High percentage of non‐diagnostic compression ultrasonography results and the diagnosis of ipsilateral recurrent proximal deep vein thrombosis: a rebuttal

Abstract

8See also Tan M, Velthuis SI, Westerbeek RE, van Rooden CJ, van der Meer FJ, Huisman MV. High percentage of non‐diagnostic compression ultrasonography results and the diagnosis of ipsilateral recurrent proximal deep vein thrombosis. J Thromb Haemost 2010; : 848–50; Tan M, Huisman MV. High percentage of non‐diagnostic compression ultrasonography results and the diagnosis of ipsilateral recurrent proximal deep vein thrombosis: reply to a rebuttal. This issue, pp 417–8. See also Tan M, Velthuis SI, Westerbeek RE, van Rooden CJ, van der Meer FJ, Huisman MV. High percentage of non‐diagnostic compression ultrasonography results and the diagnosis of ipsilateral recurrent proximal deep vein thrombosis. J Thromb Haemost 2010; : 848–50; Tan M, Huisman MV. High percentage of non‐diagnostic compression ultrasonography results and the diagnosis of ipsilateral recurrent proximal deep vein thrombosis: reply to a rebuttal. This issue, pp 417–8. In a recently published study, Tan et al. [1Tan M. Velthuis S.I. Westerbeek R.E. Van Rooden C.J. van der Meer F.J. Huisman M.V. High percentage of non‐diagnostic compression ultrasonography results and the diagnosis of ipsilateral recurrent proximal deep vein thrombosis.J Thromb Haemost. 2010; 8: 848-50Crossref PubMed Scopus (24) Google Scholar] reported a high degree of uncertainty (32%) in the diagnosis of ipsilateral recurrent proximal deep vein thrombosis (DVT) by compression ultrasonography (CUS). One of the explanations was the high inter‐ and intraobserver variability in the measurement of thrombus diameter. In order to improve the accuracy of diagnosis, these authors proposed alternative methods such as magnetic resonance. Another approach might be to increase the quality of CUS by improvement of thrombus measurement. In fact, ultrasound imaging is an accurate diagnostic test for a first episode of (DVT), whether proximal or isolated infrapopliteal distal DVT [2Kearon C. Julian J.A. Math M. Newman T.E. Ginsberg J.S. for the McMasterDiagnostic imaging practice Guidelines Initiative. Noninvasive diagnosis of deep venous thrombosis.Ann Intern Med. 1998; 128: 663-77Crossref PubMed Scopus (653) Google Scholar, 3Elias A. Cadene A. Elias M. Puget J. Tricoire J.L. Colin C. Lefebvre D. Rousseau H. Joffre F. Extended lower limb venous ultrasound for the diagnosis of proximal and distal vein thrombosis in asymptomatic patients after total hip remplacement.Eur J Vasc Endovasc Surg. 2004; 27: 438-44Abstract Full Text Full Text PDF PubMed Scopus (32) Google Scholar, 4Goodacre S. Sampson F. Thomas S. Van Beek E. Sutton A. Systematic review of the diagnostic accuracy of ultrasonography for deep vein thrombosis.BMC Med Imaging. 2005; 5: 6Crossref PubMed Scopus (259) Google Scholar]. When CUS is conducted according to a standardized examination protocol, the interobserver agreement for diagnosis (presence or absence) of a first symptomatic, as well as asymptomatic DVT, is good for both proximal (reported kappa between 0.89 and 1.0) and distal DVT (reported kappa between 0.77 and 0.86 [4Goodacre S. Sampson F. Thomas S. Van Beek E. Sutton A. Systematic review of the diagnostic accuracy of ultrasonography for deep vein thrombosis.BMC Med Imaging. 2005; 5: 6Crossref PubMed Scopus (259) Google Scholar, 5Barrellier M.T. Somon T. Speckel D. Fournier L. Denizet D. L’écho‐doppler dans le diagnostic de thrombose veineuse profonde des members inférieurs. Accord entre 2 opérateurs.J Mal Vasc. 1992; 17: 196-201PubMed Google Scholar, 6Schwarz T. Schmidt B. Schmidt B. Schellong S.M. Interobserver agreement of complete compression ultrasound for clinically suspected deep vein thrombosis.Clin Appl Thromb Hemost. 2002; 8: 45-9Crossref PubMed Scopus (64) Google Scholar]. In contrast, the diagnosis of ipsilateral recurrent DVT is a much more difficult challenge, and must be clearly distinguished from incomplete resolution of the initial thrombus. The main criteria for proximal DVT recurrence is an increase of ≥ 2 mm [7Prandoni P. Cogo A. Bernardi E. Villta S. Polistena P. Simioni P. Noventa F. Benedetti L. Girolami A. A simple ultrasound approach for detection of recurrent proximal‐vein thrombosis.Circulation. 1993; 88: 1730-5Crossref PubMed Scopus (288) Google Scholar] to ≥ 4 mm [2Kearon C. Julian J.A. Math M. Newman T.E. Ginsberg J.S. for the McMasterDiagnostic imaging practice Guidelines Initiative. Noninvasive diagnosis of deep venous thrombosis.Ann Intern Med. 1998; 128: 663-77Crossref PubMed Scopus (653) Google Scholar, 7Prandoni P. Cogo A. Bernardi E. Villta S. Polistena P. Simioni P. Noventa F. Benedetti L. Girolami A. A simple ultrasound approach for detection of recurrent proximal‐vein thrombosis.Circulation. 1993; 88: 1730-5Crossref PubMed Scopus (288) Google Scholar, 8Prandoni P. Lensing A.W. Prins M.H. Bernardi E. Marchiori A. Bagatella P. Frulla M. Mosena L. Tormene D. Piccioli A. Simioni P. Girolami A. Residual venous thrombosis as a predictive factor of recurrent venous thromboembolism.Ann Intern Med. 2002; 137: 955-60Crossref PubMed Scopus (467) Google Scholar, 9Cosmi B. Legnani C. Cini M. Guazzaloca G. Palareti G. D‐dimer levels in combination with residual venous obstruction and the risk of recurrence after anticoagulation withdrawal for a first idiopathic deep vein thrombosis.Thromb Haemost. 2005; 94: 969-74Crossref PubMed Scopus (99) Google Scholar] in the anterior‐posterior diameter of the common femoral or popliteal vein measured in transverse section at a predefined anatomical site during compression, and compared with a previous CUS. However, based on these measurements, a variance of 54% between two observers in the diagnosis of residual thrombosis has been reported [10Linkins L.‐.A. Stretton R. Probyn L. Kearon C. Interobserver agreement on ultrasound measurements of residual vein diameter, thrombus echogenicity and Doppler venous flow in patients with previous venous thrombosis.Thromb Res. 2006; 117: 241-7Abstract Full Text Full Text PDF PubMed Scopus (62) Google Scholar]. Therefore, there is a need to optimize the quality of both acute and residual thrombus measurements. In order to improve the quality of CUS for acute DVT, we studied interobserver agreement of transverse section measurements at the maximal anterior‐posterior diameter for each proximal or distal vein, including muscular veins. We hypothesized that this measurement could be more accurate and reproducible than measurement at only two predefined anatomical sites as previously described [7Prandoni P. Cogo A. Bernardi E. Villta S. Polistena P. Simioni P. Noventa F. Benedetti L. Girolami A. A simple ultrasound approach for detection of recurrent proximal‐vein thrombosis.Circulation. 1993; 88: 1730-5Crossref PubMed Scopus (288) Google Scholar, 10Linkins L.‐.A. Stretton R. Probyn L. Kearon C. Interobserver agreement on ultrasound measurements of residual vein diameter, thrombus echogenicity and Doppler venous flow in patients with previous venous thrombosis.Thromb Res. 2006; 117: 241-7Abstract Full Text Full Text PDF PubMed Scopus (62) Google Scholar] and could help in determining a new more reliable diameter increase to be tested for in the diagnosis of recurrent DVT. For this purpose, 43 patients with a positive complete venous CUS examination (mean duration 20 min), performed by a board‐certified vascular medicine physician, underwent, after informed consent, a second CUS examination on the same day, always performed by the senior experienced vascular physician. As interobserver agreement did not concern DVT diagnosis, but diameter measurement, the first observer indicated the lower limb involved (but not the veins concerned) so as not to waste too much time. CUS was performed using a standardized examination protocol, with a 5–7.5 MHz linear array and a 3–5.5 MHz curved transducer, and diagnosis of thrombosis was based on vein non‐compressibility [10Linkins L.‐.A. Stretton R. Probyn L. Kearon C. Interobserver agreement on ultrasound measurements of residual vein diameter, thrombus echogenicity and Doppler venous flow in patients with previous venous thrombosis.Thromb Res. 2006; 117: 241-7Abstract Full Text Full Text PDF PubMed Scopus (62) Google Scholar]. With the patient in a supine position, the proximal veins were examined conventionally from the common femoral vein in the groin area to the distal superficial femoral vein. The popliteal vein and the calf veins were evaluated from the knee down to ankle level, with the patient in a sitting position, legs hanging down, feet resting on the observer’s knees; posterior tibial and fibular veins as well as muscular gastrocnemial and soleal veins were all explored. Maximal anterior‐posterior thrombus diameter (mm) was measured in a transverse plane during compression by the ultrasound probe, and independently recorded on a chart. The outcome measure was agreement between the vascular physicians performing CUS on the maximal anterior‐posterior thrombus diameter. Agreement between diameter measurements by both observers was assessed by a coefficient of correlation r and a coefficient of variation (CV). Agreement was based on the correlation between the two observers and interpreted as follows: < 0.2 indicates poor agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.8, good agreement; and 0.81–1, very good agreement. To explore the possibility of any systematic bias in the measurement of thrombus diameter, the difference in thrombus diameter between observers was plotted against the average thrombus diameter (Bland‐Altman plot). We included 43 patients, with a total of 83 thrombosed veins, 47 proximal and 36 distal. For proximal DVT, the mean measured vein diameter was 8.36 mm (SD 3.08 mm). There is very good interobserver agreement regarding diameter measurement of the thrombus, with r2 = 90.9% and a CV of 8.5% (Fig. 1A). The mean difference in diameter between paired measurements was 0.73 mm, with a maximum of 2.5 mm. The 95th percentile for the difference in the measurement between the two observers was 2.2 mm. For distal DVT, the mean measured vein diameter was 5.08 mm (SD 1.67 mm). The interobserver agreement was good (r² 71.8%, CV 17.6%) (Fig. 1B). The mean difference in diameter between paired measurements was 1.01 mm, with a maximum of 2.7 mm. The 95th percentile for the difference in the measurement between the two observers was 2.2 mm. In both cases, a Bland‐Altman plot of diameter of the thrombus showed no clear evidence of a systematic difference between the observers or according to mean diameter (Fig. 1C,D). Therefore, our standardized measurement of the maximal diameter for each acute thrombosed venous segment has only a small variance, from 8% to 10%, between two observers, which is much lower than the 54% reported by Linkins [10Linkins L.‐.A. Stretton R. Probyn L. Kearon C. Interobserver agreement on ultrasound measurements of residual vein diameter, thrombus echogenicity and Doppler venous flow in patients with previous venous thrombosis.Thromb Res. 2006; 117: 241-7Abstract Full Text Full Text PDF PubMed Scopus (62) Google Scholar] for residual thrombosis. There are several hypotheses to explain this discrepancy between our results and the results from Linkins regarding interobserver agreement. Firstly, the study reported by Linkins was described as having been performed on residual thrombosis. In fact, Linkins’ population was very heterogeneous, as the mean time from the DVT event to the examination was 7 months, with extremes from 0.4 months (i.e. 12 days after the acute event) to 49 months (4 years). In contrast, our trial only focused on acute DVT. Secondly, in our practise, all CUSs were performed by vascular physicians who had followed similar training and used a standardized examination protocol. The recent paper by Sevestre et al. [11Sevestre M.A. Labarère J. Casez P. Bressollette L. Taiar M. Pernod G. Quéré I. Bosson J.L. Accuracy of complete compression ultrasound in ruling out suspected deep venous thrombosis in the ambulatory setting. A prospective cohort study.Thromb Haemost. 2009; 102: 166-72Crossref PubMed Scopus (35) Google Scholar, 12Westendorf J.B. Safety of venous ultrasound in suspected DVT – still a matter of concern?.Thromb Haemost. 2009; 101: 5-6PubMed Google Scholar] underlines the importance of complete CUS carried out by trained vascular physicians using a standardized approach, although improvement in ultrasound techniques and further standardization is still needed. Thirdly, the measurements made by Linkins et al. were performed on the common femoral vein and on the popliteal vein, at specific predefined sites. This definition might be responsible for imprecise measurement if we take into account technical and anatomical considerations. In contrast, our approach, based on the measurement of the vein at its maximal diameter, appears to be more precise and reproducible. Besides, for the first time our study provides results for interobserver agreement specifically for proximal and distal acute DVT, which was also considered as good with an r² of 71.8%. In both cases, the reproducibility is underscored by the results of the Bland‐Altman analysis. For proximal and distal DVT, the 95th percentile for the difference in the measurement between the two observers was similar, 2.2 mm, with a maximum of 2.5 and 2.7 mm, respectively. A cut‐off of between 2 and 4 mm in vein diameter has been proposed for the diagnosis of DVT recurrence [2Kearon C. Julian J.A. Math M. Newman T.E. Ginsberg J.S. for the McMasterDiagnostic imaging practice Guidelines Initiative. Noninvasive diagnosis of deep venous thrombosis.Ann Intern Med. 1998; 128: 663-77Crossref PubMed Scopus (653) Google Scholar, 7Prandoni P. Cogo A. Bernardi E. Villta S. Polistena P. Simioni P. Noventa F. Benedetti L. Girolami A. A simple ultrasound approach for detection of recurrent proximal‐vein thrombosis.Circulation. 1993; 88: 1730-5Crossref PubMed Scopus (288) Google Scholar, 8Prandoni P. Lensing A.W. Prins M.H. Bernardi E. Marchiori A. Bagatella P. Frulla M. Mosena L. Tormene D. Piccioli A. Simioni P. Girolami A. Residual venous thrombosis as a predictive factor of recurrent venous thromboembolism.Ann Intern Med. 2002; 137: 955-60Crossref PubMed Scopus (467) Google Scholar, 9Cosmi B. Legnani C. Cini M. Guazzaloca G. Palareti G. D‐dimer levels in combination with residual venous obstruction and the risk of recurrence after anticoagulation withdrawal for a first idiopathic deep vein thrombosis.Thromb Haemost. 2005; 94: 969-74Crossref PubMed Scopus (99) Google Scholar]. Based on the magnitude of interobserver differences, the threshold of 4 mm appears to be more accurate so as to avoid false positives, and in most cases must be applied to both proximal and distal DVT. Hence, in clinical practise, it would appear to be preferable to implement a standardized optimized CUS protocol based on measurement of the maximal anterior‐posterior thrombus diameter, with acceptable interobserver variability, allowing good reproducibility, which could be used in the context of recurrent DVT. The authors state that they have no conflict of interest.