low risk chest pain: dangerous to admit

 a prospective observational study looked at a large number of low-risk patients admitted for chest pain and their outcomes (See chest pain low risk not admit jamaintmed 2015 in dropbox, or doi:10.1001/jamainternmed.2015.1674). data was collected from July 2008 until July 2013 from the emergency depts (EDs) of  3 academic mid-Western US hospitals. details:

--45,416 patients were seen for chest pain, and 22,457 (49.5%) were admitted, of whom 11,230 met inclusion criteria of: primary presenting symptom of chest pain, chest tightness, chest burning or chest pressure; and two negative serial troponins, 60-420 minutes apart

--primary outcome was short-term (ie, during the hospitalzation) development of life-threatening arrhythmias, inpatient ST-segment elevation MI (STEMI), cardiac or respiratory arrest, or death during hospitalization (ie, CRACE, or clinically relevant adverse cardiac events).

--mean age 58, 44.8% arrived by ambulance, 55% women. hypertension in 46%, diabetes in 20%, prior MI in 13%.

 --results (based on 5% randomized sample of the 11230 patients):

    --primary endpoint was found in 20 of 11230 patients (0.18%, 0.11%-0.27%), where mean age was 71, 50% women 

            ​--excluding patients with abnormal vital signs, EKG-ischemia, LBBB, or pacemaker rhythm, the primary endpoint was found in only 4 of 7266 (0.06%, 0.023%-0.14%). [they excluded these patients from the analysis because admission would have been appropriate, and they were looking at really low risk people to see if they really needed admission]

            ​--of these 4 events: 2 were noncardiac (one with a GI bleed from over-anticoagulation; another was a patient with a nitroglycerin-associated bradyasystolic cardiac arrest who responded to CPR and had totally normal cardiac stress test, troponins, echo) and 2 were possibly iatrogenic (one was MI post-cardiac cath; another with acute MI while doing the stress test in the hospital)

    --secondary outcome (includes non-STEMI events not resulting in CRACE)

            --62 of 11230 patients had final diagnosis of possible or definite MI (0.55%, 0.42%-0.71%)

            --decreased to 28 patients, if exclude those with abnormal initial vital signs, or nonischemic ECG findings

    ​        ​--none of these 28 patients developed a CRACE, though  26 had a cath at least one day after admission, 18 had PCI, 4 CABG, and 4 were medically managed.

as we know, chest pain is quite common: 7 million ED visits annually for chest pain in US in 2010, 5.4% of the total ED visits. In 2006, US charges for admission for nonspecific chest pain was $11 billion. And most patients are discharged with a non-cardiac diagnosis. This study was somewhat limited by looking at only a 5% sample of the 11230 patients. Also, they did not include patients with non-STEMIs in the primary analysis, which they justified by the data suggesting that those with non-STEMIs do not benefit from aggressive invasive intervention.

so, what are the implications of this observational study? the most obvious one is that admitting these low-risk patients (normal vital signs, no STEMI, negative troponins x2) revealed that there were very few CRACEs, the most important clinical outcomes, leading to their assessment that the maximum hospitalization benefit was on the order of 1 in 1817 patients (if you add in all of these 4 cases, 2 of whom had noncardiac major events). and this number is pretty dramatically eclipsed by the current estimate of 1 in 164 patients who have hospitalization-associate preventable adverse events that contribute to their death, and with serious harm being 10- to 20-fold more common. i agree with their conclusion: many of these patients who go to the ER with chest pain are sick, at high risk of potentially bad outcomes, and need very close followup. but they may be better served by aggressive outpatient than inpatient care...