One hundred and thirty eight male patients (US Armed Services Veterans) were screened; 112 consented, nine withdrew consent, and 103 completed the study. Twenty-six of 77 patients scheduled for cardiac surgery reported a history of alcohol dependence. An additional group of 26 patients without a history of alcohol dependence were included as nonsurgical controls. A subset of subjects included in the current study was recruited for other investigations [19,20]. The inclusion of the nonsurgical control group was important to account for practice effects as a result of repeated cognitive testing [21]. Prospective patients were included if they were at least 55 years of age, provided written informed consent, and were scheduled for elective coronary artery bypass graft surgery with or without valve repair or replacement using cardiopulmonary bypass. Prospective participants were excluded if they had a history of cerebrovascular accident within three years of randomization, permanent ventricular pacing, or previously documented cognitive deficits. Patients with hepatic impairment (aspartate aminotransferase or alanine aminotransferase more than twice the upper normal limit) and chronic renal insufficiency (creatinine > 2 mg/dL) were also excluded from participation.

A history of self-reported admission to a detoxification treatment facility related to alcohol consumption was used to define alcohol dependence. All participants with such a history met the criteria for alcohol dependence, but not for alcohol-induced persisting amnesic disorder or dementia using Diagnostic and Statistical Manual of Mental Disorders—Fourth Edition (DSM-IV) definitions. All participants included in the alcohol dependence group were in full remission and reported abstinence from alcohol for at least five weeks before inclusion in the study. Based on chart review, patients with a negative history of alcohol treatment did not have a diagnosis of alcohol use disorder. Nonsurgical patients without a history of alcohol dependence were recruited from various outpatient clinics by the recommendation of the head nurses who prescreened the patients for the study criteria. All participants were screened for physical health and psychological history and were excluded if there was evidence of organic brain syndrome or preexisting dementia.

Neurologic status, delirium, and neurocognitive functions were assessed within one week before surgery. A neurological examination was performed to exclude patients with potential cerebrovascular anomalies. Delirium was monitored and reassessed up to a maximum of five days after surgery by three independent psychologists. If the first two psychologists disagreed, then the third psychologist’s rating determined whether or not the patient suffered from postoperative delirium. If a patient received different delirium scores on any postoperative day from the separate observers, these scores were averaged. The first two psychologists agreed with the diagnosis of delirium 100% of the time; therefore, the interrater consistency between them was 100%. Recent verbal and nonverbal memory and executive functions were assessed as indices of neurocognitive function.

Anesthetic management and cardiopulmonary bypass were standardized as follows. Patients received midazolam (1–2 mg) for premedication, and fentanyl and etomidate for anesthetic induction. Isoflurane (0.4–1.6%) and fentanyl were used for maintenance of anesthesia. All patients underwent a standard median sternotomy for cardiac surgery. Myocardial protection was produced using antegrade and retrograde cold blood cardioplegia provided at regular intervals (15 min), topical hypothermia (slushed 0.9% saline), and systemic hypothermia (30 to 32 degrees centigrade).

A dose of continuous warm blood cardioplegia was administered during rewarming before removal of the aortic cross clamp. Cardiopulmonary bypass flows were 2.4 to 2.5 L/min/m². Mean arterial pressure was maintained between 55 and 70 mmHg during CPB. Heparin (400 U/kg) was used for systemic anticoagulation to maintain the activated clotting time (ACT) > 500 sec. Neurological status and neurocognitive functions were reassessed one week after surgery or at hospital discharge, whichever occurred first or one week apart for patients who did not undergo surgery. The psychologists performing the pre- and postoperative neurocognitive assessments and physicians performing the neurological examinations were blinded to each participant’s history of alcohol use.

Delirium was assessed with the Intensive Care Delirium Screening Checklist [22]. This is an 8-item (altered level of consciousness, inattention, disorientation, hallucination-delusion-psychosis, psychomotor agitation or retardation, inappropriate speech or mood, sleep/wake cycle disturbance, symptom fluctuation) checklist based on DSM-IV criteria and features of delirium. Raters completed the checklist based on data from the previous 24 hrs. Routinely collected data (such as orientation) were combined with short observations of obvious manifestations of described features. The eight items are scored 1 (present) or 0 (absent) for a total of 8 points. A score of 4 or greater is considered a positive screen for delirium [22].

The neurocognitive test battery was comprised of standard clinical measures that were appropriate for use with subjects in the age group studied, with minimal sensory or motor demands. The tests covered three cognitive domains: verbal recent memory, nonverbal recent memory, and executive functions. Total test administration time was less than one hour. When available, alternate forms (A and B) were used to reduce practice effects between sessions. The order of presentation was counterbalanced with half of the subjects receiving form A at the first session and the other half receiving form B.

To test verbal recent memory, Story Memory and Word List Memory subtests from the Repeatable Battery for the Assessment of Neuropsychological Status [23] were used. Story Memory measures the ability to learn and recall a narrative story in two trials (maximum score 20) and delayed free recall (maximum score 10). Word List Memory assesses the ability to learn and remember a list of 10 unrelated words across four sequential learning trials (maximum score 40) and delayed free recall (maximum score 10).

To test nonverbal recent memory, the Brief Visual Memory Test-Revised [24] was used. Participants were shown a card with six simple geometric designs (2 points per design) for ten seconds on three sequential trials. When the card was removed, the examinee drew the designs in the locations remembered (maximum score 36). A free recall trial was administered twenty minutes later (maximum score 12).

To test executive functions, Semantic Fluency [23], Phonemic Fluency [25], and Digit Span (backward) [26] were used. Semantic and Phonemic Fluency are subtests of the Delis-Kaplan Executive Function System [27] that examine executive functions related to language. Semantic Fluency measures executive speed of word generation using semantic cues such as identifying all the “fruits and vegetables” (form A) or “animals in the zoo” (form B) that the patient is able to produce in 1 min. The obtained score is the number of appropriate words generated within the time interval. Phonemic Fluency measures executive speed of word generation using phonetic cues such as identifying all the words that start with the letter “S” (form A) or “P” (form B) that the patient is able to produce in 1 min. The score is the number of appropriate words identified within the time interval. Digit Span is a subtest of the Wechsler Adult Intelligence Scale-Third Edition [26] and measures attention span, concentration, and the Backward Digit Span executive functions.

Depression was assessed with the Geriatric Depression Scale 15-item version [28]. The obtained score was the number of symptoms endorsed.

Neurological examination focused on detection of significant visual and auditory impairment, level of consciousness, cranial nerve function, motor strength, cerebellar ataxia, intention tremor, sensation, frontal lobe release signs, deep tendon and plantar reflexes, and gait ataxia. An additional total score was calculated using the Hachinski Ischemia Scale [29]. A score greater than 4 is diagnostic of vascular dementia.

Power analysis was conducted based on a pilot study of 28 subjects. POCD based on composite z-scores was observed in 57% and 14% of patients with and without a history of alcohol dependence. Assuming 80% power and a 2-sided alpha of 0.05, power analysis yielded N = 23 as the estimated number of subjects per group to detect a difference in cognitive function between those with and without a history of alcohol dependence. An attrition rate of 10% was estimated to account for those whose second neurocognitive testing could not be performed due to dropout; thus, we recruited 1.1 times 23 yielding a total of at least 26 subjects per group.

To test for between-group differences in the demographic and medical data, Chi-square or Fisher’s exact tests were used for proportions and the Student’s t-test was used for continuous outcomes.

To determine the presence of POCD, we followed Moller’s method [4]. The z-scores express by how many standard deviations the surgical patient’s performance deviates from the expected in the control patients. The z-scores were calculated as differences in all test parameters between the pre- and postoperative test sessions for the individual patients with appropriate signs (plus or minus). After subtracting the mean change between sessions in the nonsurgical control group, which represents the practice effect, the result was then divided by the standard deviation for this change in the nonsurgical control group. The use of a suitable normative population is essential to allow correction for practice effects and variability between sessions [21]. To obtain composite z-scores, we averaged all practice-effect controlled z-scores of all tasks that comprised the test battery for the individual surgical patients. The composite z-score represents a general or global measure of the domains of functioning assessed and represents their operationalized definition of POCD. Patients were defined as having cognitive dysfunction when their composite z-score was less than −1.96 (2 standard deviations). This definition took into account general postoperative deterioration in performance across all tests. Patients with or without a history of alcohol dependence who met the 2 SD criterion were compared using the Chi squared test. Cognitive performance between groups was also compared using the Mann-Whitney U-test based on composite z-scores. In addition, we conducted repeated measures of analysis of variance (Table 3) with covariates of smoking status, age, and sleep disorder to determine the significance of group change on the individual measures. None of the covariates revealed significance; therefore we dropped them from the model. To account for multiple comparisons we used the Bonferroni correction. To reveal which neurocognitive tests were indicative of most profound impairment in patients with a history of alcohol dependence, effect sizes (Cohen’s d) were calculated. Because depression scores are not normally distributed, these values were expressed as median and confidence intervals (CI), and the Mann-Whitney U-test was used for between group comparisons. Multiple linear regression analysis was performed to examine the possible effects of demographic, medical, and perioperative conditions on the cognitive test results (composite z-scores). Significant (p < 0.05) demographic variables, perioperative predictors that showed p = 0.1 or less significance in a univariate regression model, and age were entered into the final multivariate model.

All comparisons were made using 2-sided tests at α < 0.05. All errors are reported as SD. Statistical calculations were performed using NCSS 2001 (NCSS, Kaysville, UT) and STATA/IC 10.0 (StataCorp LP, College Station, TX) software.