Original ArticleFree PreviewArchive
A Predictive Instrument to Improve Coronary-Care-Unit Admission Practices in Acute Ischemic Heart Disease — A Prospective Multicenter Clinical Trial
List of authors.Abstract
Each year 1.5 million patients are admitted to coronary-care units (CCUs) for suspected acute ischemic heart disease; for half of these, the diagnosis is ultimately "ruled out." In this study, conducted in the emergency rooms of six New England hospitals ranging in type from urban teaching centers to rural nonteaching hospitals, we sought to develop a diagnostic aid to help emergency room physicians reduce the number of their CCU admissions of patients without acute cardiac ischemia. From data on 2801 patients, we developed a predictive instrument for use in a hand-held programmable calculator, which requires only 20 seconds to compute a patient's probability of having acute cardiac ischemia.
In a prospective trial that included 2320 patients in the six hospitals, physicians' diagnostic specificity for acute ischemia increased when the probability value determined by the instrument was made available to them. Rates of false-positive diagnosis decreased without any increase in rates of false-negative diagnosis. Among study patients with a final diagnosis of "not acute ischemia," the number of CCU admissions decreased 30 per cent, without any increase in missed diagnoses of ischemia. The proportion of CCU admissions that represented patients without acute ischemia dropped from 44 to 33 per cent.
Widespread use of this predictive instrument could reduce the number of CCU admissions in this country by more then 250,000 per year. (N Engl J Med 1984; 310: 1273–8.)
Appendix
The predictive instrument can be used with a number of programmable hand-held calculators. The formula and coefficients listed in the Statistical Methods section can be used to program the calculator, or a user may request step-by-step instructions from one of us (H.P.S.). To obtain a probability value, the user presses the 1 or 2 key to indicate, respectively, whether a given clinical feature is present or absent. This is done sequentially for each of the seven clinical variables. The calculator display prompts the user by displaying the coefficient for the variable that should be entered next. On the basis of their order as listed in the Statistical Methods section, the first 1 or 2 will correspond to the presence or absence of chest pain or pressure, the second 1 or 2 to whether chest pain is the patient's most important symptom, and so on, for all seven variables. The number then displayed will be the calculated probability that the patient has acute cardiac ischemia. for example, if a patient presents to an emergency room with a chief symptom of chest pain, without prior cardiac history or use of nitroglycerin, and with electrocardiogram ST-segment depression and T-wave inversion, one would sequentially enter, 1, 1, 2, 2, 1, 1, 1, after which the calculator would display 0.78, indicating that the patient's probability of having acute cardiac ischemia was 78 per cent. Typically, the entire computation process requires less than 20 seconds.
There are several ways that the predictive instrument can be used. An emergency room triage nurse can compute the probability value and simply write it on the clinical record for the physician's use, in a manner similar to that in which vital signs are commonly recorded. Alternatively, a physician can carry a pocket-sized calculator to allow quick access to the probability value when the instrument is not available. Another method might be the integration of the instrument into a self-interpreting electrocardiograph so that, once given the four non-electrocardiographic clinical variables, the machine can compute a given patient's probability of having acute ischemia. Any method that gives the physician access to the probability value during diagnosis and triage in the emergency room should result in the type of improvements described in this study.
Print Subscriber? Activate your online access.