Interrater agreement for two systems used to determine the probability of heparin-induced thrombocytopenia.

PURPOSE: The interrater reliability of the 4T's method and the HIT expert probability (HEP) score for clinical evaluation of suspected heparin-induced thrombocytopenia (HIT) was investigated.

METHODS

Patients hospitalized over a three-year period who were tested for HIT via anti-platelet factor 4 (anti-PF4) antigen assay were identified using laboratory data; 127 patient cases met the study inclusion criteria. Nine clinical pharmacists with expertise in HIT management evaluated the 127 cases using two pretest scoring systems: the 4T's score and the HEP score. Each case was independently evaluated using both 4T's and HEP scores. The primary endpoint was interrater agreement of overall 4T's and HEP scores and individual item scores.

RESULTS

Raw agreement of values assigned by the two raters for each of the four items comprising the 4T's score ranged from 0.54 to 0.86, with agreement of 0.63 for final patient categorizations. Raw agreement of rater weightings of the eight HEP scoring items ranged from 0.34 to 1.0; for dichotomization of patients at the suggested screening cutoff value (>2.0), agreement was 0.65. Kappa coefficients were 0.15-0.45 for 4T's item scores and 0.17-0.70 for HEP score item scores. With both scoring systems, low rater agreement mainly related to determination of the timing of thrombocytopenia and possible other causes of the disorder.

CONCLUSION

In a retrospective study, inter-rater agreement in scoring of HIT probability via the 4T's and HEP scoring systems was relatively low. The HEP score did not increase interrater reliability or correlation with anti-PF4 antibodies compared with the 4T's score.