Perceived Usefulness of Cardiac Computed Tomography as Assessed by Referring Physicians and Its Effect on Patient Management
Article Outline
Despite the growing use of computed tomographic angiography (CTA), the effect on patient management is less clear. We sought to determine the perceived usefulness of the results provided by CTA and to assess whether and how it influences patient management. Comprehensive prospective data were collected from 184 consecutive patients who presented for clinical CTA for the evaluation of coronary artery disease from March to July 2008. In addition, a detailed survey was sent to each referring physician for each patient examined to assess whether they found the results of the CTA useful and whether it had any influence on subsequent patient management. Of 184 CTA examinations, which had been ordered by 82 different providers, 108 surveys (59%) were completed by 53 different physicians. No significant differences were found in either the patient or provider characteristics for the completed versus noncompleted surveys. Of the 184 CTA examinations, the severity of coronary disease detected by CTA was severe for 26%, mild to moderate in 47%, and not present in 27% of the patients. Clinicians considered the test results to be useful in virtually all cases and thought the results led to significant risk reclassification in 58% of the patients. If CTA had not been available, the clinicians indicated that they would have ordered an invasive test for 46% of the patients and noninvasive tests for 32%. After CTA, changes in medical therapies were made for 31%, invasive angiography was planned for 19%, and noninvasive testing was scheduled for 6% of the patients. In conclusion, of 53 different referring clinicians from different medical specialties, CTA was considered to almost always be useful; however, the effect on subsequent medical management was more variable.
To use computed tomographic angiography (CTA) efficiently, it is important to determine that it does not merely result in yet another test in our diagnostic algorithm but rather has a significant effect on patient management or that it prevents the use of additional testing. Although most computed tomographic angiographic studies to date have examined the technical advancements and diagnostic accuracy of this modality, the value of CTA and whether it is a useful test in patient management remain unknown. Thus, the aim of our study was to describe physicians' perceptions of the utility of CTA and its effect on patient management.
Methods
From March 2008 to July 2008, data were prospectively collected for all consecutive patients referred for CTA at the Massachusetts General Hospital. Before each examination, information, including demographics, baseline cardiac history, results of previous cardiac testing, symptoms, and the indication for the examination were obtained from the medical records, a self-administered questionnaire that had been completed by each patient, and from an interview conducted by a physician performing the computed tomographic angiographic examination.
Within 1 week after CTA, all referring physicians were contacted by e-mail and were provided with an electronic copy of the final examination report. At that time, they were asked to complete a brief on-line questionnaire that was designed to assess the usefulness of the results of the cardiac computed tomographic examination in the treatment of the given patient (see Appendix 1). The survey was administered using secure, on-line, survey software (SurveyMonkey; available at: www.SurveyMonkey.com). To encourage participation, the referring physicians were promised a small token of appreciation (a Starbucks gift certificate) for each survey they completed. The clinicians who did not respond to the initial e-mail were sent a follow-up reminder e-mail. Those who failed to respond to the reminder e-mail were then sent a printed copy of the survey by United States postal mail, together with the printed final report. As a part of each survey, information relating to the referring physician's specialty and training was obtained.
No patient underwent more than one CTA; thus, the comparisons among patient populations included no duplicate subjects; however, several physicians had ordered more than one CTA (average 2.2 per provider). The clinical characteristics of the patients with and without completed surveys were compared to determine whether the patients for whom survey data were available might differ from the overall study population.
The pretest probability of coronary artery disease (CAD) was calculated by choosing the most appropriate risk score according to the presence or absence of symptoms. For asymptomatic patients, the Framingham risk score was calculated.1 Although this score is intended to determine a patient's risk of developing coronary heart disease events, it has also been shown to correlate with the burden of clinical or subclinical atherosclerosis.2, 3 It was thus an appropriate risk score for predicting the presence of CAD in asymptomatic patients.
After calculating the Framingham risk score for each patient, their risk level was reclassified using 2 methods. First, using the Adult Treatment Panel III criteria, any patient who had a known coronary heart disease risk equivalent (ie, history of known CAD, coronary artery bypass grafting surgery, percutaneous coronary intervention, diabetes mellitus, peripheral vascular disease, or stroke) was reclassified as being at high risk.1 Any patient who had had abnormal myocardial perfusion imaging findings was also reclassified as being at high risk. This second reclassification was determined from data showing that patients with abnormal myocardial perfusion imaging findings have an annualized hard event rate equivalent to high Framingham risk (ie, >2% annually).4
For symptomatic patients, the estimated pretest probability of significant CAD was calculated using the Duke clinical score.5 Typical angina was defined as chest discomfort that was (1) precipitated by physical exertion or emotion and (2) relieved with rest or nitroglycerin. Atypical angina was defined as chest discomfort that was associated with either physical exertion or emotion or relieved with rest or nitroglycerin, but not both, or by having dyspnea on exertion that was clinically suspected to represent the patient's anginal equivalent. Nonanginal chest pain was characterized as chest discomfort that lacked any of these associations. The pretest probability of significant CAD was calculated using the previously validated Duke Clinical Score model integrating the type of chest pain, together with clinical variables of age, gender, history of myocardial infarction, smoking, hyperlipidemia, and diabetes.6 The model used for this calculation is given in Appendix 2. From the calculated pretest probability of significant CAD, each patient was then categorized as having low risk (<30%), intermediate risk (30% to 70%), or high risk (>70%).
CTA was performed using the Definition dual-source computed tomography scanner (Siemens Medical Systems, Munich, Germany) with a gantry rotation time of 330 ms and standard detector collimation of 2 × 32 × 0.6 mm. To reduce the patient's radiation exposure, tube current modulation was used whenever possible. In addition, for younger patients (eg, age <50 years), axial acquisition using prospective electrocardiographic triggering (Siemens sequential scanning) was selected, when appropriate.
Axial and double-oblique images viewed in thin, maximal intensity projections and multiplannar reformation settings were used for image analysis. Clinical reporting was then used to categorize each vessel as normal (no CAD), mild to moderate CAD (visually estimated at <70% stenosis), or severe CAD (visually estimated at >70% stenosis for any coronary artery or >50% stenosis for the left main coronary artery). The disease extent was determined by the number of vessels with severe disease (1, 2, or 3 vessels). Left main coronary artery disease was classified as equivalent to 2-vessel disease.
Statistical analysis was performed using Stata IC, version 10.0 (StataCorp LP, College Station, Texas). All continuous variables are expressed as the mean ± SD, and categorical variables are expressed as percentages. Differences in continuous variables were assessed using Student's unpaired t tests. Differences in dichotomous variables were assessed using the chi-square test or Fisher's exact test. For ordinal variables, the rank sum test was performed to assess for significance. A 2-tailed p value of <0.05 was considered statistically significant.
Results
Clinical data were obtained from 184 consecutive computed tomographic angiographic examinations that had been ordered by 82 different physicians for an evaluation of CAD as the primary indication. Of the 184 examinations, the follow-up survey was completed for 108 (59%) Of these 108 surveys, 85 (79%) were completed by cardiologists and the rest were completed predominantly by primary care physicians.
Table 1 lists the baseline patient characteristics, primary indication for CTA, frequency of previous cardiac imaging studies, and the results of the computed tomographic angiographic examination for all 184 patients and the 108 for whom the provider survey was completed. A comparison of the patient and provider characteristics for examinations with completed versus noncompleted surveys (Supplement Tables 1 and 2) did not detect any significant differences between the 2 groups.
Table 1. Baseline patient characteristics and results of computed tomographic angiography (CTA)
| Variable | Patients With Indication for CAD Evaluation (n = 184) | Patients With Indication for CAD Evaluation and Completed Survey (n = 108) |
|---|---|---|
| Clinical characteristics | ||
 Women | 73(40%) | 36(33%) |
| Age (years) | 57.2±14.3 | 58.5±15.1 |
| Coronary heart disease risk equivalent⁎ | 65(35%) | 39(36%) |
| Coronary heart disease risk equivalent†‡ | 86(47%) | 47(44%) |
| Diabetes mellitus | 28(15%) | 15(14%) |
| Hypertension§ | 106(58%) | 58(54%) |
| Hyperlipidemia§ | 115(63%) | 71(66%) |
| Smoker | 21(11%) | 13(12%) |
| Patient symptoms | ||
| Asymptomatic | 52(28%) | 26(24%) |
| Dyspnea only | 2(1%) | 2(2%) |
| Chest pain syndrome | 132(72%) | 82(76%) |
| Dyspnea thought to represent angina pectoris | 29(16%) | 23(21%) |
| Nonanginal chest pain | 47(26%) | 26(24%) |
| Atypical chest pain | 27(15%) | 16(15%) |
| Typical chest pain | 29(16%) | 17(16%) |
| Previous testing (within 6 months) | ||
| Echocardiography | 53(29%) | 31(29%) |
| Stress test | ||
| Nuclear stress test | 78(42%) | 42(39%) |
| Stress echocardiography | 6(3%) | 5(5%) |
| Exercise electrocardiographic stress test | 16(9%) | 8(7%) |
| Cardiac catheterization | 9(5%) | 6(6%) |
| Primary indication | ||
| Native coronary artery disease evaluation | 153(83%) | 89(82%) |
| Bypass graft/stent evaluation | 31(17%) | 19(18%) |
| Computed tomographic angiographic results | ||
| Normal | 49(27%) | 29(27%) |
| Mild to moderate coronary artery disease | 87(47%) | 48(44%) |
| Severe coronary artery disease | 48(26%) | 31(29%) |
⁎Coronary heart disease risk equivalent included history of known CAD, myocardial infarction, coronary artery bypass grafting surgery, percutaneous coronary intervention, diabetes, peripheral vascular disease, and/or cerebrovascular accident. |
†Coronary heart disease risk equivalent expanded to also include all patients with abnormal findings from single photon emission computed tomographic myocardial perfusion imaging. |
‡Included abnormal findings from nuclear stress testing. |
§Patients classified as having hypertension or hyperlipidemia if diagnosis present on medical record or if they had been told by physician that they had the condition. |
Of the 184 patients referred for CTA, 132 (72%) were symptomatic and 52 (28%) were asymptomatic (Table 2). CTA showed that 48 (26%) had no evidence of CAD and 73% had CAD. These proportions did not vary between the patients who were symptomatic and asymptomatic.
Table 2. Characteristics of symptomatic versus asymptomatic patients referred for computed tomographic angiography (CTA)
| Variable | Total (n = 184) | Symptomatic (n = 132) | Asymptomatic (n = 52) | p Value |
|---|---|---|---|---|
| Clinical characteristics | ||||
| Women | 73(40%) | 58(44%) | 15(29%) | 0.06 |
| Age (years) | 57.2±14.3 | 56.9±14.9 | 57.9±12.9 | 0.68 |
| Diabetes | 28(15%) | 24(18%) | 4(8%) | 0.07 |
| Coronary heart disease risk equivalent⁎ | 65(35%) | 45(34%) | 20(38%) | 0.58 |
| Coronary heart disease risk equivalent†‡ | 86(47%) | 60(45%) | 26(50%) | 0.58 |
| Hypertension§ | 106(58%) | 80(61%) | 26(50%) | 0.19 |
| Hyperlipidemia§ | 115(63%) | 83(63%) | 32(62%) | 0.87 |
| Previous testing (within 6 months) | ||||
| Any previous noninvasive testing | 116(63) | 88(67%) | 28(54%) | 0.11 |
| Any previous stress testing | 95(52%) | 75(57%) | 20(38%) | 0.03 |
| Stress echocardiography | 6(3%) | 4(3%) | 2(4%) | 0.78 |
| Exercise electrocardiographic stress test | 16(9%) | 13(10%) | 3(6%) | 0.38 |
| Nuclear stress test | 78(42%) | 62(47%) | 16(31%) | 0.045 |
| Echocardiography | 53(29%) | 37(28%) | 16(31%) | 0.71 |
| Cardiac catheterization | 9(5%) | 7(5%) | 2(4%) | 0.68 |
| Pretest probability of significant coronary artery disease¶ | 0.64 | |||
| Low | 53(29%) | 37(28%) | 16(31%) | |
| Intermediate | 24(13%) | 17(13%) | 7(13%) | |
| High | 107(58%) | 78(59%) | 29(56%) | |
| Framingham risk score∥ | 0.68 | |||
| <6% | 59(45%) | 42(45%) | 17(46%) | |
| 6–10% | 22(17%) | 15(16%) | 7(19%) | |
| 10–20% | 43(33%) | 31(33%) | 12(32%) | |
| ≥20% | 7(5%) | 6(6%) | 1(3%) | |
| Reclassification of risk by addition of Adult Treatment Panel III coronary heart disease risk equivalents⁎∥ | 0.60 | |||
| <6% | 46(35%) | 33(35%) | 13(35%) | |
| 6–10% | 14(11%) | 10(11%) | 4(11%) | |
| 10–20% | 17(13%) | 15(16%) | 2(5%) | |
| ≥20% | 54(41%) | 36(38%) | 18(49%) | |
| Reclassification of risk by addition of nuclear imaging results†‡∥ | 0.31 | |||
| <6% | 34(26%) | 25(27%) | 9(24%) | |
| 6–10% | 11(8%) | 8(9%) | 3(8%) | |
| 10–20% | 14(11%) | 13(14%) | 1(3%) | |
| ≥20% | 72(55%) | 48(51%) | 24(69%) | |
| Computed tomographic angiographic results | 0.46 | |||
| Coronary artery disease severity | ||||
| Normal | 49(27%) | 36(27%) | 13(25%) | |
| Mild to moderate | 87(47%) | 64(48%) | 23(44%) | |
| Severe | 48(26%) | 32(24%) | 16(31%) | |
| Number of coronary arteries with stenosis | 0.32 | |||
| 0 | 136(73%) | 100(76%) | 36(69%) | |
| 1 | 25(14%) | 18(14%) | 7(13%) | |
| 2 | 17(9%) | 10(8%) | 7(13%) | |
| 3 | 6(3%) | 4(3%) | 2(4%) |
⁎Coronary heart disease risk equivalent included history of known coronary artery disease, myocardial infarction, coronary artery bypass grafting surgery, percutaneous coronary intervention, diabetes, peripheral vascular disease, and/or cerebrovascular accident. |
†Coronary heart disease risk equivalent expanded to include all patients with abnormal findings from single photon emission computed tomographic myocardial perfusion imaging. |
‡Included abnormal findings from nuclear stress testing. |
§Patients classified as having hypertension or hyperlipidemia if diagnosis present on medical record or if they had been told by physician that they had the condition. |
¶Duke clinical scoring system. |
∥Framingham risk score calculated for 131 of 184 patients with lipid panel findings available. |
Figure 1 illustrates the perceived usefulness of the computed tomographic angiographic examination results as assessed by the referring physician. Most respondents (106 of 108) indicated that the test results were helpful; and 85% rated it ≥8 on a scale of 1 to 10, with 1 indicating “not helpful” and 10, “extremely helpful.” The reasons for the unhelpful category included a “technically limited study” and when the “results did not lead to any change in practice because the test did not add any new information.”
Figure 1.
Usefulness of CTA as assessed by referring clinician.
Of the survey respondents, 46% indicated they would have ordered an invasive test and 32% indicated they would have ordered a noninvasive test if CTA was hypothetically not available (ie, responded “yes” or “probably yes”; Figure 2).
Figure 2.
Cardiac testing that would have been performed if CTA was hypothetically not available.
Comparing the characteristics of the symptomatic versus asymptomatic patients (Table 2), no difference was found in age, prevalence of hypertension, or prevalence of hyperlipidemia; however, a nonsignificant trend was seen toward a greater proportion of women as symptomatic patients. Symptomatic patients were more likely to be current smokers and to have undergone a stress test in the previous 6 months.
No significant difference was found between the symptomatic and asymptomatic patients in the pretest probability of CAD using either the Duke clinical score or the Framingham risk score. Also, no difference was found in the extent or severity of CAD identified by CTA between the symptomatic and asymptomatic patients.
When assessing the effect of the computed tomographic angiographic results for symptomatic versus asymptomatic patients (Table 3), a trend was seen for the test to more often be considered “helpful” for the symptomatic patients. Although the presence or absence of symptoms had no influence on the post-test probability of CAD, significantly more clinicians indicated that they would have ordered invasive angiography if CTA was hypothetically not available for symptomatic patients than for asymptomatic patients (51% vs 31%, p = 0.05). In contrast, given the same scenario of CTA not being available, no difference was seen in the clinician predicted use of a noninvasive stress test between these 2 groups. Furthermore, although the presence or absence of symptoms did not have any influence on the planned changes to medical therapy after CTA, a trend was seen such that symptomatic patients were more likely to be referred for invasive angiography after CTA (23% vs 8%, p = 0.08).
Table 3. Effect of computed tomographic angiography (CTA) stratified by presence or absence of symptoms
| Variable | Total (n = 108) | Symptomatic (n = 82) | Asymptomatic (n = 26) | p Value |
|---|---|---|---|---|
| Helpful | 106(98%) | 82(100%) | 24(92%) | 0.06 |
| Usefulness (scale 1–10) | 8.9±1.5 | 9±1.3 | 8.5±1.9 | 0.16 |
| Change in coronary artery disease probability | 0.21 | |||
| Significantly increased | 32(30%) | 28(34%) | 4(15%) | |
| Slightly increased | 5(5%) | 4(5%) | 1(4%) | |
| No change | 18(17%) | 11(13%) | 7(27%) | |
| Slightly decreased | 4(4%) | 3(4%) | 1(4%) | |
| Significantly decreased | 30(28%) | 24(29%) | 6(23%) | |
| If computed tomographic angiography unavailable | ||||
| Probably/would have ordered invasive angiography | 50(46%) | 42(51%) | 8(31%) | 0.05 |
| Probably/would have ordered noninvasive stress test | 35(32%) | 25(30) | 10(38%) | 0.96 |
| Effect on treatment | ||||
| Initiate new medical therapy | 17(16%) | 14(17%) | 3(11%) | 0.45 |
| Increase dose of existing medical therapy | 19(18%) | 15(18%) | 4(15%) | 0.73 |
| Discontinue/reduce dose of existing medical therapy | 3(3%) | 2(2%) | 1(4%) | 0.70 |
| Order noninvasive test | 6(6%) | 4(5%) | 2(8%) | 0.59 |
| Plan for catheterization with or without percutaneous coronary intervention | 21(19%) | 19(23%) | 2(8%) | 0.08 |
| Plan for surgical revascularization | 1(1%) | 1(1%) | 0(0%) | 0.57 |
| Recommend weight loss/exercise | 21(19%) | 17(21%) | 4(15%) | 0.54 |
| Patient reassurance | 57(53%) | 43(52%) | 14(54%) | 0.90 |
| Other (Supplement Table 3) | 30(28%) |
Figure 3 displays the change in the patient's probability of CAD after CTA as assessed by the referring clinician. According to the questionnaire results, risk reclassification occurred in >1/2 of all examinations, because 28% of physicians reported a significant decrease in the probability of CAD after CTA and 30% reported a significant increase.
Figure 3.
Change in probability of CAD after CTA.
The follow-up survey after CTA revealed that new medical therapies were initiated for 16% of the patients (most commonly dyslipidemia therapy or aspirin) and for 18% of the patients, the dosage of existing therapies was increased (Table 4). Less frequently (3%), medical therapy was discontinued or the dosage of existing medications was decreased. Clinicians indicated that they planned to order another noninvasive test 6% of the time, and 19% had plans to refer their patients for invasive angiography and/or percutaneous coronary intervention. Other common actions after CTA included patient reassurance (53%) and a recommendation for weight loss or exercise (19%). Patients with CAD found by CTA were more likely to have an increase in medical therapy or to have plans for additional invasive testing, and patients without CAD were more likely to be reassured. Finally, 28% of the clinicians indicated that other actions ensued after CTA (Supplement Table 3 provides additional description).
Table 4. Effect of computed tomographic angiography (CTA) on patient management stratified by coronary artery disease (CAD)
| Variable | Total (n = 108) | CAD | p Value | |
|---|---|---|---|---|
| Yes (n = 79) | No (n = 29) | |||
| Initiate new medical therapy | 17(16) | 16(20) | 1(4) | 0.034 |
| Increase dose of existing medical therapy | 19(18) | 19(24) | 0(0) | 0.004 |
| Discontinue/reduce dose of existing medical therapy | 3(3) | 2(3) | 1(3) | 0.80 |
| Order noninvasive test | 6(6) | 6(8) | 0(0) | 0.13 |
| Plan for catheterization with or without percutaneous coronary intervention | 21(19) | 21(27) | 0(0) | 0.002 |
| Plan for surgical revascularization | 1(1) | 1(1) | 0(0) | 0.54 |
| Recommend weight loss/exercise | 21(19) | 16(20) | 5(17) | 0.72 |
| Patient reassurance | 57(53) | 34(43) | 23(80) | 0.001 |
| Other (Supplement Table 3) | 30(28) | |||
Figure 4 shows the CTA results stratified by the pretest probability of CAD. For the 132 symptomatic patients in this group, the pretest probability of CAD was calculated using the Duke Clinical Score. Of the 32 patients who had significant CAD by CTA, 27 (84%) were predicted to have a high probability of significant CAD. In contrast, of the 37 patients who were predicted to have a low probability of significant CAD, 14 had mild to moderate disease and 2 had significant disease.
Figure 4.
Incidence and severity of CAD stratified by pretest probability of disease among (A) asymptomatic and (B) symptomatic patients referred for evaluation of CAD.
Of the 52 asymptomatic patients, the Framingham risk score was available for 37, because some patients referred by providers from outside our institution did not have data available of their lipid profiles. Of these 37 patients, 17 (46%) had a very low Framingham risk score (ie, <6% risk), 7 (19%) had low risk, 12 (32%) had intermediate risk, and 1 (3%) had high risk. Two patients from the very low risk group and 10 patients from the intermediate risk group were found to have significant CAD. When the risk score was reclassified by incorporating the Adult Treatment Panel III coronary heart disease risk equivalents, 17 patients were reclassified as having a high risk. Of these 17 patients, 13 had significant CAD and 5 had mild to moderate CAD. When the risk score was again reclassified by adding the patients with perfusion abnormalities as at high risk, 6 more patients were reclassified as at high risk. Of these, 4 had mild to moderate CAD and 2 had no CAD.
Of the 4 components of the typical cardiac computed tomographic examination performed at our institution, 96% of respondents found information on the coronary anatomy to be useful (76% extremely useful, 15% very useful, and 5% somewhat useful), and the corresponding frequencies for the usefulness of the evaluation of left ventricular function, noncontrast calcium scoring, and noncardiac findings were 70%, 56%, and 44% (Figure 5).
Figure 5.
Usefulness of reported components from CTA.
Discussion
In this first-ever survey of the effect of CTA, we observed that the vast majority of physicians surveyed thought that the results provided by CTA were useful and had a significant effect on subsequent patient management.
Realizing the subjective nature of physicians' opinions about the usefulness of an examination they have ordered, we also used other metrics of examination usefulness, including the percentage of tests in which the post-test probability of CAD significantly increased or decreased and how often the test led to changes in medical management for a given patient. In our study, we observed that the probability of CAD was significantly increased or decreased in >1/2 of all patients. Subsequent changes to medical therapy were made for 31% of the patients, and invasive angiography was planned for 19% and noninvasive testing for 6% of the patients.
A key question is whether these rates represent a significant effect. Because it has been suggested that CTA, with its high negative predictive value, is best suited as a modality to rule out the presence of CAD, one goal of CTA-based imaging should be to exclude disease in appropriately selected patients. In such cases, no new therapy should be started. In contrast, when coronary plaque is identified, escalating the aggressiveness of medical therapy and/or obtaining additional testing to evaluate the physiologic significance of the disease might be appropriate options.
A logical question that follows from our results is whether this initiation or escalation of medical therapy could have been based on other clinical data, including the Framingham risk score, exercise treadmill testing results, or the coronary calcium score. Our analysis of the results from CTA, stratified by the pretest probability of disease (Figure 4), suggests that most of the 78 symptomatic patients who had a high pretest probability of CAD did have it. In this group, only 8% of the patients had no CAD. The concept that CTA is of limited value in symptomatic patients with a high pretest probability of CAD has also been shown by Meijboom et al.7 In contrast, 53% of the intermediate-risk group had no CAD. Thus, if the goal of CTA is to exclude disease, it would be most efficiently used in the intermediate-risk group.
Previous studies have shown that in appropriately selected patients, CTA can avoid the use of invasive angiography.8, 9 In our study, 46% of clinicians surveyed stated that they would have, or probably would have, ordered invasive angiography if CTA were hypothetically not available; after CTA, only 19% still planned to order invasive angiography.
A noteworthy finding in our study was that 28% of the patients referred for CTA for an evaluation of CAD were asymptomatic, even though the current guidelines suggest that CTA is not an appropriate modality for most such patients. In our cohort, however, no difference was found in the pretest probability of CAD or the severity and extent of CAD between symptomatic and asymptomatic patients (Table 3). Although a recent study by Choi et al10 found that among asymptomatic self-referred South Korean patients with no history of CAD, the prevalence of CAD was 22%, 75% of asymptomatic patients in our study had CAD and 31% had severe CAD. Thus, asymptomatic patients referred for CTA in our center were either at high risk of CAD or had known CAD.
Our study had several limitations. First, an inherent limitation to any survey design is the external validity of the results, because the survey responders might not be representative of the entire population to which the results are applied. In addition, because our study was performed at a large tertiary institution, our findings might be less generalizable to a community setting.
Given that physicians are known to be poor responders to surveys, we conducted a review of the published data to identify strategies to increase the response rate.11, 12 With the use of such methods as promised rewards and the use of repeated follow-up with nonresponders, we were able to achieve a very good response rate (the typical response rate for physician surveys can be as low as 20%13). To further strengthen the external validity of the present study, we compared both the patient and provider characteristics among the examinations with completed versus noncompleted surveys; however, we were unable to find any notable differences. Furthermore, no significant difference was found in the perceived usefulness of the examination among the physicians who completed the survey immediately versus those who needed several reminders.
Another potential source of bias inherent in the survey design is that responders could have a specific motivation to complete the survey (ie, relevant examples of this include extreme satisfaction or dissatisfaction with the results from CTA or self-gain from promoting CTA as a new modality). However, we were unable to identify any patterns of extreme satisfaction or dissatisfaction among the clinicians surveyed and also did not find any differences in the frequency of technically limited scans among responders versus nonresponders. In addition, because 53 different providers participated in the study and, of them, only 3 had any training in cardiac computed tomography (only 1 responder was clinically involved in the cardiac computed tomographic readings at our institution—he ordered <3% of the total scans), it seems unlikely that the clinicians surveyed would have had any incentive to overstate the usefulness of the results offered by CTA.
Finally, another possible source of bias in our study was that only clinicians who had found CTA useful would generally refer patients for CTA. Although this limitation could not be avoided in the present study, the large number and diverse background of the referring physicians who participated in our study made it less likely that such a bias was substantial in influencing the results.
Supplementary data
Appendix 1.
Appendix 2.
Supplement Tables.
References
- . Executive Summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA. 2001;285:2486–2497
- . Coronary artery calcium score combined with Framingham score for risk prediction in asymptomatic individuals. JAMA. 2004;291:210–215
- . Association of coronary atherosclerosis detected by multislice computed tomography and traditional risk-factor assessment. Am J Cardiol. 2008;102:316–320
- . Value of stress myocardial perfusion single photon emission computed tomography in patients with normal resting electrocardiograms: an evaluation of incremental prognostic value and cost-effectiveness. Circulation. 2002;105:823–829
- . ACC/AHA 2002 guideline update for the management of patients with chronic stable angina: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Update the 1999 Guidelines for the Management of Patients with Chronic Stable Angina). http://www.acc.org/clinical/guidelines/stable/stable.pdf2002;Accessed on October 24, 2009
- . Value of the history and physical in identifying patients at increased risk for coronary artery disease. Ann Intern Med. 1993;118:81–90
- . 64-Slice computed tomography coronary angiography in patients with high, intermediate, or low pretest probability of significant coronary artery disease. J Am Coll Cardiol. 2007;50:1469–1475
- . Usefulness of multislice computed tomographic coronary angiography to identify patients with abnormal myocardial perfusion stress in whom diagnostic catheterization may be safely avoided. Am J Cardiol. 2007;100:1605–1608
- . Prevalence of obstructive coronary artery disease in an outpatient cardiac CT angiography environment. Int J Cardiol. 2008;129:32–36
- . Coronary computed tomography angiography as a screening tool for the detection of occult coronary artery disease in asymptomatic individuals. J Am Coll Cardiol. 2008;52:357–365
- . Physician response rates to a mail survey by specialty and timing of incentive. Am J Prev Med. 2004;26:234–236
- . Response rates and representativeness: a lottery incentive improves physician survey return rates. Pharmacoepidemiol Drug Saf. 2005;14:571–577
- . In a mailed physician survey, questionnaire length had a threshold effect on response rate. J Clin Epidemiol. 2005;58:103–105
Drs. Blankstein, Nasir, and Shturman received support from grant 1T32 HL076136-02 from the National Institutes of Health, Bethesda, Maryland. This work was partially funded by an internal grant from Partners HealthCare, Boston, Massachusetts. None of the authors have any relation to industry that might pose a conflict of interest.
PII: S0002-9149(09)02915-4
doi:10.1016/j.amjcard.2009.12.040
© 2010 Elsevier Inc. All rights reserved.
