The results of this prospective cohort study confirm that low preoperative executive function and depression are significant risk factors for the development of postoperative delirium in elderly patients undergoing major non-cardiac surgery. Specifically, we found that a rapid measure of executive function (Trails B) and a simple screening tool for depression (GDS-SF) were highly predictive of postoperative delirium. Moreover, the combination of these two tests was a robust predictor of delirium in our patients having a specificity of 99% and a positive predictive value of 83% in our patient population ().
Depression, as measured by standardized depression scales, has been found to have a significant association with delirium across numerous studies.8
The findings of three recent studies in elderly surgical and medical patients support the notion that preexisting depression plays an important role in the development of delirium.11-13
The GDS-SF was used in our investigation and in two previous studies. Several of these studies demonstrated that depression was associated with both the incidence11,13
of delirium. The GDS-SF is a 15-item questionnaire that uses a yes/no format and is simple to administer and complete. This short form eases the burden on elderly and ill patients and has been validated as a screening instrument in preoperative patients.25
The GDS-SF is especially useful in surgical populations because it eliminates the somatic items contained in other depression scales that could lead to an increased rate of false-positive depression diagnoses in physically ill populations.25
It can be completed in approximately 5 minutes and is free for use in the public domain.
Cognitive impairment, as measured by global cognitive testing such as the MMSE, is also a well-known risk factor for delirium.8,9,29
The 3MS, an extended version of the MMSE, was not predictive in our sample. Kalisvaart et al.29
enrolled 603 elderly patients undergoing hip surgery (both acute fractures and elective hip surgeries) in a prospective cohort study validating a risk factor model for predicting postoperative delirium. The MMSE was an independent predictor of delirium in this study and patients who developed delirium had a significantly lower mean MMSE of 21.7 compared to 25.7 in nonaffected patients.29
A second study by Rudolph et al.30
also found an association between poor performance on the MMSE and an increased risk for delirium in elderly patients (mean age 75 years) undergoing coronary artery bypass graft surgery. The raw scores for MMSE were not reported in this study but the subjects had a 50% incidence of delirium suggesting that they had low preoperative MMSE scores making them high-risk for delirium. A MMSE score less than 24 (out of a possible 30) is generally considered to be the threshold for cognitive impairment.31
We extracted the MMSE scores from the 3MS and found that the mean preoperative MMSE scores in our study were 26 for patients who developed delirium and 28 for the patients who did not have this problem. It is likely that the 3MS and the MMSE were not predictive in our study because our patients had better preoperative global cognitive function than the patients in previous studies and the MMSE was not sensitive enough to predict postoperative delirium in our cohort with relatively normal cognition. As such, the predictive ability of executive tasks as observed in this study and in our companion study14
among relatively non-impaired samples is intriguing.
Our finding that preoperative executive functioning predicts delirium is in agreement with a study by Rudolph and colleagues.30
This study enrolled 80 elderly subjects undergoing coronary artery bypass graft surgery. The authors defined memory and executive function composites using confirmatory factor analysis. After adjusting for preoperative risk factors, they found that poor performance on measures of executive function were independently associated with the development of delirium.30
These studies collectively suggest that executive dysfunction may be a strong indicator of postoperative risk, even among subjects who are considered cognitively intact preoperatively. The finding that global measures of cognition did not predict delirium in either sample, despite its common clinical usage, may be due to the relative insensitivity of general screening measures for identifying very mild cerebral dysfunction,32
particularly among highly educated subjects. In addition, executive measures may better reflect aspects of central nervous system integrity and the underlying pathophysiology related to the development of delirium.14
One of the limitations of this paper is that the majority of patients (91%) were male limiting the generalizability of these results to females. However, a companion paper in this journal14
replicates our analysis in a large cohort of patients undergoing noncardiac surgery in which the majority (63%) were female and confirmed our findings in this second patient population. Together these two manuscripts strongly suggest that preoperative depressive symptoms and low executive scores may be important predictors of individuals at risk for postoperative delirium regardless of the gender of the patient. Other factors limiting the generalizability of these results is the preponderance (69%) of Caucasians in the sample and the inclusion of only major intra-abdominal, thoracic or orthopedic surgical procedures in the study. Future studies should target racially diverse populations and other types of surgical procedures. Another potential limitation of this paper is the inclusion of four preoperative covariates in our multivariate logistic regression model, possibly overfitting the model. The general rule of thumb is that there should be 10 events per variable in the model. However, more recent studies suggest that commonly used rules of thumb regarding number of events may be overly conservative in the “analysis of casual influences of observational data.”33
In contrast to previous publications,8
we did not find an association between age or preoperative comorbidity (Charlson and American Society of Anesthesiologists scores) and delirium. However, our study had a relatively small sample size of 100 patients and included only individuals who were 50 years or older. It is likely that, with the limited range of scores on these variables, our study was underpowered to show differences in these variables. However, preoperative executive skills and depression status correlated with postoperative delirium suggesting that cognitive measures are much more sensitive than measures of physical status for predicting delirium.
Patients who developed delirium had lower scores on tests of executive function suggesting that, although they were functioning normally, they may have had less preoperative cognitive reserve. Cognitive reserve is a concept that attempts to explain why individuals with similar degrees of cerebral insult often have significant differences in the degree of cognitive symptoms. Different levels of baseline cognitive capacity between patients may explain differences in cognitive outcomes associated with traumatic brain injury,34
and the development of dementia36,37
in the elderly. Cognitive reserve may also protect against delirium as evidenced by a study that correlated higher educational attainment with a lower risk of developing delirium.38
Executive function may be an ideal marker of cognitive reserve as it reflects higher order complex cognition. Preoperative executive skills may be a marker of the integrity of the patient’s central nervous system.
Preoperative identification of patients at risk for delirium will allow the clinician to implement behavioral or pharmacologic interventions to minimize this problem once effective intervention strategies have been identified. A recent Cochrane database review39
examined the literature on delirium prevention. This review identified only six randomized controlled trials assessing the effectiveness of interventions to prevent delirium and concluded that “the research evidence on the effectiveness of interventions to prevent delirium is sparse.” The only study that was considered to have adequate power to determine effectiveness was a trial by Marcantonio et al.40
concluding that a proactive geriatrics consultation can significantly decrease postoperative delirium in elderly patients following hip fracture repair. A second study compared two anesthetic approaches (halothane versus epidural anesthesia)41
while the remainder of the trials focused on a variety of pharmacological interventions including daily prophylactic haloperidol administration42
and the use of doneprezil, an acetyl cholinesterase inhibitor used in the treatment of Alzheimer’s disease.43
The Cochrane review concluded that there was inadequate evidence available on the effectiveness of pharmacological strategies and that “future trials of delirium prevention are urgently needed.”39
In summary, we found that preexisting executive dysfunction and symptoms of depression, as measured by the Trails B and GDS-SF respectively, were independent predictors of postoperative delirium in elderly patients. Combining the results of these screening tests improved their predictive accuracy. A measure of global cognitive ability (3MS) did not distinguish between those who developed delirium and those who did not. This suggests that less sensitive global measures of cognitive function may misidentify some patients as being low risk for postoperative delirium. The combination of a test of executive function and a depression index as an effective screening tool for patients at risk of postoperative delirium requires validation and further refinement in large patient populations. The development of a rapid simple screening tool that identifies elderly surgical candidates at risk for delirium and appropriate intervention strategies to prevent this complication should be a research priority.