Secondary Logo

Journal Logo

Geriatric Anesthesia: Original Clinical Research Report

Preoperative Salivary Cortisol am/pm Ratio Predicts Early Postoperative Cognitive Dysfunction After Noncardiac Surgery in Elderly Patients

Han, Yuan MD, PhD*,†; Han, Liu MD*; Dong, Meng-Meng MD*; Sun, Qing-Chun MD*; Zhang, Zhen-Feng MD*; Ding, Ke MD*; Zhang, Yao-Dong MD†,‡; Mannan, Abdul MD*,†; Xu, Yi-Fan MD; Ou-Yang, Chang-Li MD§; Li, Zhi-Yong MD; Gao, Can MD, PhD; Cao, Jun-Li MD, PhD*,†

Author Information
doi: 10.1213/ANE.0000000000003740



  • Question: Are preexisting abnormal circadian rhythms that are related to cortisol disorders associated with postoperative cognitive dysfunction (POCD)?
  • Findings:
  • The preoperative salivary cortisol am/pm ratio was significantly associated with the presence of early POCD and may serve as an ideal biomarker to aid in the identification of patients at elevated risk.
  • Meaning: The findings are not only beneficial for improving the understanding of the etiology of POCD but also for identifying surgical candidates at risk for POCD.

Postoperative cognitive dysfunction (POCD) is a common complication that follows noncardiac surgery, particularly in elderly patients.1–3 The reported prevalence of POCD ranges from 8.9% to 46.1%.4 Despite decades of investigation, the pathophysiology of POCD remains elusive,5,6 and its diagnosis is often delayed and requires complicated neuropsychological testing. Thus, it is crucial to explore the mechanism(s) underlying this pathological phenomenon and identify reliable, noninvasive, and convenient biomarkers that might aid in the prediction of this condition and individuals who are at an increased risk of developing it.

Cortisol is one of the most well-studied stress hormones or glucocorticoids (GCs). As documented in both experimental and clinical settings, prolonged high levels of GCs have been associated with abnormal circadian rhythms, trauma, aging, depression,7–12 and, of particular relevance here, increased cognitive impairment in neurodegenerative diseases.13,14 Importantly, clinical studies have reported a positive association between surgery-induced cortisol circadian rhythm disorder and cognitive dysfunction.15–17 This suggests that increased cortisol might be an important biological mechanism underlying cognitive disruption. Surgery in elderly patients who have a reduced ability to cope with daytime stress or disturbed or irregular patterns of sleep may result in lasting changes in cortisol levels. Therefore, exploring the relationship between cortisol and POCD in an aged population could prove to be conducive.

In addition, in the clinical setting, it is markedly more difficult to assess the relationship between POCD and GCs because no sampling method can ever fully evaluate cortisol levels, which fluctuate throughout the day. The biological activity of cortisol in blood is believed to be a function of the small fraction of cortisol that is not bound to the serum protein, corticosteroid-binding globulin. Corticosteroid-binding globulin can be affected by several physiological or pathological conditions such as drugs including salicylates or oral contraceptives, which compete with cortisol for its binding sites.18 Fortunately, repetitive saliva sampling has been demonstrated to validly and reliably reflect unbound cortisol levels and may serve as an ideal method for the purposes of assaying dynamic cortisol levels.19,20

We hypothesized that preoperative salivary cortisol dysregulation would be associated with POCD in elderly patients undergoing major noncardiac surgery. The primary objective of this study was to investigate the relationship between the preoperative salivary cortisol am/pm ratio and the development of POCD. In addition, we evaluated morning and evening salivary cortisol levels and their independent association with POCD.


Study Design

This study was approved by the clinical research ethics committee of the Affiliated Hospital of Xuzhou Medical University, Jiangsu, China (Certification No. XYFY2016-KL019-01, approved date: August 24, 2016), and written informed consent was obtained from all subjects participating in the trial. The participants were recruited from a prospective cohort as registered before patient enrollment at (clinical trial registration number: NCT02992600; principal investigator: J.-L.C.; date of registration: December 14, 2016). The present trial was registered at (number: ChiCTR-ROC-17012008; principal investigator: J.-L.C.; date of registration: July 16, 2017). This article adheres to all applicable CONSORT guidelines.

Subject Enrollment

Nonneurologically impaired, elderly patients (≥60 years of age), who had been referred for major noncardiac and nonneurological surgery under general anesthesia and expected a hospital stay of ≥5 days, were screened and enrolled between December 2016 and July 2017.

Patients were excluded if they met any of the following criteria: significant impairment of vision, hearing, or motor skills (such as hemiplegia); psychological illness or abnormal personality that might prevent them from completing baseline neuropsychological testing; history of neurological disease (eg, senile dementia, Parkinson’s syndrome, multiple sclerosis, schizophrenia, and depression); severe trauma or surgery within the past year; adrenal gland disease; previous GC therapy for >7 consecutive days within the past year; severe physical illness, drug dependence, and/or alcoholism; or vital organ dysfunction. Furthermore, preoperative Mini-Mental Sate Examination (MMSE) scores were classified according to the following point cutoffs: illiteracy as <17 points, primary school as <20 points, and higher secondary education as <24 points.20,21 Those with a Geriatric Depression Scale grade of >2 or who were unwilling to comply with the protocol or procedures were also excluded.

Control Group

In studies that use repeated neuropsychological testing, practice effects and natural variation in cognitive test performances can result in misinterpretation of outcomes.22 In the present study, 30 age- and sex-matched volunteers were recruited from the friends and family members of patients in the trial group using the same exclusion criteria. These volunteers underwent the same neuropsychological testing, by the same investigators, and at the same time intervals as the trial participants, but they had not had any surgical procedures or anesthesia within the past year.

Neuropsychological Testing

Each patient underwent a standard battery of neuropsychological tests 1 day before surgery (baseline) and 1 week after surgery. Testing was performed in a quiet room in the general care wards. Neuropsychological tests were selected on the basis of their recommendation in the 2 International Studies of Postoperative Cognitive Dysfunction (ISPOCD 1 and 2).1,23,24 The neuropsychological testing battery used here was designed to measure memory, psychomotor speed and dexterity, physical motor speed, attentional capacity, and perceptual-spatial functioning. It included 9 tests with 11 subscales. The specific tests used were as follows: the Short Story Module of the Randt Memory Test (immediate and delayed recall), Verbal Fluency Test, Trail Making Test (Part A), Digit Symbol Subtest of the Wechsler Adult Intelligence Scale-Revised (Chinese edition), Digit Span (forward and backward) Subtests of the Wechsler Memory Scale (Chinese edition), finger-tapping task, Grooved Pegboard Test (dominant and nondominant hand), and Block Test. The cognitive domains covered by these different tests are listed in Table 1.

Table 1.
Table 1.:
Cognitive Domains and Neuropsychological Tests

Research investigators were extensively trained to administer this specific battery of tests as well as in relevant interview techniques by an experienced psychiatrist. These investigators then conducted all neuropsychological tests, referring to a strict and standardized written test protocol to minimize interexaminer variability. Cognitive assessments for each patient were repeated after surgery by the same examiner. The full testing battery required approximately 40 minutes to complete.

Salivary Cortisol Levels

Saliva samples were collected via the Salivette device (Sarstedt, Rommelsdorf, Germany) the day before surgery.25 Participants were instructed to take the first sample 45 minutes after waking in the morning. The second sample was then taken in the evening on the same day immediately before going to sleep. In general, the morning samples were collected between 6:00 and 7:30 am and the evening samples were collected between 9:30 and 11:00 pm. Participants were instructed not to brush their teeth before each sampling, to refrain from eating and drinking until after the morning sample was taken, and to allow for at least 30 minutes between eating or drinking and providing the evening sample, as described previously.10 Briefly, saliva collection procedures involved patients chewing on a cotton plug (Cortisol Salivette, 51.1534.500, Sarstedt, Germany), which was developed specifically for cortisol assessments, for 2–5 minutes, which stimulated saliva production such that the plug absorbed a sufficient amount of saliva to become moist. The cotton plugs were then placed in salivettes (sample collection tube, Sarstedt) and stored at 4°C. The salivettes were then centrifuged, and the centrifuged samples were stored at −80°C until further assessment.

To assess salivary cortisol levels, the Roche Cobas Cortisol assay (German, Catalog Number: 11875116 122), which is a competitive electrochemiluminescence immunoassay, was performed using Cobas 8000 (Roche Diagnostics, Mannheim, Germany). Briefly, this procedure involves incubation of a 20-μL sample with a cortisol-specific biotinylated antibody and a Tris (2,29-bipyridyl) ruthenium(III)--labeled cortisol derivative. After this, streptavidin-coated paramagnetic microparticles are added because biotinylated antibodies bind to streptavidin on the microparticles. This complex is magnetically captured by the surface of an electrode in an individual measurement cell. Application of current to the electrode induces chemiluminescence of the ruthenium complex, and this light signal is quantified by a photomultiplier. The lower quantitation limit of this method is 0.5 nmol/L. Pools of assays were discarded if the quality controls were >2 standard deviations (SDs) from the accepted mean. The intracoefficients of variation for 4.68, 11.50, 15.10, and 19.80 nmol/L of the standard cortisol samples were 6.1%, 2.7%, 4.0%, and 2.8%, respectively. The intercoefficients of variation for 8.05, 13.1, 34.6, and 42.5 nmol/L of the standard cortisol samples were 11.5%, 7.1%, 4.9%, and 4.1%, respectively. Assays were discarded if their internal quality controls were >2 SDs from the accepted mean.

To determine the morning and evening salivary cortisol values for each patient, we calculated a ratio between them, the so-called am/pm ratio.

Primary and Secondary Outcomes

The primary outcome was the presence of POCD. The primary objective of this study was to assess the relationship between the ratio of am (morning) to pm (evening) salivary cortisol level and the presence of POCD. The secondary objective was to assess the relationship between POCD and salivary cortisol absolute values in the morning or in the evening.

Calculation of POCD

POCD was defined here according to the same ISPOCD 1 definition utilized in previous studies.1,23,24 The frequency of POCD was calculated using the following formula: (ΔXi − ΔXC)/SDΔXC. The change was calculated by subtracting the preoperative score from the postoperative score, yielding ΔXi for each participant for a given task. The averaged difference for the controls, or ΔXC, which was assumed to represent the systematic error, was also calculated in the same manner. ΔXC was then subtracted from the individual change (ΔXi) to account for any practice effects. This score was then divided by the control group’s SD of the change (SDΔXC) to control for expected variability. Individuals with POCD were defined as those with a Z-score of ≤−1.96 on at least 2 different tests (20% of the individual neuropsychological tests).

Anesthesia Protocols

Anesthesia protocols were standardized, and no anticholinergic agents were used to premedicate the patients. On arrival, all patients received standard monitoring, including electrocardiogram, blood pressure, oxygen saturation by pulse oximeter, end-tidal carbon dioxide, temperature, and spectral entropy indices. Patients were administered midazolam to alleviate anxiety and etomidate, cisatracurium, and fentanyl to induce anesthesia. Approximately 5 minutes after drug delivery, patients were intubated and ventilated to maintain a end-tidal carbon dioxide of 35 ± 5 mm Hg. Anesthesia was maintained via inhaled sevoflurane, propofol, intravenous remifentanil, and cisatracurium. During anesthesia maintenance, the spectral entropy index was maintained between 40 and 60. All aspects of clinical care were documented in each patient’s electronic medical record.

Statistical Analyses

The normality assumption was assessed using the Kolmogorov–Smirnov test in all analyses. Continuous variables were presented as a mean (SD) or median (interquartile range), and categorical variables were presented as the number of patients (%). The approximate normal distribution method was used to calculate the binomial 95% confidence interval (CI). Group comparisons were made using 2 independent sample t tests for continuous variables with a normal distribution, the Mann-Whitney U test for continuous variables with a nonnormal distribution, or the χ2 test or Fisher exact test for dichotomous and ranked data. Associations between salivary cortisol secretion levels and the occurrence of early POCD were determined using a multivariable logistic regression analysis.

The forward likelihood ratio method was used in multivariable logistic regression. We included the morning, evening, and am/pm salivary cortisol ratio, significant perioperative variables (P ≤ .1), as well as the well-recognized POCD risk factors such as preoperative MMSE score, age, sex, and education levels in existing studies1,24 in the multivariable logistic regression model. At first, 7 variables were involved in the univariable analysis for calculating odds ratio (OR). The univariable P value of morning value, am/pm ratio of salivary cortisol, and MMSE is <.05. Then, the variance inflation factor was used to assess the multicollinearity between morning and am/pm ratio, whose value was 1.16. Thus, only am/pm ratio and MMSE was included in the following multivariable logistic regression model.

The am/pm salivary cortisol ratio, which allowed for determination of whether patients developed POCD, was assessed using receiver operating characteristic (ROC) curve analysis. The optimal cutoff value was defined as the point of the am/pm ratio exhibiting the greatest sum of sensitivity and specificity. All hypothesis testing was 2 tailed. P < .05 was considered to indicate significance.

All statistical analyses were completed using SPSS, version 21.0 (IBM, New York, NY). A power analysis addressed the primary hypothesis that the baseline am/pm salivary cortisol ratio is significantly different in the 2 groups of patients studied here; those who developed cognitive dysfunction 1 week after the surgery and those who did not. A sample size calculation was performed using PASS (version 11.0; NCSS, Kaysville, UT) and a 2 independent samples t test and a common SD value of 2. According to previous studies and our own preliminary testing, we assumed that POCD occurred in 20% of patients and that the am/pm salivary cortisol ratio would be 5 in the POCD patients and 3.5 in the non-POCD patients. On the basis of a .05 level of significance with a power of 0.80, we sought to enroll at least 90 patients in the investigation to achieve sufficient statistical power. To compensate for lack of follow-up data, we aimed to recruit 120 patients.


Study Population

During the study period, 2761 patients were assessed for eligibility, of whom 879 satisfied the inclusion criteria. Among those screened, 759 patients were excluded from the study for various reasons (Figure 1). In total, 120 qualified patients provided their written informed consent and were enrolled in the study. During the postoperative period, 19 patients refused to undergo postoperative neuropsychological testing and 7 failed to provide saliva samples. As a result, these 26 patients were also excluded from the final data set. Thus, a total of 94 patients were included in the final data analyses. A flow chart of patient enrollment in the study is shown in Figure 1.

Figure 1.
Figure 1.:
Enrollment flow chart for the study population.

The control group consisted of 30 age- and sex-matched volunteers. The average age of the controls, of whom 40.0% were women, was 70.17 years. Supplemental Digital Content 1, Appendix Table 1-e,, summarizes the demographic and clinical characteristics of all participants.

Cognitive Outcomes

POCD, which was diagnosed using the same definition as used in ISPOCD 1 and the other previous studies,1,23,24 was detected in 16 of 94 patients (17.02%; 95% CI, 9.28%–24.76%) 7 days postoperation. Patient’s baseline demographic, clinical, and surgical characteristics are presented in Table 2. The MMSE scores were lower in the POCD group (P = .045), and there were no significant demographic differences between the 2 groups. Postoperative neuropsychological testing was performed after 7.31 ± 1.89 (95% CI, 6.31–8.32) days in patients with POCD and 7.26 ± 1.59 (95% CI, 6.90–7.61) days in patients without POCD (Table 2). The results of neuropsychological testing across all tasks in both groups are presented in Supplemental Digital Content 2, Appendix Table 2-e, There were no significant differences in performance on each individual neuropsychological test task between the POCD and the non-POCD groups.

Table 2.
Table 2.:
Demographic, Clinical, and Surgical Characteristics

The neuropsychological test results from all patients at the 1-week follow-up are listed in Supplemental Digital Content 3, Appendix Table 3-e, Patients with POCD performed worse primarily in the Short Story module of the Randt Memory (delayed recall) and in the finger-tapping task. The neuropsychological testing scores among the control group at baseline and 1 week later are listed in Supplemental Digital Content 4, Appendix Table 4-e,

Primary Outcome: Relationship Between the am/pm Salivary Cortisol Ratio and POCD

The median (interquartile range) am/pm salivary cortisol ratio was significantly higher (P = .006; Table 3) in patients who developed POCD (5.16 [2.31–8.27]; 95% CI, 3.85–7.58) than in those who did not (2.60 [1.68–4.39]; 95% CI, 2.81–3.57). The multivariable logistic regression analysis identified that a high preoperative am/pm salivary cortisol ratio remained an independent predictor of the occurrence of POCD 7 days postoperation even after adjusting for MMSE scores (OR, 1.55; 95% CI, 1.19–2.02; P = .001; Table 4).

Table 3.
Table 3.:
Salivary Cortisol Levels by Neuropsychological Testing
Table 4.
Table 4.:
Associations Between Salivary Cortisol and POCD
Figure 2.
Figure 2.:
Receiver operating characteristic curve for predicting the incidence of postoperative cognitive dysfunction after noncardiac surgery in elderly patients. Receiver operating characteristic curve analysis was performed for the am/pm cortisol ratio data. ↓The area under the curve for the am/pm cortisol ratio was 0.72. The optimal cutoff value was found to be 5.69 with a sensitivity of 50% and specificity of 91%. ∇Setting the am/pm cortisol ratio cutoff at 7.6 resulted in a sensitivity of 25% and a specificity of 99%. *Setting the am/pm cortisol ratio cutoff at 1.66 resulted in a sensitivity of 94% and a specificity of 23%.

ROC curve was created for the am/pm salivary cortisol ratio. The area under the ROC curve of the duration of the salivary cortisol am/pm ratio in the POCD group was 0.72 (95% CI, 0.56–0.88; P = .006). The optimal cutoff value was found to be 5.69 with a sensitivity of 50% and specificity of 91% (Figure 2).

Secondary Outcomes: Salivary Cortisol Absolute Values and POCD

The median (interquartile range) morning salivary cortisol level was 15.93 nmol·L−1 (9.97–23.29; 95% CI, 12.32–22.58) in POCD patients and 12.47 nmol·L−1 (7.96–16.79; 95% CI, 11.63–14.68) in non-POCD patients. The median (interquartile range) evening salivary cortisol level was 3.18 nmol·L−1 (1.68–5.30; 95% CI, 2.27–6.49) in POCD patients and 4.23 nmol·L−1 (2.91–6.19; 95% CI, 4.18–5.39) in non-POCD patients. The morning cortisol levels were significantly higher in patients who developed POCD than in those who did not (P = .035; Table 3). Although the OR of the median morning cortisol value was 1.07 (95% CI, 1.00–1.15; P = .042), no significant difference was found by multivariable logistic regression analysis. In addition, there was no statistically significant difference found in the evening cortisol levels between the 2 groups. We also found that the morning salivary cortisol concentrations were significantly higher than the evening concentrations in both groups (P < .001; Table 3).


In this study, elevated preoperative salivary cortisol levels were associated with an increased risk of cognitive dysfunction after surgery in elderly patients. Critically, we found that the am/pm cortisol ratio, rather than simply the morning or evening cortisol levels, had a strong relationship with cognitive decline, even after accounting for several established risk factors for POCD. Our findings indicate an association between preexisting, preoperative neuroendocrine disorders and POCD and also improve the scientific understanding of the etiology of POCD. The clinical implications of our findings are that salivary cortisol testing may allow for the prediction of the risk of cognitive decline after surgery.

Cortisol in Diurnal Rhythms and Memory Impairment

Cortisol, a multifunctional GC hormone released by the hypothalamic–pituitary–adrenal (HPA) axis, exhibits a marked diurnal rhythm secretion pattern. This pattern is characterized by a rapid increase in cortisol levels on waking, which peaks approximately 30–45 minutes after waking,26 followed by a later decline and reaching a cycle nadir in the evening.27 Given its liposoluble characteristics, cortisol can easily cross the blood–brain barrier and access the brain where it binds to receptors in the hippocampus, amygdala, and frontal lobes, among other regions. Critically, these are among the important brain areas involved in learning processes.

While short-term, stress-induced cortisol release may be adaptive and result in learning and the acquisition of behavioral strategies for coping with stress, chronically high levels of GCs can cause both cognitive and emotional impairments as well as structural changes to the brain.8,28 For example, a cross-sectional analysis involving 1140 Baltimore, Maryland, residents aged 50–70 years reported that elevated cortisol was associated with poorer cognitive function.29 This finding is consistent with that of another study with a larger study cohort of 4244 participants.10 In addition, high cortisol levels can impair cognitive function and may eventually lead to the development of dementias, including Alzheimer’s disease.30

Consistent with these findings, we found that the participants who had preexisting lower MMSE scores and steeper am/pm salivary cortisol ratios experienced greater cognitive decline after surgery. A similar relationship between diminished cognitive function and POCD has also been reported elsewhere in the literature.31

Association Between POCD and Cortisol

We found that a steeper am/pm ratio was strongly related to POCD, but the morning absolute value was not. There are 2 potential explanations for this finding. First, compared to the morning absolute value, the am/pm cortisol ratio is determined by the morning and evening absolute values, and thus, it can normalize the levels of cortisol among different individuals. Second, whereas the morning cortisol value has been implicated in recovery from sleep, providing an “energetic boost” for cognitive function and regulation of the immune system, the am/pm cortisol ratio is thought to reflect healthy HPA axis function and is associated with better health outcomes.32–34

In our study, a steeper am/pm cortisol ratio indicates that an individual’s HPA axis may overrespond and release more cortisol when coping with trauma (ie, surgery), thus impairing their cognitive function. Although we did not test for postoperative changes in cortisol levels, there are several studies that support an association between elevated levels of cortisol and POCD. For instance, the HPA axis response to major surgery includes a robust increase in cortisol levels and flattening of the am/pm cortisol secretion ratio, which has also been associated with early cognitive impairment.5,23 Moreover, excess exogenous GCs may result in selective declarative memory impairments and deficits in hippocampal and temporal lobe function. In humans, administration of the exogenous GC dexamethasone may increase the incidence of POCD.35,36 Further, abnormal circadian cortisol secretion rhythms, which are reflected by the am/pm ratio, appear to be more closely related to postoperative cognitive impairment in elderly populations.9,37

Reasons for the Conflicts Between Our Results and Previous Findings

Our results demonstrated that preexisting elevated cortisol levels were significantly associated with POCD, a result that was not consistent with a previous study.23 Reasons for this discrepancy may include the fact that in this previous study, the am/pm cortisol ratios were not quantified relative to key sleep/wake events throughout the day, as was the case here. Rather, this group used “fixed time point slopes,” in which samples were gathered at fixed times throughout the day (8:00 am and 4:00 pm), rather than in relation to the individual’s time of waking. Here, we used “peak-to-bed slopes,” which examine an absolute change or rate of change in cortisol from the peak of an individual’s cortisol waking response to their bedtime level. Because these peak-to-bed slopes are relative to each individual’s sleep–wake schedule, it is likely that these are more robust measures of the change in cortisol levels in an individual, as previously suggested.33 An additional difference between our study and this previous work is that the previous study included patients who underwent surgical procedures under both general and regional anesthesia, whereas we only enrolled participants who underwent general anesthesia. A meta-analysis based on 21 trials reported that general anesthesia, compared to other forms of anesthesia, may increase the risk for developing POCD, particularly in otherwise cognitively vulnerable individuals such as the elderly.38 This may be another important explanation for the negative results reported in the previous study.

Advantages of the Salivary Cortisol Test

In the present study, we used a salivary cortisol test rather than a more invasive blood test. We assessed samples using an electrochemiluminescence immunoassay kit, which is the most accurate, sensitive, and specifically designed test kit for the detection of cortisol concentrations in a given sample. There are several advantages to using this particular assay technique. First, there is a strong correlation between plasma and salivary cortisol concentrations; thus, salivary cortisol levels are a valid indicator of plasma free cortisol concentrations.18,39 Second, salivary sampling is economical, simple, noninvasive, and acceptable to most patients. It is further possible to assess diurnal variability using only a few samples a day. Patients can collect saliva samples at home or in the hospital under stress-free conditions and without the assistance or supervision of medical personnel. Blood samplings require mildly invasive techniques that may cause stress and thereby alter HPA axis activity.40 A third advantage of this technique is that salivary cortisol can be easily collected and remain stable at room temperature for 1–2 days and at 0°C–4°C for 1 week.19 Thus, many clinical studies have used salivary cortisol testing instead of plasma cortisol testing and have reported that it is accurate, convenient, and more acceptable as a clinical indicator.

Limitations and Future Studies

The present study has several limitations. One important limitation was that long-term follow-up neuropsychological testing was not performed. A similar study assessing the postoperative am/pm cortisol ratio and POCD did not find a significant relationship between changes in the cortisol ratio and POCD at 3 months postoperation despite an association at 7 days.23 This might be due to the etiology of POCD, which is likely multifactorial. Furthermore, a longer follow-up period allows for additional conditions and/or life events to influence cognitive function, obscuring any effect of surgery.

A second limitation of the present study is that the specificity of the optimal am/pm cortisol ratio had a cutoff value of 91% and a sensitivity of only 50%. Therefore, if the am/pm cortisol ratio is used as the diagnostic index, the probability of misdiagnosis would be smaller, but the probability of missed diagnosis could be higher. Although the ideal situation is for a 100% accurate test, it is unrealistic. Actually, by shifting the cutoff point, we can change the sensitivity and specificity characteristics of the test. For example, in the present trial, setting the am/pm cortisol ratio cutoff at 7.6 resulted in a sensitivity of 25% and a specificity of 99%, while setting the am/pm cortisol ratio cutoff at 1.66 resulted in a sensitivity of 94% and a specificity of 23%. A good alternative is to subject patients who are initially positive to a test with high sensitivity/low specificity and to a second test with low sensitivity/high specificity. In this way, nearly all of the false positives may be correctly identified as disease negative. In addition, it would be better to combine several other biomarkers, such as the plasma brain-specific protein glial fibrillary,41 to reduce the rate of missed POCD diagnoses and then enhance the specificity and sensitivity.

A final, third limitation of the present study is that cortisol was assessed 1 day before surgery. To reduce the measurement error, it has been recommended that saliva samples be averaged over several days to assess more stable, trait-like cortisol levels.42,43 Future studies should include a larger sample or repeatedly measure cortisol levels.

In future studies, the mechanisms by which cortisol is related to postoperative cognitive impairment in aged individuals warrant further exploration. One hypothesis is that age-associated brain volume reductions can lead to a diminished ability to inhibit cortisol, which then further causes brain atrophy as persistently high cortisol levels have neurotoxic effects in the hippocampus. Because it is believed that the HPA axis is inhibited by the hippocampus,44 a cascade might be initiated in which age-associated hippocampal atrophy leads to a diminished ability to inhibit cortisol and, in turn, further atrophy of the hippocampus,45 eventually leading to the development of clinical dementia. Alternatively, HPA axis dysregulation may have initiated this process. Furthermore, another study demonstrated that this phenomenon may not specifically act on the hippocampus, but rather more diffusely throughout the gray matter.10 Thus, it would be beneficial to investigate the link between perioperative cortisol levels and brain structure changes, for instance, using functional magnetic resonance imaging.


In conclusion, the relationship between preoperative elevations in am/pm cortisol secretion ratios and early POCD may provide valuable mechanistic insights into the etiology of POCD. Because the diagnosis and the treatment of POCD are rather challenging, the discovery of a simple and reliable predictive biomarker could have broad utility. The salivary cortisol test does not require venipuncture, is easily administered, and can be performed in an outpatient setting. Thus, these findings also suggest that, instead of costly and lengthy neuropsychology testing, a combination of salivary cortisol testing with other POCD predictors may be sufficient to identify high-risk persons.


Name: Yuan Han, MD, PhD.

Contribution: This author helped design and conduct the study; collect, analyze, and interpret the data; prepare and critically review the manuscript; and approve the final version of the manuscript.

Name: Liu Han, MD.

Contribution: This author helped design and conduct the study; collect, analyze, and interpret the data; prepare and critically review the manuscript; and approve the final version of the manuscript.

Name: Meng-Meng Dong, MD.

Contribution: This author helped design and conduct the study; collect, analyze, and interpret the data; prepare and critically review the manuscript; and approve the final version of the manuscript.

Name: Qing-Chun Sun, MD.

Contribution: This author helped design and conduct the study, collect the data, and critically review the manuscript.

Name: Zhen-Feng Zhang, MD.

Contribution: This author helped design and conduct the study, collect the data, and critically review the manuscript.

Name: Ke Ding, MD.

Contribution: This author helped design and conduct the study, collect the data, and critically review the manuscript.

Name: Yao-Dong Zhang, MD.

Contribution: This author helped design the study, analyze and interpret the data, and critically review the manuscript.

Name: Abdul Mannan, MD.

Contribution: This author helped design the study and prepare and critically review the manuscript.

Name: Yi-Fan Xu, MD.

Contribution: This author helped design and conduct the study, collect the data, and critically review the manuscript.

Name: Chang-Li Ou-Yang, MD.

Contribution: This author assisted with the sample testing and data analysis.

Name: Zhi-Yong Li, MD.

Contribution: This author assisted with the sample testing and data analysis.

Name: Can Gao, MD, PhD.

Contribution: This author helped design the study and prepare and critically review the manuscript.

Name: Jun-Li Cao, MD, PhD.

Contribution: This author helped design the study, critically review the manuscript, and approve the final version of the manuscript.

This manuscript was handled by: Robert Whittington, MD.


1. Moller JT, Cluitmans P, Rasmussen LS. Long-term postoperative cognitive dysfunction in the elderly ISPOCD1 study. ISPOCD investigators. International Study of Post-Operative Cognitive Dysfunction. Lancet. 1998;351:857–861.
2. Paredes S, Cortínez L, Contreras V, Silbert B. Post-operative cognitive dysfunction at 3 months in adults after non-cardiac surgery: a qualitative systematic review. Acta Anaesthesiol Scand. 2016;60:1043–1058.
3. Mashour GA, Woodrum DT, Avidan MS. Neurological complications of surgery and anaesthesia. Br J Anaesth. 2015;114:194–203.
4. Androsova G, Krause R, Winterer G, Schneider R. Biomarkers of postoperative delirium and cognitive dysfunction. Front Aging Neurosci. 2015;7:112.
5. Krenk L, Rasmussen LS, Kehlet H. New insights into the pathophysiology of postoperative cognitive dysfunction. Acta Anaesthesiol Scand. 2010;54:951–956.
6. Vutskits L, Xie Z. Lasting impact of general anaesthesia on the brain: mechanisms and relevance. Nat Rev Neurosci. 2016;17:705–717.
7. Gaffey AE, Bergeman CS, Clark LA, Wirth MM. Aging and the HPA axis: stress and resilience in older adults. Neurosci Biobehav Rev. 2016;68:928–945.
8. Carroll D, Ginty AT, Whittaker AC, Lovallo WR, de Rooij SR. The behavioural, cognitive, and neural corollaries of blunted cardiovascular and cortisol reactions to acute psychological stress. Neurosci Biobehav Rev. 2017;77:74–86.
9. Cuneo MG, Schrepf A, Slavich GM. Diurnal cortisol rhythms, fatigue and psychosocial factors in five-year survivors of ovarian cancer. Psychoneuroendocrinology. 2017;84:139–142.
10. Geerlings MI, Sigurdsson S, Eiriksdottir G. Salivary cortisol, brain volumes, and cognition in community-dwelling elderly without dementia. Neurology. 2015;85:976–983.
11. Heaney JL, Phillips AC, Carroll D. Aging, health behaviors, and the diurnal rhythm and awakening response of salivary cortisol. Exp Aging Res. 2012;38:295–314.
12. McEwen BS, Bowles NP, Gray JD. Mechanisms of stress in the brain. Nat Neurosci. 2015;18:1353–1363.
13. Ennis GE, An Y, Resnick SM, Ferrucci L, O’Brien RJ, Moffat SD. Long-term cortisol measures predict Alzheimer disease risk. Neurology. 2017;88:371–378.
14. Hatfield CF, Herbert J, van Someren EJ, Hodges JR, Hastings MH. Disrupted daily activity/rest cycles in relation to daily cortisol rhythms of home-dwelling patients with early Alzheimer’s dementia. Brain. 2004;127:1061–1074.
15. Mu DL, Li LH, Wang DX. High postoperative serum cortisol level is associated with increased risk of cognitive dysfunction early after coronary artery bypass graft surgery: a prospective cohort study. PLoS One. 2013;8:e77637.
16. Ji MH, Shen JC, Gao R. Early postoperative cognitive dysfunction is associated with higher cortisol levels in aged patients following hip fracture surgery. J Anesth. 2013;27:942–944.
17. Mu DL, Wang DX, Li LH. High serum cortisol level is associated with increased risk of delirium after coronary artery bypass graft surgery: a prospective cohort study. Crit Care. 2010;14:R238.
18. Vining RF, McGinley RA, Maksvytis JJ, Ho KY. Salivary cortisol: a better measure of adrenal cortical function than serum cortisol. Ann Clin Biochem. 1983;20(pt 6):329–335.
19. Turpeinen U, Hämäläinen E. Determination of cortisol in serum, saliva and urine. Best Pract Res Clin Endocrinol Metab. 2013;27:795–801.
20. Kirschbaum C, Hellhammer DH. Salivary cortisol in psychoneuroendocrine research: recent developments and applications. Psychoneuroendocrinology. 1994;19:313–333.
21. Liu HC, Chou P, Lin KN. Assessing cognitive abilities and dementia in a predominantly illiterate population of older individuals in Kinmen. Psychol Med. 1994;24:763–770.
22. Jacobson NS, Truax P. Clinical significance: a statistical approach to defining meaningful change in psychotherapy research. J Consult Clin Psychol. 1991;59:12–19.
23. Rasmussen LS, O’Brien JT, Silverstein JH, et al. ISPOCD2 Investigators. Is peri-operative cortisol secretion related to post-operative cognitive dysfunction? Acta Anaesthesiol Scand. 2005;49:1225–1231.
24. Rasmussen LS, Larsen K, Houx P, Skovgaard LT, Hanning CD, Moller JT. ISPOCD Group. The International Study of Postoperative Cognitive Dysfunction. The assessment of postoperative cognitive function. Acta Anaesthesiol Scand. 2001;45:275–289.
25. Fowkes RC, Moradi-Bidhendi N, Brancaleone V. Annexin-A1 protein and its relationship to cortisol in human saliva. Psychoneuroendocrinology. 2013;38:722–727.
26. Pruessner JC, Wolf OT, Hellhammer DH. Free cortisol levels after awakening: a reliable biological marker for the assessment of adrenocortical activity. Life Sci. 1997;61:2539–2549.
27. Hucklebridge F, Hussain T, Evans P, Clow A. The diurnal patterns of the adrenal steroids cortisol and dehydroepiandrosterone (DHEA) in relation to awakening. Psychoneuroendocrinology. 2005;30:51–57.
28. Brown ES, Rush AJ, McEwen BS. Hippocampal remodeling and damage by corticosteroids: implications for mood disorders. Neuropsychopharmacology. 1999;21:474–484.
29. Lee BK, Glass TA, McAtee MJ. Associations of salivary cortisol with cognitive function in the Baltimore memory study. Arch Gen Psychiatry. 2007;64:810–818.
30. Elgh E, Lindqvist Astot A, Fagerlund M, Eriksson S, Olsson T, Näsman B. Cognitive dysfunction, hippocampal atrophy and glucocorticoid feedback in Alzheimer’s disease. Biol Psychiatry. 2006;59:155–161.
31. Silbert B, Evered L, Scott DA. Preexisting cognitive impairment is associated with postoperative cognitive dysfunction after hip joint replacement surgery. Anesthesiology. 2015;122:1224–1234.
32. Stone AA, Schwartz JE, Smyth J. Individual differences in the diurnal cycle of salivary free cortisol: a replication of flattened cycles for some individuals. Psychoneuroendocrinology. 2001;26:295–306.
33. Adam EK, Kumari M. Assessing salivary cortisol in large-scale, epidemiological research. Psychoneuroendocrinology. 2009;34:1423–1436.
34. Šupe-Domić D, Milas G, Hofman ID, Rumora L, Klarić IM. Daily salivary cortisol profile: insights from the Croatian Late Adolescence Stress Study (CLASS). Biochem Med (Zagreb). 2016;26:408–420.
35. Ottens TH, Dieleman JM, Sauër AM, et al. DExamethasone for Cardiac Surgery (DECS) Study Group. Effects of dexamethasone on cognitive decline after cardiac surgery: a randomized clinical trial. Anesthesiology. 2014;121:492–500.
36. Fang Q, Qian X, An J, Wen H, Cope DK, Williams JP. Higher dose dexamethasone increases early postoperative cognitive dysfunction. J Neurosurg Anesthesiol. 2014;26:220–225.
37. Saridjan NS, Kocevska D, Luijk MPCM, Jaddoe VWV, Verhulst FC, Tiemeier H. The prospective association of the diurnal cortisol rhythm with sleep duration and perceived sleeping problems in preschoolers: the Generation R Study. Psychosom Med. 2017;79:557–564.
38. Mason SE, Noel-Storr A, Ritchie CW. The impact of general and regional anesthesia on the incidence of post-operative cognitive dysfunction and post-operative delirium: a systematic review with meta-analysis. J Alzheimers Dis. 2010;22suppl 367–79.
39. Yaneva M, Mosnier-Pudar H, Dugué MA, Grabar S, Fulla Y, Bertagna X. Midnight salivary cortisol for the initial diagnosis of Cushing’s syndrome of various causes. J Clin Endocrinol Metab. 2004;89:3345–3351.
40. Stahl F, Dörner G. Responses of salivary cortisol levels to stress-situations. Endokrinologie. 1982;80:158–162.
41. Rappold T, Laflam A, Hori D. Evidence of an association between brain cellular injury and cognitive decline after non-cardiac surgery. Br J Anaesth. 2016;116:83–89.
42. Hellhammer J, Fries E, Schweisthal OW, Schlotz W, Stone AA, Hagemann D. Several daily measurements are necessary to reliably assess the cortisol rise after awakening: state- and trait components. Psychoneuroendocrinology. 2007;32:80–86.
43. Pruessner JC, Gaab J, Hellhammer DH, Lintz D, Schommer N, Kirschbaum C. Increasing correlations between personality traits and cortisol stress responses obtained by data aggregation. Endocr Rev. 1997;22:615–625.
44. Jacobson L, Sapolsky R. The role of the hippocampus in feedback regulation of the hypothalamic-pituitary-adrenocortical axis. Endocr Rev. 1991;12:118–134.
45. Sapolsky RM. Glucocorticoids and hippocampal atrophy in neuropsychiatric disorders. Arch Gen Psychiatry. 2000;57:925–935.

Supplemental Digital Content

Copyright © 2018 International Anesthesia Research Society