There are over six million visits annually to emergency rooms for chest pain that generate nearly 1.7 million hospital admissions for acute coronary syndromes (ACS).1,2 In many teaching hospitals, patients with suspected ACS are admitted to services staffed by residents under the supervision of faculty. The program requirements set forth for residency training in internal medicine by the Accreditation Council for Graduate Medical Education residency set limits on the number of admissions that an individual resident can accept in a 24-hour period as well as the total number of patients cared for by an individual resident.3 Given the large volume of admissions for chest pain, these limits can be met quickly. Once met, programs and hospitals must come up with alternative ways to provide care for additional patients. Also, a large proportion of residents’ services may consist of chest pain admissions at the expense of more complex admissions, thereby limiting the diversity of the educational experience.
Two strategies have emerged to address the issues created by the large volume of chest pain admissions. The first involves the development of chest pain observation units that are designed to briefly “rule out” patients for myocardial infarction (with serial cardiac biomarkers and ECGs), often obviating the need for hospital admission.4 The second strategy involves the development of clinical algorithms to identify higher-risk patients in need of coronary care unit resources and to distinguish them from lower-risk patients who do not require intensive monitoring.5 Although this strategy has assisted with allocating resources, it has not helped reduce the volume of chest pain patients admitted to resident-based services.
Our hospitalist group wanted to develop a new strategy to reduce the number of chest pain admissions to resident-based services and to promote more efficient care for these patients. The strategy involved determining the thrombolysis in myocardial infarction (TIMI) risk scores of patients presenting to the emergency department (ED) with chest pain and then admitting low-risk patients (TIMI score 0 or 1) to a nonteaching service staffed by nurse practitioners who were supervised by hospitalists. The TIMI risk score is a prognostic score that has been validated as being predictive of the risk of adverse cardiac outcomes (death, reinfarction, or recurrent severe ischemia requiring revascularization) within 14 days of presentation for patients with unstable angina or non-ST segment elevation myocardial infarction (NSTEMI).6 (See List 1.)
Although the score was developed from clinical trials enrolling patients with NSTEMI/ACS, we were interested in evaluating its performance as a risk stratification tool in an unselected population of patients with chest pain. The TIMI risk score has recently been shown to be predictive of 30-day outcomes in an unselected ED chest pain population.7 We anticipated that the nonteaching service patients would have a reduced length of stay because the nurse practitioners and hospitalists on this service are not required to participate in an educational program and can therefore evaluate patients in a more timely manner.
We carried out this study from September 2003 through June 2004 at the Hospital of the University of Pennsylvania (HUP), an urban 700-bed teaching hospital. The institutional review board at the university approved the study. From July 2002 to June 2003, the hospitalist group had a total of 5,056 admissions, of which 1,371 (27%) were for patients with an admitting diagnosis of chest pain. HUP is a closed-staff hospital, which means that patients without HUP-based providers are admitted to a hospitalist service rather than a private attending service. Each hospitalist supervises and conducts daily teaching rounds for a team composed of one internal medicine resident, two interns, and one medical student.
In July 2003, a nonteaching service was created for patients with chest pain or suspected ACS admitted from the ED. This service required the hiring of three nurse practitioners, who staffed the nonteaching service from 7 am until 6 pm Monday through Thursday and from 7 am until 1 pm on Friday. That service was covered nights and weekends by housestaff but accepted new admissions only during the times when nurse practitioners were present. An adequate volume of chest pain patients for the nonteaching service was calculated and predicted based upon historical data of the TIMI risk scores of patients presenting to our ED with chest pain (Figure 1). Patients presenting to the ED with chest pain during weekday, daytime hours were triaged using the TIMI risk score. Patients with TIMI risk scores of 0 or 1 who had not used cocaine in the past week were considered to be at low risk for ACS and eligible for admission to the nonteaching service. Patients with scores of 2 or higher were considered to be at higher risk for coronary artery disease (CAD) and were admitted to a resident-based service. Low-risk patients admitted after 6 pm were admitted to the resident-based service and transferred to the nonteaching service the following morning. A total of 318 low-risk patients were included in our study: 113 on the nonteaching service and 205 on the resident-based service.
Chest pain patients were evaluated by an ED resident, ED attending, hospitalist nurse practitioner, and a hospitalist attending physician. Admission decisions were made by the ED attending. Patient care was then transferred to either the resident-based service or the nonteaching service. Patients on both services were evaluated by a hospitalist attending prior to discharge or further diagnostic testing. Clinicians were educated on the eligibility criteria outlined above. Any further clinical evaluation (e.g., stress testing, cardiology consultation) was left to the discretion of the hospitalist. There was no stress testing protocol as a part of this intervention. Depending on bed availability, patients were either discharged from a medical floor or directly from the ED.
To determine differences between the types of patients in the nonteaching service cohort and the resident-based service cohort, the chi-square test and the Fisher exact test (for 2 × 2 tables) were used for categorical data and the Student t test was used for continuous data. To assess differences between groups for the main outcomes of length of stay and hospital charges, the Wilcoxon rank sums test was used. All analyses were performed using standard statistical software. Data are presented as frequencies and percents for categorical data and as medians with interquartile ranges for length of stay and cost.
For our study, a total of 318 low-risk patients were triaged: 113 to the nonteaching service and 205 to the resident-based service (the control group). Baseline patient characteristics are shown in Table 1. The average age of the members of our study population was 50 years. There were 196 (62%) women and 222 (70%) black patients in the population. There were significantly more white patients in the control group (p = .009). Fewer than 1% of patients in each group reported a prior history of coronary artery disease or myocardial infarction.
The two cohorts were similar with respect to the following medical history items: incidence of hypertension, diabetes mellitus, tobacco use, prior CAD, recent cocaine use, and prior episodes of chest pain. The patients in the nonteaching service cohort had a higher incidence of hypercholesterolemia and family history of premature CAD. Those in the resident-based service cohort had a higher frequency of prior congestive heart failure. Clinical events during hospitalization were also similar between groups. Fifty-three (26%) and 23 (20%) of the patients in the resident-based service and nonteaching service groups, respectively, received an inpatient stress test, of which two (1%) and three (2.7%) patients, respectively, had stress tests interpreted as positive for CAD. A total of ten patients (3.1%) received a cardiac catheterization during the index hospitalization, six (2.9%) in the resident-based service versus four (3.5%) in the nonteaching service. Three (1%) patients received percutaneous cardiac interventions. Two (1.8%) patients in the nonteaching service group received a final diagnosis of ACS.
There was a significant difference in the overall utilization of hospital resources between the two services. Median length of stay was significantly lower on the nonteaching service (23 hours versus 32 hours; p < .0001), as were total hospital charges ($8,545 versus $14,133; p < .0001). Much of the difference in hospital charges was accounted for by bed charges and pharmacy charges, both of which are heavily affected by length of stay. The number of direct discharges from the ED was greater in the nonteaching service group than in the resident-based service group: 35 (31.0%) versus 25 (12.2%); p< .001.
Discussion and Conclusion
Our hospitalist-directed nonteaching service was successful in reducing the number of suspected ACS admissions to resident-based services by about one third. Also, we demonstrated a reduction in length of stay and hospital charges in an academic medical center that did not have a chest pain observation unit. The reduction in length of stay on the nonteaching service was likely due to the increased availability of both the hospitalists and the nurse practitioners on the nonteaching service. Although time to patient evaluation was not measured, the hospitalist attendings on the nonteaching service evaluated patients more quickly than did the hospitalist attendings on the resident-based service. There were several reasons for this: a lack of teaching responsibilities on the nonteaching service; timely patient evaluation by the nurse practitioners in the ED; and a reduced volume and complexity of admissions on the nonteaching service. In many instances, length-of-stay reductions were accomplished through direct patient discharges from the ED. These were rare occurrences prior to the creation of the nonteaching service, primarily due to delays in patient evaluation by residents.
We recognize that a formal evaluation of resident and attending satisfaction with the nonteaching service would have provided valuable information to assist academic medical centers in their decisions to adopt similar programs. However, the informal feedback that we received from our ED and medicine house officers about the effects of the nonteaching service on education and patient flow was positive. A similar experience with a nonphysician practitioner cardiovascular service at another academic medical center (with a nurse practitioner inpatient care model for a cohort of cardiac inpatients) demonstrated increased resident and attending satisfaction and improved efficiency.8 There was a perceived improvement in the quality of the interns’ and residents’ educational experiences as a result of the nonteaching service through decreased exposure to low-risk chest pain admissions and increased concentration on more clinically complex admissions.
Although our study is not a randomized controlled trial, our analyses support its internal validity. Patient demographics, historical characteristics, and TIMI risk scores were similar in both study groups. Despite the fact that there was a higher proportion of patients in the nonteaching service group with a history of hypercholesterolemia and a family history of premature CAD, both of which could have increased the risk of CAD/ACS or increased the proportion of patients requiring inpatient cardiac evaluations, reductions in length of stay were still achieved. There was a significant racial difference between the two groups, with a larger proportion of white patients in the resident-based service cohort. It is possible that because admission to the nonteaching service was encouraged but not enforced, physicians could have preferentially triaged patients with low TIMI risk scores to the resident-based service if their clinical suspicion for ACS was higher in an individual patient based on race. Racial bias in the evaluation of cardiac syndromes has been previously described.9,10
There are several limitations to this study. First, our generalizability is limited given that not all academic hospitals have hospitalists and nurse practitioners to create a nonteaching service devoted to chest pain. Second, the ED staff was familiar with the TIMI risk score from data collection for prior studies performed at our institution. This may have facilitated the success of the risk score for triage and translated into greater overall service efficiency. Third, we did not have independent observers calculating the TIMI risk score so the risk scores utilized by the nonteaching service may have been inaccurate, potentially resulting in nonuniform admission practices. However, such errors would not have changed the quality of care that the patients received because all chest pain patients included in this study were admitted to an internal medicine team led by a hospitalist attending.
This study demonstrates that nonteaching service models of care for a low risk group of patients can reduce the number of admissions to resident-based services thereby enabling better compliance with the internal medicine residency program requirements. We also believe that this study adds to the growing body of literature surrounding the utility of the TIMI risk score by demonstrating that TIMI risk stratification is an accurate way to identify low-risk chest pain patients who can be targeted to nonteaching services. Future research should assess the impact of nonteaching services on residency education and the role of nurse practitioners and hospitalists in helping residency programs meet their program requirements.
The authors acknowledge the Penn Hospital Care Physicians and Nurse Practitioners for their contributions to service excellence and residency education.
1 Hollander JE. Risk stratification of emergency department patients with chest pain: the need for standardized reporting guidelines. Ann Emerg Med. 2004;43:68–70.
2 Reilly BM, Evans AT, Schaider JJ. Impact of a clinical decision rule on hospital triage of patient with suspected acute cardiac ischemia in the emergency department. JAMA. 2002;288:342–50.
4 Gaspoz JM, Lee TH, Cook EF, et al. Outcome of patients who were admitted to a new short-stay unit to “rule-out” myocardial infarction. Am J Cardiol. 1991;68:145–49.
5 Goldman L, Cook EF, Johnson PA, Brand DA, Rouan GW, Lee TH. Prediction of the need for intensive care in patients who come to the emergency department with acute chest pain. N Engl J Med. 1996;334:1498–504.
6 Antman EM, Cohen M, Bernink PJ, et al. The TIMI risk score for unstable angina/non-ST elevation MI. JAMA. 2000;284:835–42.
7 Pollack CV, Jr., Sites FD, Shofer FS, Sease KL, Hollander JE. Application of the TIMI Risk Score for unstable angina/non-ST elevation acute coronary syndrome to an unselected emergency department chest pain population. Acad Emerg Med. 2006;13:13–18.
8 Nishimura RA, Linderbaum JA, Naessens JM, Spurrier B, Koch MB, Gaines KA. A nonresident cardiovascular inpatient service improves residents’ experiences in an academic medical center: a new model to meet the challenges of the new millennium. Acad Med. 2004;79:426–31.
9 Venkat A, Hoekstra J, Lindsell C, et al. The impact of race on the acute management of chest pain. Acad Emerg Med. 2003;10:1199–208.
10 Ferguson JA, Tierney WM, Westmoreland GR, et al. Examination of racial differences in the management of cardiovascular diseases. J Am Coll Cardiol. 1997;7:1707–13.