Predictors of cognitive recovery after cardiac surgery

Abstract

Background: Postoperative neurocognitive decline occurs frequently. Although predictors of cognitive injury have been well examined, factors that modulate recovery have not. We sought to determine the predictors of cognitive recovery after initial injury following cardiac surgery.

Methods: Two hundred eighty-one patients previously enrolled in cognitive studies who experienced cognitive decline 6 weeks after cardiac surgery were retrospectively evaluated. Eligible patients completed a battery of neurocognitive measures and quality-of-life assessments at baseline, 6 weeks, and 1 year after surgery. Factor analysis was conducted to calculate the cognitive index (CI), a unified, continuous measure of cognitive function. Cognitive recovery was defined as 1-year CI greater than baseline CI. Potential predictors of cognitive recovery including patient characteristics, quality-of-life factors, comorbidities, medications, and intraoperative variables were assessed with multivariable regression modeling; P<0.05 was considered significant.

Results: Of the 229 patients in our final data set, 103 (45%) demonstrated cognitive recovery after initial decline in CI at 6 weeks. Multivariable analyses revealed that more education (odds ratio [OR] 1.332 [1.131-1.569], P<0.001), baseline CI (OR 0.987 [0.976-0.998], P=0.02), less decline in CI at 6 weeks (OR 1.044 [1.014-1.075], P=0.004), and greater activities of daily living at 6 weeks (OR 0.891 [0.810-0.981], P=0.02) were significant predictors of cognitive recovery.

Conclusion: Cognitive recovery occurred in approximately one half of the cardiac surgical patients experiencing early decline. The association between cognitive recovery and Instrumental Activities of Daily Living scores at 6 weeks merits further investigation as it is the only potentially modifiable predictor of recovery.

Figure 1

The association between Instrumental Activities of Daily Living at 6 weeks and predicted probability of cognitive recovery. The predicted probability of recovery is derived from the multivariable logistic regression model, which estimates a greater likelihood of cognitive recovery with increasing IADL scores after adjusting for years of education, baseline function, and amount of decline at 6 weeks. That is, a patient with a higher IADL score is more likely to recover cognitive function than a patient with a lower IADL score, even if the two patients have the same baseline score, the same education level, and experienced the same amount of decline. Lower IADL scores indicate enhanced function and the dashed lines represent 95% confidence intervals. The figure footnote indicates the percentage of the sample which had the corresponding IADL score or a lower score; for example, 75% of the sample scored 12 or lower.