AHRQ Annual Meeting September 28, 2010

1. Missed Diagnoses of Acute Myocardial Infarction in the Emergency Department: An Exploration Using HCUP Data AHRQ Annual Meeting September 28, 2010

2. Team • Agency for Healthcare Research and Quality • Ernest Moy, MD, MPH • Thomson Reuters • Cheryl Kassed, PhD, MSPH • Marguerite Barrett, MS • Rosanna Coffey, PhD • Anika Hines, PhD, MPH 2

3. Outline • Background • Specific Aims • Methods • Results • Conclusion • Implications 3

4. Background • Some patients with acute myocardial infarction (AMI) are mistakenly released from the emergency department (2-5%) • Such patients may have increased mortality • Failure to hospitalize may be related to race, gender, and the absence of typical cardiac symptoms • Little work comparing rates across institutions

5. Specific Aims • To explore the use of administrative data to identify missed diagnoses of AMI • How do HCUP estimates compare to the literature? • How do rates of missed diagnosis of AMI vary across subgroups? • How do rates of missed diagnosis of AMI vary across hospitals?

6. Data: HCUP • Healthcare Cost and Utilization Project (HCUP) is a family of health care databases developed through a Federal-State-Industry partnership sponsored by the Agency for Healthcare Research and Quality (AHRQ) • SID: State Inpatient Databases = universe of inpatient discharge records from 42 states • SEDD: State Emergency Department Databases = hospital-affiliated emergency departments visits that do not result in hospitalizations

7. Methods • Sample: HCUP data for 9 states with reliable person linkages and race/ethnicity data—AZ, FL, MA, MO, NH, NY, SC, TN, UT • Design: Cross-sectional analysis of adults • 18 years or older • First AMI admission between Feb and Dec 2007 • Analysis: Subgroup estimates compared using t-tests (p-value<0.05)

8. Methods • Key Measure: • Percentage of patients with an AMI admission who were seen in the ED within the prior 2 to 7 days for a cardiac-related issue • cardiac diagnosis/symptom • abdominal pain

9. Percent of patients with an ED visit with likely missed AMI—patient attributes Medicaid *Medicare *Uninsured Private insurance (ref) *p<0.05

10. Percent of patients with an ED visit with likely missed AMI—hospital attributes *<100 beds *p<0.05

11. Percent of patients with an ED visit with likely missed AMI—other attributes *Weekend Jul-Aug *Mar-Apr Sept-Oct *Jan-Feb Nov-Dec Weekday (ref) *Busy ED day May-Jun (ref) Slow ED day (ref) *Crowded ED day *Moderate ED day *Hospital without cardiac cath Hospital with cardiac cath (ref) *p<0.05

12. Conclusions • Study rate of AMI missed diagnosis=1.85% • Pope et al, study found 2.1% • Administrative data are a reasonable source for estimating missed AMI diagnoses

13. Conclusions • Unsurprising results • Vulnerable populations have higher rates of missed diagnoses for AMI—minorities, the uninsured, those with low-income, and those visiting hospitals in rural areas • Surprising results • Busy hospitals have lower rates of AMI missed diagnoses (i.e. hospitals with higher occupancy rates, higher bed volume, and residency programs) • Weekend visits and slow ED days have higher rates of AMI missed diagnoses

14. Limitations • Administrative data  lower estimates of missed diagnoses • Data not representative—9 states

15. Implications • Administrative data may be useful for studying other types of missed diagnoses. • Reporting on variation in missed diagnoses could lead to better quality of care.