Background The use of high-sensitivity troponin (hs-cTn) is established in guideline-directed clinical practice to facilitate the diagnosis or exclusion of myocardial infarction (MI). The manufacturer provided 99th percentile for this assay is derived from a small population of relatively healthy individuals but is used as an “upper limit of normal” (ULN) in routine hospital practice. This raises the question: does this ULN actually represent the 99th centile of the distribution of the assay in an all comers hospital population, in most of whom there is no clinical suspicion of an MI?

Methods Twenty thousand consecutive patients attending this large University Teaching Trust who had biochemistry tests for any clinical reason, either as in- or outpatient had an hs-cTn assay as part of this study. In those patients in whom hs-cTn was not requested by their supervising clinician, the result of the assay was nested, and never revealed to doctor or patient. Ethical approval for this method required special permission from the national Confidentiality Advisory Group following support from the British Cardiac Patients Association. The Beckman Coulter Access AccuTnI+3 assay was used throughout. These data were combined with the source location for the blood test, as well as the presence of coded co-morbidity since the year 2000.

Results There were 18171 patients left once patients discharged with a diagnosis of MI (122) and those suspected of MI (1707) by the clinical team were excluded. There were 4759 inpatients, 9280 outpatients and 4132 patients in the emergency department with 7.3%, 2.0% and 7.4% of these having an hs-cTn level above the manufacturer-provided ULN respectively. Table 1 demonstrates the prevalence of hs-cTn levels above the ULN in hospitalised patients by speciality. In terms of outpatients; cardiothoracic, renal and oncology/haematology patients were the most likely to have an hs-cTn above the ULN with 7.4%, 4.9% and 2.6% respectively above this level. There were 4265 (46%) outpatients in whom there was at least one coded co-morbidity present. These patients were more likely to have an hs-cTn above the ULN than those without a coded co-morbidity (2.8% versus 1.2%, P<0.0001). Table 2 demonstrates the prevalence of outpatients with an hs-cTn above the ULN by coded co-morbidity.

Conclusion These data suggest that application of a manufacturer-derived 99th centile for hs-cTn assay to a hospital population may be flawed, particularly if any assumption is made that a result above this level may represent acute MI in patients without a classical history. Caution and a good understanding of the assay is imperative for accurate diagnosis and management of hospital patients with an apparently elevated level, but these results also raise interesting questions about what the levels actually indicate in patients in whom the suspicion f acute Mi is low. More data are required.

Conflict of Interest Educational support from BAYER