This article is more than 5 years old. Information may no longer be current.
Statistical risk estimate a better method to predict mortality after cardiac surgery
Statistical risk estimate is more effective than a physician’s subjective assessment for the prediction of short- and long-term mortality among patients undergoing cardiac surgery, according to findings from a recent study.
Researchers identified for evaluation 5,099 patients who underwent cardiac surgery at Minneapolis VA Medical Center between 1993 and 2010. The VA Continuous Improvement in Cardiac Surgery Program (CICSP) was used to collect data for risk estimates.
For the study, statistical risk estimate was calculated using a multivariable logistic regression model, and estimates ranging from 0% to 100% were assigned to each patient. Physician’s risk estimates consisted of the cardiac surgeon’s subjective evaluation of 30-day mortality risk, and physicians were instructed to assign a numeric mortality estimate ranging from 0% to 100% based on their clinical observations.
One month after surgery, the overall mortality rate was 3.3% (168 deaths). At 1 year, the rate was 7.1% (360 deaths), and at 5 years, the rate was 18.5% (942 deaths).
The researchers noted modest correlation between actual mortality and both the physician’s risk estimate (mean, 5.6%) and statistical risk estimate (mean, 4.3%; c-index, 0.56; P<.001).
However, both methods of estimation slightly overestimated the risk for operative mortality.
Statistical risk estimate demonstrated significant superiority over physician’s risk estimate in distinguishing patients who died from those who survived at 30 days (c-index, 0.78 vs. 0.73; P=.003), at 1 year (c-index, 0.72 vs. 0.61; P<.001) and at 5 years (c-index, 0.72 vs. 0.64; P<.001) after surgery. Physicians rendered higher mortality risk estimates than those generated by statistical estimate for all subgroups except high-risk patients, for whom they underestimated risk.
“Risk scores have been used to predict adverse outcomes and aid medical decision-making in clinical settings,” the researchers wrote. “Risk scores provide an objective measure of risk stratification as a supplement to physician’s clinical intuition, thereby improving the quality of medical decision making and ultimately patient outcomes.
“In patients undergoing cardiac surgery, short- and long-term mortality could be better predicted by statistical risk estimate compared with physician’s subjective risk estimate. However, both methods modestly overestimate observed mortality,” the researchers concluded.
Disclosure: One researcher reports receiving investigator-initiated research grants from Medtronic and Boston Scientific. The other researchers report no relevant financial disclosures.