This article is more than 5 years old. Information may no longer be current.
Pooled Cohort Equations predict lower risk when CVD events surveillance excluded
CV risk prediction with the Pooled Cohort Equations was lower when surveillance for atherosclerotic CVD events was not included, according to a study published in JAMA Internal Medicine.
The addition of events from surveillance of CMS data caused the risks to be better aligned, according to the study.
Women’s Health Initiative
Samia Mora, MD, MHS, associate professor of medicine at Brigham and Women’s Hospital, and colleagues analyzed data from 19,995 women (mean age, 64 years) from the Women’s Health Initiative aged 50 to 79 years. Women with a medical history of atrial fibrillation or HF were excluded.
Information on statin and aspirin use was collected at baseline and at 3 years for patients from the observational study and at baseline, 1, 3, 6 and 9 years in the clinical trial. All patients underwent medication assessment at the close-out of the study.
Women completed questionnaires annually on whether they were hospitalized since the last report, which was then adjudicated by medical records. Medicare files for patients older than 65 years were also analyzed.
Outcomes of interest were the first atherosclerotic CVD event, which was defined as the first occurrence of CHD, nonfatal MI, death, or fatal or nonfatal stroke.
In 10 years, 1,236 first atherosclerotic CVD events occurred and were adjudicated by the Women’s Health Initiative. Adjudicated observed risks were lower than originally predicted.
The observed (predicted) incident rates were 1.7 (2.8) for women with a baseline 10-year predicted risk of less than 5%, 4.4 (6.2) for risk between 5% and less than 7.5%, 5.3 (8.8) for risk between 7.5% and less than 10% and 12.4 (18.2) for risk of 10% or more.
After adjusting for time-dependent changes in aspirin and statin use, minimal changes were noted.
Medicare enrollees
Women older than 65 years who were enrolled in Medicare had lower adjudicated risks than predicted. After events that were linked to CMS were included, observed (predicted) risks were aligned: 1.7 (2.8) in women with risk less than 5%, 7.1 (6.4) in those with risk between 5% and less than 7.5%, 8.3 (8.7) in those with risk between 7.5% and less than 10% and 18.9 (18.7) in those with a risk of 10% or more.
Results were similar across ethnic/racial groups.
The C statistic was 0.726 (95% CI, 0.714-0.738).
“Because discrimination was not the issue affecting the performance of the equations, the differences in the predicted:observed rates may result from apparent or real differences in disease prevalence,” Mora and colleagues wrote. “This issue is well-recognized in the literature, where it is recommended that scores should be recalibrated if incidence rates differ substantially in the new settings. It is well-known that CVD incidence rates have been declining, which suggests that it would be appropriate to align the predicted rates with more contemporary cohorts as long as they have comprehensive surveillance for events.”
In a related editor’s note, Gregory D. Curfman, MD, assistant professor of medicine in the division of pharmacoepidemiology and pharmacoeconomics at Brigham and Women’s Hospital, wrote: “The central message of the study by Mora et al is that accurate measurement of ASCVD risk in populations depends on comprehensive surveillance of events, but whether this provides a complete explanation of the overestimation of risk by the [Pooled Cohort Equations] remains uncertain. A new study suggests that risk equations may become outdated over time and require updating and recalibration to improve their accuracy.” – by Darlene Dobkowski
Disclosures: Mora reports she received research grant support from Atherotech Diagnostics and NHLBI and personal fees for participating in a scientific advisory board for Amgen, Lilly, Pfizer and Quest Diagnostics. Curfman reports no relevant financial disclosures. Please see the study for all other authors’ relevant financial disclosures.