Since its introduction in 1993,1 contrast-enhanced magnetic resonance angiography (ce-MRA) has been considered the technique of choice for angiographic examination of most regions of the body. This is because of several technical and clinical advantages in comparison with non-contrast-enhanced MRA (nce-MRA) imaging. In this article, we focus on patients with suspected peripheral arterial occlusive disease (PAOD) who were indicated for both MRA approaches; therefore, the clinical, technical and cost evaluation is restricted to this patient group. Specific nce techniques known as time-of-flight (TOF) methods have been investigated for peripheral MRA, despite their general and well-known limitations in clinical routine. These limitations are related to two main concerns. First, additional acquisition time is needed to obtain adequate images, thus leading to a reduced patient throughput, and second, more artefacts occur. Compared with ce-MRA, TOF MRA is more time-consuming and hence often less appropriate for imaging body regions with larger fields of view (FoV). Among other reasons, time consumption is extended by the necessity of saturation pulses orientating the imaging slice perpendicular to the blood flow. As a result, TOF MRA is particularly sensitive to patient motion. Additionally, signal losses are encountered when flows are complex or turbulent, e.g. in stenotic areas, leading to overestimation of the degree and length of stenoses. Furthermore, saturation effects and ghosting artefacts due to vessel pulsatility may be disadvantageous.2–4 These limitations in terms of motion sensitivity and long acquisition times for larger FoV are less critical for intracranial MRA. In addition, compared with other body regions, a short time window between arterial and venous phase in ce-MRA is obtained, making ce-MRA correspondingly more demanding. Consequently, nce-MRA can be routinely used for many intracranial indications; however, the situation is different for peripheral MRA for diagnosing PAOD.
Recently, technical progress has been achieved in terms of more sophisticated nce-MRA sequences that may reduce artefacts and thus allow the imaging of peripheral vessels with acceptable resolution. These technical aspects, as well as the recently observed possible link between occurrence of nephrogenic systemic fibrosis (NSF) and the use of gadolinium-based MR contrast agents, have motivated equipment manufacturers to develop more advanced nce techniques. For example, novel nce-MRA techniques based on T2 effects using ultrafast spin-echo sequences (fresh blood imaging [FBI], nce-MRA/FBI) have been published and promoted as potential alternatives to standard ce-MRA.5–9 A more recent review of nce-MRA methods with focus on these newer, commercially available techniques can be obtained in the reference list.10
The major equipment manufacturers have used different technical development strategies for their nce-MRA applications. Toshiba Medical Systems has promoted three nce techniques – nce-MRA/FBI, contrast-free improved angiography (CIA) and time–spatial labelling inversion pulses (time-slip) – and has implemented them in the fourth generation of the Toshiba Vantage system. Recent developments by Philips Medical Systems have comprised technologies termed triggered angiographic non-contrast enhancement (TRANCE) and phase-contrast angiography (PCA- SENSE). TRANCE was introduced with the 11th release of software by Philips, while PCA-SENSE has been available for more than eight years, starting with the ninth release. Siemens has actively promoted new techniques for peripheral MRA (NATIVE truFi), head and neck MRA (NATIVE SPACE) and cerebral perfusion MRA (ASL Perfusion). General Electric (GE) Healthcare also offers a number of more recent pulse sequences on their currently available HDx and HDe systems.
Taken together, the most prominent techniques today for nce-MRA are TOF, PCA, FBI, balanced steady-state free precession (bSSFP), arterial spin-labelling (ASL) and dark-blood imaging. However, of these, TOF and nce-MRA/FBI are most often used for MRA of the peripheral vessels. The following discussion will therefore be restricted to these methods.
The increasing expectation of healthcare payers that providers will deliver cost-effective services heightens the demand for greater procedural efficiency. The aim of this study was to compare the technical and clinical utility of the TOF and nce-MRA/FBI techniques with those of ce-MRA in the specific context of diagnosing peripheral arterial occlusive disease in the lower extremities. In addition, the competitiveness in costs of ce-MRA compared with nce-MRA/FBI was evaluated, the main focus being on costs of the investigation. The three main questions that our analysis was intended to answer were: what are the technical and clinical advantages of the different MRA procedures? How do differences in structure and duration of the procedures influence the costs accruing to the service provider? What are the costs of conducting a diagnosis on a patient with suspected PAOD when ce-MRA is used, compared with nce-MRA/FBI?
Clinical and Technical Utility of TOF, nce-MRA/FBI and ce-MRA – A Mini-review
The respective technical and clinical strengths and disadvantages of nce- and ce-MRA techniques differ in nature. The techniques under consideration are those used in an indication for peripheral MRA. These are mainly TOF, nce-MRA/FBI and ce-MRA. The comparison of clinical and technical utility performed here was based on published literature.
Compared with other nce-MRA methods, the advantage of TOF-based techniques is their relatively short acquisition time. However, this acquisition time depends on the size of the FoV, the spatial resolution, the magnetic field strength and the specific sequence used, and also on the possibility of using additional saturation pulses to selectively depict arteries and veins. The radiologist can also achieve suppression of tissue background, which provides good vessel contrast. Usually, operators are familiar with TOF techniques because of their wide use in intracranial MRA. On the other hand, standard TOF images are sensitive to the patient’s movements and to variations in the velocity and direction of blood flow. This may lead to a loss of signal for in-plane vessels, oblique vessels, vessels forming loops, slow-flow regions and turbulent or retrograde flow.11 Saturation of blood imposes size limitations on the FoV, which in turn leads to inefficient coverage of long vessels and incomplete suppression of fresh thrombi. Most clinically used TOF methods do not provide dynamic information. The strength of TOF is the possibility of arterio-venous separation and single-station unidirectional flow.
nce-MRA/FBI is a more recent technical development. It requires well-trained personnel and long scan times. If used correctly, nce- MRA/FBI provides large anatomical coverage, good spatial resolution even in the peripheral vessels (at least better than TOF) and good separation of arteries and veins.12 Like other spin-echo sequences, nce-MRA/FBI is less sensitive to susceptibility artefacts than gradient- echo sequences. nce-MRA/FBI can be combined with ASL. On the other hand, extensive scan preparation requirements, including electrocardiogram (ECG) trigger and determination of appropriate spoiler gradients, are time-consuming and make extensive demands on the operator’s time. Therefore, the operator needs to be experienced and specifically trained. This technique is also unsuitable for patients with arrhythmia and, especially, tachyarrhythmia. Ultrafast spin-echo can show low signal or signal voids, leading to impairment of image quality. Furthermore, there is limited robustness in vessels with varying flow patterns or collateral vessels, and T2 blurring artefacts may also be experienced. nce-MRA/FBI allows multistation MRA and a section thickness of 4mm (interpolated to 2mm). For parallel imaging, 1mm is feasible; however, the signal and arterio-venous separation depend on flow changes during the cardiac cycle; if flow changes are small (e.g. in connection with extended occlusions or reverse blood flow), a reduced signal is to be expected.
Unlike nce techniques, ce-MRA allows large FoVs with an imaging plane along the vessels of interest.13 It reduces or eliminates most of the artefacts of TOF by avoiding flow-related problems caused by flow direction, flow rate, turbulent flow and blood saturation. The images can be acquired rapidly within the breath-hold capacity of the patient, which is important for abdominal and pelvic MRA in many clinical settings. Along with valuable options of dynamic imaging, ce-MRA has proved to be robust, with high diagnostic output.2–4 In patients with arteriovenous malformations in particular, time-resolved multiphase ce-MRA of the hand can deliver valuable diagnostic information and supports surgical treatment planning.14
Nevertheless, except for time-resolved ce-MRA techniques, proper timing of the start of the scan relative to the contrast injection is essential in order to ensure high-quality images of the arterial system without venous overlay. Generally, the timing can be determined by fluoroscopic bolus techniques or by administration of a test bolus. For first-pass imaging there is a trade-off between spatial resolution and acquisition time. Easy-to-use protocols such as Smart Prep or Care Bolus are available from respective MRI device manufacturers to ensure optimal and convenient bolus timing.15
The strengths of ce-MRA are its high sensitivity for slow flow and its increased sensitivity and specificity in particular when parallel imaging is used. A true voxel size of 0.6mm3 is feasible in lower limbs, especially when parallel imaging such as SENSE (Philips), GRAPPA (Siemens) or ASSET (GE) is being used, while time-resolved imaging of contrast kinetics (TRICKS) or time-resolved echo-shared angiography (TREAT) for the distal run-off vessels may help to identify target vessels for surgery in cases of asymmetrical flow. The limitation is the acquisition time for abdominal and lower leg. In order to minimise potential risks of NSF for patients with severely impaired renal function, exact adherence to the manufacturer’s instructions is imperative.
Irrespective of the imaging method used, in acute ischaemia the main challenge of the diagnostic work-up is the depiction of occlusions and collateral vessels because of slow flow and/or small rudimentary vessel segments due to limited spatial resolution. The main challenges in critical limb ischaemia are reliable stenosis grading, the depiction of collateral and patent distal vessels and the depiction and quantification of stenosis length.16
In clinical routine, digital subtraction angiography (DSA) is widely used and is frequently employed as the reference standard in clinical studies to evaluate new imaging procedures. Especially for peripheral angiography, DSA is regarded as the gold standard. However, for critical limb ischaemia, Kreitner et al.17 have shown that, despite a lower spatial resolution compared with DSA, ce-MRA is valuable in the diagnostic work-up of the most challenging subgroup of PAOD patients; that is, those with chronic ischaemia. These authors find “...significant superiority of MR angiography to DSA for the detection of patent pedal vessels...”.17 The main advantage in this context is the ability of ce-MRA to detect blood flow as slow as 2cm/second, whereas DSA shows limitations because of its much lower contrast sensitivity. Clinical studies comparing ce-MRA with conventional catheter angiography show convincing overall sensitivities between 91 and 94% for the detection of vascular disease, and specificities between 90 and 93%.18 Large studies demonstrate that ce-MRA of the whole body has the potential to change the treatment of patients due to additional detected pathologies.19,20
The major characteristics of TOF, nce-MRA/FBI and ce-MRA for peripheral MRA are summarised in Table 1, which is based on our experience (author personal communication).
Overall, the main limitations of peripheral nce-MRA are the need for complex techniques and the relatively long acquisition times required for high resolution. Separation of arteries and veins with nce-MRA may be difficult depending on the technique used, and often makes high technical demands on the equipment and the operator. For nce- MRA/FBI sequences, an ECG trigger has to be prepared and the flow- gradient strength has to be optimised. The latter depends both on the body region and the grade of proximal stenosis.
The use of nce-MRA has been established in only one body region. For imaging of the intracranial arteries, nce techniques are preferred because ce-MRA is difficult owing to the small time window between the arterial and venous phases. On the other hand, typical disadvantages of nce-MRA resulting from patient movement are reduced or absent, so that longer nce sequences can be carried out.
Awareness of the potential risk of NSF in some patients has given momentum to the promotion of nce-MRA techniques, mainly by Toshiba Medical USA and to a lesser extent by Siemens, Philips and GE.21 Nevertheless, despite all improvements, nce-MRA still suffers from fundamental disadvantages. These are:
- more examination time is needed, resulting in reduced patient throughput;
- more artefacts are obtained, which reduces diagnostic confidence; and
- more technical difficulties arise, reducing reproducibility and robustness.
Therefore, ce-MRA remains the method of choice for most indications, except for certain applications such as imaging of the intracranial arteries. ce-MRA adds diagnostic value with less effort because of its greater robustness, high spatial resolution, which allows the imaging of patent distal vessels, reliable stenosis grading, fewer technical failures, fewer non-diagnostic images and a broader range of applications. ce-MRA is superior from the technical and clinical perspectives. Therefore, we investigated whether ce-MRA is also competitive in terms of costs of investigation.
Evaluation of Investigation Costs Methods
The costs were assessed on the basis of published literature and an earlier cost study for ce-MRA in the lower extremities, in which costs of a ce-MRA with manual table-feed technique and a standard DSA investigation were compared from the hospital’s perspective. The cost study referred to (‘Study 1’) was conducted by Bayer Schering Pharma AG in German hospitals with a sample of 11 observations for each procedure.22 The main emphasis of Study 1 was the observation of time per work step to run the MRA, the association of these times with costs and the determination of fixed costs for personnel and for equipment use. Investigation costs for ce-MRA were found to be ~20% lower than those of standard DSA (€142.38 and €176.63, respectively), excluding the costs of consumables such as contrast agents and catheters.
For nce-MRA/FBI, no direct or published cost data were available for use in the study in question. Therefore, the results for ce-MRA were adapted to nce-MRA/FBI by re-calculations based on data from recent publications,5–9 as explained below. Investigation costs were the direct costs of the examination plus overheads, the latter comprising costs indirectly linked to the MRA examination ‘Examination’ costs were assigned according to the experience gained in Study 1. These comprised costs for personnel and equipment on the basis of linear depreciation, including interest on a loan for the capital outlay, electricity costs (adjusted for the duration of use), costs for the examination and resting rooms (adjusted for the duration of use) and a fixed sum for administration. The overhead costs (‘other’) are for other personnel, furniture, cleaning, waiting room, insurance, maintenance of equipment, scientific literature, travel, seminars, telephone, literature for the waiting room and so forth; by their nature, these costs were independent of the imaging technique.
Personnel costs for the clinicians were calculated on the basis of a published analysis of salaries.23 For the technician (radiographer) the tariff of the German civil service (Tarifvertrag für den öffentlichen Dienst [TVöD, Kommunen West]) was used, assuming (as in Study 1) an average of three years’ job experience (Entgeltgruppe 6, Stufe 3).24,25 The personnel costs were based on the average working time that the various staff members (head/registrar, assistant physician, technician or radiographer) spent on the examination process.22 It was assumed that the working time of the technicians increases in proportion to the usage time of the equipment; a high standard of qualification for the technician is needed. The working time of the clinicians for nce-MRA/FBI was assumed to be at least as long as for ce-MRA (anamnesis, etc., would be independent of the technique, while the greater time needed for diagnosis in nce-MRA/FBI would more than outweigh that required for placing the needle for contrast-medium injection in ce-MRA).
Occupation times of the MR suite are the sum of pre- and intra- diagnostic, acquisition and post-diagnostic times. Pre-diagnostic times include the preparation and placing of the patient in the scanner and documentation. For ce-MRA, extra time is required for preparing the contrast-medium injection. The intra-diagnostic times cover the start of the examination, the preparatory sequences (scout, TOF imaging for vessel localisation) and the imaging of four sequences that will later be subtracted from the contrast-enhanced sequences. For ce-MRA, extra time for the application of the contrast agent is added. Post- diagnostic times include the time between the end of the acquisition and patients leaving the MR room, cleaning the equipment, data handling, printing, archiving or other preparation of images, plus for ce-MRA the time needed to remove the injection needle. In Study 1, each of the four ce-MRA sequences was assumed to take one to 30 minutes, giving a total acquisition time of six minutes for ce sequences. Overall, the MR suite occupation time added up to 31–55 minutes.22 Published data on nce-MRA/FBI5–9 suggest that the acquisition time needed to run the nce-MRA/FBI sequences requires the scanning of at least four body regions at three to eight minutes each, adding up to 12–32 minutes in all (average 22 minutes).
Purchase and maintenance costs for the equipment (MR machine and power injector) were considered to make up the equipment’s total costs. Standard operating hours were taken from the German Uniform Value Scale (Einheitlicher Bewertungsmaßstab [EBM]26) as seven x 12 hours = 84 hours per week. Occupation times for the equipment and MR suite were derived from scan and preparation times. On the basis of experience, an average MR suite was assumed to have an area of 15m2 with a rent of €7.20/m2.22 Administrative costs were taken to be €10 per investigation.22 The other overheads were set at €40 per examination.22 Costs of medication used by <5% of all patients were not included. In cases of technical failure, an MRA procedure would be followed by standard DSA, the cost of which was likewise taken into account. Costs for electricity were obtained from a reference hospital that meters electricity consumption at the MRA site, assuming energy costs of €0.20/kWh with operation and stand-by times of 84 hours per week each and 52 weeks per year.
For this analysis, variable costs for individual consumables such as contrast agents and syringes were not considered at first because of price differences across regions and different accounting policies across hospitals. In addition, Germany was chosen as the example to evaluate the total costs including consumables. The list prices were considered and comprised €55.99 for the most commonly used gadolinium-based MRI contrast agent (Magnevist®, Bayer Schering Pharma AG, list price at July 2009 for a single dose of 15ml) and €3.31 for other consumables, totalling €59.30 as maximum costs according to the German medication index (Ifap® Index).27 No discounts were considered. If, for example, accounting is based on the tariff of the German Hospital Society (Tarif der Deutschen Krankenhausgesellschaft, DKG-NT),28 costs of the contrast agents are declared and reimbursed separately as add-ons to the reimbursement of the MRA. While this tariff is used for accounting within and between departments of hospitals in Germany, it is also used for calculating the procedure costs in reference hospitals in order to determine appropriate fixed costs. Tariff accounting is the preferred way to show any potential cost-related effects of alternative treatment pathways in the hospital setting.
A sensitivity analysis was performed in terms of the proportions of technical failures. The nce-MRA/FBI sequence is assumed to take on average 22 minutes in addition to the preparatory pulse sequences (scout, localiser, MRA sequences for vessel localisation, etc.). For the initial nce-MRA, the proportions of non-diagnostic images in clinical studies were reported to range from 6 to 34% of cases (average 20%).29–31 For the nce-MRA/FBI sequence, a similar proportion of non-diagnostic images is expected because of its technical disadvantages, such as the patient’s long stay in the machine and movement artefacts. These limitations can be considered as costs that must be incurred if a better image (higher resolution) is to be obtained. In cases of non-diagnostic images obtained by nce-MRA/FBI, it is assumed that a standard DSA for a final diagnosis will be performed in a different session. Therefore, calculations were performed using the average non-diagnostic proportion of 20%.29–34 Sensitivity analyses using 6 and 34% were conducted to explore the effect of different proportions on the total costs of nce-MRA with nce-MRA/FBI. Similar calculations were performed for ce-MRA on the basis of reported values between 1 and 3%;29–31 the average of 2% was used for the main calculation and the limits of 1 and 3% for sensitivity analyses. Costs of DSA for all calculations were obtained from Study 1,22 and total investigation costs were €176.63. As for ce-MRA, the costs of consumables for DSA (~€128 per examination according to list prices) were not taken into account.
In the analysis, the costs per minute were calculated as decimals with five digits for precision of results. All averages are expressed as arithmetic means. The data management and evaluation was performed with Microsoft® Office Excel 2003.
Results
Our cost analysis was strictly limited to the costs of routine investigations by two MRA techniques: ce-MRA and nce-MRA/FBI (as currently promoted by Toshiba) as an example for nce-MRA. The focus was on imaging procedures of lower extremities in a hospital setting (in- or outpatient). Both procedures in our study, i.e. ce-MRA and nce-MRA with nce-MRA/FBI, differ in terms of personnel time and overall duration of the procedure. Figure 1 shows the occupation times of the MR suite. With an assumed acquisition time of 22 minutes for nce-MRA/FBI, this modality requires on average 10 minutes more than ce-MRA. The personnel costs were included in the costs for ‘examination’ and were based on daily working hours and 220 working days per year. The total income per year for the head/registrar was estimated as €178,75023 with a working day of 11.88 hours,22 and for the assistant physician as €64,00023 with a working day of 9.5 hours.22 The total income of the technician (radiographer) was estimated as €28,06324,25 with a working day of eight hours.22 The working hours of physicians per examination were assumed to be the same for nce-MRA/FBI and ce-MRA (head/registrar 3:08 minute, assistant physician 11:09 minute).22 The time per examination of technicians has been reported to be 37–39 minutes for ce-MRA.22 Based on these data, the occupation time of the MRI facility with nce-MRA/FBI was assumed to be 32–22 to 52–22 minutes with nce- MRA/FBI sequences of 12 (shortest) and 32 minutes (longest).
The total costs for the equipment consist of purchase and maintenance of the machine and amounted to €527,479 per annum (€2.01 per operating minute) for the MR scanner and €10,144 per annum (€0.04 per operating minute) for the power injector, assuming an average usage time of 7.25 years.22 The average power consumption of an MR scanner was estimated to be 8kW in stand-by and 30kW in use. The electricity costs added up to €33,197 per year (in use €26,208, stand-by €6,989) leading to €0.13 per operating minute.22
Costs for ce-MRA and for various acquisition times with nce-MRA/FBI are displayed in Table 2. The costs for nce-MRA/FBI vary with acquisition time. Examination costs for nce-MRA/FBI exceed those for ce-MRA, even if the shortest reported nce-MRA/FBI acquisition time of 12 minutes is assumed: differences range from -€12 to -€61 (€102.49–€114.31 and €102.49–€163.47). Each additional minute spent on nce-MRA/FBI increases the difference by about €2.46. Some European countries (e.g. Austria, Belgium, Switzerland and Greece) deal with costs of contrast agents as consumables additional to the procedure costs. In other countries (e.g. Poland, Finland and Sweden), contrast agent costs form part of the all-inclusive sum reimbursed for the overall procedure. However, from the perspective of the radiological department the investigation costs are typically of interest when looking for cost-saving potential.
The investigation costs of ce-MRA, i.e. costs including consumables, total a maximum of €201.79 when no discount is considered, leading to a difference in costs between ce-MRA and nce-MRA/FBI of +€47 and -€2 (€201.79–€154.31 and €201.79–€203.47, respectively).
Figure 2 displays the results of the sensitivity analyses, taking into account costs for technical failures that result in standard DSA being performed to obtain a final diagnosis. Technical failures were expected in 6–34% (average 20%) of cases for nce-MRA/FBI and 1–3% (average 2%) for ce-MRA.
Considering the cost variance incurred through technical failures, ce-MRA shows a cost advantage over nce-MRA/FBI ranging from -€17 to -€118 (€147.79–€164.91 and €145.49–€263.53, respectively). When additionally including maximum costs for consumables (i.e. list prices), the difference in costs range from +€42 to -59 (€207.09– €164.91 and €204.79–€263.53, respectively).
Discussion
nce- and ce-MRA of the lower extremities are dominant over (i.e. cheaper and better than) standard angiography, as shown by Berry et al.32 Furthermore, in their study Watt et al. state that “there are good reasons both on a cost and time basis to use ce-MRA” instead of standard angiography.33 This is supported by the results of a cost- impact analysis by Hay et al. comparing ce-MRA and standard DSA.34
In our study, we focused on the comparison of ce-MRA and nce-MRA in the diagnosis of PAOD in a subset of patients who are indicated for both procedures. In the first section, technical and clinical differences between ce-MRA and nce techniques TOF and nce-MRA/FBI were presented. So far, the use of nce-MRA has been limited owing to a restricted FoV, difficulties in performing multistation imaging, artefacts with turbulent, slow or reverse blood flow and the lack of dynamic information. Today, the only established routine application of nce-MRA is intracranial MRA, for which in most cases TOF and PCA techniques are employed. nce techniques are superior in indications with small FoV, where motion artefacts are not present and where the time window for imaging is very short. One example is imaging of the head, with large experience in the radiology community in the use of nce techniques.
The technique of choice in peripheral vessels is ce-MRA, because of its advantages over nce techniques in patients with suspected PAOD in whom both MRA approaches are indicated. First, its robustness results in fewer cases of images being inadequate for diagnosis, thus reducing additional diagnostic work-up. Second, less time is needed for imaging, which is particularly important for severely ill patients who cannot undergo a long stay in the scanner. Third, with its high spatial resolution, ce-MRA provides reliable stenosis grading. Fourth, ce-MRA also allows multistation imaging and imaging with a large FoV. This is an essential feature for supra- aortic arteries and can be applied to whole-body MRA as well. In addition, flow artefacts from turbulent, slow or reverse blood flow are avoided, which allows a reliable grading of stenoses and diagnosis of aneurysms. Moreover, ce-MRA provides high sensitivity in small vessels with slow blood flow, such as arteries of the foot. By generating high-quality images, ce-MRA allows excellent visualisation of a complex vessel anatomy, such as collateral vessels. Finally, ce-MRA allows dynamic imaging for subclavian steal syndrome or arterio-venous malformations, for example.
A cost comparison of nce-MRA/FBI and ce-MRA was performed using the example of radiological departments in German hospitals. This analysis shows lower investigation costs for ce-MRA in all scenarios evaluated compared with nce-MRA/FBI, the latter having typical acquisition times ranging from 12 to 32 minutes. Thus, cost differences of between €12 and €61 per examination were found in favour of ce-MRA when excluding costs for consumables. For German hospitals, for example, the differences in total costs for ce-MRA including consumables (list price of €59) and nce-MRA/FBI were found to be in range of +€47 to -€2.
When costs for technical failures – adding investigation costs for standard DSA – were also considered, differences of €17 and up to €118 were found in favour of ce-MRA (excluding consumables). Total individual costs, including local reimbursement practice for consumables, can easily be obtained by adding such costs as far as applicable. When considering maximum costs for consumables for our example, Germany, we found scenarios with cost savings in favour of nce-MRA/FBI of up to €42, but also scenarios with cost savings in favour of ce-MRA of up to €59.
The calculations presented above do not take into account the loss of income consequent upon the lower patient throughput that results from the longer occupation times when nce-MRA/FBI is used compared with ce-MRA. Many radiological departments (excluding research centres) operate their facilities with time windows of 20–40 minutes per patient. If the imaging of a patient with nce-MRA/FBI takes longer than this, fewer patients can be scheduled, and consequently fewer procedures can be reimbursed. The same holds true for repeated examinations. During a second examination the scanner is occupied once again; therefore, in terms of efficiency, more robust procedures are preferred in order to minimise the number of unwanted repetitions.
Our cost analysis has certain limitations. ce-MRA and nce-MRA/FBI have been treated as directly competing techniques, without contraindications being taken into account, Such contraindications would include intolerance to the contrast agent for ce-MRA, and arrhythmia or instability of the patient for nce-MRA/FBI. There are instances where nce-MRA techniques can be clinically advantageous or more cost-effective. For example, in patients at risk of NSF, nce- MRA may be the only clinical alternative. Almost one-third of the MRA studies are performed on patients with no clinically significant occlusive arterial disease, or in a population with a relatively low risk of having PAOD. In those patients, nce-MRA may actually perform better in ruling out disease with fewer technical errors. These patient subgroups were not evaluated in our study. Some studies have shown that the sensitivity and specificity of ce-MRA, while good, is not as excellent as is often assumed. In a relatively large study of 400 subjects, Vasbinder et al.35 found that the sensitivity for detecting renal artery disease is around 60% even for experienced physicians. These data would have an effect on the cost–benefit evaluation. On the one hand, Leiner and Schoenberg36 stated that “most important contributors to the poor sensitivity were insufficient spatial resolution, motion artefacts due to long acquisition times and a large number of cases of fibromuscular dysplasia”. However, the spatial resolution improved significantly during the last few years. On the other hand, we focused on the peripheral artery occlusive disease and used high-quality clinical trials in this indication as the basis for estimating the diagnostic performance of ce- and nce-MRA. Several assumptions had to be made in the absence of published data. Acquisition times for nce-MRA/FBI have been reported in several publications as being three to eight minutes per sequence. However, personnel times can only be calculated on an indirect basis. We estimated that occupation times for the clinicians are comparable for nce-MRA/FBI and ce-MRA. nce-MRA/FBI does not require a clinician’s presence for placing the needle and supervision of the injection, but it might require more of the clinician’s time for reading the images. Once a procedure fails, DSA, as the reference standard is assumed to deliver the final diagnosis. However, in clinical routine, the diagnostic tree has several paths: repetition of the same procedure, applying another MRA technique, clinical follow-up, computed tomography angiography (CTA) or DSA. Personnel costs were calculated on the basis of the staff member’s gross income, which is usually much lower than the total corresponding cost to his/her employer. This may bias results in favour of the more time- consuming procedure. Costs for false-positive and false-negative results of the two procedures compared were not taken into account at all.
Conclusion
In patients with suspected PAOD and indication for ce- and nce-MRA, peripheral nce-MRA/FBI incurred higher investigation costs in terms of the sole acquisition of MR images than ce-MRA in most scenarios evaluated when excluding costs for the contrast agent.
The total costs including costs for consumables were found to be higher for ce-MRA not considering technical failures as in our initial example using German list prices. Depending on the proportion of technical failures and acquisition time of the FBI, scenarios with cost savings were found for either modality. On the other hand, several technical and clinical advantages of ce-MRA in patients with suspected PAOD were presented and discussed, such as high robustness and the shorter time needed for acquiring adequate diagnostic quality images. These increase the probability for adequate images after a single examination compared with nce-MRA/FBI in the peripherals.