Author Affiliations: Division of General Internal Medicine (Drs Persell, Cameron, and Zielinski and Mr Zei), Institute for Health Care Studies (Drs Persell and Cameron), Division of Cardiology (Dr Lloyd-Jones), and Department of Preventive Medicine (Dr Lloyd-Jones), Feinberg School of Medicine, Northwestern University, Chicago, Illinos.
Data are sparse regarding how physicians use coronary risk information for prescribing decisions.
We presented 5 primary prevention scenarios to primary care physicians affiliated with an academic center and surveyed their responses after they were provided with (1) patient risk factor information, (2) 10-year estimated coronary disease risk information, and (3) 10-year and lifetime risk estimates. We asked about aspirin prescribing, lipid testing, and lipid-lowering drug prescribing.
Of 202 physicians surveyed, 99 (49%) responded. The physicians made guideline-concordant aspirin decisions 51% to 91% of the time using risk factor information alone. Providing 10-year risk estimates increased concordant aspirin prescribing when the 10-year coronary risk was moderately high (15%) and decreased guideline-discordant prescribing when the 10-year risk was low (2 of 4 cases). Providing the lifetime risk information sometimes increased guideline-discordant aspirin prescribing. The physicians selected guideline-concordant thresholds for initiating treatment with lipid-lowering drugs 44% to 75% of the time using risk factor information alone. Selecting too low or too high low-density lipoprotein cholesterol thresholds was common. Ten-year risk information improved concordance when the 10-year risk was moderately high. Providing lifetime risk information increased willingness to initiate pharmacotherapy at low-density lipoprotein cholesterol levels that were lower than those recommended by guidelines when the 10-year risk was low but the lifetime risk was high.
Providing 10-year coronary risk information improved some hypothetical aspirin-prescribing decisions and improved lipid management when the short-term risk was moderately high. High lifetime risk sometimes led to more intensive prescription of aspirin or lipid-lowering medication. This outcome suggests that, to maximize the benefits of risk-calculating tools, specific guideline recommendations should be provided along with risk estimates.
Current guidelines for the primary prevention of cardiovascular disease recommend using information from multiple risk factors to guide decision making.1- 5 Patients with higher estimated risk generally should receive more intensive risk-reduction interventions such as aspirin or pharmacotherapy to lower cholesterol levels. Physicians have difficulty accurately estimating cardiovascular risk when provided with patient risk factor information alone.6 Multivariable risk estimation methods, such as the Framingham Risk Score7 or the European SCORE,8 produce estimates of risk over the next 10 years. These tools require that clinicians either perform a manual risk estimation process using tables or enter risk variables into a computer application to calculate risk.9 Providing this kind of risk information to physicians or physicians and patients, either manually or using automated methods, can improve the appropriate use of preventive therapies in some cases.10- 15 Calculating cardiovascular risk routinely also could improve aspirin prescribing for primary prevention. Although major guidelines differ to some degree with respect to aspirin prescribing for primary prevention, all include the use of some form of risk assessment to guide therapeutic decision making.2,3,5 Prior work has identified aspirin underuse by some patients with increased risk and potentially inappropriate use by some with low risk.16- 18 However, provision of short-term cardiovascular risk estimates has not demonstrated a strong effect in most experimental studies. Important treatment gaps remain, and guidelines still are not followed for significant numbers of eligible patients.10- 15,19,20
Recently, methods have been developed to estimate an individual's cardiovascular risk over his or her remaining lifetime.21,22 Knowledge of lifetime risk could help inform physicians and patients as they make decisions about therapies to reduce cardiovascular risk. Physicians who are not inclined to treat patients whose cholesterol levels are above a guideline goal might make different decisions for patients with a high lifetime risk for cardiovascular disease. However, this information could also lead physicians to prescribe treatments inappropriately, such as recommending aspirin therapy to a patient with a low short-term risk of cardiovascular disease, in whom the risk for bleeding complications would exceed the expected benefit in cardiovascular risk reduction. We sought to examine whether and how providing calculated 10-year risk estimates influenced preventive cardiology decision making and to determine the additional impact of adding lifetime risk information.
We surveyed all of the 202 primary care physicians who had an academic affiliation with Northwestern University's Feinberg School of Medicine, Chicago, Illinois, and who cared for adults in the fall of 2008. The majority, 82.2%, were affiliated clinical faculty members whose predominant academic activity is office-based teaching of medical students or residents, and the rest were members of the full-time medical school faculty and were internists practicing in the outpatient setting. The university's institutional review board approved the study. We mailed physicians a survey with a $2 bill and also contacted them by e-mail up to 3 times. Physicians could respond using either the mail or the e-mail version of the survey. The survey presented specific clinical scenarios (described in Table 1) in the following order: first with hypothetical patients' coronary heart disease risk factor information alone, then with additional information about their estimated 10-year coronary disease risk (cardiac death or nonfatal myocardial infarction) based on continuous models from the Framingham Study,7 and then with both 10-year and lifetime coronary disease risk estimates. Lifetime risk estimates were derived from the Cardiovascular Lifetime Risk Pooling Project, an ongoing study using pooled data from 17 major longitudinal epidemiologic cohort studies in the United States that takes into account both the risk of developing coronary disease and the risk of dying of other causes using methods similar to those used previously in single cohort studies.21,22
The physicians were instructed to complete the survey only with the information provided at each step and not to skip ahead or to go back to previously answered questions (although there was nothing to prohibit them from answering questions out of order). In each case, we asked physicians (1) whether they would prescribe low-dose aspirin, (2) when they would repeat lipid testing (at 5 years, 1 year, 6 months, or 6 weeks or start therapy without retesting), and (3) at what level would the low-density lipoprotein (LDL) cholesterol need to remain above on retesting (with confirmation by repeating a second time if necessary) (190, 160, 130, 100, or 70 mg/dL [to convert to millimoles per liter, multiply by 0.0259]) for them to prescribe cholesterol-lowering drug therapy. We asked them to respond to identical multiple-choice questions 3 times per patient scenario: based on risk factor data alone, after providing the 10-year risk estimate, and after providing both 10-year and lifetime risk estimates. In all cases, there was a family history of 1 parent with a myocardial infarction after the age of 70 years and no medical problems that would represent a contraindication to treatment with aspirin or cholesterol-lowering medication.
We assessed whether prescribing thresholds were concordant or discordant with contemporary guidelines. For aspirin prescribing, we considered the response to be concordant if aspirin was prescribed in cases in which the 10-year coronary risk was 15% and not prescribed in cases with a 10-year risk of 2% to 4%.2 For lipid monitoring and prescribing, we considered responses to be guideline concordant if they were consistent with the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) recommendations1 or with the optional thresholds described in 2004.4
For aspirin-prescribing decisions, we used the McNemar test to determine the agreement of responses before and after the addition of additional pieces of risk information among physicians who answered all 3 questions within a single case. For cholesterol monitoring or thresholds for drug prescribing, we used the Bowker test of symmetry because there were more than 2 possible responses.23 We performed exploratory analyses to determine whether the survey results varied by physician sex, full-time medical school faculty compared with affiliated clinical faculty, hours of direct patient care per week (≤20 h/wk vs >20 h/wk), years since completion of training (<10 vs ≥10), and whether or not physicians reported using a risk-calculating tool for aspirin prescribing or cholesterol-lowering drug treatment decisions (never or rarely vs sometimes or more often) using the Fisher exact test. We used SAS version 9.2 (SAS Institute Inc, Cary, North Carolina) for all analyses.
One hundred four of 202 physicians returned surveys, 99 of which contained usable data (49% response rate). Among responders, 28% were full-time medical school faculty members (compared with 8% of nonresponders; P < .001), and 59% were male (compared with 49% of nonresponders; P = .16). All respondents were practicing primary care internists. The number of physicians who responded to all clinical questions within each of the 3 sets of risk information (risk factors only, risk factors with 10-year risk, and risk factors with 10-year and lifetime risk estimates) ranged from 90 to 95. The respondents' characteristics are provided in Table 2. Thirty-seven percent of respondents reported that they rarely or never used a risk-calculating tool when making prescribing decisions for the primary prevention of cardiovascular disease.
The physicians made guideline-concordant aspirin-prescribing decisions 51% to 91% of the time when they were provided with risk factor information alone. The provision of 10-year risk information significantly improved guideline-concordant aspirin prescribing when the 10-year risk was moderately high (the 10-year risk of cardiac death or nonfatal myocardial infarction was 15% [Table 3, case 2]). In 2 of the 4 clinical scenarios in which the 10-year risk was low, explicitly providing physicians with the 10-year risk reduced aspirin prescribing that was not guideline concordant (Table 3, cases 3 and 4). Providing lifetime risk information increased aspirin prescribing in all 3 scenarios in which the 10-year risk was low and the lifetime risk was moderate or high (Table 3, cases 1, 4, and 5). Such prescription would not be concordant with contemporary guideline recommendations based on the 10-year risk estimates.2,3
In the 2 cases in which the LDL cholesterol level was not at the NCEP ATP III goal (cases 1 and 2), physicians often selected a wait time longer than that recommended by the guideline before repeating lipid testing (NCEP ATP III recommends instituting therapeutic lifestyle change and repeating testing in 6 weeks) (Table 4). Providing lifetime risk information increased immediate prescribing without repeated testing from 2% to 12% in the case of a patient who was not at goal and had low 10-year and high lifetime risks (case 1), a decision that is not concordant with guideline recommendations. Providing 10-year risk information in case 2, in which both the 10-year and the lifetime risks were high, increased immediate prescribing without retesting, a decision concordant with the optional 2004 guideline.4 In cases in which LDL cholesterol levels were at goal (cases 3, 4, and 5), physicians rarely waited the 5 years permitted in the NCEP ATP III guideline before retesting (Table 4). Neither provision of 10-year risk estimates nor provision of lifetime risk estimates significantly changed physician recommendations for the timing of retesting.
Physicians selected LDL cholesterol thresholds for initiating treatment with lipid-lowering drugs that were consistent with the NCEP ATP III or the optional 2004 guideline 44% to 75% of the time with risk factor information alone (Table 5). In the case with low 10-year and high lifetime risks (case 1), physicians' LDL cholesterol treatment thresholds were the same as the ATP III guideline 44% of the time, whereas similar numbers of physicians selected higher or lower thresholds. Providing the lifetime risk estimate led more physicians to select an LDL cholesterol threshold for drug treatment that was lower than the guidelines currently recommend (Table 5).
When both 10-year and lifetime risks were elevated (case 2), 32% of physicians selected an LDL cholesterol threshold for drug prescribing of 160 mg/dL or higher when presented with just risk factor information, responses that are above guideline recommendations. The addition of 10-year risk information or 10-year and lifetime risk information led to an increase in guideline-concordant responses (Table 5).
When the 10-year risk was low and the lifetime risk was low or moderate (cases 3, 4, and 5), physicians selected an LDL cholesterol threshold for drug therapy below the optional 2004 guideline recommendation of 160 mg/dL or higher after a therapeutic lifestyle change 16% to 46% of the time. Neither provision of 10-year risk estimates nor provision of lifetime risk estimates significantly changed prescribing thresholds in these cases (Table 5).
In secondary analyses, several physician characteristics were associated with guideline-concordant responses. When provided risk factor information alone, female physicians were more likely than male physicians to prescribe aspirin when the 10-year risk was moderately high (case 2) (94% for female physicians vs 71% for male physicians, P = .007); this difference was reduced by the provision of 10-year risk information (97% for female physicians and 89% for male physicians) and was no longer statistically significant (P = .24).
Physicians with less than 20 hours of direct patient care per week, physicians who were full-time medical school faculty members, and physicians with less than 10 years since completion of their training were more likely in some cases to provide NCEP ATP III guideline-concordant responses regarding when to perform repeated lipid testing among patients who were at their LDL cholesterol goal; ie, they were more likely to report that they would wait 5 years to retest lipid levels, compared with physicians who had more than 20 hours per week of patient care, physicians who were affiliated clinical faculty members, and physicians with 10 or more years since they completed training, respectively. Physicians who reported that they rarely or never used a cardiovascular risk–calculating tool to aid with prescribing decisions were significantly less likely to select an LDL cholesterol threshold for prescribing cholesterol-lowering therapy concordant with contemporary guidelines for a woman with a low short-term, moderate lifetime risk and an LDL cholesterol level of 158 mg/dL (case 4) (P = .03). Providing additional risk information had little impact on this association.
Practicing primary care physicians—many of whom do not routinely use a risk-estimation tool for prescribing decisions—frequently made preventive cardiology treatment choices that were inconsistent with major prevention guidelines when they were provided with patient risk factor data alone. Physicians often recommended aspirin therapy in cases in which the short-term coronary risk was low, yet 20% did not recommend aspirin therapy to a male patient with a 10-year coronary risk of 15%. They recommended repeated lipid testing less promptly than the guideline recommends for individuals who were not at their guideline goal for LDL cholesterol levels and recommended testing more frequently than the guideline recommends for patients who were at goal at the time of screening. It was not uncommon for physicians to select LDL cholesterol thresholds for drug prescribing that were higher or lower than guideline-recommended thresholds. These findings are consistent with prior studies examining adherence to or agreement with cardiovascular disease prevention guidelines for lipid management,6,24,25 hypertension treatment,6,26,27 or aspirin prescribing,6 which have shown that considerable numbers of practicing physicians choose management recommendations that are inconsistent with major contemporary guidelines.
Inaccurate appraisal of cardiovascular risk may be an important explanatory factor regarding guideline-discordant choices.6,28 Providing estimates of 10-year coronary risk using the method derived from the Framingham Heart Study led to several important changes in physicians' choices. Physicians made aspirin-prescribing decisions that were more concordant with evidence-based guidelines; they prescribed aspirin more often in the case in which the short-term risk was moderately high; and, in some cases, they reduced prescribing when the short-term risk was low. Providing 10-year risk information also increased the number of physicians who made guideline-concordant cholesterol management choices for an individual with multiple risk factors and a moderately high 10-year risk (≥10% risk of cardiac death or nonfatal myocardial infarction). For the most part, though, respondents who provided guideline-discordant answers when they were given risk factor data alone did not change their response when they were given the 10-year coronary risk estimate—the technique recommended by contemporary US guidelines to inform clinical decision making.1- 4 These findings suggest that even when physicians obtain a 10-year risk estimate, through the use of a risk table or the National Heart, Lung, and Blood Institute’s online risk calculator,29 this step alone may not result in appropriate clinical decision making. The results also are consistent with the modest or negative results of clinical studies aimed at increasing physicians' use of short-term cardiovascular risk estimates for clinical decision making.11,15,19,20
Recent prevention guidelines advise clinicians to take patients' lifetime risk for coronary disease into consideration.1,3 Individuals with a low short-term but high lifetime risk, who represent a substantial proportion of the population younger than 50 years,30 should attempt intensive lifestyle modifications and consider using lipid-lowering medication when their LDL cholesterol levels remain high.1 Providing lifetime risk estimates for coronary disease to physicians generally did not alter their prescribing decisions substantially with regard to aspirin therapy. It did, however, lead physicians to prescribe aspirin more often in cases in which the lifetime risk was 26% to 43% but the 10-year risk was low. Such a decision would be inconsistent with current guideline recommendations based on 10-year risk estimates. This finding suggests that specific education may be needed to improve physicians' awareness that aspirin prescribing should be based on a determination that the potential short-term benefits clearly exceed a patient's risks of treatment. Making lifetime risk information available in addition to 10-year risk information did alter physicians' cholesterol monitoring and drug therapy–prescribing decisions in 1 case (case 1, involving a man with a low short-term, high lifetime risk and an LDL cholesterol level of 135 mg/dL). However, instead of improving guideline concordance by increasing the number of physicians who would initiate lifestyle changes and retest in 6 weeks, providing the lifetime risk information led more physicians to recommend immediate drug treatment—at a lower LDL cholesterol threshold than is currently recommended. Whereas clinical trial data suggest that relative risk reductions with statin therapy are similar over 5 years regardless of baseline LDL cholesterol levels,31 the absolute level of risk determines the number needed to treat to prevent an event and substantially influences short-term cost-effectiveness. Data are sparse regarding the long-term effectiveness and cost-effectiveness of statin therapy for individuals with low short-term but high lifetime risks for coronary heart disease.
Our findings could inform the development of intervention strategies aimed at improving preventive cardiology care for the prescribing of aspirin or cholesterol-lowering medication. Many physicians in our study reported a lack of routine use of a risk-calculating tool for preventive cardiology decision making. Automating the risk assessment process using computerized decision support within electronic health records offers the potential to make cardiac risk information more readily available to clinicians.32,33 However, since guideline-discordant responses were common even when short-term or lifetime risk estimates were provided, risk estimation tools intended to increase guideline-recommended treatments should also provide recommendations for specific clinical actions. One particularly successful study of computerized decision support aimed at improving preventive cardiology practice for lipid management generated cardiovascular risk assessments from data within electronic health records and provided clinicians with both estimates of risk and specific guideline-based recommendations. Among the patients who were served by physicians receiving automated alerts, 66% of those requiring treatment received it, compared with 36% of the corresponding control group.10
The results of this study should be viewed with several limitations in mind. Respondents included proportionally more full-time medical school faculty members than the overall population. There may also have been other unmeasured differences between responders and nonresponders (eg, physicians who were more interested or familiar with prevention guidelines may have been more likely to respond). These differences may have led to a greater proportion of guideline-concordant responses than had the entire population responded. We surveyed practicing internists from multiple practices, but all were affiliated with a single institution in 1 city. We do not know how generalizable these findings are to other groups of practicing primary care physicians. The size of the study was fairly small, and there might have been additional significant findings if more physicians had been surveyed. The order in which the cases were presented may have influenced how physicians made their choices. Physicians may have been less likely to use risk estimates to inform their hypothetical decision making if they felt committed to the original choice that they made using risk factor data alone. It is not known whether providing these forms of risk estimates in actual practice would influence clinical decision making in the same way.
In summary, providing 10-year coronary risk information to physicians improved hypothetical prescribing decisions for aspirin, but lifetime risk information led physicians to overtreat with aspirin for primary prevention, compared with contemporary guideline recommendations. Guideline-discordant choices for cholesterol management were fairly common and were usually not influenced by the provision of risk information. Providing estimated 10-year risk information improved concordance with current cholesterol management recommendations when short-term risk was moderately high. Providing lifetime risk estimates may lead some physicians to prescribe lipid-lowering medications more intensively and in a manner inconsistent with contemporary guidelines based solely on short-term levels. These findings suggest that, to maximize their benefits, risk estimation tools should be coupled with clinical decision support that provides specific guideline recommendations along with risk assessment.
Correspondence: Stephen D. Persell, MD, MPH, Division of General Internal Medicine, Northwestern University, 750 N Lake Shore Dr, 10th Floor, Chicago, IL 60611 (email@example.com).
Accepted for Publication: September 19, 2009.
Author Contributions: Dr Persell had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Persell, Cameron, and Lloyd-Jones. Acquisition of data: Persell, Zei, and Zielinski. Analysis and interpretation of data: Persell, Cameron, and Lloyd-Jones. Drafting of the manuscript: Persell. Critical revision of the manuscript for important intellectual content: Zei, Cameron, Zielinski, and Lloyd-Jones. Statistical analysis: Persell. Obtained funding: Persell and Lloyd-Jones. Administrative, technical, and material support: Zei and Cameron. Methodology/survey design: Persell and Cameron.
Financial Disclosure: None reported.
Funding/Support: This study was supported by Mentored Clinical Scientist Development award 1 K08 HS015647-01 from the Agency for Healthcare Research and Quality (Dr Persell). Dr Lloyd-Jones and the Cardiovascular Lifetime Risk Pooling Project were supported by grant R21 HL085375 from the National Heart, Lung, and Blood Institute.
Role of the Sponsors: The sponsors had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.
Additional Contributions: Heather L. Heiman, MD, and Toshiko Uchida, MD, provided valuable feedback during the development of the survey instrument, and the REACH Network coordinated the data collection. We are grateful to all the physicians who completed the survey.
Thank you for submitting a comment on this article. It will be reviewed by JAMA Internal Medicine editors. You will be notified when your comment has been published. Comments should not exceed 500 words of text and 10 references.
Do not submit personal medical questions or information that could identify a specific patient, questions about a particular case, or general inquiries to an author. Only content that has not been published, posted, or submitted elsewhere should be submitted. By submitting this Comment, you and any coauthors transfer copyright to the journal if your Comment is posted.
* = Required Field
Disclosure of Any Conflicts of Interest*
Indicate all relevant conflicts of interest of each author below, including all relevant financial interests, activities, and relationships within the past 3 years including, but not limited to, employment, affiliation, grants or funding, consultancies, honoraria or payment, speakers’ bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued. If all authors have none, check "No potential conflicts or relevant financial interests" in the box below. Please also indicate any funding received in support of this work. The information will be posted with your response.
Some tools below are only available to our subscribers or users with an online account.
Download citation file:
Web of Science® Times Cited: 23
Customize your page view by dragging & repositioning the boxes below.
Enter your username and email address. We'll send you a link to reset your password.
Enter your username and email address. We'll send instructions on how to reset your password to the email address we have on record.
Athens and Shibboleth are access management services that provide single sign-on to protected resources. They replace the multiple user names and passwords necessary to access subscription-based content with a single user name and password that can be entered once per session. It operates independently of a user's location or IP address. If your institution uses Athens or Shibboleth authentication, please contact your site administrator to receive your user name and password.