The Management of Chest Pain in an Acute Care Setting an Application of ePhysiciansOffice.com
The health care industry has wrestled with unnecessary hospital admissions for chest pain, accounting for greater than 5 billion dollars per year. In addition, between 2 and 8 percent of patients with myocardial infarctions are released from the emergency department yearly. Furthermore, an estimated 30 percent of malpractice money awards against emergency physicians are related to misdiagnosis and mistreatment of acute coronary syndromes.
ePhysicians Office, Inc., has developed a real-time interactive predictive tool for management of indeterminate chest pain in the emergency department. The tool is a Bayesian-based information technology system that provides the user the capacity to create, via a patented web-enabled 'e' infrastructure, a customizable self-accruing evidence-based template. Side by side multivariable analysis of all etiologies of chest pain (cardiac ischemia, cardiac nonischemia, and non-cardiac) simultaneously in an acute care setting is now possible with results depicted as ranked probabilities.
The chest pain application offers web-based technology support for triage purposes in an urgent care setting, with objectives as follows:
To reduce the number of patients admitted to rule out myocardial infarction with resultant significant savings in operating expenses to a facility.
To decrease the number of patients with acute coronary ischemia who are mistakenly diagnosed.
To expedite the decision-making process using a standardized database with comprehensive clinical application, leading to 'best practice
Because this application can be integrated into standard emergency workflow processes, including ACC/AHA guidelines, transaction with physicians and other emergency personnel is facilitated. Y et the program is compatible with numerous information technology platforms, including handheld devices, allowing for a multitiered approach to clinical information management.
A prospective double-blind study was devised to database patients in real time at the ED bedside. The patented template was functionalized, utilizing Bayesian probability techniques, from data automatically configured from the study population itself, and hence was purely evidence based. Each patient was analyzed against the customized software.
The data in this study revealed 100% sensitivity/100% negative predictive value for diagnosis of ischemia. All patients in 5 out of 6 diagnostic categories were diagnosed in first place. Only 8 percent of patients were unnecessarily hospitalized. These results, when extrapolated to just a region-wide application would save hundreds of unnecessary admissions and millions of dollars per year. Service, outcome and cost are all greatly impacted.
The patented predictive tool and associated proprietary knowledge and 'e' infrastructure has several advantages over existing methodologies. It allows complete scalability and data mining capability for use with numerous disease management and education applications.
Other benefits to use of this technology:
- Increases the probability of a positive radionuclide study (or coronary computed tomography angiogram study) in groups undergoing testing while still in ED.
- Decreases the number of patients admitted for observation.
- Clearly establishes a greater severity index for those patients eventually admitted as inpatients, thereby reducing the potential for denial of payments.
- Facilitates appropriate outpatient workups, thus reducing extraneous non-reimbursable inpatient workup (upper endoscopy, psychiatric, bone scans, etc.)
- Improves hospital care mix by reducing the number of patients admitted for chest pain, a diagnostic related group (DRG) that is poorly reimbursed.
- Automatic assembly of data for immediate interpretation. A variation/fluctuation over time in the presentation of a certain diagnosis and in the weighted contribution of a specific sign/symptom is easily detectable by real time monitoring.
- Frees up beds for better ED and inpatient flow, resulting in reduced need for ED closure to ambulances.
The academic world is looking for this type of technical innovation. A recent position paper in JAMA states, 'the identification of a group at low risk of acute MI for short-term mortality and morbidity and reproducible identification of that group within a nonexperimental framework remains an important area of future research.'
Finally, the practical view from the inside: In an article in Health Care Informatics, disease management via predictive modeling is discussed. 'If we can speak to a member right after discharge from a heart attack, we have a much better success rate than when that time is increased by as little as a couple of weeks.' The product/service available at ePhysicians Office, Inc., pushes the envelope right up to 'real time.' Hence, its value to patient, provider and payer is without peer.
A Typical Example of ePhysicians Office Inc.'s Chest Pain Application
A hospital system has evaluated its medical and financial management of patients requiring ER evaluation of chest pain over the previous calendar year. In agreement with published nationally derived data, the evaluation indicates that close to 70% of subsequent hospitalizations were unnecessary, and 'acute MI' or 'unstable angina' was confirmed in no more than 40%. These patients accounted for millions of dollars in unnecessary expenditures, thereby threatening the viability of several risk contracts.
The physicians and hospitals involved have decided to adopt a decision-making tool as an adjunct to typical decision making processes, that accomplishes several objectives:
To reduce the number of patients admitted to rule out myocardial infarction with resultant significant savings in operating expenses to a facility.
To decrease the number of patients with acute coronary ischemia who are mistakenly diagnosed.
To expedite the decision-making process using a standardized database with comprehensive clinical application, leading to 'best practice
To assure the public, patients, and payors that prompt evaluation and effective care for noncardiac diagnoses are maintained.
To provide for data collection in a format that is useful for education/Quality assurance and policy making.
Creating a Database
All patients entering the emergency room with a chief complaint of chest pain are first evaluated with respect to vital signs, general appearance, EKG changes, and cardiac enzymes elevations. These initial questions will identify certain patients that require immediate treatment and disposition. Such patients are not deemed appropriate for inclusion in the database. These questions eventually will be utilized as 'early triage' prior to utilization of the predictive modeling tool.
A series of consecutive patients with acute chest pain are chosen from the remaining group, who now fit into the specific tier of ACC/AHA guidelines termed 'indefinite/indeterminate for ischemia' as determined by the ER.
For each patient, the available ER chest pain algorithm, containing a series of clinical questions, is completed and recorded, along with diagnosis/management recommendations.
The questions in the algorithm are assembled in standard clinical format: HPI, PMH, meds, FH, SH, ROS, PE, labs, and CXR.
Potential diagnostic outcomes of the decision tool include the 29 most common diagnoses of patients presenting with chest pain in an ER setting, according to the literature.
Assembling the Tool Online, By the User, 'In the field'
Data from a requisite number of patients may be entered onto the HIPAA-compliant web site without the need for IT input. Once completed, the 'tool', which is essentially a web-enabled neural network, is created. It is now ready for use.
Results/Capabilities and Functionality
The initial questions are to be answered by physicians prior to entry into the predictive modeling tool (so called 'early triage'). Access to the algorithm is denied to patients experiencing an acute cardiac event or the critically ill.
If early triage of the patient suggests that he falls into the 'indeterminate for ischemia' category, he would be considered a candidate for the 'tool.' Utilizing the program, ancillary personnel can now determine which diagnosis/management scheme is appropriate for a patient with a specific set of signs and symptoms.
The tool's questions can be altered, and supportive data added or subtracted, whenever desired, with immediate updating of the program's functionality.
Conclusion
The described chest pain tool:
- Is incorporated within standard protocols.
- Classifies differential diagnoses not simply as “present” or “absent” but as computed probabilities.
- Can utilize the probabilities to assist in evaluating the indications for more definitive intervention (admit, consultation, radionuclide study, cardiac CT scan, etc.)
Physicians make the decision to admit, discharge or test further.
- Gives disease probability thresholds that prompt recommendations which can be customized according to perspectives of providers, IDN or payors.
- Offers additional applications including EMR incorporation, quality assurance and education.