Accuracy of a prediction model for heparin-induced thrombocytopenia (HIT): An analysis based on individual patient data

Background: Heparin-induced thrombocytopenia (HIT) is an adverse drug reaction associated with thrombosis, and its paradoxical nature is a challenging issue for the diagnosis. The '4Ts' scoring system represents a simple and efficient way to improve clinical diagnoses of the syndrome. This system classifies patients as having high, intermediate, and low clinical probability for HIT. However, uncertainty remains concerning its clinical meaning, thus weakening the diagnostic value of this screening instrument. Methods: We analyzed the diagnostic test accuracy based on individual patient data extracted from published primary scientific studies. This study focused on 186 cases of treatment with heparin, which later evolved into a clinical suspicion of HIT. Upon choosing the most appropriate reference laboratory, the accuracy of the 4Ts was analyzed using the receiver operator characteristic curve analysis. Results: Half of the positive cases (57.1%) were classified as having a high score, while 25.5% of the negative cases were classified as a having low score for HIT. Slightly more than half of all patients (53.2%) were classified as having an intermediate score. As such, the pre-test instrument would most likely fail to distinguish between diseased and nondiseased patients in a relevant number of cases. The calculated accuracy of the summary indicates that the 4Ts can be considered a good, but not a defining, test. Conclusion: Further studies are warranted regarding clinical score systems, either alone or in combination with laboratory tests, in an attempt to improve the early diagnosis of this adverse drug reaction and to provide better care for at-risk patients. (C) 2011 Elsevier B.V. All rights reserved.