risk of CVD death. In a representative community-based sample, we found that the perception of the severity of risk as well as the potential willingness to take medication therapy to lower that risk varied by which calculator was used as well as how the calculator was presented. The severity of risk perception and willingness to take therapy were highest when individuals were shown a 10-year lifetime ASCVD risk of 50% and lowest when shown a 10-year risk of CVD death of 4%, even though these estimates could represent the same patient. When the same disease risk estimates were shown with a face pictogram compared with a bar graph or no graphics, participants had significantly lower disease risk perception and willingness to take therapies.

Although the survey was piloted for understanding, many individuals (up to 1 in 4) were unable to answer or understand the questions about risk. While we cannot fully determine the proportion of patients who skipped these questions because of a lack of understanding vs a desire to complete the survey more quickly, few (<2%) skipped the demographic questions about insurance or education, which were placed after the risk questions in the survey. The higher rate of missing responses for these questions may have been due to poor comprehension of the concept of risk estimation. Individuals with missing responses were older; had lower education levels, income, and numeracy; and were less likely to have private insurance. However, more than half of those who skipped the risk questions had at least some college, and there was considerable overlap in numeracy scores between the 2 groups. Thus, clinicians cannot rely on objective measures of education to determine who understands concepts of risk. When discussing risk with their patients, regardless of the risk horizon estimate used, clinicians should explain risk in qualitative terms (eg, high vs low) or put risk into context using vascular age or standardized risk percentiles.15

Individual characteristics, such as age, education, and numeracy, may affect not only risk comprehension but also the qualitative interpretation of risk. Across all risk horizons estimates, therapy willingness was generally higher in those who had at least some college compared with no college and in those in the top half of numeracy compared with the bottom half. Younger adults generally reported a greater perceived risk severity than older adults, but this did not translate into differences in the willingness to take therapy by age group. Although adults were asked to consider a hypothetical patient, those with prior ASCVD had a consistently higher perceived risk and therapy willingness than those without.

Perceived disease risk varied considerably when individuals were presented risk estimates that were generated from different risk tools. This underscores how the qualitative perception of risk can vary even when “accurate” or “unbiased” risk estimates are used. A clinician who communicates that an individual has a 50% chance of a heart attack or stroke in his or her lifetime may be as correct as a clinician who informs the same individual about a 4% chance of dying of a heart attack or stroke in the next 10 years, but the effect of these “equivalent” pieces of information on the patient may be markedly different. In general, a lifetime risk estimate in a person will be much higher than 10-year CVD risk up to age 55 years, which will be higher than the 10-year risk of CVD mortality. Our data suggest that individuals are most affected by the estimate that produces the highest absolute number.

Participants with a higher risk perception were more likely to report a higher willingness to take drug therapy to lower their risk for all risk estimates presented. As a result, the proportion of adults who reported a willingness to take drug therapy to lower their risk was highest for lifetime risk, followed by 10-year risk, and then mortality risk. Nearly 15% more participants reported a willingness to take drug therapy when shown a 50% lifetime risk than when shown a 4% chance of CVD death in the next 10 years. However, the differences in willingness to take drug therapy were lower than the differences in risk perception. Although only 844 of 3280 (25.7%) who were shown a 4% 10-year risk of CVD death reported risk to be high to very high, 2097 of 3324 (63.1%) reported a willingness to take therapy to lower risk. This demonstrates that perceived risk is only one of many factors that individuals consider when deciding whether to initiate preventive therapies.

Beyond the risk calculator that was used, the differences in how that risk is communicated to patients may influence patient-perceived risk and the willingness to engage initiate drug therapy. Across all scenarios, when participants were shown a risk estimate with a corresponding pictogram, the perceived risk and willingness for therapy was lower than when the risk estimate was shown as a bar graph. While pictograms may help individuals better understand the concept of a proportion, the number of “happy” faces in the diagram may have led to qualitatively lower risk estimates. This reinforces the need to test the influence of decision aids not only on patient satisfaction and risk understanding, but also on therapy uptake and adherence.16 In the future, guidelines around risk estimation may also consider providing evidence-based guidance around risk communication.

Limitations

This study has several limitations. First, all participants received the same 10-year and lifetime risk estimates that were meant to correspond with a high-risk patient rather than personalized risk scores. Second, we evaluated willingness for drug therapy based on a hypothetical medication that would “lower risk by about a third.” We did not specify a particular medication to prevent individual preconceived biases about specific therapies from influencing the results. Third, to focus on risk perception, we did not test different formats for showing therapeutic effectiveness (eg, relative vs absolute risk reduction or the number needed to treat), which may influence willingness for drug therapy. Fourth, we presented the risk estimates individually to patients, whereas in clinical practice clinicians may present multiple risk estimates. Finally, nearly 1 in 4 survey respondents skipped the questions about risk, with nonresponders having lower numeracy and education levels, higher age, and less likely to have private insurance. Thus, the generalizability of these findings to populations with less education, insurance, and overall numeracy may be limited.

Conclusions

Integrating risk-based treatment paradigms into clinical practice requires effective strategies to accurately communicate risk with patients. Individuals may perceive their risk to be higher and be more willing to engage in therapy when shown a lifetime risk compared with 10-year fatal or nonfatal risk estimates, regardless of education level, numeracy level, or age. Using face pictograms to display risk may lead to lower qualitative assessments of perceived risk. Effective risk communication tools should consider which risk score is used and how risk estimates are displayed.

Back to top
Article Information

Accepted for Publication: September 20, 2018.

Corresponding Author: Ann Marie Navar, MD, PhD, Duke Clinical Research Institute, Duke University School of Medicine, 2400 Pratt St, Durham, NC 27715 (ann.navar@duke.edu)

Published Online: November 7, 2018. doi:10.1001/jamacardio.2018.3680

Author Contributions: Dr Navar had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Wang, Robinson, Virani, Roger, Goldberg, Peterson.

Study concept and design: Navar.

Acquisition, analysis, or interpretation of data: Navar, Wang, Mi, Robinson, Roger, Wilson, Goldberg, Peterson.

Drafting of the manuscript: Navar.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Mi.

Obtained funding: Wang, Peterson.

Administrative, technical, or material support: Navar, Wilson.

Study supervision: Navar.

Supervision: Wang, Peterson.

Conflict of Interest Disclosures: Dr Navar reports receiving research support from Amgen, Sanofi, Amarin, Janssen, and Regeneron and consulting fees from NovoNordisk, Amarin, Amgen, and Sanofi. Dr Wang reports receiving research support from AstraZeneca, Boston Scientific, Bristol Myers Squibb, Daiichi Sankyo, Eli Lilly, Gilead, Pfizer, Regeneron and consultant/advisory/education fees from Merck, Gilead, and Sanofi. Dr Mi reports no relevant disclosures. Dr Robinson reports receiving research support from Amarin, Amgen, Astra-Zeneca, Eli Lilly, Esai, GlaxoSmithKline, Merck, Pfizer, Regeneron/Sanofi, and Takeda and consulting fees for Amgen, Eli Lilly, Merck, Pfizer, Regeneron, and Sanofi. Dr Virani reports receiving research support from the American Diabetes Association, American Heart Association, and the US Department of Veterans Affairs and honorarium from the American College of Cardiology and the National Lipid Association. Dr Goldberg reports receiving research support from Amarin, Amgen, Pfizer, and Regeneron/Sanofi; consulting fees from Regeneron/Sanofi and Esperion; and honorarium for editorial work for Merck Manual. Dr Peterson reports receiving research support from Amgen, AstraZeneca, Merck, and Sanofi and consulting fees for Amgen, AstraZeneca, Merck, and Sanofi Aventis.

Funding/Support: This study was supported by Sanofi Pharmaceuticals and Regeneron Pharmaceuticals. Dr Navar is supported by the National Institutes of Health, National Heart, Lung, and Blood Institute grant K01HL133416.

Role of the Funder/Sponsor: The sponsors contributed to the design of the study, interpretation of the data, and critical review of the manuscript, but played no role in data collection, management, or analysis. The decision to submit the manuscript for publication was controlled by authors at the Duke Clinical Research Institute.

Additional Contributions: Peter Hoffmann, BA, Duke Clinical Research Institute, provided editorial assistance. He was not compensated for his contributions.

References
1.
The Expert Panel.  Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults.  Arch Intern Med. 1988;148(1):36-69. doi:10.1001/archinte.1988.00380010040006PubMedGoogle ScholarCrossref
2.
Stone  NJ, Robinson  JG, Lichtenstein  AH,  et al; American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines.  J Am Coll Cardiol. 2014;63(25 Pt B):2889-2934. doi:10.1016/j.jacc.2013.11.002PubMedGoogle ScholarCrossref
3.
Whelton  PK, Carey  RM, Aronow  WS,  et al.  2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines.  Hypertension. 2018;71(6):1269-1324. doi:10.1161/HYP.0000000000000066PubMedGoogle ScholarCrossref
4.
Catapano  AL, Graham  I, De Backer  G,  et al; Authors/Task Force Members.  2016 ESC/EAS guidelines for the management of dyslipidaemias: the Task Force for the Management of Dyslipidaemias of the European Society of Cardiology (ESC) and European Atherosclerosis Society (EAS) developed with the special contribution of the European Association for Cardiovascular Prevention & Rehabilitation (EACPR).  Atherosclerosis. 2016;253:281-344. doi:10.1016/j.atherosclerosis.2016.08.018PubMedGoogle ScholarCrossref
5.
Zipkin  DA, Umscheid  CA, Keating  NL,  et al.  Evidence-based risk communication: a systematic review.  Ann Intern Med. 2014;161(4):270-280. doi:10.7326/M14-0295PubMedGoogle ScholarCrossref
6.
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(3):213-217. doi:10.1370/afm.795PubMedGoogle ScholarCrossref
7.
Edwards  A, Thomas  R, Williams  R, Ellner  AL, Brown  P, Elwyn  G.  Presenting risk information to people with diabetes: evaluating effects and preferences for different formats by a web-based randomised controlled trial.  Patient Educ Couns. 2006;63(3):336-349. doi:10.1016/j.pec.2005.12.016PubMedGoogle ScholarCrossref
8.
Stacey  D, Légaré  F, Lewis  K,  et al.  Decision aids for people facing health treatment or screening decisions.  Cochrane Database Syst Rev. 2017;4:CD001431.PubMedGoogle Scholar
9.
Koelewijn-van Loon  MS, van der Weijden  T, Ronda  G,  et al.  Improving lifestyle and risk perception through patient involvement in nurse-led cardiovascular risk management: a cluster-randomized controlled trial in primary care.  Prev Med. 2010;50(1-2):35-44. doi:10.1016/j.ypmed.2009.11.007PubMedGoogle ScholarCrossref
10.
Krones  T, Keller  H, Sönnichsen  A,  et al.  Absolute cardiovascular disease risk and shared decision making in primary care: a randomized controlled trial.  Ann Fam Med. 2008;6(3):218-227. doi:10.1370/afm.854PubMedGoogle ScholarCrossref
11.
Karmali  KN, Persell  SD, Perel  P, Lloyd-Jones  DM, Berendsen  MA, Huffman  MD.  Risk scoring for the primary prevention of cardiovascular disease.  Cochrane Database Syst Rev. 2017;3:CD006887.PubMedGoogle Scholar
12.
Navar  AM, Wang  TY, Li  S,  et al.  Lipid management in contemporary community practice: results from the Provider Assessment of Lipid Management (PALM) Registry.  Am Heart J. 2017;193:84-92. doi:10.1016/j.ahj.2017.08.005PubMedGoogle ScholarCrossref
13.
Navar  AM, Wang  TY, Goldberg  AC,  et al.  Design and rationale for the Patient and Provider Assessment of Lipid Management (PALM) registry.  Am Heart J. 2015;170(5):865-871. doi:10.1016/j.ahj.2015.08.002PubMedGoogle ScholarCrossref
14.
Fagerlin  A, Zikmund-Fisher  BJ, Ubel  PA, Jankovic  A, Derry  HA, Smith  DM.  Measuring numeracy without a math test: development of the Subjective Numeracy Scale.  Med Decis Making. 2007;27(5):672-680. doi:10.1177/0272989X07304449PubMedGoogle ScholarCrossref
15.
Navar  AM, Pencina  MJ, Mulder  H, Elias  P, Peterson  ED.  Improving patient risk communication: translating cardiovascular risk into standardized risk percentiles.  Am Heart J. 2018;198:18-24. doi:10.1016/j.ahj.2017.12.005PubMedGoogle ScholarCrossref
16.
Mann  DM, Ponieman  D, Montori  VM, Arciniega  J, McGinn  T.  The statin choice decision aid in primary care: a randomized trial.  Patient Educ Couns. 2010;80(1):138-140. doi:10.1016/j.pec.2009.10.008PubMedGoogle ScholarCrossref
×