Deep Vein Thrombosis Diagnosis with D-Dimer Adjusted to Clinical Probability
Bibliographic record
Abstract
Introduction Diagnostic testing for deep vein thrombosis (DVT) is a multi-step and time-consuming process. Testing starts with clinical pretest probability (C-PTP) assessment. A negative D-dimer in combination with low C-PTP is widely used to exclude DVT; otherwise ultrasound imaging is required. When proximal vein ultrasound is used, a repeat ultrasound after a week is usually required to exclude DVT in moderate or high C-PTP patients. Ultrasound imaging is costly and can introduce delays. The goal of this study was to evaluate the safety and efficiency of a diagnostic algorithm for DVT that was designed to minimize the need for ultrasound imaging by using C-PTP-based D-dimer thresholds to exclude DVT (the 4D algorithm), rather than a standard fixed D-dimer cut-off value. Methods Consenting patients were enrolled in a Canadian prospective multicentre management study. Outpatients with symptoms or signs of DVT were eligible to be included in this study. Physicians used the 9-item Wells score to categorize the patient's C-PTP as low (Wells score, -2 to 0), moderate (1 or 2), or high (≥3). Patients with low C-PTP and a D-dimer <1,000 ng/mL or with a moderate C-PTP and a D-dimer <500 ng/mL underwent no further diagnostic testing for DVT and did not receive anticoagulant therapy. All other patients underwent proximal vein ultrasound. Patients with a single negative ultrasound but very high D-dimer (low or moderate C-PTP with D-dimer ≥3000 ng/mL, high C-PTP with D-dimer ≥1500 ng/mL) had a second proximal venous ultrasound one week later. The primary outcome was symptomatic, objectively verified, venous thromboembolism (VTE), which included proximal DVT or pulmonary embolism. All patients were followed for 90 days. A sample size of 1500 was required to establish 4D algorithm safety (90-day post-test probability of VTE <2%). Results From April 2014 through March 2020, a total of 1512 patients were enrolled and analyzed. The mean age was 60 years and 58% were female. Overall, 173 (11%) had DVT on initial or serial diagnostic testing (168 had DVT on ultrasound imaging on the day of presentation and 5 had DVT on repeat ultrasound imaging at one week). Of all 1298 patients (86% of total) who did not have DVT (at either initial presentation or at scheduled repeat ultrasound imaging) and who did not receive anticoagulant therapy, 7 had VTE during follow-up (0.5%, 95% confidence interval (CI): 0.3 to 1.1%). In the 579 patients who had low (378 patients) or moderate (201 patients) C-PTP and negative D-dimer results (i.e. <1000 or <500 ng/mL respectively) and who did not receive anticoagulant therapy, 2 had VTE during follow-up (0.4%, 95% CI: 0.1 to 1.3%). In the 572 patients with a single negative ultrasound who were low or moderate C-PTP with D-dimer <3000 ng/mL (423 patients), or high C-PTP with D-dimer <1500 ng/mL (149 patients) and who did not receive anticoagulant therapy, 3 had VTE during follow-up (0.5%, 95% CI: 0.2 to 1.5%). The difference in the mean number of ultrasound examinations with the 4D algorithm (0.72) compared with the conventional algorithm (1.36) was -0.64 (95% CI, -0.68 to -0.61), corresponding to a 47% relative reduction (1083 ultrasound scans performed with the 4D algorithm compared with 2053 ultrasound scans required for the conventional algorithm). Conclusions The 4D diagnostic algorithm ruled out DVT safely while substantially reducing the requirement for ultrasound imaging. Disclosures Wu: BMS-pfizer: Honoraria, Other: advisory board; leo pharma: Other: advisory board; Pfizer: Honoraria; Servier: Other: advisory board.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.004 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.003 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".