Abstract
Background: A total of 826 patients were included in three ‘play-the-winner’ studies to investigate the safety of prophylaxis against venous thromboembolism in digestive surgery. To characterize patients benefiting from prophylaxis with low-molecular heparin, the 445 patients allocated to enoxaparin were investigated. Methods: A training set consisting of 292 patients from 2 of the studies was analysed by using a linear discriminant model. The reliability of the results was verified on a test set consisting of 153 patients from the third study. Results: A typical ‘winner’ was a young patient, preferably female, with serum bilirubin in the lower normal range, combined with body temperature, serum sodium, creatinine, and albumin in the upper normal ranges. By using the discriminant function on the test set, 81.7% of the ‘winners’ and 21.1% of the ‘losers’ were correctly classified. Conclusion: The discriminant function for characterization of winners to enoxaparin was found adequate. No rule was acceptable for characterization of losers.