Myocardial infarction and the validation of physician billing and hospitalization data using electronic medical...

Area of Science:

• Health Services Research
• Cardiovascular Epidemiology
• Health Informatics

Background:

• Population-based identification of myocardial infarction (MI) is often limited to hospitalized acute events.
• Accurate MI ascertainment is crucial for epidemiological studies and healthcare planning.

Purpose of the Study:

• To evaluate the accuracy of combining physician billing data with hospital discharge data for identifying MI patients.
• To determine if this combined data approach improves MI case capture compared to hospital data alone.

Main Methods:

• Retrospective chart abstraction of 969 adult patients in Ontario, Canada.
• Utilized primary care physician electronic medical record data as the reference standard.
• Developed an algorithm combining physician billing codes and hospital discharge abstracts.

Main Results:

• The algorithm achieved 80.4% sensitivity and 98.0% specificity for MI identification.
• Positive predictive value was 69.5%, and negative predictive value was 98.9%.
• A kappa statistic of 0.73 indicated substantial agreement.

Conclusions:

• Combining hospital discharge abstracts and physician billing data offers a more comprehensive method for assessing MI trends.
• This integrated data approach expands MI case ascertainment beyond hospitalized patients.
• The findings support using combined data for improved population-level MI surveillance.