| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Vascular and Interventional Radiology |
1 From the Program for the Assessment of Radiological Technology, Departments of Radiology (R.O., P.M.T.P., M.G.M.H.), Epidemiology and Biostatistics (R.O., T.S., M.G.M.H.), and Vascular Surgery (M.R.H.M.v.S.), Erasmus MC Rotterdam, Room Ee 21-40a, Dr Molewaterplein 50, 3015 GE Rotterdam, the Netherlands; Department of Health Policy and Management, Harvard School of Public Health, Boston, Mass (M.G.M.H.); and Department of Radiology, Maastricht University Hospital and Cardiovascular Research Institute, Maastricht, the Netherlands (M.d.V., M.W.d.H., J.M.A.v.E.). Received June 30, 2004; revision requested September 3; revision received October 8; accepted November 15. Supported by grant 945-01-039 from ZonMw, Netherlands Organization for Health Research and Development, and in part by grant 904-66-091, Netherlands Organization for Scientific Research. Address correspondence to M.G.M.H. (e-mail: m.hunink{at}erasmusmc.nl).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Institutional review board approval and informed consent were obtained. Patients referred for diagnostic imaging work-up to evaluate the feasibility of a revascularization procedure were randomly assigned to undergo either MR angiography or CT angiography. Clinical utility was assessed with therapeutic confidence (scale of 0–10) at initial imaging and with the need for additional imaging. Patient outcomes included ankle-brachial index, maximum walking distance, change in clinical status, and health-related quality of life. Actual diagnostic and therapeutic costs were calculated from the hospital perspective. Differences between group means were calculated with unpaired t tests and 95% confidence intervals.
RESULTS: A total of 157 consecutive patients with peripheral arterial disease were prospectively randomized to undergo MR angiography (51 men, 27 women; mean age, 63 years) or CT angiography (50 men, 29 women; mean age, 64 years). For one of the 78 patients in the MR group, no data were available. Mean confidence for MR angiography (7.7) was slightly lower than that for CT angiography (8.0, P = .8). During 6 months of follow-up, 13 patients in the MR group compared with 10 patients in the CT group underwent additional vascular imaging (P = .5). Although not statistically significant, there was a consistent trend of less improvement in the MR group across all patient outcomes. The average cost for diagnostic imaging was 
359 ($438) higher in the MR group than in the CT group (95% confidence interval: 209, 511 [$255, $623]; P < .001). Therapeutic costs were higher in the MR group, but the difference was not significant.
CONCLUSION: The results suggest that CT angiography has some advantages over MR angiography in the initial evaluation of peripheral arterial disease.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Diagnostic imaging is performed when PAD becomes lifestyle limiting and a revascularization procedure is considered. Decision making prior to surgery or percutaneous intervention depends on accurate characterization of the level, multiplicity, and severity of stenoses (2,3). Both magnetic resonance (MR) angiography and computed tomographic (CT) angiography are increasingly used for noninvasive vascular imaging. Both techniques provide a precise road map for treatment planning. Disadvantages of MR angiography include the higher investment cost for equipment, the small number of cases in which the image is uninterpretable because of artifacts, and the fact that some patients are claustrophobic or have a contraindication to MR imaging. Disadvantages of CT angiography are the use of radiation, the use of potentially nephrotoxic iodinated contrast media, vessel wall calcifications that hamper image interpretation, and the time-consuming three-dimensional reconstruction techniques.
Both MR angiography and CT angiography have been shown to be sensitive and specific techniques for the evaluation of peripheral arteries (4–11). However, the clinical utility, patient outcomes, and associated costs have not yet been evaluated, and, therefore, the decision of whether MR angiography or CT angiography should be used in the diagnostic work-up of PAD remains to be clarified (12,13).
Thus, the purpose of this study was to prospectively compare outcomes following contrast material–enhanced MR angiography and multi–detector row CT angiography as the initial imaging test in the diagnostic work-up of patients with PAD: The primary outcome evaluated was total diagnostic costs, and secondary outcomes evaluated were clinical utility, patient outcomes, and therapeutic and follow-up costs.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
We recruited consecutive patients who were referred from the Department of Vascular Surgery at our tertiary care university hospital (Erasmus MC). The study was approved by the hospital institutional review board, and informed consent was obtained from all patients. The study was performed according to Good Clinical Practice guidelines (14). Data are analyzed and reported in accordance with the guidelines of the Consolidated Standards of Reporting Trials (15).
Study Design
We used an empirically based and pragmatic trial design. That is, the trial was performed to evaluate two diagnostic strategies, either of which could become routine clinical practice. For our hospital, CT angiography was the current practice and MR angiography was considered the new imaging examination.
The patients were randomly assigned to undergo either MR angiography or CT angiography as the initial imaging examination. Randomization was performed centrally and took place through the trial coordinating center by means of telephone. A computer-generated list for the strategy assignment was used. Block randomization was used with a block size of eight patients to obtain equal numbers for both strategies. Eligible patients were enrolled by one of several researchers who were unaware of the randomization sequence. After randomization, patients and clinicians were not blinded to the imaging strategy because this would have been highly impractical and inconsistent with our pragmatic study design.
Imaging Techniques and Evaluation
MR angiography was performed with a 1.5-T imager (Signa; GE Medical Systems, Milwaukee, Wis). A dedicated peripheral vascular phased-array coil (USA Instruments, Aurora, Ohio) was used for signal reception.
We used bolus-chase MR angiography with commercially available software (SmartStep; GE Medical Systems). The following parameters were used: for the aortoiliac region, 4.8/1.5 (repetition time msec/echo time msec), flip angle of 30°, field of view of 400 x 280 mm, section thickness of 2.6 mm, matrix of 384 x 192, centric phase encoding, and an acquisition time of 15 seconds; for the femoropopliteal region, 4.8/1.5, flip angle of 30°, field of view of 400 x 320 mm, section thickness of 2.0 mm, matrix of 384 x 192, centric phase encoding, and an acquisition time of 18 seconds; and for the crural region, 5.1/1.5, flip angle of 30°, field of view of 400 x 360 mm, section thickness of 2.0 mm, matrix of 512 x 512, elliptic centric phase encoding, and an acquisition time of 73 seconds. Zero interpolation was performed to improve image quality. Each patient was administered 45 mL of gadopentetate dimeglumine (Magnevist, 0.5 mmol/mL; Schering, Berlin, Germany) at a rate of 1.2 mL/sec for the first 10 mL and 0.8 mL/sec for the remaining 35 mL (total injection duration, 52 seconds), followed by a saline flush of 15 mL administered at 0.8 mL/sec.
CT angiography was performed with a 16–detector row scanner (Sensation 16; Siemens Medical Systems, Forchheim, Germany). After obtaining an initial scout image (120 kV, 100 mAs), the scanning range was planned to encompass the entire vascular system from the diaphragm to the level of the ankles. Data acquisition was performed craniocaudad with the following parameters: collimation, 0.75 mm; table feed, 18 mm per rotation; gantry rotation period, 0.5 second; pitch, 1.5; x-ray tube voltage setting, 120 kV; and current, 140 mAs. Each patient was administered 120 mL of contrast agent (Visipaque, 320 mg of iodine per milliliter; Amersham Health, Buckinghamshire, England) at a flow rate of 3 mL/sec.
Postprocessing for both MR angiography and CT angiography resulted in 12 angiogram-like images rotated over 180° for the aortoiliac, femoropopliteal, and crural arteries. Two readers with extensive experience in interpreting MR and CT angiograms, a vascular radiologist (M.G.M.H.) with 13 years of postresidency experience and a dedicated researcher (R.O.) with 2.5 years of general radiology training and 1 year of experience in vascular radiology, evaluated all images for arterial stenosis or other pathologic conditions. The following five-point ordinal scale was used to grade stenotic or occlusive disease: score of 0 for 0%–19% stenosis, score of 1 for 20%–49% stenosis, score of 2 for 50%–74% stenosis, score of 3 for 75%–99% stenosis, and score of 4 for complete occlusion. When two or more stenotic luminal lesions were detected in the same vessel segment, the most severe lesion was used for grading. We performed visual assessment of the degree of stenosis. The images were first evaluated independently and were subsequently evaluated with a consensus reading. The consensus readings were used for the data analysis. All images were evaluated without knowledge of further work-up.
Measurement of Clinical Utility
We assessed the therapeutic confidence of vascular radiologists and surgeons during the weekly vascular conference, where therapeutic decisions were made in consensus. In addition to patient history and physical examination, the findings from the initial imaging were discussed, and each clinician was asked to rate his or her individual confidence in making a well-founded therapeutic choice on a 10-point rating scale. Three radiologists (including P.M.T.P. and M.G.M.H.) and four vascular surgeons (including M.R.H.M.v.S.) who were aware of the study design completed the confidence forms during the vascular conferences.
Furthermore, we measured the recommendations for additional imaging (duplex ultrasonography [US], digital subtraction angiography, MR angiography, or CT angiography) during the vascular conference. Any additional vascular imaging performed within 60 days after the initial test was noted.
Measurement of Patient Outcomes
Health-related quality of life was assessed by using a self-administered questionnaire sent to all patients at the time of randomization and at 2 weeks, 3 months, and 6 months after initial imaging. The questionnaires contained the questions from the EuroQol-5D, the Rating Scale, the generic Medical Outcomes Study 36-Item Short Form Health Survey (SF-36), and the disease-specific VascuQol.
The EuroQol-5D covers five different health dimensions, including mobility, self care, usual activities, pain and discomfort, and anxiety and depression, which include a total of 243 different health states. With use of a published population-based utility function, a single index score was calculated for each patient (16). A value of 0 indicated death and a value of 1 indicated maximum health.
The Rating Scale is a valuation instrument that consists of one question in which a patient is asked to rate his or her current state of health on a scale from 0 to 100 for which 0 represents death and 100 represents perfect health (17).
The SF-36 is a multiple-item scale that covers eight different health dimensions (18). On the basis of a previous study, we determined that physical functioning, role-functioning limitations due to physical problems, bodily pain, and general health were the relevant dimensions to describe health status with PAD (19). Each dimension is valued on a 100-point scale for which 0 indicates death and 100 indicates maximum health.
The VascuQol is a disease-specific descriptive quality-of-life instrument used especially for patients with PAD that contains five domains (activity, symptom, pain, emotion, and social functioning) (20). Scores for these five domains add up to a total score, which is valued on a seven-point scale for which 1 indicates poor quality of life and 7 indicates maximum health.
For each patient, we compared the scores of the different quality-of-life measurements at 2 weeks, 3 months, and 6 months of follow-up with the baseline scores of that particular measurement, which resulted in a mean improvement for each quality-of-life measurement. We used standard rules for item recoding, treatment of missing items, and scoring (16–18,20). If a questionnaire was not returned because the patient had died, we gave a value of 0 for the EuroQol-5D, Rating Scale, and SF-36 scores. For the VascuQol, if the patient died the score was considered missing because this is a disease-specific questionnaire that does not cover the health status of death.
The brachial, dorsal pedal, and posterior tibial arterial systolic pressures were assessed by using a blood pressure cuff and continuous-wave Doppler US, both before starting and immediately after completion of the treadmill test, to determine the resting and postexercise ABIs. The measurements were performed by two vascular technologists who each had more than 5 years of experience at the time. To calculate the ABI, the value of the highest ankle pressure was divided by the value of the highest brachial pressure. A treadmill test based on a standard protocol (4.0 km/h at 0%) was performed to assess the maximum walking distance. The patients walked until they had to stop because of leg pain or because they reached the time limit of 5 minutes. Both ABI and maximum walking distance were measured at baseline and after 6 months of follow-up.
Furthermore, we assessed the change in clinical status during 6 months of follow-up. For this purpose we used the criteria for reporting significant change in clinical status according to Rutherford et al (21). These criteria are a combination of standard clinical categories with objective ABIs.
Improvement in ABI and change in clinical status during the trial period were assessed for the treated leg only. If both legs were treated or if neither leg was treated, we selected the leg with the most severe symptoms at baseline. If a patient had the same symptoms in both legs at baseline, we selected a leg at random.
If a patient was treated by means of amputation, the ABI and maximum walking distance were scored as 0. Patients who did not undergo a treadmill test because they had died were excluded from this analysis.
Measurement of Costs
For the cost analysis, we collected information concerning all relevant items of medical care (ie, diagnostic and therapeutic) used by each patient during the entire trial to calculate the mean cost per imaging strategy per patient. The cost of diagnostic imaging included the initial imaging examination, all additional vascular imaging, and the associated hospital admissions during the 6 months of follow-up. If initial imaging was technically inadequate, a new MR angiography or CT angiography examination was performed or an additional imaging test was performed. This would result in a more expensive initial test or in higher costs for additional imaging. The therapeutic cost included costs for percutaneous vascular interventions (ie, percutaneous angioplasty, stent placement, and thrombolysis), vascular surgery (ie, aortic bifurcation reconstruction, bypass surgery, endarterectomy, and amputation), and associated hospital admissions during the 6 months of follow-up. In addition, the number and kind of percutaneous vascular interventions and vascular surgery were compared between the groups. Futhermore, we assessed the costs for outpatient visits during 6 months of follow-up. All costs were computed from the hospital perspective according to the Dutch guidelines for cost calculations in health care (22).
Diagnostic costs can be divided into directly and nondirectly assignable costs. Directly assignable costs include personnel costs, material costs such as film, and equipment costs such as investment, servicing, and construction costs. Personnel costs were computed by using the measured time spent on a diagnostic imaging test for each involved personnel category and the mean wage rates from our hospital. Social security of 37% of the wage was added in accordance with national guidelines. Costs of supplies used in diagnostic procedures were based on cost prices and summed. The annuity costs (23) of the radiologic equipment and the annual equipment servicing costs were summed and then divided by the proportion of the total available room time (80% of a 40-hour work week) (22,23). Costs were discounted at a rate of 3% per annum (24). Nondirectly assignable costs include costs of supporting departments, housing costs, and overhead costs. Information on costs of supporting departments was obtained from records of our Financial and Economics Department. The costs for housing were computed for the involved radiologic rooms by multiplying the surface space with the housing costs of 204 per square meter per year. The overhead costs for MR angiography and CT angiography were estimated to be 15% of the directly assignable costs (22).
The costs of percutaneous vascular interventions were measured and calculated in a similar fashion. We obtained unit costs of surgery from another study with a comparable study domain and setting to calculate an overall cost per patient per surgical procedure (25). For some surgical procedures performed in our study, the unit costs were not available from the other study (25), and we had to estimate these costs (M.R.H.M.v.S., oral communication, March 1, 2004). The number of days of hospital admission and the number of outpatient visits were collected, and the associated costs were calculated by using national estimates of hospital admission, intensive care unit admission, and outpatient visits (22). All costs were reported in euros and dollars for the year 2002 (the exchange rate was 0.82 per U.S. dollar, September 2004).
Statistical Analysis
For each moment in time we calculated the response rate for the quality-of-life questionnaires. Furthermore, for 20% of both the quality-of-life data and the data of the case record form, double entry was performed to calculate the entry error.
The required sample size was estimated on the basis of the primary outcome, which was the mean estimated strategy costs (ie, total diagnostic costs) per patient. Estimates for the total diagnostic costs were obtained from a previously performed cost analysis in our hospital. To demonstrate a significant difference between the strategy costs for MR angiography (estimated to be 550, with a standard deviation of 400) and the strategy costs for CT angiography (estimated to be 350, with a standard deviation of 300), a power of 0.90, and an 
level of .05 would require at least 66 patients per strategy. To allow for some redundancy we included at least 12 extra patients per strategy.
The results were analyzed according to the intention-to-diagnose-and-treat principle. This implies that once a patient has been randomly assigned, he or she will remain in the assigned group for the analysis regardless of whether crossover to the other strategy occurred and of whether follow-up was complete or not. We calculated the means (± standard deviations) of the therapeutic confidence scores for the initial imaging test, the numbers of additional imaging tests performed, the quality-of-life scores at follow-up, the ABI, the maximum walking distance, the change in clinical status, the diagnostic costs, the therapeutic costs, the costs of outpatient visits, and the total costs for both groups. When data were normally distributed, we assessed the significance of differences between group means by using unpaired t tests and calculated 95% confidence intervals (CIs). We used the 
2 test for dichotomous outcomes and the Mann-Whitney test for ordinal outcomes.
In addition, we analyzed the differences adjusted for predictive baseline characteristics by using multivariable linear and logistic regression. On the basis of previous studies (26) and clinical experience, we assumed that severity of disease (critical ischemia vs claudication), renal disease (ie, renal insufficiency and renal transplantation), cerebrovascular disease, cardiac disease, and diabetes mellitus at baseline were potentially predictive for the outcomes. To adjust for the learning curve of the physicians and analyze trends in the outcome measures, we included the rank order of the initial imaging tests in the regression analysis. We expressed the rank order by ranking the dates when the initial imaging tests were performed. To analyze the improvement in quality of life, ABI, and maximal walking distance during follow-up, we also adjusted for the baseline scores of these outcome measures. To adjust for variability of interpretation, we normalized confidence scores from each physician (27). To analyze the trend of increasing experience over time for the confidence, we plotted the confidence scores as a function of chronologic ranking of the dates when the initial imaging tests were performed and fitted a regression line. A one-way sensitivity analysis was performed for the diagnostic costs by exploring a range of 50%–200% of the investment costs of radiology equipment. Another sensitivity analysis was done for the costs of surgical procedures, excluding the outliers. Outliers were defined as a cost more than the mean surgical costs plus 3 standard deviations.
For all outcome measures, we used mean imputation for missing values. A P value of .01 was considered to indicate a statistically significant difference for the quality-of-life outcomes and the costs because of multiple testing. For other tests, a P value of .05 was used. Calculations were performed with SPSS 11.0 for Windows (SPSS, Chicago, Ill).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
|
|
Patient Outcomes
The response rate for the quality-of-life questionnaires was 99% at baseline, 93% at 2 weeks, 89% at 3 months, and 89% at 6 months of follow-up.
The improvement in the EuroQol-5D index from baseline to 2 weeks, 3 months, and 6 months of follow-up was not statistically significant for either the MR or the CT group (Table 2). The improvement was slightly smaller in the CT group than in the MR group. With adjustment for baseline scores, potentially predictive variables, and rank order, the improvement was slightly smaller in the MR group than in the CT group, but there was no statistically significant difference between the groups (Table 2).
|
The improvement in all dimensions of SF-36 was slightly smaller in the MR group than in the CT group (Table 2). With adjustment for baseline scores, predictive variables, and rank order, we found a statistically significant difference (P = .01) in favor of CT angiography for the dimension of physical functioning (–10; 95% CI: –17, –2) at 3 months of follow-up (Table 2).
The improvement in the VascuQol was statistically significant for both the MR and the CT groups at 3 and 6 months of follow-up (Table 2). Also for the VascuQol, the improvement from baseline to follow-up was slightly smaller in the MR group than in the CT group, but there was no statistically significant difference between the groups with and without adjustment for baseline scores, predictive variables, and rank order (Table 2).
For both groups, no significant trend with increasing experience could be demonstrated for improvement in all measures of quality of life.
Both groups showed a significant improvement in resting ABI, postexercise ABI, and maximum walking distance during the follow-up period (Table 3). However, the improvement was slightly smaller in the MR group than in the CT group. This difference in improvement was not statistically significant between the groups with and without adjustment for baseline scores, potentially predictive variables, and rank order. For both groups, no significant trend with increasing experience could be demonstrated for improvement in ABI and maximum walking distance. The change in clinical status was slightly higher in the MR group than in the CT group (Table 3).
|
514 ($627) for all MR angiography examinations (range, 331–1098 [$404–$1340]) and 163 ($199) for all CT angiography examinations (range, 118–238 [$144–$290]). For the additional imaging tests, the mean unit cost was 1223 ($1492) for all diagnostic digital subtraction angiography examinations (range, 688–2474 [$839–$3018]) and 43 ($52) for all duplex US examinations (range, 40–49 [$49–$60]).
The total diagnostic costs per patient were significantly higher in the MR group than in the CT group (a difference of 359 [$438]; 95% CI: 209, 511 [$255, $623]; P < .001) (Table 4). This increase in diagnostic costs was not caused by more additional imaging but by the higher unit costs of the initial imaging in the MR group (Table 4). With adjustment for potentially predictive variables and rank order, the increase in diagnostic costs remained significant (326 [$398]; 95% CI: 174, 478 [$212–$583]; P < .001).
|
274 ($334) in diagnostic costs (95% CI: 125, 423 [$153, $516]; P < .001). A 200% increase in investment costs of CT equipment demonstrated a similar result (a difference of 343 [$418]; 95% CI: 192, 494 [$234, $603]; P < .001). Even the combination of a 50% reduction in the investment costs of MR equipment and a 200% increase in the investment costs of CT equipment resulted in a significant difference of 258 ($315) in favor of CT angiography (95% CI: 108, 407 [$132, $497]; P = .001).
One-third of the patients in each group underwent percutaneous intervention, one-third underwent surgical intervention, and one-third underwent conservative treatment. On average, more percutaneous interventions and surgical procedures per patient were performed in the MR group than in the CT group. For surgical procedures this difference was statistically significant (P < .001) (Table 5). The total costs of percutaneous interventions averaged 1379 ($1682) (standard deviation, 1834 [$2238]) per patient in the MR group and 1078 ($1315) (standard deviation, 1636 [$1996]) in the CT group, which was not a statistically significant difference (difference of 301 [$367]; 95% CI: –249, 850 [–$304, $1037]; P = .3) (Table 4). The mean cost for surgical procedures was 2118 ($2584) (95% CI: –142, 4378 [–$173, $5341]; P = .07) higher in the MR group. With one-way sensitivity analysis for the surgical costs, two outliers in the MR group were excluded, which also resulted in higher costs in the MR group than in the CT group (difference of 1100 [$1342]; 95% CI: –697, 2920 [–$850, $3562]; P = .2). With adjustment for predictive variables and rank order, similar results were demonstrated for the costs of percutaneous and surgical interventions.
|
The total cost was 2784 ($3397) higher in the MR group (95% CI: 549, 5020 [$670, $6124]), which was not a statistically significant difference when considering the multiple comparisons that we performed (P = .02). Also, with adjustment this difference was not statistically significant (P = .03) (Table 4).
Both MR angiography and CT angiography showed a significant trend of decreased diagnostic costs with increasing experience and a trend of increased costs for percutaneous interventions. The costs for surgical procedures, outpatient visits, and total costs showed no significant trend with increasing experience.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
We found that the therapeutic confidence for CT angiography was slightly higher than that for MR angiography, but both were comparable to the mean therapeutic confidence of 8.2 for digital subtraction angiography that was observed in a previous study (27). For CT angiography, the confidence remained constant with increasing experience, but for MR angiography, we found a trend of increased confidence with increasing experience. This is exactly what we expected beforehand because CT angiography is used in our current practice and MR angiography is the new test in our hospital. Probably because of the lower confidence in MR angiography, the physicians requested additional imaging more frequently in the MR group than in the CT group. Because this learning curve could have an effect on the outcomes, we adjusted for the learning curve in the regression analysis.
Although not a statistically significant difference, we observed a consistent trend of more improvement in the CT group than in the MR group for all patient outcomes. This occurred despite the fact that the MR group had lower quality-of-life scores at baseline and therefore had more to gain with treatment. As expected, with adjustment for baseline scores, potentially predictive variables, and rank order, the difference in improvement in quality of life was even larger in favor of the CT group. This could indicate that CT angiography more accurately depicts the extent and localization of disease, which results in better treatment and better patient outcomes. On the other hand, the larger number of interventions performed in the MR group could have had a negative influence on the short-term patient outcomes. Furthermore, the different measures of quality of life are correlated, and this consistent trend could simply be a chance finding.
The mean diagnostic cost was significantly higher in the MR group than in the CT group, and this is because of the higher costs of the initial imaging. MR angiography is more expensive than is CT angiography because of higher investment costs, construction costs, costs for the contrast agents, and personnel costs.
Although not a statistically significant difference, both the cost for percutaneous interventions and the cost for surgical procedures were higher for the MR group than for the CT group. One-third of the patients in each group underwent a percutaneous intervention and one-third underwent a surgical intervention, but, on average, more percutaneous interventions and surgical procedures per patient were performed in the MR group than in the CT group. For the surgical procedures this difference was statistically significant. Also, here one can reason that better identification of the extent and localization of disease led to less therapeutic interventions in the CT group, but this too could be a chance finding.
A cohort study has traditionally been used for the evaluation of new diagnostic imaging tests by performing both the new test and the reference test in all patients to determine the sensitivity and specificity. For both MR angiography and CT angiography, a sensitivity of 91%–98% and a specificity of 92%–99% have been reported (4–11). However, these results are difficult to translate into a meaningful clinical decision with respect to whether the new diagnostic strategy should actually be implemented. A decision about the usefulness of a new diagnostic strategy requires either decision analysis or a randomized controlled trial (12). Although randomized controlled trials are not frequently used to evaluate diagnostic tests, we found our pragmatic randomized trial to be both feasible and inexpensive (13,28–31).
We acknowledge several limitations of our study. One limitation of the study was that, although patients were randomized, there were differences between the groups at baseline regarding diabetes mellitus, cardiac disease, renal disease, and critical ischemia. These baseline differences were not statistically significant, but we believed it would be prudent to adjust for predictive baseline variables that may lead to differences in outcomes (32,33). Therefore, adjustment for potentially predictive variables in a multivariable or logistic regression was used to correct the estimates of the outcomes for any imbalance that by chance might have occurred between the randomized groups.
A possible limitation was that patients and physicians were not blinded for group allocation. At the same time, the goal of our study was to evaluate the outcomes of the diagnostic tests as they are used in routine clinical practice. Patients could not be blinded, and blinding of the treating physicians, for example by means of transferring the diagnostic information to a schematic drawing, would have introduced an artificial step that could have hampered diagnostic interpretation and therapeutic planning. Furthermore, although schematic drawings are used in routine clinical practice as adjunct, they are not used alone when the imaging test provides a good road map. Finally, we chose not to blind physicians for clinical findings because this would have hampered therapeutic planning and would have been inconsistent with our goal of comparing the tests as they are used in clinical practice.
Another possible limitation is related to the generalizability of the results. In our study population, 79% of the patients (123 of 156) had intermittent claudication and only 21% (33 of 156) had critical ischemia (Table 1). One could argue that this study population was very healthy and that we used an aggressive treatment strategy for this group, which raises the question of generalizability. However, the PAD populations described in several other articles about the accuracy of MR angiography or CT angiography also consisted of 20%–23% patients with critical ischemia (5,7,34,35). All these patients were referred to undergo digital subtraction angiography for treatment planning. Furthermore, the management algorithm for intermittent claudication in the TransAtlantic Inter-Society Consensus document recommends invasive therapy if a patient has severe disabling claudication and walking exercise is not successful. In addition, a review of the literature about above-knee femoropopliteal bypass procedures described a population with PAD in which 39%–45% of patients had intermittent claudication (36), which is consistent with the percentage of patients in our study who underwent surgery for intermittent claudication. Thus, our study population is comparable with study populations at other institutions, which supports the generalizability of our results.
Another limitation is that the reported costs may be unique to our institution and thus cannot be generalized to other hospitals. Therefore, we compared our cost estimates with those at two other national university hospitals and one national private hospital. We found our costs to be within the range of costs in other settings. Furthermore, we performed a one-way sensitivity analysis to analyze the effect of uncertainty in the diagnostic cost estimates. Varying the investment costs of equipment did not affect our conclusions.
Finally, the costs were calculated from a hospital perspective instead of a societal perspective. There is international consensus that an economic evaluation should be performed from the societal perspective. A societal perspective implies that not only the costs within the health care sector but also the direct costs (ie, patient costs) and indirect costs (ie, costs of production losses) outside the health care sector have to be included in the cost analysis (24). In our study, we chose the hospital perspective because results of other studies performed to assess the costs related to the management of PAD showed that patient costs were low in both Dutch and U.S. settings (37,38). Furthermore, in the setting of PAD, the costs of production losses are negligible because most patients are retired (39).
In conclusion, the results suggest that CT angiography has some advantages over MR angiography in the initial imaging evaluation of patients with PAD. There were no statistically significant differences in outcomes between the groups except for the total diagnostic costs, which were significantly lower in the CT group compared with the MR group.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: ABI = ankle-brachial index • CI = confidence interval • PAD = peripheral arterial disease • SF-36 = Medical Outcomes Study 36-Item Short Form Health Survey
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, M.G.M.H.; study concepts, all authors; study design, M.G.M.H., P.M.T.P., J.M.A.v.E.; literature research, R.O.; clinical studies, R.O.; data acquisition, R.O.; data analysis/interpretation, all authors; statistical analysis, M.G.M.H., R.O.; manuscript preparation, R.O.; manuscript editing, M.G.M.H., R.O.; manuscript definition of intellectual content, revision/review, and final version approval, all authors
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. Ouwendijk, M. de Vries, T. Stijnen, P. M. T. Pattynama, M. R. H. M. van Sambeek, J. Buth, A. V. Tielbeek, D. A. van der Vliet, L. J. SchutzeKool, P. J. E. H. M. Kitslaar, et al. Multicenter Randomized Controlled Trial of the Costs and Effects of Noninvasive Diagnostic Imaging in Patients with Peripheral Arterial Disease: The DIPAD Trial Am. J. Roentgenol., May 1, 2008; 190(5): 1349 - 1357. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. J. White and W. A. Gray Endovascular Therapies for Peripheral Arterial Disease: An Evidence-Based Review Circulation, November 6, 2007; 116(19): 2203 - 2215. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Schernthaner, D. Fleischmann, F. Lomoschitz, A. Stadler, J. Lammer, and C. Loewe Effect of MDCT Angiographic Findings on the Management of Intermittent Claudication Am. J. Roentgenol., November 1, 2007; 189(5): 1215 - 1222. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. H. Heijenbrok-Kal, M. C. J. M. Kock, and M. G. M. Hunink Lower Extremity Arterial Disease: Multidetector CT Angiography Meta-Analysis Radiology, November 1, 2007; 245(2): 433 - 439. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Roos, D. Fleischmann, A. Koechl, T. Rakshe, M. Straka, A. Napoli, A. Kanitsar, M. Sramek, and E. Groeller Multipath Curved Planar Reformation of the Peripheral Arterial Tree in CT Angiography Radiology, July 1, 2007; 244(1): 281 - 290. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. White Intermittent Claudication N. Engl. J. Med., March 22, 2007; 356(12): 1241 - 1250. [Full Text] [PDF]
|
|
|
|
|
|
R. Ouwendijk, M. C. J. M. Kock, L. C. van Dijk, M. R. H. M. van Sambeek, T. Stijnen, and M. G. M. Hunink Vessel Wall Calcifications at Multi-Detector Row CT Angiography in Patients with Peripheral Arterial Disease: Effect on Clinical Utility and Clinical Predictors Radiology, November 1, 2006; 241(2): 603 - 608. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Cina, R. Ouwendijk, and M. G. M. Hunink First Challenge of MR versus CT Peripheral Angiography: How to Identify a Winner? Radiology, July 1, 2006; 240(1): 302 - 304. [Full Text] [PDF]
|
|
|
|
 |
|
 B. L. Dolmatch Commentary Perspectives in Vascular Surgery and Endovascular Therapy, June 1, 2006; 18(2): 191 - 193. [Abstract] [PDF]
|
|
Read all eLetters
| |||||||||||||
