Study Objectives: The HEART score is a promising, new quantitative cardiac risk calculator which has been proposed as a useful aid for guiding disposition decisions for patients with potential cardiac chest pain. Derived and validated on ED patients, this simple outcome predictor has been shown to accurately quantify the likelihood of a major adverse cardiac event within 30 days of an index visit. While the HEART score has performed well in rigorously controlled clinical trials, its reliability in the hands of clinicians has never been studied. We sought to evaluate the potential variability of HEART score calculations among ED providers. Method(s): This was an IRB-approved REDCap survey of emergency medicine providers. A convenience sample of clinicians was recruited locally and through EMDocs. Participants were asked to categorize standardized patient scenarios, per the HEART score criteria, as if in clinical practice using whatever tool they normally would in the ED (eg, smartphone app, online calculator). Five clinical histories, five ECGs, five age descriptions, five risk factor profiles, and five troponin results, designed to be realistically ambiguous but not particularly challenging, were offered for categorization (0-2) along with one rather straightforward composite scenario requiring a 0-10 HEART score calculation. Result(s): Among a diverse group of clinicians who were familiar with the HEART score and have used it for ACS risk-stratification (n=41), there was substantial variability in the calculated HEART score for the composite scenario (Figure [the solid bars highlight the transition zone from low-to moderate-risk for ACS-a critical juncture where 37% of the scores tend to be grouped]). Despite the HEART score's purported objectiveness, scores ranged from 1 to 7 with some correctly classifying the patient data as low-risk (score of <3) for ACS and others classifying it as moderate-(score of 4+) to high-risk (score of >7). The results for individual HEART score components were similarly inconsistent. The study is limited by our small sample size, which restricts meaningful inference testing. Conclusion(s): Our results suggest that, absent the rigid definitions and strict oversight of a controlled clinical trial, HEART score calculations by clinicians can vary substantially, which may disrupt the expected performance characteristics of the instrument. We speculate that the results are due to nebulous criteria (ie, what defines high, moderate, and slight suspicion for the history of present illness), misinterpretation of the data (eg, ECG and troponin values), being misguided by the tool used (eg, what qualifies as an ACS risk factor and how they are defined), or a well-established, subconscious, risk-minimizing provider bias (ie, accepting a lower positive troponin threshold or upgrading the history to generate a more conservative disposition). A larger-scale study is warranted and a qualitative follow-up study is needed to better understand these findings. [Table Presented].
Villarroel, N., Mader, T. J., Soares, W., & Westafer, L. (2017). 75 Clinician HEART Score Calculation Variability: A Pilot Study. Annals of Emergency Medicine, 70(4), S31. https://doi.org/10.1016/j.annemergmed.2017.07.100