Background: Heparin induced thrombocytopenia (HIT) is a prothrombotic adverse drug reaction to heparin. The 4T scoring system is a well validated system for assessing the pre-test probability of HIT and has a high negative predictive value for excluding HIT in most clinical situations, however, only a minority of patients have been evaluated in the ICU setting. The critically ill pose different challenges when compared with other medical and surgical patients. They have increased rates of thrombotic and haemorrhagic complications. Thrombocytopenia is reported in approximately 30–50%. We examined our ICU experience to evaluate the pairing of the 4T scoring system and EIA-GTI PF4 ELISA assay to evaluate their diagnostic accuracy in confirming a diagnosis of HIT in the acute setting. The gold standard measure is the serotonin release assay (SRA), but this is not routinely available in the UK.Methods: This single centre, retrospective, observational study involved reviewing the records of 75 consecutive patients investigated for HIT on our ICU from February 2006 to April 2008. We follow the British Society of Haematology guidelines for the management of HIT. Initial testing is with the EIA-GTI PF4 assay (positive if >0.4 OD) coupled with a heparin induced platelet activation assay (HIPA).Results: During the 26 months studied, 3.5% of ICU admissions were screened for HIT. A diagnosis of HIT, defined in this study as an intermediate or high 4T pre-test probability score and an ELISA result >0.7 OD was present in 0.46%. 45% of patients were medical, 40% had cardiothoracic surgery, the remaining 15% were general surgical patients. 66% had three or more organ failure, 70% received haemofiltration. Severe sepsis and haemofiltration were alternative causes of thrombocytopenia in 97% of patients. The 4T's were used to risk stratify patients, 68% of patients had a low score, 29% were at intermediate risk and 3% had a high probability of HIT. 76% of patients with a low pre test probability had a negative ELISA and none developed a thrombosis. 4 (8%) patients with a low pre-test probability had a positive ELISA with an optical density >0.7 OD. 4 patients had a low pre-test probability but a positive ELISA, none of whom developed a thrombus. 2 patients had a positive ELISA and a positive HIPA, both had an intermediate pre-test probability. 1 patient had a negative ELISA (<0.4 OD) in the context of a high pre-test probability, he became thrombocytopenic on repeated heparin exposure. There was no correlation between a high pre-test probability and the magnitude of the ELISA result. 10 patients were treated with lepirudin, 2 developed haemorrhage requiring interruption of anticoagulation and transfusion.Discussion: The incidence of HIT recorded over the 26 month study period at GSTT is similar to the incidence of 0.3 -0.5% reported by Selleng. No diagnoses of HIT were missed after increasing our ELISA threshold from 0.4 - 0.7 OD. Neither the 4T pre-test probability nor the EIA GTI PF4 ELISA alone are sufficient to exclude a diagnosis of HIT. The 4T scoring system although well validated system may not work as well in the critically ill. The high prevalence of thrombocytopenia and alternative causes of thrombocytopenia mean the presence of thrombosis and the timing of the thrombocytopenia post heparin exposure are the major discerning features. Lo has previously reported the potential to over diagnose HIT by 100% when using the EIAGTI ELISA (Lo GK. Am J Hematol. 2007; 82: 1037–43). The positive ELISA results in those patients with a low pre-test probability may be false positives secondary to cross-reacting autoantibodies. The high prevalence of acute renal failure highlights the importance of making an accurate diagnosis of HIT. Use of lepirudin in renal failure is associated with an increased risk of haemorrhage. We use 10% of the standard dose in patients with renal failure, haemorrhage occurred despite intensive monitoring in 20% of the patients. Our study is limited by its retrospective design and the lack of SRA testing. We recommend that both tests are performed in patients at risk of HIT. Increasing the positive ELISA threshold to 0.7 OD improves the specificity of the ELISA for detecting clinically significant heparin/PF4 antibodies in critically ill patients.