Suspicion of heparin-induced thrombocytopenia in internal medicine: How appropriate is the ordering of anti-PF4/heparin antibody testing?

PMID 25275932


Thrombocytopenia is frequent in hospitalized patients, and heparin-induced thrombocytopenia (HIT) is often suspected when a decrease in platelet count is concomitant with heparin treatment. ELISA tests used for anti-PF4/heparin antibodies detection usually have high sensitivity but only fair specificity for HIT. Pre-test probability scores (such as 4 Ts or HEP scores) have been validated and a low probability score rules out HIT without anti-PF4/heparin testing. The aims of this study are to evaluate the appropriateness of anti-PF4/heparin testing according to pre-test probabilities of HIT and to compare the abilities of the 4 Ts and HEP scores to avoid inappropriate anti-PF4/heparin testing. This retrospective observational study included 74 consecutive patients hospitalized in a general internal medicine division who had anti-PF4/heparin testing for suspicion of HIT. 4 Ts and HEP scores were computed retrospectively. About 73% of patients who had ordering of an anti-PF4/heparin were at low risk according to the 4 Ts score, and 46% according to the HEP score. Heparin was discontinued in 61% and 62% of low-risk patients according to 4 Ts and HEP scores and switched to alternative anticoagulant in 31% and 32% of them, respectively. No case of HIT was diagnosed in patients with a low-risk score. One major bleeding and no thrombosis were observed. For the 4 Ts score, the sensitivity was 100%, the negative predictive value (NPV) was 100%, the specificity was 77%, and the positive predictive value (PPV) was 20% (95% CI: 7-44). For the HEP score, the sensitivity was 100%, the NPV was 100%, the specificity was 49%, and the PPV was 10%. In conclusion, pre-test probability scores were vastly underused in this internal medicine population despite their ability to rule out HIT without laboratory testing in a large proportion of patients. Appropriate use of those instruments should be actively promoted.