Abstract 210: Determining In-Hospital Myocardial Infarction Rate from Administrative Data in a Large Tertiary-Care Hospital
Background & Objective: The PULSE Trial is a 5-year multisite randomized clinical trial to evaluate implementation of American Heart Association Practice Standards for Electrocardiographic Monitoring on nurses' knowledge, quality of care, and patient outcomes. The primary intervention was an online ECG monitoring education program for nurses. A key indicator in assessing patient outcomes is the incidence of myocardial infarction (MI) while on a monitored unit. Determination of MI while in hospital can be challenging, requiring time-consuming medical record review. We propose a method of reducing the number of medical records necessary for review to determine occurrence of an in-hospital MI by using an algorithm that incorporates administrative and laboratory data.
Methods: Administrative data from a large tertiary-care hospital in the US were obtained for a 6-month period. Fields included patient identifier, admission identifier, demographic data, unit transfers, length of stay, procedural codes, and diagnostic codes. Troponin and CK-MB levels were matched with administrative data. Step 1: We deleted all patient encounters where there were no troponin or CK-MB values above the upper reference limit (URL). We then deleted all encounters where all values occurred <24 hours after admission and the patient did not have coronary artery bypass graft (CABG) surgery or percutaneous coronary intervention (PCI). Finally, we deleted all patient encounters for which all values >24 hours after admission were either decreasing or below the URL and the patient did not have CABG or PCI. This allowed us to exclude all those who did not have a MI, or had one prior to being admitted to a unit. Step 2: If patients had an elevated biomarker <24 hours after admission, we examined if there was an increase of >20% in value following the index event or CABG or PCI procedure. If yes, they were considered to have an MI. If patients had an elevated biomarker >24 hours after admission, we determined if the values were 3x and 5x the URL for PCI and CABG, respectively. If yes, they were considered to have an MI; those who did not required further record review.
Results: We had 7,691 admissions. In step 1 we were able to rule out MI in 6,849 admissions. In step 2 we determined 596 had elevated biomarkers within 24 hours of admission. We determined 193 had an MI. The remaining 403 required further review. We determined that 245 had elevated biomarkers >24 hours after admission. Of those, 21 had an elective PCI or CABG, 4 of whom had 3x or 5x the URL of biomarkers and were consider to have a MI. The 224 who did not have an elective PCI or CABG required further review.
Conclusions: This algorithm reduced the number of medical record reviews from 7,691 to 627 and allowed for determination of an outcome that is not easily assessed, thereby precluding the need for lengthy medical record reviews. The algorithm can be adjusted to reflect changing criteria for MI.
Author Disclosures: M. Funk: None. K. Fennie: None. K. Stephens: None. J. May: None.
- © 2014 by American Heart Association, Inc.