Abstract
INTRODUCTION
Patients with elevated 4Ts scores for suspected heparin-induced thrombocytopenia (HIT) require immediate non-heparin anticoagulation until negative results of PF4/heparin immunoassay or, if positive, confirmatory serotonin release assay (SRA). These patients are at high risk of thrombosis and bleeding complications. Traditional ELISA-based immunoassays have a 1-3 day turnaround time (TAT). Available rapid HIT assays provide results within 1 hour but require an upfront investment for expensive coagulation analyzers. We aimed to determine if rapid PF4 testing would be cost-effective by minimizing risk exposure time to therapeutic anticoagulation compared to traditional testing in patients with suspected HIT.
METHODS
In this independent analysis free of industry influence, we built a Markov simulation model of patients with intermediate/high pretest probability for HIT hospitalized in the intensive care unit undergoing the following testing strategies: (1) rapid assays (TAT; <1 hour) with chemiluminescence immunoassay (CLIA) or latex immunoturbidimetric assay (LIA), vs 2) ELISA-based assays at 1-day and 3-day TAT (at both low and intermediate thresholds). All patients in an ELISA-based strategy were switched to non-heparin anticoagulation whereas patients in a rapid-based strategy were switched only if the assay was positive. Both bivalirudin (base-case) and argatroban (scenario analysis) were examined. For patients testing positive, we also evaluated variability in confirmatory SRA testing at both 3- and 5-day SRA TAT in scenario analyses. Transition probabilities for clinical HIT diagnosis and management were informed from a real-world bivalirudin-treated cohort of patients with suspected HIT, known HIT assay test characteristics and cost. Conservatively, zero quality-of-life benefit was assumed for faster diagnostic adjudication of HIT. The primary outcome was the incremental cost-effectiveness ratio or, if the intervention was found to be cost saving, its reformulation as the incremental monetary net benefit (iNMB), reported across all accepted willingness-to-pay thresholds from the US health system perspective. In additional scenario analysis, we accounted for the maximum coagulation analyzer upfront and extended warranty cost (i.e., for rapid testing), reporting results across a 5-year budget time-horizon. The secondary outcomes were threshold analyses for minimum annual number of rapid HIT tests sent (for a rapid testing strategy to be cost-effective) and two-way sensitivity analysis on number of annual tests and coagulation analyzer cost. We concluded by conducting probabilistic analyses to capture uncertainty in all parameters over 10,000 Monte Carlo iterations.
RESULTS
Rapid testing (with CLIA or LIA) was the cost-saving (and more effective) strategy in 100% of all iterations in the base-case and all scenarios. Cost savings increased further with longer TAT for ELISA-based testing. Specifically, iNMBs per-patient tested were $740 and $1,940 in favor of CLIA versus low-threshold ELISA-based testing at 24- and 72-hour TAT, respectively, and $530 and $1,750 versus intermediate-threshold ELISA-based testing at 24- and 72-hour TAT, respectively. This benefit was also seen for LIA rapid testing over ELISA-based testing, which although slightly less than with CLIA, always favored rapid testing. 5-day SRA TAT increased the iNMB further with rapid testing in all cases, again regardless of non-heparin anticoagulant used. Between rapid diagnostic strategies, CLIA was favored over LIA >80% of the time across scenario analyses. There would need to be a minimum of at least 29 CLIA assays run annually, when compared to low-threshold ELISA-based 72-hour TAT, to offset the combined upfront and extended warranty coagulation analyzer cost over 5 years.
CONCLUSIONS
Rapid assay testing for HIT patients with elevated 4T scores produces significant cost savings per patient tested and empirically treated with either bivalirudin or argatroban. This is achieved by reducing the costs associated with unnecessary non-heparin anticoagulation and major bleeds in patients who test negative for HIT. In addition, our model allows individual stakeholders to assess cost-effectiveness based on their own quoted coagulation analyzer price and expected annual volume of HIT diagnostic tests.
