Inter-Observer Variability in the Assessment of the 4Ts Score for the Diagnosis of Heparin Induced Thrombocytopenia (HIT) in Patients Undergoing Cardiac Surgery (CS).

HIT is a potentially serious complication of heparin therapy. Because of the multiple conditions potentially causing thrombocytopenia, particularly in clinical scenarios such as patients undergoing CS or admitted to intensive care units, its diagnosis is frequently not straightforward and depends on a combination of clinical suspicion and laboratory tests. The 4Ts score has been proposed as a tool to assess the pretest probability of a HIT diagnosis (as low, intermediate or high) and it comprises 4 clinical parameters: severity and timing of onset of thrombocytopenia, development of thrombosis, and the presence of alternate causes for thrombocytopenia. This score has several handicaps including the fact that the temporal profile of thrombocytopenia onset is frequently unclear, the history of previous heparin exposure is not always available, the diagnosis of new thrombosis (including extensions of previous ones) is often difficult, and the judgment about the likelihood of an alternate cause for the thrombocytopenia is entirely subjective. The clinical usefulness of a scoring system depends mainly on its robustness and reproducibility and therefore inclusion of difficult to evaluate or subjective components might compromise its clinical performance.

We evaluated the inter-observer agreement for classifying the probability of HIT in a population of patients undergoing CS.

We conducted a retrospective study of patients admitted for cardiac surgery to our institution between January 2006 and December 2008 and in whom HIT was suspected. Clinical information including all necessary data for calculating the 4 components of the 4Ts score was collected in a standardized database which was used by 2 independent observers (blinded to serological tests results) to calculate the score and the pretest probability for HIT as low (≤3 points), intermediate (4-5 points), or high (≥6 points). Scores assigned by both observers were compared using a Wilcoxon signed ranks test and the inter-observer agreement for scores and pretest probabilities was evaluated using a Kappa statistic. 95% confidence intervals for proportions were estimated using the Wilson score method.

73 patients were included in the analysis. The score assigned by both observers differed in 40 cases (54.8%; 95% CI 43.4, 65.7; p=0.001). The value of the Kappa statistic for the inter-observer agreement for scores was 0.266 (95% CI 0.111, 0.421) and for pretest probabilities was 0.400 (95% CI 0.218, 0.582).

In this study, we found that the 4Ts score has an inter-observer agreement ranging between slight and moderate when applied to patients undergoing CS. The variability observed in the assessment of the score raises doubts about its usefulness when evaluating the possibility of HIT in this patient population. Limitations of this study include a relatively small sample size, inclusion of a single clinical setting, and its retrospective nature. Further studies in this and other populations are needed to assess the reproducibility of the 4Ts score.

No relevant conflicts of interest to declare.