Why Risk Assessment Models are Ineffective in Predicting Venous Thromboembolism in Plastic Surgery Patients

Shaikh et al1 claim that combining the American Society of Anesthesiologists Physical Status Classification System (ASA) and Caprini scores improves the ability to predict venous thromboembolism (VTE) in plastic surgery patients.The authors' Table 9 reports the sensitivity and specificity for both risk assessment models individually, and then a combined test using ASA and Caprini scores. The authors report a sensitivity of 86.6% for the combined test. No data are presented to support this dramatically improved index. According to the authors,1 such a combined screening test would be positive if the patient had both an ASA score ≥3 and a Caprini score ≥5. In fact, combining methods can only decrease the test sensitivity because the subset of patients meeting high-risk criteria for both models would be less than the number identified using either individually. The maximum possible number of true positives for the combined test would be 13, for a sensitivity of 54.2%, and this level would only be achieved in the unlikely event that all 13 patients with an ASA score ≥3 also had Caprini scores ≥5.Similarly, the authors …