Change in Care Transition Practice for Patients with Non-Specific Chest Pain after ED Evaluation 2006–2012

Abstract

Objectives

From 2005–2010 healthcare financing shifts in the United States may have affected care transition practices for emergency department (ED) patients with non-specific chest pain (CP) after ED evaluation. Despite being less acutely ill than those with myocardial infarction, these patients’ management can be challenging. The risk of missing acute coronary syndrome is considerable enough to often warrant admission. Diagnostic advances and reimbursement limitations on the use of inpatient admission are encouraging the use of alternative ED care transition practices. In the setting of these health care changes, we hypothesized there is a decline in inpatient admission rates for patients with non-specific CP after ED evaluation.

Methods

We retrospectively used the Nationwide ED Sample (NEDS) to quantify total and annual inpatient hospital admission rates from 2006–2012 for patients with a final ED diagnosis of non-specific CP. We assessed the change in admission rates over time, and stratified by facility characteristics including: safety-net hospital status, US geographic region, urban/teaching status, trauma-level designation, and hospital funding status.

Results

The admission rate for all patients with a final ED diagnosis of non-specific CP declined from 19.2% in 2006 to 11.3% in 2012. Variability across regions was observed, while metropolitan teaching hospitals and trauma centers reflected lower admission rates.

Conclusion

There was a 41.1% decline in inpatient-hospital admission for patients with non-specific CP after ED evaluation. This reduction is temporally associated with national policy changes affecting reimbursement for inpatient admissions.

INTRODUCTION

Several healthcare financing changes impacting Untied States inpatient hospital admission reimbursement may have affected admission practices for emergency department (ED) patients without a clear cause for chest pain.¹ These patients often receive an extensive work-up, yet most are eventually discharged with non-cardiac diagnoses.⁴ Only 0.4–1.8% will have myocardial ischemia identified as the cause.^{1, 5}

The prevalence of non-specific CP visits is approximately 3.8% among ED patients.^{2, 4} Historically, a hospital admission was the reflexive approach to test for signs of myocardial infarction (via serial cardiac biomarkers) and primary or secondary signs of myocardial ischemia (via non-invasive testing). This approach extends the evaluation beyond a traditional ED visit into observation or admission.^{3, 5}

Patients with non-specific CP have the widest variability in admission rates (median 17.6% [interquartile ranges (IQRs) 10.7-30.1%]) in comparison to other high-volume diagnoses such as pneumonia, congestive heart failure (CHF), septicemia, and chronic obstructive pulmonary disease (COPD).⁴ Previous studies have not yet described the magnitude of the admission practice shift in this high-volume ED sub-population or how the shift may differ across EDs. We quantify changes in inpatient admission practices which may be temporally associated with national health financing changes between 2005 and 2010. We specifically examined the admission disposition of non-specific CP from 2006–2012. We hypothesized there would be a decline in inpatient admission rates.

METHODS

Study Design and Data Source

This was a multi-year descriptive retrospective cohort study from 2006 to 2012 using the Nationwide ED Sample (NEDS). NEDS is a publically-available database from the Healthcare Cost and Utilization Project (HCUP) sponsored by the US Agency for Healthcare Research and Quality. It is the largest all-payer ED database in the US, enables analyses of ED utilization, and supports clinician, policymaker and provider understanding of health resource allocation and utilization. NEDS includes approximately 950 hospitals and 31 million ED encounters per year from 30 states. Beginning in 2006 it is constructed annually from the HCUPS State Emergency Department Databases (SEDDs) and the State Inpatient Databases (SIDs). The sample represents 20% of US EDs, and sampling weights are used to estimate national ED utilization and care.⁶

Setting and Population

We included all ED visits for adult patients with a primary final ED diagnosis of “chest pain” (ICD-9 codes: 786.50-786.59). We excluded those with an explanatory secondary diagnosis (positions 2-5) that would make an ongoing evaluation for myocardial ischemia unnecessary. These included: myocardial infarction (410.00-410.92), pulmonary embolism (415.10, 415.11, 415.12, 415.13, 415.19), aortic dissection (441.00-441.9), myocarditis (429.00, 429.10), cardiac arrest (427.50), pneumonia (480.00-487.8), endocarditis (391.1, 391.2, 391.8, 391.9, 391.00), pericarditis (420.90-420.99, 423.80, 423.90) and chest wall injury (959.11).

Primary Outcome

Our primary outcome was inpatient admission status. Of note, NEDS does not record observation admissions. This challenged our ability to separate ED discharges from in-hospital observation, but permitted a binary determination of disposition as inpatient verses non-inpatient (including observation admissions and ED discharges).

ED Characteristics

To uncover ED-level variation associated with institutional differences, we analyzed visits by US geographic region, specialist intensity (represented as trauma-level designation), urbanicity and teaching status (combined in NEDS), and safety-net status. Safety-net EDs provide a disproportionate amount of care to low income, uninsured, and Medicaid patients. Based on prior definitions, these were defined as: >30% Medicaid, >30% self-pay (the uninsured), or the combined Medicaid and uninsured pool >40%.⁸ Safety-net status was calculated from NEDS patient data aggregated by ED facility.

Analysis

We quantified total and annual inpatient hospital admission rates from 2006-2012 for a final ED diagnosis of non-specific chest pain as a proportion of total visits with 95% confidence intervals accounting for the NEDS database design, including the sampling weights. An odds ratio (OR) was calculated for the likelihood of admission per year. We assessed the change in admission rates over time, and stratified the data by facility characteristics subgroups. All statistical analyses were performed using the SAS System for Linux, Version 9.4 (SAS Institute, Inc., Cary, NC).

RESULTS

The admission rate for patients with a final ED diagnosis of non-specific chest pain declined from 19.2% in 2006 to 11.3% in 2012, a relative reduction of 41%. This correspond to a 10.3% reduction in the odds of admission per year (OR=0.897, 95% CI=0.885, 0.910). Safety-net hospitals had higher admission rates than non-safety-net (16.6% vs 15.0%), with the most notable difference (17.0% vs. 15.0%) in 2009. A marked downward inflection was noted in overall admission rates over time, and for each facility-type subgroups, in 2007. In 2012 all regions, except the Northeast, had admission rates ranging 10.3–11.5%. The variability among regions was significantly lower, and with less variability, than observed in 2006 (range 17.3–21.5%). The Midwest and Northeast had the lowest and the highest rates, respectively, while the West demonstrated the greatest decline from 21.5% in 2006 to 10.6% in 2012. Non-urban EDs had admission rates close to the total study population mean for each year. For all years, metropolitan teaching hospitals had lower admission rates than metropolitan non-teaching institutions. Non-trauma centers had higher admission rates than trauma centers (Figure 1).

Figure 1.

Hospital Admission Rates for Patients with Non-Specific Chest Pain after ED Evaluation (2006–2012) by Trauma Level Designation¹

¹ “All” includes all patients with a final ED diagnosis of Non-specific chest pain.

DISCUSSION

The reduction in inpatient admission rate for patients with non-specific chest pain suggests increased use of ED, in-hospital observation, or out-of-hospital follow up for the continued evaluation of possible non-infarction myocardial ischemia. The marked 2007 downward inflection in admission rates may reflect a rippled practice shift from financing reforms initiated in 2005.

Increasing ED visits, ED overcrowding, and hospital consolidations have placed pressure on providers to find alternative care transitions for patients previously admitted. In 2013 the Centers for Medicare and Medicaid Services introduced the “Two Midnight Rule” where Medicare beneficiaries receiving reasonable and necessary hospital services beyond two midnights were defined as appropriate for inpatient admission. Early analyses suggest this new rule may further increase hospital visits classified as observation status.

From 2006 to 2012, observation visits rose dramatically. Among Medicare beneficiaries, the Medical Payment Advisory Commission (MedPAC) estimated outpatient observation stays increased by 88% during this period. This accounted for approximately 40% of the decline in short-stay inpatient admissions.⁷ Patients with CP are the most common reason (19.1%) for observation.⁴ A recent Medicare analysis reported a significant increase in CP related observation stays during our study period.⁹ More recent health policy changes are likely to sustain our observed decline in hospital admission rates.

While these pressures encourage a reduction in inpatient admission, it is difficult to know if observation increases are due to more frequent classification of observation status without a noticeable change in setting or model of care management, versus increased use of dedicated observation units using CP protocols. There is likely a significant contribution from both of these factors with some facility-level variation.¹⁰ Since NEDS does not differentiate between discharges and observation status, the relative contributions are unclear. It may also reflect less follow-up testing in the setting of more sensitive diagnostic strategies.

The Northeast had higher overall admission rates compared to other US geographic regions. This has been previously observed, however, the reasons are not well understood. Higher admission rates in safety-net hospitals may reflect reduced access to outpatient alternatives. Non-trauma centers had higher admission rates than trauma centers. One potential explanation for this difference is that level I centers have been found to experience more inpatient crowding. This places a higher premium on inpatient beds, and may encourage more intensive outpatient evaluations (including extended ED and observation), thus decreasing the admission rate over typically less specialist resourced non-trauma centers. This dynamic may also explain lower inpatient admission rates in metropolitan teaching facilities compared to non-teaching.

LIMITATIONS

These findings must be interpreted within the limitations of our study design and data source. This is a descriptive study limited to univariate analyses. The association between 2005–2010 reforms and a decline in inpatient admission from 2006 to 2012 is temporal. Given the data available we are limited to examining changes in care transition practice relative to inpatient admission.

Using NEDS precludes direct analysis of observation services. However, we leveraged the dichotomization of inpatient admission and non-inpatient associated ED visits (ED discharges or observation stays) to compare inpatient admission to other dispositions. Despite our careful selection of ICD-9 codes for inclusion and exclusion, the dataset does not permit us to directly validate diagnosis accuracy. NEDS is a de-identified administrative claims dataset which lacks clinical granularity. Using this national database, however, has elucidated trends difficult to study with primary data collection.

CONCLUSION

From 2006 to 2012 there was a 41.1% decline in inpatient-hospital admission for patients with non-specific chest pain after ED evaluation. This reduction is temporally associated with US health financing changes affecting the use of inpatient admission status.