Within the US, chest pain is the most common etiology for observation and short inpatient stays.  Patients are admitted to assess for the presence of serious pathology including acute coronary syndrome (ACS). To decrease short inpatient stays, institutions have transitioned to observational status and those patients are considered outpatients. The impact of this change on resource utilization and patient outcomes continues to be unclear.

Clinical Question

What is the effect of placing an adult chest pain patient in observation on healthcare utilization (angiography, PCI, rehospitalization) and on 30-day acute MI as compared to a short inpatient admission.


All patients > 18 years of age who presented to an ED with a primary diagnosis of acute chest pain based on ICD-9 codes who were placed in observation, admitted for < 2 days or discharged home. The data source was OptumLabs Data Warehouse, a database from privately insured and Medicare advantage enrollees throughout the US from Jan 1, 2010 and Dec 31, 2014


Healthcare utilization within 30 days (defined as invasive procedures – coronary angiography, PCI or CABG)
Hospitalization within 30 days
Acute Myocardial Infarction within 30 days


Retrospective observational study using propensity score matching.
– Observational studies are not randomized. Propensity score matching attempts to estimate the treatment’s effect by accounting for covariates that predict receiving treatment (in this case characteristic like gender, age, CAD, diabetes, hx of acute MI) and tries to reduce bias due to confounders


Patients with acute myocardial infarction during ED visit, patients who did not have health insurance for the 12 months prior to presentation or for at least 1 month after presentation

Primary Results

Used one-to-one propensity score without replacement in 3 models to construct 3 cohorts of matched patients – controlled for age, gender, race, comorbidity burden, and year of ED visit

  • 51,072 Obs vs. short inpatient stay
  • 235,129 Obs vs. discharged home
  • 51,072 Short inpatient vs. discharged home

Critical Results

  • AMI within 30 days
    • Observation vs. Admitted Short Inpatient
      • 0.23% vs. 0.21%
      • OR 1.09 (95% CI = 0.84 to 1.42)
      • No statistically significant difference
    • Observation vs. Discharged Home
      • 0.17% vs. 0.23%
      • OR 0.72 (95% CI = 0.71- 0.91)
    • Admitted Short Inpatient vs. Discharged Home
      • 0.21% vs. 0.36%
      • OR = 0.57 (95% CI = 0.45 to 0.73)
  • Health care utilization during index visit and within 30 days
    • Lower for patients placed in observation in comparison to short in-patient admission
    • Coronary angiography
      • 10.9% vs. 24.4%
      • OR 0.38 (95% CI = 0.36 to 0.39 )
    • PCI
      • 1.8% vs. 7.6%
      • OR 0.23 (95% CI 0.21-0.24)
    • No difference in rate of CABG
  • Patients placed in observation had lower all-cause 30-day hospitalization compared to short inpatient admission


  • Study asks a clinically relevant question that is patient centered
  • The sample size is large
  • Propensity matching was performed to attempt to control for confounders


  • Did not specify which outcome investigated was the primary outcome
  • Administrative claims susceptible to coding errors
  • History, EKG and troponin not available for matching
  • All patients were insured, not necessarily generalizable to our population to all populations
  • Observation class was a catch all term. We cannot differentiate between a general observation area versus a protocol driven unit
  • Retrospective analysis of database
    • Lacks randomization. Propensity matching used to help overcome this limitation
    • Bias could arise because the difference in an outcome could be due to the characteristics that led them to receive treatment (placement in obs vs. floor) and not the disposition itself
    • Unclear why certain patients admitted versus discharged versus placed in observation

Author's Conclusions

“There were higher rates of cardiac catheterization and PCI among those admitted as a short inpatient compared to observation, while the incidence of subsequent AMI within 30 days was similar”

Our Conclusions

We agree with the authors conclusions that in this study, short term admission was associated to higher advanced treatment utilization without improving outcomes. However, a randomized trial is necessary to demonstrate causality.

Bottom Line

Patients admitted to higher level of care received “more care” but not necessarily better care. The additional care seen in this cohort did not result in better patient outcomes.