Effects of Changing Guidelines on Prescribing Aspirin for Primary Prevention of Cardiovascular Events

Jennifer Hissett, MD, Brittany Folks, MD, Letoynia Coombs, MS, William LeBlanc, PhD, and Wilson D. Pace, MD


J Am Board Fam Med. 2014;27(1):78-86. 

In This Article


To observe how recorded aspirin use changed over time, we used an eNQUIRENet database originally created as part of a CVD learning community grant.[20] The original limited dataset included information on all individuals >17 years old from 33 primary care practices in 11 different clinical organizations spread across 6 states. This dataset was fully de-identified by converting dates of service to date ranges corresponding to the dates of the availability of new information or guideline changes related to aspirin for the primary prevention of CVD. From this de-identified data set, information on 131,050 patients was abstracted to create cohorts of patients who met various criteria for primary prevention.

The de-identified dataset included ≥4 years of data for the following information extracted from electronic health records (EHRs): diagnoses (rare diagnoses suppressed) with initial date and follow-up dates recorded in the EHR, medications recorded in the EHR, year of birth (suppressed to 95 years if >89 years of age), blood pressure, weight, height, smoking status, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, hemoglobin A1c, creatinine, and urine protein-to-creatinine ratio.

From the dataset described above, a retrospective cohort study was conducted comparing (1) patients with a known diagnosis of CVD and (2) patients without CVD but with increased risk of CAD. Patients were identified as having a >10% 10-year risk of CVD using the Framingham scoring system[21] based on age, sex, blood pressure, cholesterol, smoking status, and a diagnosis of diabetes. Cohorts were conservatively developed to ensure a calculated Framingham 10-year CVD risk score of >10% for the entire cohort. For instance, men older than 50 with diabetes and one other risk factor, such as hypertension or hyperlipidemia, all have a Framingham 10-year risk score of >10%; therefore, this cutoff was used (full cohort definitions for each group are available from the authors). Information used to create the cohorts included diagnoses, average blood pressures (last 3), average non–high-density lipoprotein total cholesterol (last 3 measurements), smoking status, age, and sex.

The full cohort also was organized into subgroups by diagnosis (diabetes, chronic kidney disease, peripheral vascular disease and aortic aneurysm, hyperlipidemia, and hypertension) for presentation and to observe any variance by underlying diagnoses. Changes in recorded aspirin use were observed among patients who had either previous or new diagnoses of CAD risk equivalents within a 4-year period. The periods correspond with the publication of DOPPS in 2007 (time 1: January 1, 2007, to December 31, 2007); POPADAD and JPAD in 2008 (time 2: January 1, 2008, to December 31, 2008); ATTC in 2009 (time 3: January 1, 2009 to May 31, 2010), and the ADA/AHA/American College of Cardiology guideline changes in mid-2010 (time 4: June 1, 2010, to May 31, 2011). Aspirin usage was determined by electronic review of medication lists. Aspirin was identified through review of National Drug Codes (NDCs) from the Medispan drug database[22] and by reviewing text entries for medications with no associated NDCs. More than 60% of all aspirin users were identified through NDC matches, and more than 95% of these were for low-dose aspirin. The text-based aspirin records were grouped and the top 250 records were manually reviewed, representing more than 80% of all non-NDC-matched aspirin options. Of these, again more than 95% were for low-dose aspirin. Thus the vast majority of individuals taking aspirin in this study were using it for some type of prevention, not for pain or anti-inflammatory indications.

A total of 131,050 individuals met the criteria for inclusion in one or more of the cohorts for analysis. Individuals could change cohorts from one time period to another based on new diagnoses or clinical data within a given period. Individuals could be in more than one primary prevention subanalysis cohort, but once they received a diagnosis of CVD they remained in this cohort for the remainder of the periods analyzed.

Each cohort by period was broken into 2 groups: met cohort definition before the time period (current) or entered the cohort during the time period (new). This allowed for an analysis of the percentage of patients starting aspirin therapy based on new clinical information over time. For each cohort, the population was determined based on clinical data. Medication lists were subsequently analyzed to determine the percentage of the cohort receiving aspirin therapy. For all known CVD cohorts, clopidogrel was included as an aspirin equivalent. The study was approved by the Colorado Multi-Institutional Review Board.