Feasibility of second-generation ultrasound contrast media in the detection of active sacroiliitis




To determine whether a recently available contrast-enhanced ultrasound (CEUS) technique using second-generation microbubbles allows for the detection of active sacroiliitis, and to measure CEUS enhancement depth at the dorsocaudal part of the sacroiliac (SI) joints in healthy volunteers compared with patients with sacroiliitis.


Forty-two consecutive patients (84 SI joints) presenting with a clinical diagnosis of sacroiliitis in 50 SI joints and 21 controls (42 SI joints) were investigated by CEUS using a standardized low mechanical index ultrasound protocol. Detected vascularity was used to retrospectively measure the enhancement depth in the dorsocaudal part of the SI joints.


CEUS detected enhancement in all clinically active SI joints, showing an enhancement depth into the dorsal SI joint cleft of 18.5 mm (range 16–22.1), which was significantly higher compared with both inactive joints of patients (3.6 mm, range 0–12; P < 0.001) and healthy controls (3.1 mm, range 0–7.8; P < 0.001). All inactive joints were correctly classified based on a lack of deep enhancement in patients with sacroiliitis and controls (42 of 42, 100% sensitivity, 100% specificity; Cohen's κ = 1).


CEUS allowed the differentiation of active sacroiliitis from inactive SI joints, and proved to be a feasible method for the detection of vascularity in clinically active sacroiliitis by showing deep contrast enhancement into the SI joints not detectable in inactive joints of patients or controls. If this technique might add information to the earlier detection of sacroiliitis, it should be addressed in further studies.


Sacroiliitis is a hallmark of many forms of spondylarthropathies, including reactive arthritis, psoriatic arthritis, undifferentiated spondylarthropathy, and ankylosing spondylitis. As a group, the prevalence of these inflammatory arthropathies is as high as 0.5–1.9% (1). The purpose of an earlier diagnosis is emphasized by the need for a better management. Newer diagnostic methods include magnetic resonance imaging (MRI) and ultrasonography (US), because clinical diagnosis and physical examination are not very specific (2, 3).

MRI can demonstrate early predestructive alterations of sacroiliitis, and therefore can provide an earlier diagnosis of sacroiliitis (4, 5). Although MRI can shorten the interval between the onset of symptoms and the radiographic diagnosis of sacroiliitis, the availability of MRI is limited in many countries and the technique is relatively time consuming and cost-intensive (6, 7), which might be the reason why MRI in clinical practice is not routinely used in all patients presenting with inflammatory low back pain or suspicion of sacroiliitis.

Muche et al (8) reported that the dorsocaudal synovial parts of the sacroiliac (SI) joints were the most frequently inflamed structures in early disease. The dorsal joint capsule, the dorsal enthesis, and the dorsal cavum can be well detected using US, as described in 4 studies using US to detect sacroiliitis or to guide injections into the SI joint (9–12). However, until now, different US techniques were used for the detection of sacroiliitis. First, color Doppler US (CDUS), a technology widely used for the detection of blood flow, was used to detect vascularity at the dorsal SI joint (9, 10). A subsequent study used contrast-enhanced CDUS to improve vascularity detection in inflamed SI joints, because CDUS without contrast is limited in the detection of slow flow and flow in small vessels (i.e., neovessels) as it occurs in inflammatory angiogenesis. The US contrast media used was a so-called “first-generation” contrast agent, evaluated by CDUS and a high mechanical index examination protocol, showing an improved detection of vascularity in patients with sacroiliitis when compared with controls (11).

Meanwhile, second-generation US contrast media are available by applying low mechanical index protocols and gray-scale contrast-enhanced US (CEUS). The low mechanical index US technique is based on nonlinear acoustic effects and the interaction with microbubbles that makes microbubbles more stable and durable, therefore offering the possibility of performing continuous scanning over a specific period of time, which allows for a prolonged US examination.

Furthermore, CEUS maximizes contrast and spatial resolution, thereby leading the evolution of contrast US from vascular imaging to real-time imaging of perfused tissue at the microvascular level (13–15). This newer contrast technique has been used until now in the peripheral joints of patients with rheumatoid arthritis (14), as well as recently in osteoarthritic knees (16). To our knowledge, the value of CEUS has not been proven in any study until now in SI joints. Therefore, the purpose of this study was to evaluate the potential of a second-generation contrast agent for the detection of vascularity at the dorsocaudal part of the SI joints in patients with sacroiliitis compared with controls.


Study population.

Forty-two consecutive patients (84 SI joints in 27 men and 15 women, mean age 29.8 years, range 21–37) with spondylarthritis according to the European Spondylarthropathy Study Group criteria (17), or with ankylosing spondylitis according to the modified New York criteria (18), presented with inflammatory low back pain at the rheumatologic outpatient clinic of the Department of Internal Medicine, and were referred for CEUS examination the same day or the following day between 2003 and 2006. Twenty-one volunteers (42 SI joints in 9 women and 12 men, mean age 26.3 years, range 20–35) without suspicion of inflammatory low back pain served as a control group. Oral and written informed consent according to the Declaration of Helsinki was obtained in all patients and healthy volunteers. The examiner performing CEUS was blinded to all clinical findings, as well as to the clinical results of the retrospective analysis of collected CEUS data.

Clinical examination.

To determine SI joint activity, eligible patients presented with inflammatory lower back pain for more than 3 months with pain localization over the SI joints, including at least 3 of the 4 following positive parameters: stiffness and pain in the morning for >30 minutes, awakening because of back pain during the second half of the night only, improvement with exercise but not with rest, and alternating buttock pain (19). Patients with active disease, defined as a score of >3 according to the Bath Ankylosing Spondylitis Disease Activity Index, were included (20). Clinical assessment included SI joint mobility, pain provocation tests as lateral pelvic compression, prone sacral pressure, pressure over the second sacral foramen, upward pressure on the ischial bone, and superior iliac glide test.

Mean ± SD erythrocyte sedimentation rate (ESR) was 23.47 ± 15.0 mm/hour (normal value <15), mean ± SD C-reactive protein (CRP) level was 0.97 ± 0.66 mg/dl (normal value <0.07), and 18 (42.8%) of 42 patients were HLA–B27 positive.

Baseline US technique.

We used an MPX Technos Unit (Esaote, Genoa, Italy) fitted with a curved array abdominal probe, using a frequency of 4–6 MHz (CA430) for US examination. As previously described by Klauser et al (21), the curved array allows for better delineation of sonoanatomic landmarks than a linear transducer using higher frequencies (7–12 MHz), which is explained due to the penetration depth (4–6 cm) needed. Accordingly, care was taken to place the focal zone on a penetration depth of 4–6 cm, just where the hypoechoic cleft of the dorsal SI joint is located, depending on the body habitus of the patient. The depth was measured by placing calipers on the bony surface of the SI joint toward the skin for patients and controls.

US scanning was performed by posterior placement of the transducer with the patient in prone position. First, baseline US was performed to identify bony landmarks by depiction of the bony counters of the posterior superior iliac spine laterally and the spinous process of the fifth lumbar vertebra medially by axial transducer positioning. Then the transducer was moved caudally to depict the dorsal surface of the sacrum and the median and lateral sacral crest, the gluteal surface of the ilium, and the first posterior sacral foramen. From this level, the transducer was moved downward until the second posterior sacral foramen was visualized (21).

Distinction of these landmarks was readily available. Attention was paid to visualize the hypoechoic cleft to prove sufficient US beam penetration at the level of the first and second sacral foramen by shifting the transducer downward and upward in paraxial planes. The transducer was positioned at the level of the first sacral foramen and contrast material was injected intravenously.

To avoid observer bias of our data, the examining radiologist was not permitted to ask the volunteers or the patients about symptoms. The only information provided was a written request from the referring rheumatologist that the patient should be examined for the presence of vascularity at the SI joint. In addition, care was taken to not press the transducer over the SI joint in terms of sonopalpation to avoid a provocative pressure and unblinding of the sonographer.

Contrast-enhanced gray-scale US.

Initially, a dedicated fundamental B-mode scanning was carried out to obtain an overview of bony landmarks and the direction of the dorsal SI joint clefts. Because this may greatly vary from patient to patient, slight transducer tilting might be necessary. After contrast administration, a sweep to obtain an overview over both SI joints was performed first, followed by a sweep with adapted slight transducer tilting over one and then the other SI joint by paying attention to the limited contrast-enhancing time, allowing for a scanning window of ∼4–5 minutes. Images and cine loop sequences were recorded and stored digitally to be used at the end of the examination for placing calipers to measure contrast enhancement.

The second-generation contrast agent SonoVue (Bracco, Milan, Italy) used in this study is an aqueous suspension of microbubbles, composed of peripheral phospholipid monolayers filled with sulfur hexafluoride. The bubble size varied between 1 and 10 μm, allowing the microbubbles to pass after intravenous injection in the circulation through the pulmonary capillary bed and to serve as strong reflectors for the US beam. Contraindications for the application of SonoVue are known hypersensitivity to sulfur hexafluoride or to any other components of SonoVue, right-to-left shunt, severe pulmonary hypertension, uncontrolled systemic hypertension or adult respiratory distress syndrome, chronic obstructive pulmonary disease, severe congestive heart failure, severe arrhythmia, or pregnancy, representing the exclusion criteria in our study.

The agent was prepared in a standard manner with a dose of 2.4 ml, administered as an intravenous bolus and flushed with 5 ml of saline. Subsequently, US scanning was performed using a low mechanical index (≤0.1) technique by sweeping the probe cranially and caudally over the lower back at the level of the first and second sacral foramen covering both SI joints. US protocol settings were kept constant during the examination, and standardized machine settings for the CEUS examination with a single focal zone were used. According to the bubble lifetime, the examination time window lasted for up to ∼4–5 minutes.

Vascularization in the dorsal SI joint was defined as gray-scale enhancement due to microbubbles in the area of the SI joints. Measurement of enhancement depth was obtained at the area with maximal dorsocaudal enhancement, located in the hypoechoic cleft of the dorsal SI joint space in order to differentiate patients with active sacroiliitis from healthy controls. For each SI joint, the area with the highest amount of enhancement at the posterocaudal portion was used for documentation. Images and clips were stored digitally for retrospective evaluation.

Maximal intraarticular enhancement depth was measured in millimeters by placing calipers at the level of the dorsal bony border of the entrance to the SI joint, as described in the study by Klauser et al, toward the maximal detectable enhancement in the SI joint (21). To obtain measurement from CEUS enhancement, the area at the edge of the sacral bone just lateral to the first sacral foramen toward the second sacral foramen was carefully reviewed in the cine loop to finally place the calipers where maximum enhancement was detected. Care was taken to place the dorsal caliper in order to include the enhancement toward the extension of the dorsal joint capsule and ligaments, if vascularized (Figures 1–4).

Figure 1.

A 21-year-old male patient with clinically bilateral active sacroiliitis. Bony landmarks are marked as the following: equation image is located at the median sacral crest; ▴ points toward the bony contour of the sacrum. The thin arrow shows enhancement in the first sacral foramen. Thick arrows show bilateral enhancement in the dorsocaudal parts of the sacroiliac (SI) joints.

Figure 2.

A 27-year-old female patient with clinically left-sided sacroiliitis. Bony landmarks are marked as the following: equation image shows the bony contour of the ilium; ▴ shows the bony contour of the sacrum. Contrast-enhanced ultrasound demonstrates vascularity on the left sacroiliac (SI) joint (arrows) with 18.8-mm enhancement between calipers. Note that the caliper placement for measurement includes dorsal joint capsule and ligament extension, which is seen outlined as a slightly hypoechoic thickening overlaying the dorsal cleft, where the dorsocranial caliper is placed.

Figure 3.

A 21-year-old male patient with clinically right-sided sacroiliitis. Bony landmarks are marked as the following: equation image shows the bony shadowing of the ilium; ▴ shows the bony contour of the sacrum. Contrast-enhanced ultrasound demonstrates vascularity on the right sacroiliac (SI) joint (arrows) with 21.1-mm enhancement between calipers. Note that the ventrocaudal caliper is placed for better understanding slightly to the left at the level of the white line, representing the deepest contrast enhancement.

Figure 4.

A 31-year-old male patient with clinically left-sided sacroiliitis. Bony landmarks are marked as the following: equation image shows the bony shadowing of the ilium; ▴ shows the bony contour of the sacrum. Contrast-enhanced ultrasound demonstrates vascularity on the left sacroiliac (SI) joint (arrows) with 19.2-mm enhancement between calipers.

Intrareader variability was calculated by remeasurement of enhancement at the dorsal cleft by a second radiologist blinded to clinical and previous imaging findings to test reproducibility.

In addition, mean examination time for US and CEUS was calculated. The calculated examination time includes careful fundamental B-mode scanning, mandatory to be orientated before switching to the contrast mode (CEUS), where the information of the fundamental B- mode image is mainly suppressed to better detect contrast arrival and enhancement.

Statistical analysis.

Statistical analysis was performed using SPSS software, release 14.0 (SPSS, Chicago, IL). Normal distribution of data was tested with the Kolmogorov-Smirnov test. Standard descriptive statistics were used to summarize the characteristics of the study patients and controls, including means and SDs for the continuous variables and counts and percentages for the categorical variables. The presence of vascularity and the amount of maximal intraarticular enhancement depth from CEUS were compared with the clinical diagnosis as the gold standard. The diagnostic accuracy of CEUS (sensitivity, specificity, positive predictive value, and negative predictive value) compared with the clinical evaluation was calculated. The enhancement depth in active SI joints was compared with the inactive joints of the patients and with the inactive joints of the controls using the independent-samples t-test. The interobserver agreement of measured enhancement depth was calculated using Pearson's correlation coefficients.



The evaluated mean ± SD depth from the skin to the bony surface of the SI joint was 5.1 ± 0.7 cm in patients and 5.5 ± 0.6 cm in controls (P > 0.05). In 42 patients, 50 (59.5%) of 84 SI joints (8 patients bilaterally) were clinically active, 23 at the left side and 27 at the right side. The mean ± SD enhancement depth at the right SI joint was 13.4 ± 7.1 mm (range 0–21), whereas for the clinically active right SI joint, an enhancement depth of 18.4 ± 1.4 mm (range 17–21) was measured (Table 1). The mean ± SD enhancement depth at the left SI joint was 11.4 ± 8.2 mm (range 0–22.1), whereas for the clinically active left SI joint, an enhancement depth of 18.7 ± 1.6 mm (range 16–22) was calculated.

Table 1. Enhancement depth in patients and volunteers*
CEUSOverall, mmActive joints, mmInactive joints, mm
  • *

    Values are the mean ± SD (range). CEUS = contrast-enhanced ultrasound.

Patients (n = 42)12.4 ± 7.7 (0–22.1)18.5 ± 1.5 (16–22)3.6 ± 2.5 (0–12)
 Right (n = 42)13.4 ± 7.1 (0–21)18.4 ± 1.4 (17–21)4.5 ± 3.3 (0–12)
 Left (n = 42)11.4 ± 8.2 (0–22.1)18.7 ± 1.6 (16–22)2.7 ± 1.7 (0–6)
Volunteers (n = 21)3.1 ± 2.4 (0–7.8)  
 Right (n = 21)2.9 ± 2.2 (0–7.7)  
 Left (n = 21)3.2 ± 2.6 (0–7.8)  

In the clinically inactive SI joints of the patients, the mean ± SD enhancement depth was 3.6 ± 2.5 mm (range 0–12 mm), and was significantly lower compared with clinically active joints (P < 0.001).

Laboratory data among our patients revealed a mean ± SD CRP level of 0.95 ± 0.65 mg/dl (range 0.64–2.99; normal value <0.07 mg/dl) and a mean ± SD ESR of 21.08 ± 17.33 mm/hour (range 1.89–51; normal value <15 mm/hour). Thirty-five (83.3%) of 42 patients were HLA–B27 positive. The mean ± SD disease duration was 2.9 ± 2.7 years.


The 21 volunteers without any evidence of inflammatory lower back pain showed a mean ± SD enhancement depth at the right SI joint of 2.9 ± 2.2 mm (range 0–7.7), and at the left SI joint of 3.2 ± 2.6 mm (range 0–7.8). The mean ± SD enhancement depth in volunteers (3.1 ± 2.4 mm, range 0–7.8 mm) was significantly lower compared with the clinically active joints of patients (P < 0.001). Controls demonstrated only vascularity at the very peripheral zone, which is consistent with vascularity close to the joint capsule. Maximal dorsal enhancement in 1 volunteer was 7.8 mm (Figure 5).

Figure 5.

A 25-year-old male control. Bony landmarks are marked as the following: equation image shows the bony contours/shadowing of the ilium; ▴ shows the bony contour of the sacrum. Contrast-enhanced ultrasound demonstrates vascularity on the left sacroiliac (SI) joint (arrows) of 7.8-mm enhancement between calipers. Note that slight transducer tilting was necessary, which explains why the distance between the second sacral foramen and the SI joint seems smaller than, e.g., in Figure 1. The influence of transducer tilting can also be seen when comparing the dimension of the left sacral surface (arrow heads) with the right-sided sacral surface (line tilted toward 4:00).

CEUS demonstrated a sensitivity of 100% (50 of 50; 95% confidence interval [95% CI] 92.9–100), a specificity of 100% (76 of 76; 95% CI 95.2–100), a positive predictive value of 100% (50 of 50; 95% CI 92.9–100), and a negative predictive value of 100% (76 of 76; 95% CI 95.2–100) in the detection of clinically active SI joints. The agreement between CEUS and the clinical rating of SI joints was excellent (100%; Cohen's κ = 1). No side effects were noted by the US contrast agent administration. The mean ± SD examination time for US and CEUS was 11 ± 8.7 minutes (range 8.2–24.1).

The interobserver correlation for enhancement depth was excellent, with correlation coefficients of 0.80 for the right side of symptomatic patients, 0.97 for the right side of asymptomatic patients, 0.73 for the left SI joint of symptomatic patients, and 0.88 for the left side of asymptomatic patients (P < 0.001 for all measurements).


The first study of unenhanced CDUS in a group of 21 patients with active sacroiliitis found vascularity around or inside of the SI joints in 10 patients (48%) (12). By using CEUS, we found vascularity in all SI joints of patients with active sacroiliitis, extending deep into the dorsal SI joint cleft over several millimeters (mean 18.7 mm, maximum 22.1 mm). These findings are in line with the second study using contrast-enhanced CDUS in 43 patients (70 of 206 SI joints), where contrast-enhanced CDUS (Az = 0.89) was significantly better than unenhanced CDUS (Az = 0.61) for the detection of vascularity of early active sacroiliitis as diagnosed by MRI (P < 0.0001) (11). However, in this study, an older contrast technique was used that is known to be affected by artifacts, shorter contrast duration, and overestimation of contrast enhancement caused by early bubble disruption by the use of the high mechanical index technique. Vascularity overestimation due to artifacts from capsular vessels could be an explanation of why a higher rate of hypervascularity in the SI joints of patients with active sacroiliitis was detected when compared with MRI, and why no explanation for these false-positive results could be presented. In our study, no false-positive results were obtained in a patient's asymptomatic SI joints when only deeply located enhancement was believed to be positive for sacroiliitis. However, we found vascularity at the dorsal superficial SI joint cleft in clinically inactive joints of the patients (mean ± SD 3.6 ± 2.5 mm, maximum 12.0 mm) and in our controls (mean ± SD 3.1 ± 2.4 mm, maximum 7.8 mm) using CEUS. Vascularization near the SI joints caused by branches of sacral arteries has already been described by Pekkafahli et al (12) in 13 healthy volunteers, whereas 10 did not show any vascularity, which is in line with our findings where no vascularity at all was detected in 9 (42.8%) of 21 healthy controls and in 4 (11.7%) of 34 asymptomatic joints of the patients. Therefore, hypervascularity at the superficial dorsal SI joint can be seen in healthy and asymptomatic patients, but extension to a depth of more than 16 mm was only found in symptomatic patients. Because both patients and controls can show vascularity at the dorsal border, which might reflect pericapsular perfusion, a depth measurement of the enhancement has to be obtained to ensure extension of intraarticular vascularity at the dorsocaudal joint space in patients with active sacroiliitis.

The measurement of enhancement is relatively simple and quick to obtain because the time window of second-generation contrast media is enlarged and artifacts are diminished, which allows for a clear delineation of the enhancement. However, in older patients or in patients with long-standing disease duration, there might be a limited US beam penetration into the dorsal part of the SI joint when bony spurs or ankylosis are present. Advantages of second-generation contrast media have already been demonstrated in recent studies showing a significantly improved detection of blood flow in the inflamed synovium of joints with rheumatoid arthritis and osteoarthritic knees (14, 15). Our results demonstrate that CEUS is a feasible technique for the detection of active sacroiliitis when compared with clinical examination.

Early diagnosis is essential for the optimal management of patients with spondylarthropathy because new therapeutic options, especially tumor necrosis factor (TNF) antagonists, are already demonstrating good clinical efficacy and safety in patients with axial spondylarthritis and pre–radiographically defined sacroiliitis (22). In our study population, only a nonsignificant proportion (n = 2) was receiving such therapy at the time of the US scanning, and there was no difference in contrast enhancement between patients treated with TNF and those not receiving TNF therapy. However, further studies should investigate the value of this technique for the early detection of sacroiliitis and for changes after TNF therapy, which could have implications for therapeutic concepts.

As stated by Deyo and Weinstein (2), there is excessive MRI performed in patients with low back pain. MRI is expensive and may be not available as a screening tool for all patients with low back pain. MRI is limited in patients with metal implants, certain pacemakers, or claustrophobia. Nevertheless, MRI is used as a gold standard for the early detection of sacroiliitis in most centers because of its advantages over plain radiography and computed tomography. How CEUS performs for early sacroiliitis detection and compared with MRI should be proven in further studies. Plain radiography and computed tomography may not demonstrate evidence of active inflammatory disease until years after the onset of symptoms, and are not useful for the early detection of sacroiliitis (23–26). As opposed to these modalities, contrast-enhanced US is a relatively simple, portable, and less expensive imaging modality that could serve as a valuable tool for patients with suspected inflammatory low back pain (11).

Sieper et al have stressed that only 50–70% of patients with active ankylosing spondylitis have increased levels of CRP or ESR (27). A similar lack of correlation between disease activity and CRP/ESR can be expected for the other spondylarthropathies. Increased ESR and CRP levels may also be caused by infection and are therefore not specific for inflammatory disease. Therefore, the lack of a sensitive/specific laboratory test for inflammatory spondylarthropathy supports the need for imaging modalities in patients with suspected inflammatory low back pain.

We have to note several limitations of this study. First, a single investigator performed CEUS examinations and we do not have data on intra- or interobserver variability.

As a major limitation, we have to mention that the gold standard used is not ideal (28). However, very recently a new approach using clinical parameters for inflammatory back pain classification criteria has been shown to be robust, easy to apply, and have good face validity (29). Although MRI is a valuable diagnostic tool, it is not part of any existing classification criteria (30). Several studies address the value of MRI with or without contrast application, but it is not confirmed as a gold standard until now, and the value of scintigraphy is also under debate (31). Williamson et al found incongruent findings between MRI and clinical evaluation in a patient with psoriatic arthritis, and discussed that sacroiliac symptoms and signs relate to abnormalities that are not detectable on MRI. They stated that there may be sources of pain other than bone marrow edema and cartilage erosions, as seen by MRI (32). Further studies comparing the value of CEUS findings with MRI are planned.

In fact, only a pathohistologic correlation would have been the true gold standard (33, 34) to compare the vascularity detected by CEUS in the SI joints, because US contrast media are real intravascular enhancers and different in behavior than, e.g., MRI contrast media. Therefore, we correlated our findings with clinical diagnosis, but a correlation with other imaging modalities, especially for early sacroiliitis detections, should be performed in further studies.

A further limitation could represent an increased body mass index–reducing diagnostic accuracy similar to abdominal US, but we did not encounter such a case in our patients or controls. Furthermore, joint space narrowing and bony spur formation will inhibit US beam penetration in a more advanced disease stage, but we did not test such cases in this study. Even a partly ventrally located inflammation of the SI joints might not be detected by CEUS; however, this special condition should be compared with MRI findings in further studies and did not seem to affect our results because we did not have false-negative findings in our patient population.

From the economic perspective, the cost of CEUS is estimated to be less than one-third of that for contrast-enhanced MRI. The application of CEUS could reduce the numbers of MRI studies in patients with lower back pain. Since the early diagnosis of definite sacroiliitis allows early treatment and avoids unnecessary further examinations (33), we believe that CEUS might be as cost-effective as an initial imaging modality in the evaluation of inflammatory lower back pain.

In summary, even if our results are preliminary, obtained in a relatively small population group, and further larger imaging studies are required, CEUS is a promising and readily available imaging modality for the detection of active sacroiliitis and might be an interesting tool to add for early diagnosis.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Klauser had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Klauser, De Zordo, Feuchtner, Gruber.

Acquisition of data. Klauser, De Zordo, Bellmann-Weiler, Feuchtner, Gruber.

Analysis and interpretation of data. Klauser, Bellmann-Weiler, Feuchtner, Sailer-Höck, Sögner, Gruber.