The utility of preoperative hemostasis screening to predict complications is uncertain. The authors quantified the screening rate in US neurosurgery patients and evaluated the ability of abnormal test results as compared with history-based risk factors to predict hemostasis-related and general outcomes.
Eleven thousand eight hundred four adult neurosurgery patients were identified in the 2006–2009 American College of Surgeons National Surgical Quality Improvement Program database. Multivariate logistic regression modeled the ability of hemostatic tests and patient history to predict outcomes, that is, intra- and postoperative red blood cell [RBC] transfusion, return to the operating room [OR], and 30-day mortality. Sensitivity analyses were conducted using patient subgroups by procedure.
Most patients underwent all 3 hemostatic tests (platelet count, prothrombin time/international normalized ratio [INR], activated partial thromboplastin time), but few had any of the outcomes of interest. The number of screening tests undergone was significantly associated with intraoperative RBC transfusion, a return to the OR, and mortality; an abnormal INR was associated with postoperative RBC transfusion. However, all tests had low sensitivity (0.09–0.2) and platelet count had low specificity (0.04–0.05). The association between patient history and each outcome was approximately the same across all tests, with higher sensitivity but lower specificity. Combining abnormal tests with patient history accounted for 50% of the mortality and 33% of each of the other outcomes.
This is the first study focused on assessing preoperative hemostasis screening as compared with patient history in a large multicenter sample of adult neurosurgery patients to predict hemostasis-related outcomes. Patient history was as predictive as laboratory testing for all outcomes, with higher sensitivity. Routine laboratory screening appears to have limited utility. Testing limited to neurosurgical patients with a positive history would save an estimated $81,942,000 annually.