Is documentation of risk assessment associated with positive imaging in patients with craniocervical trauma or suspected venous thromboembolism?.

This study aimed to determine whether documentation of the result of standardised risk assessment of emergency department patients with suspected head or cervical spine injury or venous thromboembolism was associated with a higher rate of positive imaging. A pilot audit of three metropolitan emergency departments (EDs) was conducted over a 7-week period between February and May 2009. On randomly selected days that were not known by ED medical staff, electronic referrals for imaging were read by two radiology trainees with the aim of determining whether or not the result of risk assessment using previously implemented, standardised risk assessment tools (NEXUS, NICE, Kline and Wells, Royal Children's Hospital, Melbournel head injury guideline) had been documented on the request for imaging. After this was recorded, the report was read to determine whether the study (CT brain adult and child, CT cervical spine adult and child, lower limb venous ultrasound, CT pulmonary angiogram) showed any clinically significant pathology. Scoring of referrals as positive or negative for documentation of risk assessment outcome was checked by a consultant radiologist for the first 2-week period with feedback to the trainees regarding coding of requests. A total of 160 examinations were audited. There was >85% agreement between the trainees and consultant for coding of referrals with regard to risk assessment. At Hospital 1, 8.2% (95% CI: 3.0-17.2%) of imaging in 61 patients was positive and 50% of patients had a valid risk assessment documented. For Hospital 2, 17% (95% CI: 8.6-29%) of 52 studies were positive with an 87% risk assessment documentation. For Hospital 3, 15% (95%CI: 6.8-27.3%) of imaging was positive with 81% risk assessment documentation. Imaging was positive substantially more often in EDs where risk assessment documentation occurred most frequently. Small sample size may be the reason for this difference being non-statistically significant at this early point in the