The Impact of Accelerated Diagnostic Protocol Implementation on Chest Pain Observation Unit Utilization

Abstract

Background:

Data evaluating the impact of the history, ECG, age, risk factors, and troponin (HEART) Pathway on observation unit (OU) use is limited. The objective of this study is to determine how HEART Pathway implementation affects OU use.

Methods:

An analysis of OU registry data from October 2012 to October 2016, 2 years before and after HEART Pathway implementation at an academic medical center, was conducted. Adult patients placed in the OU for chest pain were included. The proportion of patients placed in the OU chest pain protocol per total OU volume and hospitalization and myocardial infarction (MI) rates were determined. Proportions before versus after implementation were compared using χ² tests and age was compared using a Mann-Whitney U test.

Results:

During the study period, 1688 patients with chest pain before HEART Pathway implementation and 1692 after were included. The proportion of chest pain patients in the OU per total OU volume decreased following implementation from (57% [1688/2968] to 43.6% [1692/3882]; P < 0.001). Before HEART Pathway implementation, the hospitalization rate was 10.4% (175/1688) versus 12.4% (210/1692) after (P = 0.07). More patients were diagnosed with MI following implementation (0.8% [14/1665] vs. 2.0% [33/1686]; P = 0. 008). Median age was older postimplementation (52 years [IQR: 45–59 years] vs. 54 years [IQR: 48–64 years]; P < 0. 001).

Conclusions:

HEART Pathway implementation resulted in management of higher risk patients in the OU. Following implementation, OU chest pain patients were older and were more likely to be hospitalized or diagnosed with MI.

Chest pain is the second most common reason people visit the Emergency Department (ED) and accounts for more than 8 million visits annually.¹ The history, ECG, age, risk factors, and serial troponins (HEART) Pathway is a well-validated accelerated diagnostic protocol (ADP) designed to risk stratify ED patients with chest pain concerning for acute coronary syndrome (ACS).^2-4 Low-risk patients with negative cardiac biomarkers are candidates for early ED discharge while non low–risk patients with negative initial cardiac biomarkers are often admitted to observation units (OU) for further risk stratification with objective cardiac testing (OCT). The evaluation of patients with non low–risk chest pain is the most common patient complaint cared for in OU in the United States, and OU-based care has been found to significantly reduce patient length of stay (LOS) compared with traditional hospital observation admission.^5,6

The impact of the implementation of a well-validated ADP such as the HEART Pathway on OU volumes, demographics/risk, and outcomes is limited. Thus, the primary objective of this study is to determine how HEART Pathway implementation affects OU utilization in patients that present to the ED with symptoms concerning for ACS. We hypothesize that HEART Pathway implementation will decrease the relative proportion of chest pain patients cared for in the OU but will increase the diagnostic value of an observation stay—specifically, there will be an increase in the percentage of patients requiring admission and that are diagnosed with myocardial infarction (MI).

METHODS

Study Design and Setting

Observation unit registry data were analyzed from 2 years before (October 2012–November 2014) and after (November 2014–October 2016) HEART Pathway clinical decision support (CDS) implementation at an academic medical center. This study was approved by the Biomedical Institutional Review Board of Wake Forest University with a waiver of informed consent and registered with clinicaltrials.gov (NCT02056964). Methods of the HEART Pathway Implementation study, a prospective prepost interrupted time series, were previously described.² This study included a single OU located in the Wake Forest Baptist Medical Center Emergency Department that has approximately 114,000 visits annually.

Selection of Participants

All adult ED patients (≥21 years old) placed in the OU chest pain protocol for OCT were included in this analysis. These patients were required to be chest pain free, have a stable nondiagnostic ECG, and a contemporary troponin less than the 99th URL. In addition, they could not have imaging contraindications, be hemodynamically unstable or have history of CAD, significant left ventricular dysfunction, active heart failure, or significant valvular heart disease. Finally, patients needed to be free of other significant, uncontrolled medical, surgical, or psychiatric complaints. Index encounter data (from initial ED presentation through OU stay) were extracted from the health system’s EHR data (Clarity-Epic Systems Corporation, Verona, WI). Prevalidated structured EHR variables and diagnoses (ICD-9 and ICD-10) were used to obtain patient demographics, past medical history, cardiovascular risk factors, comorbidities, and discharge diagnoses (including MI).²

HEART Pathway Intervention

Before HEART Pathway CDS implementation, clinicians would utilize gestalt, ECG, biomarker testing, and history of CAD to determine if a patient was suitable for further evaluation in the ED OU for serial biomarkers and OCT.⁷ HEART Pathway implementation consisted of fully integrating the chest pain CDS tool within a clinician’s workflow in the EHR. Providers were automatically prompted by a best practice alert to utilize the CDS if a patient had a chief complaint of chest pain and a cardiac troponin (cTn) was ordered. The HEART Pathway CDS could also be manually entered by a clinician for any patient with a presentation concerning for ACS. Providers were educated on the HEART Pathway CDS tool through in person and online modules. Further, provider use of the CDS tool was monitored and providers were given feedback on their compliance in using the tool. Due to this education and the feedback mechanisms in place, a run-in phase after the implementation of the tool was not performed.

The HEART Pathway CDS is a validated ADP designed to risk stratify ED patients with chest pain concerning for ACS (Appendix 1). Following the completion of the HEART Pathway algorithm, a numerical score and risk level with recommendations were assigned to the patient. Patients with a HEAR score of 0–3 were deemed low risk, and a recommendation for discharge from the ED was made if serial zero and 3-hour cTn were both less than the 99th URL (referred to as “negative”). Non low–risk patients with HEAR scores greater than 3 and a negative cTn were eligible for the chest pain protocol and further evaluation in the OU. Patients with elevated cTn were admitted to the hospital for further evaluation and treatment, as were patients that developed hemodynamic instability, ECG changes, or had evidence of cardiac ischemia on OCT.

Outcomes

The proportion of patients placed in the OU chest pain protocol per total OU volume is the primary outcome for this analysis. Secondary outcomes included hospital admission rate as defined as an inpatient admission, transfer or observation stay (outside of the ED OU), and diagnosis of index MI defined by diagnosis codes validated by prior cardiovascular trials.^8-12 Patient median age and the percentage of patients with hyperlipidemia, diabetes, smoking history, and hypertension placed in the OU before and after HEART Pathway implementation were determined from health records.

Statistical Analysis

Descriptive statistics were used to characterize the pre- and postimplementation cohorts using median and interquartile range for continuous variables and percentages with 95% confidence intervals (95% CI) for categorical data. Age was treated as a continuous variable, while sex, race/ethnicity, hyperlipidemia, diabetes, smoking history, and hypertension were treated as categorical variables. Proportions were compared using χ² and Mann-Whitney U test was used to compare nonparametric continuous data. An alpha of 0.05 was held as representing statistical significance, and no adjustments were made for multiple comparisons. Formal power calculations were not performed as the sample size was fixed by the duration of the study period. Stata 11.2 (College Station, TX) was used for statistical analyses.

RESULTS

During the study period, a total of 3380 patients with suspected ACS were included. The preimplementation cohort had 1688 patients, while 1692 patients were in the postimplementation cohort. Patient characteristics for each cohort are provided in Table 1. Age and MI status were available on 3351 patients. Age and MI status was not accounted for on 3 patients pre HEART Pathway implementation and 6 patients post HEART Pathway implementation. The total observation patient census in the preimplementation period was 2968, while the number of observation patients in the same protocols increased to 3882 in the postimplementation phase. The proportion of chest pain patients in the OU per total OU volume (using only protocols present in the preimplementation period) decreased pre- and postimplementation with 1688/2968 (56.9%, 95% CI, 55.1%-59.7%) and 1692/3882 (43.6%, 95% CI, 42.0%-45.2%), respectively (P < 0. 001).

TABLE 1.

Prepost HEART Pathway OU Chest Pain Patient Characteristics^*

Patient Characteristic	Precohort n = 1688 (%, 95% CI)	Postcohort n = 1692 (%, 95% CI)
Age (IQR)	52 (45–59)	54 (47–64)
Female	1025 (60.7%, 58.4-63.1)	1001 (59.2%, 56.8-61.5)
Race
White or Caucasian	982 (58.2%, 55.8-60.5)	991 (58.6%, 56.2–60.9)
Black or African	612 (36.3%, 34.0-38.6)	602 (35.6%, 33.3-37.9)
American
Other	91 (5.4%, 4.4-6.6)	98 (5.8%, 4.7-7.0)
Ethnicity
Hispanic or Latino	76 (4.5%, 3.6-5.6)	69 (4.1%, 3.2-5.1)
Risk factor
Hypertension	1104 (65.4% 63.1-67.7)	1244 (73.5%, 71.4-75.6)
Smoking	589 (34.9%, 32.6-37.2)	592 (35.0%, 32.7-37.3)
BMI
Hyperlipidemia	773 (45.8%, 43.4-48.2)	958 (56.6%, 54.3-48.2)
Diabetes	529 (31.3%, 29.1-33.6)	585 (34.6%, 32.3-36.8)
Admission rate	175 (10.4%, 8.9-11.8)	210 (12.4%, 10.8-14.0)

^*Continuous variables are expressed with medians and interquartile ranges, and categorical variables are presented as counts, percentages, and 95% confidence intervals. Race data were missing in 2 cases and ethnicity in 1, and 1 refusal to answer in the precohort. There was 1 refusal to answer and no missing race or ethnicity data in the postcohort.BMI indicates body mass index; CI, confidence intervals; IQR, interquartile range.

Before HEART Pathway implementation the admission rate from the OU for patients with chest pain was 10.4% (175/1688, 95% CI, 9.0%-11.9%) versus 12.4% (210/1692, 95% CI, 10.9%-14.1%) after implementation (P = 0. 07). More patients were diagnosed with MI during the index presentation following implementation; with 0.8% (14/1665, 95% CI, 0.5%-1.4%) with MI preimplementation compared with 2.0% (33/1686, 95% CI, 1.4%-2.7%) postimplementation (P = 0. 008). Patients in the postimplementation cohort were older (54, 95% CI, 47%-64%) and higher rates of hypertension (73%, 95% CI, 71.4%-75.6%) and hyperlipidemia (56.6%, 95% CI, 54.3-48.2%).

DISCUSSION

Following implementation of the HEART Pathway ADP, the proportion of chest pain patients evaluated in the OU decreased, and the patients managed in the OU were older, more likely to have comorbidities, and were more likely to be diagnosed with MI. The HEART Pathway has been shown to increase early discharge of low-risk chest pain patients that historically would have been cared for under the chest pain observation protocol. However, even with a reduction in the proportion of chest pain patients in the postimplementation cohort, the chest pain protocol remained the top protocol managed in the OU. In fact, the absolute number of patients with chest pain who were cared for in the OU was similar before and after implementation. This suggests that the main impact of the HEART Pathway ADP was a shift towards managing patients at higher risk for ACS in the OU.

Overall, the OU dramatically increased the total number of patients under the same protocols in the OU post HEART Pathway implementation, from 2968 to 3882. While the cause of this is multifactorial, OU chest pain patients are resource intense and tend to have longer stays relative to other OU protocols. Patients admitted to the OU under chest pain protocols in the morning for OCT have been shown to have prolonged stays compared with those admitted at night.⁶ A reduction in the chest pain demographic could allow OUs to focus on protocols that are less resource and time intensive, enabling more total patients to be treated under observation care. Unfortunately, metrics on overall LOS for OU patients in our study were not available.

The proportion of patients with coronary artery disease (CAD) risk factors including age, hypertension, and hyperlipidemia increased in the postimplementation cohort. The risk profile of patients in the OU post HEART Pathway implementation cohort was higher with a relatively unchanged total number of patients. It may be presumed that before explicit risk stratification with the HEART pathway, a patient with a substantial number of risk factors would have been admitted to the hospital due to “gestalt, versus evaluated in the OU.¹³” In addition, the number of ED OU patients diagnosed with index MI increased significantly (0.8% vs. 2.0%). This suggests ED providers were more comfortable taking care of higher risk patients in the OU, potentially preventing these patients from being admitted.

With chest pain being the most common cause of observation and short inpatient stays across US hospitals, this analysis has significant implications for how hospital systems across the United States manage their chest pain patients. In-hospital admission and observation LOSs have been shown to be longer than what is seen in an ED-based OU.¹⁴ Some studies have shown chest pain consists of more than 10% of all observation stays and yields negative margins per stay.¹⁵ However, Sheehy et al’s study was performed in a type IV observation setting (with no geographic cohorting and no explicit protocols), which is the least efficient way of managing observation care. By reducing the proportion of chest pain patients in the ED OU and shifting in-hospital chest pain patients to the ED OU, significant operation improvements could be made to the observation stay patient population.

LIMITATIONS

Our study design has limitations. This study was done at 1 large academic center, and results may not be generalization to other health systems. However, our ED OU is not managed by resident trainees, but by advanced practice providers, consistent with nonacademic centers who utilize ED-based OUs. Data were drawn from the electronic health record and depended on the accuracy of nursing and provider input. The diagnosis of MI was based on ICD-9 and ICD-10 codes. We were not able to distinguish between type-1 and type-2 MI in our analysis. While providers received recommendation on patient disposition post HEART Pathway implementation, they were not forced to follow the recommendation and could still use the chest pain protocol for patients with HEAR scores of 0–3. Before HEART Pathway implementation, the tool was prospectively validated at our academic center.⁵ This validation period could have caused providers in our preimplantation cohort to use the HEART Pathway in the care of their patients. The registry data used in this study excluded patients who did not have an explicit order for observation care. This could have led to a discrepancy of missed index MI cases in our data since a positive troponin in the ED in a patient with a plan for observation, but who ruled in before the initiation of observation, could have led to the termination of a serial troponin order and excluded the patient.

CONCLUSIONS

HEART Pathway implementation reduced the proportional use of the OU for acute chest pain. Following HEART Pathway implementation OU chest pain patients were older, more likely to be diagnosed with MI, and more likely to have comorbidities associated with CAD. These findings suggest HEART Pathway implementation results in the management of higher risk patients in the OU and improve the diagnostic yield for MI. HEART Pathway implementation at this study site showed an improved overall OU efficiency in the ability to increase the usage of non-chest pain-related protocols. Downstream effects of the HEART Pathway’s impact on the OU may be significant from an operational, financial, and clinical standpoint.

APPENDIX 1. History Electrocardiogram Age Risk Factor Troponin Pathway

History
High-risk features	Low-risk features
Middle- or left-sided	Well localized
Heavy/tight/pressure chest pain	Sharp pain
Diaphoresis	Nonexertional
Radiation	No diaphoresis
Nausea/vomiting	No nausea/vomiting
Exertional
Relief of symptoms with sublingual nitrates
Mostly high-risk features	2 points
Mixture of high and low-risk features	1 point
Mostly low-risk features	0 points
Electrocardiogram
nonspecific changes	1 point
Repolarization abnormalities
nonspecific T wave changes
nonspecific ST-segment depression/elevation
Bundle branch blocks
Pacemaker rhythms
Left ventricular hypertrophy
Early repolarization
Digoxin effect
Normal	0 points
Age
≤65 yr old	2 points
45–64 yr old	1 point
<45 yr old	0 points
Risk factors
Obesity (body mass index ≥30)
Current or recent (≤90 d) smoker
Currently treated diabetes mellitus
Family history of coronary artery disease (first degree relative <55 yr old)
Diagnosed and treated hypertension
Hypercholesterolemia
3+ risk factors listed above OR prior CVA OR peripheral arterial disease	2 points
1–2 risk factors	1 point
No risk factors	0 points
HEAR score (total points)	_________
HEART Pathway
Low risk	(i) No history of coronary artery disease (i) No new ischemia on electrocardiogram (ii) HEAR Score <4 (iii)) Negative troponin at 0 and 3 hours
Recommendation	→Patient is safe for discharge with outpatient follow up.
Non low-risk	(i) History of coronary artery disease (ii) New ischemia on electrocardiogram (iii) HEAR score ≥4 (iv) Negative troponin at 0 and 3 hrs
Recommendation	→Patient should receive further objective cardiac testing before discharge

HEAR indicates history electrocardiogram age risk factor; HEART, history electrocardiogram age risk factor troponin.