Coronary artery disease is the single largest cause of death in developed countries, andone of the leading contributors to death in the developing world [1, 2]. Coronary artery bypass graft (CABG) surgery is a lifesaving treatment forsevere ischemic heart disease. However, this procedure is associated withneuropsychiatric complications. These complications include delirium, whichsubstantially worsens postoperative recovery and prognosis [3, 4].

According to recent studies, the most prominent factors contributing to postoperativedelirium include comorbid load (atrial fibrillation, prior stroke, anemia, peripheralvascular disease) as well as psychiatric comorbidity such as cognitive impairment andpreoperative major depressive disorder (MDD) [57]. The pathological association between MDD and postoperative delirium isunclear. These disorders have been proposed to be linked by a greater rise in plasmacortisol, interleukins and abnormalities in amino acids [5, 7, 8]. However, few studies have attempted to or been able to identify thepathogenesis of delirium following cardiac interventions, although two recent importantstudies suggest an association with raised postoperative cortisol levels [9, 10], whilst Plaschke and colleagues have additionally implicated increased levelsof IL-6 [10]. These authors hypothesize that the increased cortisol level is a stressmarker. However, although current thinking implicates cortisol and cytokineabnormalities in both MDD and cognitive impairment, neither of these was screened for inthe studies cited above [9, 10]. As such, the precise delineation as to whether this was related to surgicalstress rather than additional neuropsychiatric comorbidities remains unclear. Thefailure to assess for comorbidities such as MDD, cognitive impairment and impairedexecutive function may therefore represent a confound in the accurate interpretation ofprior studies.

In light of this, the primary objective of the current study was to investigate theassociation between preoperative and postoperative plasma cortisol concentrations andthe development of postoperative delirium. The secondary objective was to assess whetherany association between cortisol and delirium is stress related or mediated by way ofMDD or cognitive impairment. We hypothesized that: delirium after CABG surgery isindependently associated with increased preoperative cortisol levels; these raisedcortisol levels may be related to pre-existing conditions, such as MDD, cognitivedisturbances and aging; increased reactivity of the hypothalamus-pituitary-adrenal (HPA)axis associated with MDD results in a greater cortisol response postoperatively ascompared with patients without MDD; and patients with MDD are at a greater risk ofdelirium postoperatively as a consequence of these mechanisms.

Materials and methods


The study was approved by the Ethics Committee of the Medical University of Lodz,Poland and was performed in accordance with the ethical standards of the Declarationof Helsinki. The study was conducted in the 14-bed cardiac surgical intensive careunit (ICU) of a university teaching hospital (University Hospital, Central VeteransHospital, Poland) between May and September 2011. The subjects signed an informedconsent the day before their operation. The inclusion criteria were: consecutiveadult patients scheduled for CABG surgery with cardiopulmonary bypass. The exclusioncriteria were as follows: concomitant surgery other than CABG; history of adrenalgland disease; history of glucocorticoid therapy within the last year;non-Polish-speaking subjects; illiteracy; and patients with pronounced hearing and/orvisual impairment.

Preoperative psychiatric and psychological procedures

The study population was examined by a psychiatrist (JK) on the day prior to thescheduled operation using the Montreal Cognitive Assessment (MoCA) and the TrailMaking Test Part B (TMT-B) to assess global cognition, and executive functions,respectively. The Mini International Neuropsychiatric Interview was additionallyemployed to assess for a diagnosis of MDD.

The MoCA was designed as a rapid screening instrument for mild cognitive dysfunction.This instrument assesses different cognitive domains: attention and concentration,executive functions, memory, language, visuoconstructional skills, conceptualthinking, calculations, and orientation [11]. The TMT-B is a widely used paper-and-pencil task that evaluates theexecutive functions and cognitive flexibility [12]. The Mini International Neuropsychiatric Interview is a structureddiagnostic interview, developed jointly by psychiatrists and clinicians in the UnitedStates and Europe for The Diagnostic and Statistical Manual of MentalDisorders, Fourth Edition and for International Classification of Diseases,Tenth Revision psychiatric disorders [13].

Anesthesia and surgery

For premedication, midazolam 7.5 mg per orally 1 hour before surgery was used. Beforeinducing anesthesia in the patients, routine monitoring was installed:electrocardiography leads II and V5, invasive radial arterial blood pressuremonitoring, central venous pressure monitoring, cerebral oxygen saturation, andperipheral oxygen saturation. A standard anesthesia technique was used for allpatients. Induction of anesthesia involved fentanyl 5 to 10 μg/kg, midazolam 0.1to 0.15 mg/kg, and rocuronium 0.6 to 0.8 mg/kg. Medication during maintenance was asfollows: fentanyl in continuous intravenous infusion of dose 1 to 2 μg/kg,midazolam (0.1 to 0.2 mg/kg), and interrupted doses of rocuronium. Ventilation wasprovided with a breathing mixture of FiO2 0.5 and air to maintainend-tidal carbon dioxide at 35 mmHg. From surgical incision to cardiopulmonary bypassconnection, sevoflurane 0.5 to 1.5 vol.% was used. Intraoperative monitoringadditionally included end-tidal expiratory carbon dioxide, nasopharyngealtemperature, bladder temperature, and urine output. Pulmonary artery catheter wasinserted when necessary. In cases of hypotension, norepinephrine was employed tocounteract profound vasodilatation, at a rate 0.05 to 1 μg/kg/minute, tomaintain mean arterial pressure above 60 mmHg.

All patients underwent CABG surgery through a median sternotomy. During the studyperiod, the surgical and cardiopulmonary bypass procedures remained similar. Thepatients were operated on under normothermia using antegrade cold crystalloid StThomas' Hospital cardioplegic solution No. 2 (4 to 6°C). After surgery, allpatients were transferred to the ICU and were placed on mechanical ventilation. Untilextubation, 102 (90%) study patients were sedated with midazolam in continuousinfusion of 0.075 to 0.2 mg/kg/hour, plus additional interrupted doses of 0.1 to 0.2mg/kg morphine, while the remaining participants were sedated with propofol perfusionat a rate of 1 to 2 mg/kg/hour, targeting Ramsay Sedation Scale scores of -4 to -5.The acceptable levels of arterial blood gases (maintained at pH 7.35 to 7.45, partialpressure of carbon dioxide in arterial blood 35 to 45 mmHg, partial pressure ofoxygen in arterial blood > 90 mmHg) and oxygen saturations > 90% were necessarycriteria for extubation. Additionally, patients were required to be awake andcooperative, hemodynamically stable with a body temperature > 36.5°C (preferablynormothermic), have no active bleeding (< 400 ml/2 hours) nor coagulopathy and todemonstrate a return of muscle strength (defined by > 5 seconds head lift/strong handgrip). As a standard, weaning from mechanical ventilation and extubation inuncomplicated cases took place 4 to 6 hours after the operation.

Intraoperative and postoperative measures were recorded on the basis of localprotocols pertaining to postoperative management of patients on the cardiac ICU. Thelowest intraoperative hemoglobin concentration was entered into current analysis.During the surgery and postoperatively, the presence of atrial fibrillation wasrecorded by 24-hour electrocardiography monitoring. One-time and multiple increasesof partial pressure of carbon dioxide in arterial blood ≥ 45 mmHg and a drop ofpartial pressure of oxygen in arterial blood ≤ 50 mmHg were recorded andentered into the analysis.

Measurement of serum cortisol and IL-2 levels

The venous blood samples were taken twice during the study period: the day prior tothe surgery (baseline measurement) and on the first postoperative day, between thehours 08:00 and 09:00 a.m. The blood samples were centrifuged at 7,000 rpm for 10minutes and were refrigerated for a maximum of 1 month at -20°C untilbiochemical parameters were determined. The serum cortisol concentration was measuredwith a competitive electrochemiluminescent enzyme immunoassay in a calibrated Elecsys2010 analyzer (Roche Diagnostic GmbH, Mannheim, Germany). The normal range of plasmacortisol according to the laboratory where measurements were performed is 171 to 536nmol/l in the morning and 63 to 327 nmol/l in the evening. The venous blood samplesfor IL-2 were taken on the first day postoperatively, between the hours of 08:00 and09:00 a.m. The serum IL-2 concentration was determined by chemiluminescentimmunoassay technology. The normal concentration of plasma IL-2 according to thelaboratory where the measurement was performed is < 710 U/ml. The tests wereconducted by investigators that were blinded to clinical data.

Delirium diagnosis

None of the patients had preoperative delirium while being assessed according to theConfusion Assessment Method. Following surgical interventions, the ConfusionAssessment Method for the Intensive Care Unit was used to diagnose delirium [14]. Each individual was assessed by one of the study psychiatrists twice aday (from 08:00 to 10:00 a.m. and from 08:00 to 10:00 p.m.) within the first 5 daysafter surgery. Before each administration of the Confusion Assessment Method for theIntensive Care Unit, the level of sedation/arousal was assessed using the RichmondAgitation Sedation Scale [15]. If the patient was deeply sedated or was unarousable (-4 or -5 on theRichmond Agitation Sedation Scale), evaluation was stopped and repeated later. If theRichmond Agitation Sedation Scale was above -4 (-3 through +4), assessment with theConfusion Assessment Method for the Intensive Care Unit was administered.

Statistical analysis

Quantitative variables are expressed as medians and interquartile ranges (IQRs). Forcategorical variables, the number of observations (n) and fraction (%) werecalculated. Normality was tested using the Shapiro-Wilk's test for normality.Differences between two independent samples for continuous data were analyzed usingthe Mann-Whitney U test (since the distributions of variables were different fromnormal).

For categorical variables, statistical analysis was based on the chi-squared test orthe chi-squared test with Yates' adjustment. Spearman's rank correlation coefficientswere calculated to assess the correlation between two quantitative variables. Theminimum study sample size was calculated using the power analysis, estimating theexpected effects from the pilot data and assuming an alpha level of 0.10 and a powerof 80% (minimum sample size for each group is 37 patients).

Distributions for postoperative cortisol levels were different from normal in bothdepression and nondepression groups (P < 0.001). Similarly, theassumption of homogeneity of variance was not satisfied for postoperative cortisollevels (P < 0.01). The nonparametric Friedman's version of analysis ofvariance was thus used to compare cortisol before and after CABG surgery consideringdepression. Initially, baseline and perioperative variables were evaluated forunivariate association with postoperative delirium. For quantitative variables(preoperative and postoperative cortisol concentration), significantly associatedwith the occurrence of delirium, receiver operating characteristic curves were drawnand decision thresholds were found. The sensitivity, specificity, positive predictivevalue and negative predictive value were calculated. Odds ratios with 95% confidenceintervals and standard errors were also presented. Factors significant in univariatecomparisons (P < 0.10) were included in a forward stepwise logisticregression model to identify the set of the independent risk factors for delirium.The results were considered significant for P < 0.05. All of thecalculations were performed using STATISTICA (version 9, 2009; StatSoft, Inc., Tulsa,OK, USA) and SPSS (SPSS Statistics, version 19; IBM, Armonk, NY, USA) software.


One hundred and eighty-two patients underwent CABG surgery during the study period; ofthese, 59 subjects did not meet the inclusion criteria (Figure 1).Baseline demographic characteristics and patients' comorbidities are presented in Table1. Postoperative delirium developed in 36% (41 of 113) ofpatients. The median duration of delirium was 3.5 days (IQR = 2 to 4). The frequency ofdiagnosis of delirium decreased with an increasing number of postoperative days (day 1,n = 22, 54%; day 2, n = 13, 32%; day 3, n = 4, 10%; day 4,n = 1, 2%; day 5, n = 1, 2%). Patients with postoperative deliriumhad a significantly longer stay in the ICU (6 vs. 2 days; P < 0.0001) and alonger total duration of hospitalization (19 vs. 11 days; P < 0.0001)compared with patients who did not develop delirium.

Figure 1
figure 1

Number of patients excluded and included in the data analysis. CABG,coronary artery bypass graft.

Table 1 Demographic characteristics and comorbidities of all 113 patients enrolled in thestudy

The results of the univariate analysis of variables related to the condition ofparticipants, anesthesia and surgical procedures are shown in Tables 2, 3, and 4. The unadjusted riskof postoperative delirium was higher both for patients with increased preoperative andpostoperative cortisol concentrations (odds ratio = 1.004, P = 0.006; oddsratio = 1.002, P < 0.0001, respectively). Subjects with higher preoperativeand postoperative cortisol level remained at increased risk of developing delirium aftercontrolling for the following variables significant in univariable analysis: age,gender, cognitive performance (MoCA and TMT-B scores), preoperative urea, creatinine,hemoglobin concentration, peripheral vascular disease, duration of surgery, dose ofmidazolam, intraoperative hemoglobin level, partial pressure of oxygen, partial pressureof carbon dioxide, atrial fibrillation, and IL-2 concentration. However, aftercontrolling for preoperative depression, only preoperative cortisol concentrationremained significant, irrespective of the cortisol level after surgery (Table 5).

Table 2 Biomarkers and variables related to demography and mental condition of patientsanalyzed in univariate analysis
Table 3 Variables related to physical condition of patients analyzed in univariateanalysis
Table 4 Variables related to anesthesia and surgery analyzed in univariate analysis
Table 5 Factors independently associated with delirium after CABG surgery revealed inmultivariate stepwise logistic regression analysisa

According to receiver operating characteristic analysis, the most optimal cutoff valuesthat predict the development of delirium were as follows: preoperative cortisolconcentration ≥ 353.55 nmol/l, with sensitivity of 65.85% and specificity of63.89%, positive predictive value of 50.94% and negative predictive value of 76.67%(odds ratio = 3.41) (area under the curve = 0.66; 95% confidence interval = 0.55 to0.76; standard error = 0.05); and postoperative cortisol concentration ≥ 994.10nmol/l, with sensitivity of 65.85% and specificity of 69.44%, positive predictive valueof 55.10% and negative predictive value of 78.13% (odds ratio = 4.38) (area under thecurve = 0.72; 95% confidence interval = 0.63 to 0.82; standard error = 0.05).

The median preoperative and postoperative cortisol concentrations in the wholepopulation were 335.6 nmol/l (IQR = 247.5 to 459.5) and 940.7 nmol/l (IQR = 783.8 to1,273), respectively. According to the Mann-Whitney U test, the median preoperativecortisol concentration was higher in patients with depression than in nondepressionsubjects: 483.1 nmol/l (IQR = 388.4 to 612.3) versus 318.8 nmol/l (IQR = 235.5 to 439.8)(P = 0.001), respectively. Similarly, the median postoperative cortisolconcentration was higher in patients with depression compared with the nondepressiongroup: 1,194.5 nmol/l (IQR = 936 to 1438) versus 908.4 nmol/l (IQR = 709 to 1,256)(P = 0.009), respectively. However, according to nonparametric analysis ofvariance, the interaction between the presence of depression and preoperative andpostoperative cortisol concentration was not statistically significant (P =0.447). This suggests that the postoperative cortisol concentration was higher than thepreoperative, regardless of depression occurrence. The Spearman's rank correlationcoefficients between preoperative cortisol and MoCA scores and between postoperativecortisol and MoCA scores were -0.21 (P = 0.025) and -0.14 (P = 0.130),respectively. The Spearman's rank correlation coefficients between preoperative cortisoland age and between postoperative cortisol and age were 0.18 (P = 0.049) and0.25 (P = 0.007), respectively.


This study investigated the impact of increased preoperative and postoperative cortisolconcentration in relation to a diagnosis of preoperative MDD and cognitive impairment onthe risk of developing postoperative delirium.

Among 113 patients undergoing CABG, 36% (41) developed delirium. The effect of bothpreoperative and postoperative cortisol concentration on the risk of developing deliriumwas significant after controlling for demographic, physical, cognitive, surgical andanesthetic-related factors. However, when cortisol levels were controlled for MDD, onlythe preoperative cortisol concentration remained significant. The final multivariateregression analysis revealed that the preoperative cortisol level, MDD, impairedexecutive functions, higher creatinine and IL-2 concentrations and a higher dosage ofmidazolam independently increase the risk of postoperative delirium.

The incidence of postoperative delirium reported in the present study is in line withfindings of similar contributions related to cardiac surgery (the reported estimatesvary from 3 to 50%) [1618]. In our previous study conducted in cardiac surgery patients [19], however, the incidence of delirium was lower (11.5%) compared with thosecurrently reported (36%). This discrepancy may be due to differences in the groupsstudied, diagnostic approaches and the assessment tools used. In the first contribution,The Diagnostic and Statistical Manual of Mental Disorders Fourth Editioncriteria were used to diagnose delirium and the participants were younger (mean age 62years). Moreover, screening for delirium was conducted once a day starting from thesecond postoperative day. All of these factors may have affected the finalestimates.

In the current study, the incidence of MDD and cognitive impairment (MoCA score < 25)were 16% and 32%, respectively. This is higher than the prevalence of unipolardepression in the general population (5 to 9% for females, 2 to 3% for males) and isconsistent with the prevalence of MDD in CABG patients reported in other studies (15 to20%) [2022]. Recent studies confirmed that cognitive impairment is common among olderpatients undergoing major surgery, including cardiac interventions [5, 6, 18, 23]. Depending on the diagnostic measures employed, the prevalence of cognitivedisturbances ranges between 17 and 44%, and this is consistent with the resultspresented here (32%) [5, 6, 18]. Worthy of notice is that our previous study revealed cognitive disturbancesin 100 out of 563 patients (17%). In the current study, however, we used more sensitivediagnostic instruments, which additionally enabled us to diagnose milder forms ofcognitive impairment [11].

A number of studies have investigated risk factors for postoperative delirium, butfindings have been heterogeneous. This inconsistency of the results may be, in part,related to the multifactorial etiology of delirium. Unfortunately, the pathophysiologyand biological processes underlying this neuropsychiatric syndrome are poorly described- but it is possible that different mechanisms involved in delirium act through the samefinal pathways. This might explain the heterogeneity of research findings.

The most consistently reported independent associations with delirium in recent studiesof cardiac surgery to date include older age, preoperative cognitive impairment and MDD [5, 6, 18]. MDD and advanced age may contribute to delirium through the elevated levelof cortisol, secondary to activation of the limbic HPA axis in such individuals [7]. In this model, glucocorticoids inhibit endothelial cell proliferation andturnover in the hippocampus and prefrontal cortex [24], whilst HPA axis dysregulation results in decreased hippocampus volume [25]. Two recent studies investigated the association between plasma cortisol anddelirium among cardiac surgery patients [9, 10]. Plaschke and colleagues reported the association between increased cortisolconcentration and delirium among a heterogeneous population of cardiac surgery patientsin a univariate analysis that did not control for other factors - as such, theassociation with and significance of raised cortisol levels were not determined [10]. Mu and colleagues showed an independent association between elevatedcortisol levels and postoperative delirium in individuals who underwent CABG surgery [9]. Both of these groups propose that increased cortisol concentration may be amarker of stress response, with the caveat that surgery-related stress is probably notthe only factor contributing to elevated cortisol levels. Neither group was able todetermine whether hypercortisolemia was a cause or an effect of postoperative deliriumin the absence of baseline cortisol measurements, samples only being collectedpostoperatively. Furthermore, preoperative screening for potentially confoundingneuropsychiatric disorders that were associated with altered cortisol levels and withdelirium were not performed.

According to the results of present study, major depression prior to surgery is stronglyand independently associated with an increased risk of postoperative delirium.Interestingly, high postoperative cortisol level also increases the risk of delirium,but this association lost significance once preoperative MDD was controlled for.Moreover, according to univariate analysis, the concentration of cortisol after surgeryis significantly higher among patients suffering from depression when compared withnondepression subjects. These data suggest that, regarding delirium, depression is theprimary factor affecting the condition of the ICU patients. Hypercortisolemia may be thefactor that mediates the impact of MDD on postoperative cognition. This interpretationshould be treated with caution, however, since the postoperative cortisol concentrationwas higher than the preoperative one regardless of depression occurrence, according toanalysis of variance. On the contrary, our analysis revealed that a higher cortisolconcentration measured the day prior to surgery independently increases the risk ofdelirium, even after controlling for depression, cognitive performance and age. Theconcentration of preoperative cortisol was higher among individuals with depressioncompared with patients without this diagnosis; however, this difference was observedonly in univariate comparisons. MDD and the associated increase in HPA axis reactivityand postoperative hypercortisolemia is probably not the only pathophysiologicalmechanism involved in the development of postoperative delirium. For example, a higherpreoperative cortisol concentration may be another contributing factor. Unfortunately,the etiology of preoperative hypercortisolemia is unknown. It may be linked topreoperative MDD or reflect other, separate and undiscovered pathologies. According torecent publications, an increased cortisol level carries a predictive value in thedevelopment of mild cognitive impairment [26]. Moreover, higher cortisol measures have also been reported in Alzheimer'sdisease and are associated with poorer memory performance in subjects with cognitivedecline [27, 28] and alterations in HPA axis activity frequently accompany aging [29].

The current analysis revealed that delirium was significantly more frequent amongpatients with advancing age and with lower MoCA scores. However, older age and lowerMoCA scores did not maintain significance in a multivariate analysis. Older patientssignificantly differed from younger participants in relation to the both preoperativeand postoperative cortisol concentration. Furthermore, there was a correlation betweenlower MoCA scores and higher preoperative cortisol level. These findings suggest thathigher cortisol levels prior to surgery that act as an independent risk factor forpostoperative delirium may be associated with advanced age and the impaired cognitiveperformance in these participants.

Strengths and limitations

This study has several advantages. The study population was homogeneous, and thesubjects were consecutive, prospectively enrolled and examined by an experienced,well-trained investigator. The analysis included a variety of factors associated withthe mental and physical condition of participants, as well as those related toanesthesia and surgery. Therefore, while investigating the association betweencortisol and delirium, both traumatic stress-related and psychiatric pathways weretaken into consideration. To the knowledge of the authors, this represents the firststudy to investigate whether hypercortisolemia is a cause or an effect of deliriumafter cardiac surgery. This in turn allowed the analysis to control for possibleconfounders such as MDD and cognitive impairment, as well as factors associated withanesthesia and surgery (duration of surgery and aortic cross-clamping, dose ofmidazolam).

However, the study is not without limitations. The present findings cannot beconsidered definitive since other circulating hormones, mediators and inflammatoryfactors were not included in the analysis. In addition, not all prescribedmedications and anesthetic agents were taken into account in the analysis, whichfocused on the association between the dose of midazolam and delirium. We decided toinclude midazolam into the analysis since this medication may decrease the level ofcortisol perioperatively [30, 31]. Moreover, the association between midazolam and postoperative deliriumhas been frequently reported [32, 33]. However, the impact of other anesthetic agents that may play a role indelirium development, post-surgery sedation, as well as the impact of postoperativecomplications on the incidence of delirium was not assessed in this study. This beingsaid, a recent study suggested that none of 20 different drug classes investigated(including antihypertensives, diuretics, antiplatelets and psychiatric agents) wereassociated with delirium after elective surgery [34].


On the basis of the current analysis we can conclude that patients with raised levels ofcortisol prior to surgery are at significantly increased risk of postoperative delirium.This higher level of preoperative cortisol may be associated with MDD, aging andcognitive decline.

Secondly, patients with increased HPA axis reactivity secondary to pathologies such asMDD are characterized with higher postoperative cortisol concentrations compared withpatients without MDD and, possibly as a consequence, are more likely to developpostoperative delirium. These observations suggest that an increased level of cortisolmay be a cause rather than an effect of postoperative delirium. Preoperativeneuropsychiatric screening and monitoring of cortisol levels of cardiac surgery patientscombined with postoperative surveillance may improve the early detection of deliriumand, indirectly, the prognosis.

Key messages

  • Cardiac surgery patients with raised concentration of plasma cortisolprior to surgery are at significantly increased risk of postoperative delirium.

  • A higher level of preoperative cortisol may be associated with MDD,advanced age and cognitive impairment.

  • Preoperative diagnosis of MDD is an independent predictor of deliriumafter CABG surgery.

  • Patients with a preoperative diagnosis of MDD have higherpostoperative cortisol levels compared with patients without MDD, which may contributeto the development of delirium postoperatively.