Objective: Trauma centers that perform more duplex ultrasounds report more deep vein thromboses (DVT). However, it is uncertain if this is due to variations in hospital practices or patient characteristics and case mix. We hypothesize that admission to trauma centers that use duplex ultrasound more frequently, independently predicts DVT reporting for individual patients, controlling for patient-level risk factors.
Methods: We analyzed patients from trauma centers reporting at least one vascular ultrasound and one DVT to the National Trauma Data Bank (v6.2). Because National Trauma Data Bank contains no data on hospital duplex surveillance practice, we defined "screening" trauma centers as those performing ultrasound on more than 2% of patients. The primary outcome measure was DVT diagnosis. Multiple logistic regression was performed, using patient-level risk factor covariates as well as hospital duplex rate to compare patients at "screening" versus "non-screening" trauma centers. Sensitivity analysis was performed by varying duplex rate cutoff, outcome measure, and patient population.
Results: Approximately half of 492,496 patients were admitted to "screening" trauma centers. Unadjusted DVT rate was threefold higher in "screening" trauma center patients (1.18% vs. 0.35%, p < 0.001). Age > or = 40 years, extremity injury, head injury, ventilator days > or = 3, venous injury and major surgery were independently associated with DVT diagnosis. "Screening" trauma center admission was independently associated with a higher likelihood of DVT reporting (odds ratio, 2.16; 95% confidence interval, 1.07-4.34). No qualitative differences were identified on sensitivity analyses.
Conclusions: Trauma center ultrasound practice is an independent predictor of DVT diagnosis for individual patients, controlling for patient-level risk factors. Elevated DVT rates at these centers are due to surveillance bias. In the absence of standardized surveillance, hospital DVT rate is an inappropriate quality of care measure after trauma.