Design: Prospective clinical cohort study.
Patients: 193 sequentially enrolled patients evaluated for suspected symptomatic deep vein thrombosis of the leg, reporting to the LDS Hospital or Intermountain Medical Center Peripheral Vascular Laboratory. Vascular technicians will screen patients for study entry with a simple questionnaire. Those meeting screening criteria will then be interviewed by the study coordinator to verify they meet inclusion criteria, have no exclusions, and provide informed consent.
Demographic and medical history information:
Study clinical variables are obtained from the enroller interview and electronic medical record at entry and included in a computerized clinical database. They include age, sex, weight, inclusion and exclusion criteria, and clinical information necessary to create the pre-test clinical score. Demographic information and basic biometrics (gender, height, weight, leg circumference) will be obtained and recorded by the study coordinator. Informed consent documents will be obtained and stored according to the policies of the Institutional Review Board. All female patients will undergo urine or serum pregnancy testing prior to study entry.
Measurements: A research clinician will perform a brief clinical assessment specific for determining pre-test probability of DVT prior to evaluation with diagnostic tests. This formalized scoring system has been validated in a prior study. Patients with a clinical score of "DVT unlikely" will undergo comprehensive duplex ultrasonography. If normal, patients will be followed for clinical events as described below.
Patients not studied(comprising a group with lower pre-test probability of DVT) will go on to diagnostic testing and further care by the referring physician. Demographic information will be used in the descriptive portions of the study to demonstrate risk prevalence, but patients will not be followed for clinical outcome.
The scoring system to determine pre-test probability is that described by Wells et al. Patients with a score corresponding to "DVT Unlikely" will be tested and treated according to the treating clinicians, and are not followed for outcomes.
Comprehensive, real-time B-mode ultrasonography with color Doppler analysis (CDU) will be performed on patients with a pre-test probability of "DVT Likely."
A standardized technique will be used for the CDU examination. Compressibility of the veins will be assessed. The results will be categorized as normal if all imaged venous segments are fully compressible, as abnormal if a noncompressible segment is identified, or as inadequate for interpretation.
Interpretation will be performed by experienced vascular surgery staff according to protocols currently in use at LDS Hospital and Intermountain Medical Center. Findings will be redacted to specific data fields by the interpreting surgeon according to a form generated for this purpose. The form will be completed by the interpreting surgeon and added to the patient logbook for entry into the database by the data entry clerk.
Interobserver variability will be limited by using a priori criteria for a positive and negative study and by using only interpreters privileged by Intermountain Healthcare to interpret vascular studies, who have substantial experience and expertise. Interobserver agreement for this technique has been found to be high.
If the result of initial CDU is normal (no DVT identified), anticoagulation will be withheld, regardless of symptoms.
If the result of testing is abnormal (DVT identified), anticoagulation will be given unless contraindicated, according to established protocols.
Subjects whose results are classified as inadequate for interpretation will have reasons logged and be excluded from analysis. They will receive further management according to the discretion of the emergency physician.
Long-Term Follow-up Patients in the observation cohort (negative CDU) will be undergo three months of clinical follow-up, a strategy used in many diagnostic trials for DVT. Patients will be instructed to return immediately to our emergency department if they have symptoms or signs of venous thrombosis or pulmonary embolism. A thromboembolic disease specialist will assess them at 3 months by telephone interview.
At the follow-up assessment, an interval history will be taken with emphasis on specific symptoms (including leg pain, tenderness and swelling, chest pain, dyspnea, hemoptysis, and syncope), hospitalization, interval evaluation for DVT or PE and use of anticoagulants. For all patients who die, the cause of death will be determined from autopsy or by independent clinical review if autopsy cannot be obtained.
Results of the follow-up assessment will be entered on a data sheet by the assessing clinician created for this purpose. This sheet will be added to the logbook by the study coordinator and information appended to the database by the data entry clerk. The evaluating clinician will refer patients for appropriate testing for thromboembolic disease based on clinical suspicion. Patients suspected of having deep vein thrombosis will be referred for compression ultrasonography. If CDU is abnormal, patients will be treated with anticoagulation therapy. If normal, patients will be referred for venography (the current gold standard test) to exclude the diagnosis of DVT. Patients suspected of having pulmonary embolism will undergo objective testing with ventilation-perfusion lung scanning and, if indicated, pulmonary angiography or high-resolution computed tomography. Qualified physicians blinded to the subject's cohort will perform interpretation of all testing.
Main Outcome Measures and Statistical Comparisons:
Primary outcome measure will be venous thrombosis, pulmonary embolism (VTE) and death from thromboembolic disease during follow-up confirmed by objective testing in the 3 months following enrollment.
The study will be designed to estimate the event rate of thromboembolic disease and death attributable to thromboembolic disease in the normal cohort during the three-month follow-up period. An exact 95% confidence interval that excludes a rate of VTE of 3% will be defined as clinically acceptable, in accordance with previous trials of DVT diagnosis.
Adjudication of outcomes:
A panel of three independent physicians with established expertise in venous thromboembolism will be named to adjudicate all suspected events of VTE and all diagnostic tests for VTE occurring during the follow-up period. Simple majority will resolve disputes.
Methodological Issues and Avoidance of Bias Entering consecutive patients into the study will avoid selection bias. To avoid bias during the initial testing period, we have established pre-defined criteria for negative and positive studies. Further diagnostic testing on patients with negative results will not be undertaken unless dictated by follow-up evaluation. Patients in the observation cohorts will not be treated and therapy will be given to all patients with positive results. Diagnostic suspicion bias will be avoided by objectively testing all patients who return during follow-up with symptoms or signs suggestive of deep venous thrombosis or pulmonary embolism. Interpretation bias will be avoided by obtaining independent interpretation of follow-up testing in patients with suspected disease. The interpreter will be blinded to the results of the initial ultrasound and the cohort to which the patient belongs. All deaths will be independently reviewed. The reviewer will be blinded to the cohort to which the patient belongs. Patient confidentiality will be maintained by keeping the patient identity log separate from the database, which will use only unique numerical identifiers. Complete data tracking will be assured by comparison of enrollment log and database by independent reviewer.
Statistical Analysis and Sample Size
Descriptive statistics for age, sex, symptoms at presentation (pain, tenderness, duration of symptoms), clinical conditions (recent surgery, recent hospitalization, cancer, congestive heart failure, recent immobilization, recent childbirth, cellulitis, superficial phlebitis and family history of thromboembolic disease) and the pre-test probability score will be calculated to characterize the study population.
The event rate of thromboembolic complications and death from thromboembolic disease will be calculated for patients in the two observation cohorts.
A two-sided confidence interval for the event rate (objectively verified venous thromboembolism) will be calculated for the observation cohort by exact methods. If this confidence interval excludes the commonly accepted threshold event rate of 3%, we will conclude that the diagnostic strategy is clinically valid.
We will also describe the clinical characteristics of clots and other endpoints when those data are available.
The sample size of 193 patients who meet eligibility criteria was chosen so that an exact 95% confidence interval would exclude an event rate of venous thromboembolism in the observation cohort of 3%. Excluding an event rate of 3% is the commonly accepted standard by which diagnostic strategies are deemed clinically acceptable.
