Cognitive performance after cardiac operation: implications of regression toward the mean.

OBJECTIVE: The purpose of this study was to assess the influence of regression toward the mean on different definitions of cognitive dysfunction after coronary artery bypass graft operation. METHODS: A total of 120 patients who underwent nonemergency coronary bypass operation and who were involved in a randomized trial of an anti-inflammatory agent were assessed prospectively with a battery of 10 psychometric tests covering a variety of cognitive domains. The battery was administered before the operation and 5 days and 3 months after the operation. Data from 2 representative tests, the Rey Auditory Verbal Learning Test and the Trail Making Test (part A), were used. The influence of regression toward the mean on 3 commonly used single-case definitions of cognitive impairment (1 SD method; one-half SD method; 20% method) was analyzed. RESULTS: Group mean performance deteriorated on the Rey Auditory Verbal Learning Test at the discharge assessment (P <.001) and remained below baseline at 3 months (P =.03). Mean performance on the Trail Making Test (part A) showed a near-significant decline at discharge (P =. 06), followed by improvement at 3 months (P <.01). Regression toward the mean was demonstrated on both tests by classifying the preoperative scores into low, moderate, and high-performance categories. Applying the different definitions of dysfunction resulted in substantially larger numbers of patients in the high-performance group being classified as impaired. CONCLUSION: Single-case definitions of cognitive dysfunction are influenced strongly by regression toward the mean. Disproportionate numbers of high-baseline performers are classified as impaired, thereby questioning the validity of established definitions. Group mean analysis with controls is potentially the most reliable method for detecting real change or differences.