Chest pain prediction tool

The second article on the ED evaluation of chest pain involved an instrument to predict/stratify cardiac risk, finding it to be quick, reliable and efficient (see MI HEART score in ER AIM2017 in dropbox, or doi:10.7326/M16-1600).

Details:

--Nine Dutch hospitals assessed the HEART instrument prospectively to evaluate unselected patients presenting to EDs with chest pain, in a sequence where every 6 weeks, 1 hospital was randomly switched to using the instrument (see MI HEART score in ER AIM2017 in dropbox, or doi:10.7326/M16-1600). Publicly-funded study.

--3648 patients (1827 receiving usual care, 1821 HEART care)

--Exclusion criteria included evident ST-segment elevation MI.

--The HEART score is based on History, Electrocardiogram, Age, Risk Factors, and Troponin levels , with each having a score range of 0-2 (see https://www.mdcalc.com/heart-score-major-cardiac-events for HEART score calculator):

   --score of 0-3 is low-risk, and the patient was to be discharged with reassurance

   --score of 4-6 is intermediate-risk, with recommendation for hospitalization for observation and investigation

   --score of 7-10 is high-risk, and prompted invasive treatment

   --But, physicians could overrule the score's recommendation

--primary outcome: incidence of MACE (major adverse cardiac events) within 6 weeks, including: MI (with or without ST-segment elevation), unstable angina, percutaneous coronary intervention, CABG, >50% stenosis treated conservatively, or death from any cause

Results:

--low HEART score found in 715 (39%); intermediate in 231 861(47%), and high in 190 (11%)

-- MACE within 6 weeks, after using HEART was 1.3% lower than during usual care (ie non-inferior):

    --HEART: 18.9%

    --usual care: 22.2%

--Usual care: 405 patients (22.2%): 9 (0.5%) cardiovascular death, 400 (21.9%) with cardiac ischemia, 290(15.9%) with significant stenosis

--breakdown of MACE by HEART score:

    --low score: 14 patients (2.0%): 1 (0.1%) cardiovascular death, 10 (1.4%) with cardiac ischemia, 10 (1.4%) with significant stenosis

    --intermediate score: 175 patients (20.3%): 2 (0.2%) cardiovascular death, 162 (18.8%) with cardiac ischemia, 117 (13.6%) with significant stenosis

    --high score: 140 patients (73.7%): 2 (1.1%) cardiovascular death, 143 (75.3%) with cardiac ischemia, 102 (11.8%) with significant stenosis

--no difference in early discharge, readmissions, recurrent ED visits, outpatient visits, or visits to general practitioners

--BUT, nonadherence to protocol occurred in 313 of 1766 (18%) of patients, 291 of 715 (41%) of low-risk patients and 22 of 190 (12%) of high risk patients. Nonadherence for low-risk patients consisted of prolonged observation or hospitalization after presentation at the ED in 80% of them, a 2^nd troponin measurement in  58%, and stress exercise testing in 18%. No data were presented on outcomes of those patients who are on-protocol vs those who were outside of the protocol

-- overall, for the low-risk patients (39% of the total), the HEART score was 99% sensitive in identifying these patients as low risk and eligible for early discharge.

Commentary:

--only 20% of patients coming to the ED with chest pain have acute coronary syndrome. But one of the difficulties is that about 50% with acute coronary syndrome do not have classic symptoms. And 2-6% of patients with acute coronary syndrome are missed by current practice.

--Overall in the Netherlands (and other countries), management is conservative and 2/3 of the patients with chest pain get admitted. So, this study adds to the data that using a prescribed simple instrument (HEART in this case, hs-cTnT in the previous blog) can lead to efficient and safe risk-stratification

--one interesting contradiction in these 2 studies is that the HEART score represents much more "gray area", as opposed to the all-or-none issue with the single high-sensitivity troponin test in the last blog. And per the HEART protocol, the pretty common clinical situatione of someone who is 45yo, has a history of hypercholesterolemia/ hypertension/diabetes, some non-specific ECG changes but only a slightly suspicious cardiac history and a normal troponin (and most hospitals in this study used a high-sensitivity assay) would have a HEART score of 4, leading to admission. And a person aged 65 with the same risk factors and a normal ECG and troponin would similarly be admitted. Maybe that is reasonable, but these people would be missed by using the single hs-troponin level as in the last study. In this regard, it would be useful to know if there should be different ratings in the HEART scale: ie, if someone has a HEART count of 3 (low-risk) with a high troponin, do they really have the same risk of MI or ACS as someone with the same count but a normal troponin? Or alternatively, is there a difference between those with the same intermediate risk of 4 or 5 with a normal troponin vs a high one? It would be useful to see more granular data from the HEART study the assess post-hoc how important the troponin component was.

-- Overall, the study was impressive in that included 9 different hospitals of different types, had 99.9% follow-up, and its design allowed within-hospital comparisons. And, they captured all of the clinically relevant major cardiac adverse outcomes.

-- unfortunately, one major problem with this study was that ED physicians were hesitant to send low-risk patients home, though the final analysis showed non-inferiority to this approach. Given the high number of patients who were treated by protocol, and the rather dramatic outcome differences between the low-risk and the higher-risk categories, it seems that this tool worked quite well

So, these last 2 blogs are pretty encouraging that we may soon be able to risk-stratify patients with chest pain adequately (the accepted false negative rate on chest pain work-ups in the ED, whether appropriate or not, is in the 2% range, similar to what this study found). Would be great to have a point-of-care high-sensitivity troponin test (per the last blog), though this study suggests that using the HEART tool even without a troponin level might bring the risk above the low-risk category for some patients (leading to direct referral to the ED), or, alternatively, categorize the patient as low risk if they have 1 point or less on the HEART scale (leading to discharge and close followup), because even adding 2 points in those with an elevated troponin would not matter.  This, of course, should be tested in a well-conducted study to see if it were valid.