CAN B-MODE COMPRESSION ULTRASOUND OR DUPLEX SONOGRAPHY BE USED FOR DVT SCREENING IN HIGH RISK PATIENTS? RESULTS OF A META-ANALYSIS

Purpose: Compression ultrasound (CUS) with and without additional Doppler information (Duplex) is widely accepted for the diagnosis of symptomatic DVT, whereas for screening, guidelines recommend contrast phlebography (CP). Published data was analyzed, to evaluate ultrasound (US) for screening.

Materials and Methods: A multilingual search of Medline, Embase and Cochrane databases for studies, published between 1989 and 2004, using CP (gold-standard) and CUS and/or duplex for DVT screening in high risk patients was performed and revealed 85 papers. Thereof 62 were excluded. The remaining 23 studies formed the database. Articles were scrutinized by 2 independent reviewers using stringent quality selection criteria. Appropriate meta-analysis techniques including heterogeneity testing were applied and results were, if applicable, random-effects-model (REM) adjusted.

Results: The mean percentages (95% CI) for sensitivity, specificity, PPV, NPV and the mean positive likelihood ratio (LR+) for the detection of proximal DVT were 72(63–81), 99(99–100), 77 (70–84), 99 (98–99), 121 (86–169), respectively. Figures came to 75 (65–85),

99 (98–99), 87 (81–93), 96 (94–97), 68 (45–104) for distal DVT, respectively.

Pre-test probabilities for proximal and distal DVT were 6.8% and 16.1%. The post-test probability was calculated for corrected mean LRs (Bayes Theorem). For negative scan results the post-test probability for proximal DVT was 1.9% (1,6–2,5), for distal DVT

4.6% (3.7–5.8).

An independent group of US physicists ranked the used US devices into three categories: out-dated, modern, state-of-the-art. All but three (modern) devices classified as out-dated.

Conclusions: Insufficient diagnostic accuracy disqualified US for DVT screening, supporting the use of CP for this purpose. In contrast, taking pre-test probabilities into account, post-test probabilities for proximal and distal DVT in case of negative US-screening were acceptably low, providing evidence, that US may indeed be usable for DVT screening. The insufficient diagnostic accuracy is likely caused by the outdated US equipment, representing the standard from the late 1980ies rather than the 21st century. Considering the dramatic improvement in US techniques over the last decade the presented data are inadequate for describing the quality of present US devices. Thus, albeit ethical concern, new trials comparing CP against CUS are required to solve the conflict between CP and US.