Background

Anesthesia and/or surgery accelerate Alzheimer’s disease pathology and cause memory deficits in animal models, yet there is a lack of prospective data comparing cerebrospinal fluid (CSF) Alzheimer’s disease–related biomarker and cognitive trajectories in older adults who underwent surgery versus those who have not. Thus, the objective here was to better understand whether anesthesia and/or surgery contribute to cognitive decline or an acceleration of Alzheimer’s disease–related pathology in older adults.

Methods

The authors enrolled 140 patients 60 yr or older undergoing major nonneurologic surgery and 51 nonsurgical controls via strata-based matching on age, sex, and years of education. CSF amyloid β (Aβ) 42, tau, and p-tau-181p levels and cognitive function were measured before and after surgery, and at the same time intervals in controls.

Results

The groups were well matched on 25 of 31 baseline characteristics. There was no effect of group or interaction of group by time for baseline to 24-hr or 6-week postoperative changes in CSF Aβ, tau, or p-tau levels, or tau/Aβ or p-tau/Aβ ratios (Bonferroni P > 0.05 for all) and no difference between groups in these CSF markers at 1 yr (P > 0.05 for all). Nonsurgical controls did not differ from surgical patients in baseline cognition (mean difference, 0.19 [95% CI, –0.06 to 0.43]; P = 0.132), yet had greater cognitive decline than the surgical patients 1 yr later (β, –0.31 [95% CI, –0.45 to –0.17]; P < 0.001) even when controlling for baseline differences between groups. However, there was no difference between nonsurgical and surgical groups in 1-yr postoperative cognitive change in models that used imputation or inverse probability weighting for cognitive data to account for loss to follow up.

Conclusions

During a 1-yr time period, as compared to matched nonsurgical controls, the study found no evidence that older patients who underwent anesthesia and noncardiac, nonneurologic surgery had accelerated CSF Alzheimer’s disease–related biomarker (tau, p-tau, and Aβ) changes or greater cognitive decline.

Editor’s Perspective
What We Already Know about This Topic
  • Anesthesia and surgery have been associated with increases in Alzheimer’s disease pathology and memory deficits in animal models

  • It is unclear whether anesthesia and surgery are associated with increases in Alzheimer’s disease pathology and memory deficits in humans

What This Article Tells Us That Is New
  • In this prospective matched cohort study, there is no conclusive evidence that anesthesia and surgery are associated with increases in Alzheimer’s disease pathology or cognitive deficits in humans

Some older patients have lasting cognitive impairment after anesthesia and surgery, but it remains unclear to what extent this cognitive decline is caused by anesthesia and/or surgery versus the extent to which it reflects their natural cognitive trajectory (see review1 ). Postoperative cognitive dysfunction2  and postoperative neurocognitive disorder3  are both characterized by objectively measured cognitive decline within 1 to 12 months after surgery; postoperative neurocognitive disorder also requires the presence of subjective cognitive complaints. Since postoperative neurocognitive disorder is a relatively new term, its incidence is not yet well defined; postoperative cognitive dysfunction (as defined by a 1 SD or greater decrease in one or more cognitive domains at 6 weeks after surgery) has been reported in up to 41% of surgical patients above age 60 yr.4  Postoperative cognitive dysfunction is associated with decreased quality of life, increased workforce attrition, and increased postoperative mortality.5 

One theory for perioperative neurocognitive disorders suggests surgical trauma/stress and anesthetic drugs accelerate Alzheimer’s disease pathology, which then disrupts brain function and results in postoperative cognitive dysfunction and/or neurocognitive disorder. This theory is supported by work demonstrating that inhaled anesthetics promote amyloid β (Aβ) oligomerization6 in vitro, and tau phosphorylation and aggregation7  in mice. In humans, 24-h postoperative cerebrospinal fluid (CSF) tau levels increase after a variety of surgical procedures and anesthetic techniques.8–13  However, the largest postoperative increases in CSF tau levels have been observed after neurosurgical and otolaryngology procedures, which involve direct surgical manipulation of the brain and/or dura.9,13  Further, the absence of a nonsurgical control group in these studies makes it unclear to what extent these postoperative CSF Alzheimer’s disease biomarker changes were due to anesthesia and/or surgery versus the passage of time or other factors, such as inflammation due to repeated lumbar punctures.14,15 

To better understand whether anesthesia and surgery contribute to cognitive decline and/or an acceleration of Alzheimer’s disease–related pathology in older adults, we compared changes in cognition and CSF Alzheimer’s disease–related biomarkers from before to after surgery between older surgical patients and demographically matched nonsurgical controls who underwent identical assessments over the same time intervals as the surgical patients. This builds upon our previous work16  to determine whether surgical patients had significantly more abnormal cognitive scores or CSF Alzheimer’s disease–related biomarkers than nonsurgical community-dwelling older adults across a 1-yr study period.

This is a secondary analysis of data from Markers of Alzheimer’s Disease and neuroCognitive Outcomes after Perioperative Care (MADCO-PC), an observational cohort study registered with clinicaltrials.gov (NCT01993836) in November 2013, and approved by the Duke University institutional review board.16  The primary aim of MADCO-PC was to examine the extent to which there is a correlation between postoperative changes in both cognition and CSF Alzheimer’s disease–related biomarkers in older surgical patients, which was published last year.16  Our previous report16  included only surgical patients who returned for 6-week follow-up (n = 110), while the current study included all surgical patients who completed baseline cognitive testing (n = 137).

MADCO-PC study participants provided informed consent before enrollment. Patients were prospectively screened (and enrolled if willing) if they were 60 yr or older undergoing noncardiac, nonneurologic surgery under general anesthesia for 2 h or more, and lived within a 60-mile radius (to help ensure that transportation to the hospital for study visits would not be an issue). For additional inclusion or exclusion criteria, see the supplemental methods (https://links.lww.com/ALN/D457).

Upon receipt of additional funding, enrollment of an approximately 50-participant matched nonsurgical group began in February 2016, after surgical enrollment was complete; the nonsurgical controls underwent the same assessments as the surgical patients at the same time intervals. We used strata-based enrollment to recruit nonsurgical controls who, at a group level, matched the surgical cohort based on age, sex, and years of education (see Supplemental Table 1 for additional details, https://links.lww.com/ALN/D457). This strategy necessitated enrolling the nonsurgical controls after the demographics of the surgical patients were known; thus, surgical patients were enrolled from 2013 to 2016 (with 1-yr follow-ups completed in 2017), and nonsurgical controls were enrolled from 2016 to 2018 (with 1-yr follow-ups completed in 2019).

For strata-based matching, age was divided into four strata of 60 to 64, 65 to 69, 70 to 74, and more than 75 yr. Years of education (total years of complete schooling) was divided into four strata of less than 12 (less than high school), 12 (high school), 13 to 15 (partial college or associate’s degree), and more than 16 yr (college degree or more). Two sex strata by four age strata by four education strata produced 32 different groups (2 × 4 × 4 = 32). Into these 32 bins, we then sorted the 110 surgical patients who returned for the 6-week follow-up. We then recruited a targeted number of nonsurgical participants within each of these strata groups, such that the surgical and nonsurgical groups would be matched overall on these baseline characteristics, even though the groups differed in size (n = 110 surgical patients and n = 51 nonsurgical controls). Nonsurgical controls were recruited from research subject registries from the Duke Center for the Study of Aging (n = 32) and the Duke Bryan Alzheimer’s Disease Research Center (n = 7), or via advertising at Duke Hospital and public locations within an approximately 60-mile radius (n = 12), the same area in which surgical patients had to live to participate.

APOE genotyping was performed as described.9  Mild cognitive impairment and Alzheimer’s disease diagnoses were based on International Classification of Diseases, Tenth Revision, codes in patients’ medical records at the time of study entry.

Cognitive Testing and Analysis

At preoperative, 6-week postoperative, and 1-yr postoperative visits (and the same intervals in controls), participants completed a cognitive test battery as previously described (Supplemental Digital Content, https://links.lww.com/ALN/D457).17  Factor analysis of these tests produced four cognitive domains: verbal memory, visual memory, executive function, and attention/concentration (see Supplemental Digital Content for details, https://links.lww.com/ALN/D457).16  The continuous cognitive index was defined as the average of these four cognitive domain scores and represents a sensitive global measure of cognition that our group has used in multiple studies for more than 20 yr.17,18  Patients also completed questionnaires to assess perceived physical function, general health, instrumental activities of daily living, depression and anxiety symptoms, social support, and cognitive difficulties (Supplemental Digital Content, https://links.lww.com/ALN/D457). Patients also completed the Mini-Mental State Examination. Mild and major postoperative neurocognitive disorder were defined as previously described (see Supplemental Digital Content for details, https://links.lww.com/ALN/D457).3  Postoperative delirium assessments in these patients are described in the Supplemental Digital Content (https://links.lww.com/ALN/D457).

CSF Sampling and Alzheimer’s Disease Biomarker Assays

CSF samples (10 to 12 ml each) were obtained at preoperative baseline, and 24 h, 6 weeks, and 1 yr after surgery, and the same time intervals in nonsurgical controls. Aβ, tau, and p-tau-181p were measured via AlzBio3 assays (Innogenetics; Ghent, Belgium).16  The AlzBio3 assay was no longer in production by the time 1-yr CSF sample collection was complete, so CSF Aβ, tau, and p-tau-181p were measured in 1-yr samples with the Fujirebio Lumipulse platform (USA; see Supplemental Digital Content for details, https://links.lww.com/ALN/D457).

Statistical Analysis

We previously observed baseline to 24-h postoperative CSF tau level increases of 87 pg/ml.9  Based on this, we calculated that 85 or more surgical and 42 or more nonsurgical participants would provide 80% power with α = 0.05 to detect a 65-pg/ml smaller increase in 24-h CSF tau change among nonsurgical controls versus surgical patients (i.e., a 75% smaller increase in CSF tau levels among controls than surgical patients). Based on previous work,17  this sample size also provides 80% or greater power to detect a 0.15 or greater unit difference in continuous cognitive index change (a moderate Cohen’s d effect size of 0.55) from before to 6 weeks or 1 yr later between surgical patients and nonsurgical controls, which is even smaller than the difference in continuous cognitive index change seen between patients with versus without postoperative cognitive dysfunction in a previous study.19  Given the 51 to 53% rate of loss to follow-up observed in previous studies with multiple lumbar punctures,10,20  we enrolled 140 surgical and 51 nonsurgical participants to ensure sufficient sample size after loss to follow-up; see the supplemental methods for details (https://links.lww.com/ALN/D457).

CSF Alzheimer’s disease–related biomarker trajectories (from baseline to the 24-h and 6-week time points) were compared with nonparametric longitudinal models in R version 4.2.0 (https://www.R-project.org/, accessed December 11, 2023).21  CSF biomarker data at the 1 yr time point were analyzed using Wilcoxon rank sum tests. To reduce type I error, the 5 CSF biomarker models were Bonferroni-corrected. Hodges–Lehmann group median difference estimates were used to calculate 95% CI for all nonparametric variables, including CSF biomarkers; Hodges–Lehmann estimates do not match the absolute differences between groups because these are nonparametric, rank-based calculations.

Mild and major postoperative neurocognitive disorder rates were compared using chi-square or Fisher exact tests. A multivariable linear regression model was used to assess group differences in 1-yr continuous cognitive index change, with multivariable adjustment for baseline cognition and statistically significant baseline differences between groups. Multiple imputation and inverse probability weighting were applied to address missing data. A tipping point approach was utilized to address the possibility that the overall results may have differed if 1-yr cognitive scores were not missing at random (Supplemental Digital Content, https://links.lww.com/ALN/D457). Finally, we examined the effect on the overall study findings if we substituted the worst possible 1-yr cognitive test scores for surgical patients who died before the 1-yr visit, were institutionalized (e.g., in a nursing home), or were too sick to return for 1-yr cognitive testing.

We also performed a series of post hoc sensitivity analyses and investigated the impact of four alternative modeling approaches on our findings for cognitive function. Specifically, first, we compared cognitive outcomes between recruitment sources (for nonsurgical controls) via t test to address the possibility of confounding if nonsurgical controls recruited via aging or Alzheimer’s disease–related research registries were at higher long-term cognitive decline risk versus nonsurgical controls recruited by public flyers. Second, we used baseline attention/concentration instead of the overall continuous cognitive index in the linear regression model, given that surgical patients and nonsurgical controls trended toward a difference in this cognitive domain (table 1). Third, since different statistical modeling techniques can yield divergent results when applied to the same data,22  we also analyzed cognition as a 1-yr follow-up score rather than baseline to 1-yr change score to investigate the impact of this parameterization. Fourth, we also included a longitudinal mixed model of baseline, 6-week, and 1-yr cognitive function with time interaction terms for group and covariates that had significant main effects.

Table 1.

Baseline Demographics, Cognitive Function, and Cerebrospinal Fluid Biomarkers in Surgical Patients, Strata-matched Nonsurgical Controls

Baseline Demographics, Cognitive Function, and Cerebrospinal Fluid Biomarkers in Surgical Patients, Strata-matched Nonsurgical Controls
Baseline Demographics, Cognitive Function, and Cerebrospinal Fluid Biomarkers in Surgical Patients, Strata-matched Nonsurgical Controls

Next, we examined the possibility that a subgroup of surgical patients, such as those with postoperative delirium, would have worse cognitive dysfunction at the 1-yr time point than the other groups (surgical patients without postoperative delirium, nonsurgical controls). A Fisher exact test was used to compare the fraction of patients within each group with an overall cognitive index at least one unit below the sample mean at the 1-yr time point. Last, we analyzed a longitudinal mixed model of group (surgical patients with postoperative delirium, surgical patients without postoperative delirium, nonsurgical controls), time (baseline, 6 weeks, 1 yr), and a group-by-time interaction to determine whether surgical patients with postoperative delirium had worse cognitive scores than the other subgroups at baseline or over time. Unless otherwise specified, all statistical analyses were performed in SAS v9.4 (SAS Institute Inc., USA; see Supplemental Methods, https://links.lww.com/ALN/D457).

Figure 1 shows participant enrollment flow; model-specific sample sizes are described in the Supplemental Digital Content (https://links.lww.com/ALN/D457). Intraoperative factors in the surgical cohort are described in Supplemental Table 2 (https://links.lww.com/ALN/D457). Of 31 measured baseline characteristics, 25 did not differ between surgical patients and nonsurgical controls (table 1), including the three characteristics the groups were matched upon (age, sex, and years of education). However, as compared to nonsurgical controls, surgical patients had more hypertension (absolute difference in rates between surgical and nonsurgical groups, 25.5% [95% CI, 8.38 to 42.5%]; P = 0.004), lower self-reported physical functional capacity (Hodges–Lehmann group median difference estimate, –8.00 [95% CI, –15.2 to –1.00]; P = 0.012), worse subjective health scores on the Medical Outcomes Study 36-Item Short Form Health Survey General Health Perceptions questionnaire (Hodges–Lehmann group median difference estimate, 1.00 [95% CI, 0.00 to 1.00]; P = 0.001), and worse scores on the Social Activities (Hodges–Lehmann group median difference estimate, 2.00 [95% CI, 0.00 to 3.00]; P = 0.016) and Symptom Limitations scales (Hodges–Lehmann group median difference estimate, 1.14 [95% CI, 0.00 to 2.29]; P = 0.045). However, a higher proportion of the nonsurgical cohort had baseline CSF Aβ levels less than 250 pg/ml and/or CSF tau levels greater than 93 pg/ml (indicators of brain Aβ and tau pathology, respectively)23,24  than were seen in the surgical group (difference, 22.9% [95% CI, 6.77 to 39.1%]; P = 0.004). Among 185 participants with complete baseline cognitive data, no surgical patients or nonsurgical controls had a diagnosis of mild cognitive impairment or Alzheimer’s disease, although 13% of the surgical cohort and 10% of the nonsurgical controls had Mini-Mental State Examination scores less than 27 (which has been shown to have 87% specificity for mild cognitive impairment).25 

Fig. 1.

Participant consort diagram. Surgical patients (left) and nonsurgical controls (right).

Fig. 1.

Participant consort diagram. Surgical patients (left) and nonsurgical controls (right).

Close modal

Figure 2 shows median CSF biomarker levels in the surgical and nonsurgical cohorts at baseline, 24 h, and 6 weeks (from the AlzBio3 assay) and at 1 yr after surgery (from the Fujirebio Lumipulse platform). There were no significant differences in CSF tau, p-tau-181p, or Aβ levels or the tau/Aβ or p-tau-181p/Aβ ratios between groups, and no effects of time or group-by-time interactions for any of the CSF Alzheimer’s disease–related biomarkers measured at baseline, 24 h, or 6 weeks (P > 0.05 for all after Bonferroni correction; Supplemental Table 3, https://links.lww.com/ALN/D457; fig. 2). In a separate analysis of 1-yr CSF biomarker values, there were no significant differences between groups for Aβ (Hodges–Lehmann group median difference estimate, –72 [95% CI, –224 to 75]; P > 0.999), tau (2 [95% CI, –69 to 75]; P > 0.999), p-tau-181p (4.5 [95% CI, –4.2 to 16]; P > 0.999), tau/Aβ (0.03 [95% CI, –0.03 to 0.10]; P > 0.999), or p-tau-181p/Aβ (0.006 [95% CI, –0.001 to 0.014]; P = 0.400) after Bonferroni correction (fig. 2).

Fig. 2.

Cerebrospinal fluid levels of (A) amyloid β (Aβ), (B) tau, (C) p-tau-181p, (D) tau/Aβ, and (E) p-tau-181p/Aβ in surgical patients and nonsurgical controls. The first column represents baseline, 24-h, and 6-week data from the AlzBio3 assay platform (Innogenetics; Ghent, Belgium). Error bars represent the 25th and 75th percentiles of the data. The second column represents 1-yr cerebrospinal fluid Alzheimer’s disease biomarker levels in surgical patients (red) and nonsurgical controls (blue) from the Fujirebio Lumipulse (USA) assay platform; 1-yr data were log-transformed to reduce skew. Each dot represents data from an individual patient at a single time point; the width of the colored area indicates the data distribution. Within the boxplots, the middle line shows the median of the data, and the upper and lower edges show the interquartile range. There were no significant group differences (see the main text for analysis details). Missingness: 98 surgical patients had cerebrospinal fluid data at baseline, 90 at 24 h, 94 at 6 weeks, and 48 at 1 yr. One additional surgical patient was missing tau data at 1 yr due to assay artifact. A total of 46 nonsurgical controls had cerebrospinal fluid data at baseline, 42 at 24 h, 36 at 6 weeks, and 32 at 1 yr. One additional control was missing tau data at 24 h; two other controls were missing Aβ and tau data at 1 yr, respectively, due to assay artifact.

Fig. 2.

Cerebrospinal fluid levels of (A) amyloid β (Aβ), (B) tau, (C) p-tau-181p, (D) tau/Aβ, and (E) p-tau-181p/Aβ in surgical patients and nonsurgical controls. The first column represents baseline, 24-h, and 6-week data from the AlzBio3 assay platform (Innogenetics; Ghent, Belgium). Error bars represent the 25th and 75th percentiles of the data. The second column represents 1-yr cerebrospinal fluid Alzheimer’s disease biomarker levels in surgical patients (red) and nonsurgical controls (blue) from the Fujirebio Lumipulse (USA) assay platform; 1-yr data were log-transformed to reduce skew. Each dot represents data from an individual patient at a single time point; the width of the colored area indicates the data distribution. Within the boxplots, the middle line shows the median of the data, and the upper and lower edges show the interquartile range. There were no significant group differences (see the main text for analysis details). Missingness: 98 surgical patients had cerebrospinal fluid data at baseline, 90 at 24 h, 94 at 6 weeks, and 48 at 1 yr. One additional surgical patient was missing tau data at 1 yr due to assay artifact. A total of 46 nonsurgical controls had cerebrospinal fluid data at baseline, 42 at 24 h, 36 at 6 weeks, and 32 at 1 yr. One additional control was missing tau data at 24 h; two other controls were missing Aβ and tau data at 1 yr, respectively, due to assay artifact.

Close modal

Supplemental Table 4 (https://links.lww.com/ALN/D457) summarizes scores on each cognitive test at baseline, 6 weeks later, and 1 yr later in the surgical and nonsurgical cohorts; the continuous cognitive index and cognitive domain data in both groups over time are shown in figure 3 with statistics for group, time, and group-by-time effects. There was no significant difference between surgical patients and nonsurgical controls in the rate of mild postoperative neurocognitive disorder (n = 25 of 105 surgical patients, 18 of 46 nonsurgical controls; absolute difference in rate between groups, –15.3% [95% CI, –31.6 to 0.97%]; P = 0.110) or major postoperative neurocognitive disorder (n = 1 of 103 surgical patients, 0 of 46 nonsurgical controls; 0.97% [95% CI, –0.92 to 2.86%]; P > 0.999) between groups at 6 weeks after Bonferroni correction. Similarly, there was no group difference in rates of mild or major neurocognitive disorder at 1 yr after Bonferroni correction (mild: n = 32 of 80 surgical patients, 19 of 40 nonsurgical controls; absolute difference in rate between groups, –7.50% [95% CI, –26.3 to 11.3%]; P = 0.883; major: n = 0 of 79 surgical patients, 2 of 40 nonsurgical controls; –5.00% [95% CI, –11.75 to 1.75%]; P = 0.222).

Fig. 3.

Cognitive function by domains and overall continuous cognitive index (the average of the four domain scores) over time, in surgical patients (red) and nonsurgical controls (turquoise). Each dot represents data from an individual patient at a single time point; the width of the colored area indicates the data distribution. Within the boxplots, the middle line shows the median of the data, and the upper and lower edges show the interquartile range (see table 2 for statistical comparisons). P values are Bonferroni corrected for the five cognitive models.

Fig. 3.

Cognitive function by domains and overall continuous cognitive index (the average of the four domain scores) over time, in surgical patients (red) and nonsurgical controls (turquoise). Each dot represents data from an individual patient at a single time point; the width of the colored area indicates the data distribution. Within the boxplots, the middle line shows the median of the data, and the upper and lower edges show the interquartile range (see table 2 for statistical comparisons). P values are Bonferroni corrected for the five cognitive models.

Close modal
Table 2.

Multivariable Linear Regression Model for Continuous Cognitive Index Change from Baseline to 1 Year after Surgery with Observed Baseline Data for All Variables Listed Below (n = 72 Surgical Patients, 35 Nonsurgical Controls)

Multivariable Linear Regression Model for Continuous Cognitive Index Change from Baseline to 1 Year after Surgery with Observed Baseline Data for All Variables Listed Below (n = 72 Surgical Patients, 35 Nonsurgical Controls)
Multivariable Linear Regression Model for Continuous Cognitive Index Change from Baseline to 1 Year after Surgery with Observed Baseline Data for All Variables Listed Below (n = 72 Surgical Patients, 35 Nonsurgical Controls)

In a linear regression model for cognitive change controlling for the baseline differences observed between groups in table 1, the nonsurgical controls still had greater cognitive decline from baseline to 1 yr than the surgical patients (β, –0.31 [95% CI, –0.45 to –0.17]; P < 0.001; table 2). To address the possibility of greater loss to follow up among patients who may have been more likely to experience cognitive decline, we repeated this analysis on 1-yr cognitive data using inverse probability weighting and multiple imputation. Inverse probability weighting showed similar results to the observed data model, in which nonsurgical controls had greater 1-yr cognitive decline than surgical patients (β, –0.33 [95% CI, –0.47 to 0.18]; P < 0.001; Supplemental Table 5, https://links.lww.com/ALN/D457). Although no longer significant when using imputed data, there was a potential trend toward greater 1-yr cognitive decline in the nonsurgical controls versus surgical patients (β, –0.16 [95% CI, –0.32 to 0.01]; P = 0.071; Supplemental Table 6, https://links.lww.com/ALN/D457).

To address the potential for data missing not at random, we used a tipping point approach to calculate the shift in imputed 1-yr cognitive decline scores of surgical patients who did not return for 1-yr follow-up, which would be required in order to conclude that surgical patients had significantly greater cognitive decline at 1 yr than nonsurgical controls. The imputed mean 1-yr continuous cognitive index change among surgical patients lost to follow-up was –0.22, while the actual 1-yr continuous cognitive index change among surgical patients who returned for follow-up was –0.03. Thus, based on the tipping point analysis, in order for the surgical group as a whole to have had greater cognitive decline than the nonsurgical controls at 1 yr, every surgical patient lost to follow-up at 1 yr would have to have experienced a –1.08 further shift in their mean imputed 1-yr cognitive decline scores (i.e., beyond –0.22). This means that the surgical patients lost to follow-up would have had a mean 1-yr cognitive change of –1.30 ± 0.62, which would be approximately 4 SDs below the actual observed 1-yr cognitive change among surgical patients who did return for follow-up (mean ± SD, –0.03 ± 0.33). This equates to a Cohen’s d effect size of 2.69, which is implausibly large compared to, for instance, the Cohen’s d of 1.38 previously observed between cognitively normal versus mild cognitive impairment patients.26  While theoretically possible, it is highly unlikely that we would have observed nearly double this magnitude of an effect (i.e., a Cohen’s d of 2.69) between surgical patients who did versus did not return for 1-yr cognitive testing.

Of the 57 surgical patients who did not return for 1-yr cognitive testing (of the 137 who completed the baseline study visit), 4 passed away before the 1-yr time point, and 6 others were too ill to return at the 1-yr time point or were unable to return because they were living in nursing homes or other assisted living facilities. Hence, we examined the effect on the 1-yr cognitive change analysis of imputing the worst possible cognitive test values for these 10 surgical patients at the 1-yr time point (i.e., if they had gotten the worst score on every individual test at the 1-yr time point). For this analysis, we used the previously imputed scores for the nonsurgical and other 47 surgical patients who did not return for 1-yr follow-up, since they were still alive and living independently, and many simply did not want to return for the 1-yr study visit due to other obligations. When using this approach, the mean 1-yr cognitive change in the surgical patients was still not significantly worse than the mean 1-yr cognitive change in the nonsurgical controls (β for nonsurgical controls versus surgical controls, 0.03 [95% CI, –0.27 to 0.33]; P = 0.843; Supplemental Table 7, https://links.lww.com/ALN/D457).

Further, we performed six additional analyses to ensure that our findings were robust to possible confounding. First, we checked whether there was a confounding effect of nonsurgical control recruitment source (public flyers vs. aging-related research registries) on baseline to 1-yr changes in cognition. The recruitment source for nonsurgical controls (i.e., public flyers vs. registries) was not associated with 6-week (mean difference, 0.09 [95% CI, –0.12 to 0.30]; P = 0.391) or 1-yr cognitive change (0.23 [95% CI, –0.02 to 0.47]; P = 0.069), although we may have been underpowered to detect effect sizes in these ranges given the small sample sizes (at baseline, 12 patients recruited from flyers vs. 39 patients recruited from aging/Alzheimer’s disease registries).

Second, there remained a significant worsening of cognition in the nonsurgical group even if we included baseline attention/concentration (rather than baseline continuous cognitive index) in our linear regression model for cognitive change (β, –0.30 [95% CI, –0.44 to –0.16]; P < 0.001; Supplemental Table 8, https://links.lww.com/ALN/D457). Third, our alternative model of 1-yr cognitive index scores (rather than change in cognitive index values from before to 1 yr after surgery) also showed similar results (i.e., the surgical patients did not have worse cognition at the 1-yr time point than nonsurgical controls; in fact, the nonsurgical controls had worse cognition at 1 yr than the surgical patients; β, –0.31 [95% CI, –0.45 to –0.17]; P < 0.001; Supplemental Table 9, https://links.lww.com/ALN/D457). Fourth, a longitudinal mixed model for cognitive function showed a significant interaction between group and time for continuous cognitive index change from baseline to 1-yr follow-up. In this model, as in the primary linear regression model, the nonsurgical controls again had greater cognitive decline at 1 yr (β, –0.30 [95% CI, –0.43 to –0.18]; P < 0.001; Supplemental Table 10, Supplemental Figure 1, https://links.lww.com/ALN/D457).

Fifth, we explored the possibility that although there was a lack of evidence that the surgical group had worse cognitive decline than the nonsurgical group overall, there may have been a subgroup of surgical patients with worse cognitive dysfunction (such as those who developed postoperative delirium) than both the rest of the surgical group and the nonsurgical group. Indeed, in this exploratory analysis, the percentage of each group with an overall cognitive index at least one unit below the sample mean in this cohort (i.e., the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, objective criteria for mild neurocognitive disorder27 ) at the 1-yr time point was 42.9% (3 of 7) among surgical patients with postoperative delirium, 6.85% (5 of 73) among surgical patients without postoperative delirium, and 7.5% (3 of 40) among the nonsurgical controls (overall P = 0.037). Sixth, we explored whether surgical patients with versus without postoperative delirium had different cognitive trajectories than nonsurgical controls. This model suggested that surgical patients who later developed postoperative delirium (n = 7) started with a lower cognitive baseline than both the surgical patients who did not develop postoperative delirium (n = 73; β, –0.42 [95% CI, –0.83 to –0.01]; P = 0.047) and the nonsurgical controls (n = 40; β, –0.49 [95% CI, –0.92 to –0.05]; P = 0.029; Supplemental Figure 2, https://links.lww.com/ALN/D457). Additionally, the surgical patients who developed postoperative delirium had steeper trajectories of baseline to 1-yr cognitive decline than surgical patients who did not develop postoperative delirium (β, –0.28 [95% CI, –0.50 to –0.05]; P = 0.017). However, the surgical patients who developed postoperative delirium did not have steeper trajectories of baseline to 1-yr cognitive decline than nonsurgical controls (β, 0.01 [95% CI, –0.23 to 0.24]; P = 0.948). It is important to note that only seven surgical patients who developed postoperative delirium returned for 1-yr follow-up, so these conclusions are likely underpowered. The small number of patients with postoperative delirium emphasizes the need for caution in concluding that patients with delirium have greater cognitive decline than patients without delirium from these data.

In this prospective cohort study, with strata-based matching of nonsurgical controls to the surgical patient group based on age, sex, and education, there was no difference between groups in CSF Aβ, tau, or p-tau-181p levels at 24 h, 6 weeks, or 1 yr. Further, as compared to nonsurgical controls, the surgical patients did not have a greater incidence of postoperative neurocognitive disorder from baseline to 6 weeks or 1 yr later. In a multivariable analysis, contrary to the hypothesis that surgery would lead to long-term cognitive decline, the nonsurgical controls had greater cognitive decline than surgical patients during the next 1 yr. Further, when using imputation, inverse probability weighting, or worst-case scores for data lost to follow-up as well as other alternative modeling strategies, there remained no evidence of greater 1-yr cognitive decline in surgical versus nonsurgical participants. Additionally, a tipping point analysis showed that, in order to flip our conclusions, the required shift observed among surgical patients who did not return for 1-yr follow-up would have been approximately 4 SDs larger than the mean 1-yr cognitive decline among observed surgical patients at the 1-yr time point, which equates to an effect size nearly double what has previously been observed between memory composite scores of cognitively normal versus mild cognitive impairment groups.26 

These results stand in contrast to findings from animal studies in which anesthesia and/or surgery led to increased Alzheimer’s disease–related pathology and memory deficits,6,7,28  yet they are broadly consistent with previous work that showed delirium and critical illness (but not anesthesia and/or surgery per se) were associated with 1-yr cognitive decline.29  These results are also consistent with previous work from our group: the rate of mild and major postoperative neurocognitive disorder in the surgical patients at the 1-yr time point in this study was 40%, which is similar to the 46% rate of postoperative cognitive dysfunction (defined as a 1 SD or greater drop in one or more cognitive domains) seen previously in a similar noncardiac surgery cohort at our institution.17 

However, few previous studies have directly compared cognitive data in surgical patients to those in a matched nonsurgical control group. Thus, the lack of greater cognitive decline in the surgical patients (than nonsurgical controls) seen here can be interpreted in two ways. First, these data could reflect a true lack of greater cognitive decline in older surgical patients versus nonsurgical controls among the range of noncardiac, nonneurologic surgeries studied here, which could be explained by three factors. First, some major surgeries can lead to postoperative cognitive improvement rather than cognitive decline, especially if the surgery treats underlying medical problems that caused cognitive dysfunction.5,30–32  Second, many patients in this study underwent minimally invasive procedures such as thyroidectomies,33  which cause less tissue trauma and may have fewer detrimental cognitive effects than longer or more invasive procedures like cardiac surgery.2,18  Consistent with this idea, the postoperative delirium rate in this surgical cohort (8.8% or 7 of the 80 surgical patients who returned for 1-yr cognitive testing) was lower than that seen in studies of more invasive procedures such as cardiac surgery, in which postoperative delirium rates as high as 73% have been seen.34  Third, improvements in surgical care that occurred by the time this study was conducted, such as widespread use of continuous nerve blocks,35  epidurals, and enhanced recovery protocols,36  may have improved cognitive outcomes, although this seems unlikely in light of the fact that the rate of longer-term cognitive decline seen here is similar to that seen in a previous study from our group conducted from 2000 to 2005.17  Further, prospective randomized studies have not found lower delirium rates after regional versus general anesthesia for hip fracture repair,37,38  although less is known about the effects of nerve blocks and enhanced recovery protocols on cognitive dysfunction between 6 weeks and 1 yr after other types of noncardiac surgery.

Second, the lack of greater cognitive decline in the surgical patients (vs. the nonsurgical controls) could reflect unmeasured baseline differences between groups that may have led to greater 1-yr cognitive decline in the nonsurgical group. Indeed, although the groups were matched on 25 of 31 baseline characteristics, there were baseline differences between them in attention/concentration, Aβ and/or tau pathology, self-perceived physical function, general health, social activities, and symptom limitations. However, the surgical patients did not have greater cognitive decline in our models even when accounting for these baseline differences, suggesting that the greater cognitive decline in the nonsurgical group was not due to these baseline differences. However, it remains possible that other unmeasured baseline differences between groups may have confounded the results.

Additionally, since the majority of the surgical participants who had Aβ|tau pathology in this cohort were exclusively Aβ+ and tau (n = 15 of the 18 with any Aβ|tau pathology), the lack of correlation between amyloid deposition and cognitive function in the Alzheimer’s literature may partially explain why baseline differences in Aβ|tau pathology between the surgical patients and nonsurgical controls in this cohort did not account for their group differences in cognitive function through 1 yr.39  Future studies are needed to determine the extent to which Aβ and tau pathologies separately influence surgical versus nonsurgical cognitive trajectories over time.40  Prospective matched cohort study designs (as used here) are considered an inferior form of evidence compared to randomized controlled trials,41  since there is always a possibility of unmeasured confounding between matched cohort groups. However, studies examining cognitive change after anesthesia and/or surgery are often restricted to sampling nonsurgical controls matched to surgical patients based on demographics, because it is usually neither ethical nor practical to randomize patients to surgery versus nonoperative management.4,42–51  Our results suggest the need for careful consideration for minimizing potential confounders when using matched nonsurgical control groups for analyses such as reliable change index calculations, which present data on cognitive dysfunction solely in the surgical cohort (indexed to cognitive data from the nonsurgical group),40,52,53  since the two cohorts may not be fully matched on baseline characteristics related to cognitive function (as seen here). These imbalances could confound cognitive comparisons between groups. Thus, while variables such as baseline CSF Alzheimer’s disease biomarkers, hypertension, and APOE4 genotypes may be challenging to include as part of the matching process, thorough reporting of variables that may impact cognition is crucially important for both minimizing unmeasured confounding and enabling better comparisons in studies with matched surgical and nonsurgical groups.

This study has limitations. First, this was a single-center study at a tertiary academic medical center, which may limit generalizability. Second, selection bias may have been present among nonsurgical controls recruited from aging and Alzheimer’s disease research subject registries, since individuals may have enrolled in these registries due to a family history of dementia or personal memory concerns. While we found no difference in 6-week or 1-yr cognitive change among nonsurgical controls recruited from these registries versus public flyers, we may have been underpowered to detect this difference. Future studies should aim to recruit nonsurgical controls outside of aging or Alzheimer’s disease registries to minimize this potential for bias. For example, instead of recruiting from aging or Alzheimer’s disease registries, nonsurgical controls could be recruited from patients seen in surgery clinics who did not elect to have surgery.

Third, the control cohort size was modest relative to other postoperative cognitive dysfunction or neurocognitive disorder studies with greater than 100 controls.54,55  Of the 185 patients who completed baseline testing (table 1), 80 surgical patients and 40 nonsurgical controls completed 1-yr cognitive testing, a 35% rate of loss to follow-up. Nonetheless, this is actually smaller than the 51 to 53% loss to follow-up rates seen in similar previous studies with repeated lumbar punctures in older patients.10,20  Lower baseline continuous cognitive index and attention or concentration were each associated with loss to follow-up in the nonsurgical controls but not in the surgical patients (table 3), yet controlling for these factors did not account for the group difference in cognitive decline at 1 yr (table 2; Supplemental Table 7, https://links.lww.com/ALN/D457). Further, our results remained similar regardless of whether we utilized inverse probability weighting or multiple imputation for data lost to follow-up.

Table 3.

Baseline Demographics, Cognitive Function, and Cerebrospinal Fluid Biomarkers between Individuals Who Subsequently Remained in the Study versus Those Lost to Follow-up at 1 Year

Baseline Demographics, Cognitive Function, and Cerebrospinal Fluid Biomarkers between Individuals Who Subsequently Remained in the Study versus Those Lost to Follow-up at 1 Year
Baseline Demographics, Cognitive Function, and Cerebrospinal Fluid Biomarkers between Individuals Who Subsequently Remained in the Study versus Those Lost to Follow-up at 1 Year

Fourth, the possibility that cognitive scores in the surgical patients who did not return for 1-yr follow-up were not missing at random is an important and prominent limitation of this study. However, our tipping point analysis suggested that the mean 1-yr cognitive change scores among unobserved surgical patients would need to have been nearly 4 SD beyond the mean cognitive change scores among observed surgical patients (nearly double the effect size seen in a previous study of normal cognitive function vs. MCI26 ) in order to reverse our conclusions at 1 yr. While theoretically possible, this is highly unlikely to have occurred.

Fifth, unlike previous studies,6,13,42,56  here anesthesia and surgery was not associated with a detrimental change in CSF Alzheimer’s disease–related biomarkers or cognition. However, cognitive decline in Alzheimer’s disease typically occurs over years to decades. This study was limited to 1 yr after surgery, so it is possible there could be greater differences in Alzheimer’s disease biomarkers or cognitive change between patients who do versus do not undergo surgery over longer time periods. However, large retrospective studies have found only small surgery-related differences in long-term cognitive decline at a group level.57,58  No previous prospective study has examined CSF Alzheimer’s disease biomarker trajectories in both surgical patients and nonsurgical controls during the 1-yr time period studied here, although animal studies have found acute (not chronic) effects of anesthetics and/or surgery on Alzheimer’s disease pathology.59 

Sixth, neither retrospective studies57,58  nor the prospective data reported here rule out the presence of smaller patient subgroups who may have significant cognitive decline after surgery, such as APOE4 allele carriers or others who may be more sensitive to the detrimental effects of postoperative inflammation on cognition.60–62  This may include patients who experienced postoperative delirium, whom we found had an increased incidence of cognitive dysfunction at the 1-yr time point, although this result should be interpreted cautiously given the small total number of patients with postoperative delirium here. Indeed, there were only seven patients with postoperative delirium, three of whom had cognitive deficits more than 1 unit below the sample mean at 1 yr (the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, objective criteria for mild neurocognitive disorder), although four of these seven patients with delirium did not meet this threshold at 1 yr. These small numbers and the fact that four of the seven surgical patients with postoperative delirium did not meet Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, objective criteria for mild neurocognitive disorder at 1 yr emphasize the need for caution in concluding from these data that patients with delirium have greater cognitive decline, although the idea that patients with postoperative delirium have worse cognitive decline is consistent with numerous other studies.63–66 

In fact, the small number of patients with postoperative delirium in this cohort may suggest that this overall sample may have been more resilient and/or have experienced relatively milder surgical trauma than seen in other delirium studies in intensive care unit and/or cardiac surgery patients,34,67  which may explain why we found no evidence for greater cognitive decline in surgical patients versus nonsurgical controls in this cohort. Further, our exploratory subgroup analysis of cognition among surgical patients with versus without postoperative delirium (and compared to nonsurgical controls) suggested that the subgroup of patients who developed postoperative delirium experienced greater 1-yr cognitive decline after surgery than the surgical patients who did not develop postoperative delirium. Given the small sample size of this subgroup analysis in this cohort, appropriately powered future studies should compare cognition and Alzheimer’s disease–type biomarkers in these subgroups (i.e., surgical patients with postoperative delirium, surgical patients without postoperative delirium, and nonsurgical controls) at baseline and over time.

Seventh, the 1-yr CSF Alzheimer’s disease–related biomarker data came from a different assay than that used for the earlier time points, since the AlzBio3 assay was no longer being manufactured by the time all 1-yr samples were collected here. This limits the ability to draw conclusions about CSF Alzheimer’s disease–related biomarker changes through 1 yr in this study. However, our repeated-measures analyses from baseline, 24 h, and 6 weeks did not show group differences in postoperative CSF biomarkers between surgical patients and nonsurgical controls, and neither did a cross-sectional comparison of group differences in biomarker levels at 1 yr.

Conclusions

These data represent the first prospective comparison of cognitive and CSF Alzheimer’s disease–related biomarker trajectories among older surgical patients and matched nonsurgical controls. Although matched cohort designs cannot exclude possible selection bias and/or unmeasured confounders, the data showed no difference between groups in CSF Alzheimer’s disease–related biomarker changes or increased cognitive decline among surgical patients greater than 1 yr. These conclusions held after accounting for missingness among patients lost to follow-up and controlling for baseline group differences. Thus, despite the limitations discussed, the findings from this cohort do not support the hypothesis that anesthesia and noncardiac, nonneurologic surgery in older adults are associated with accelerated Alzheimer’s disease pathology or cognitive decline over the next year.

Research Support

This work was supported by an International Anesthesia Research Society (San Francisco, California) Mentored Research Award to Dr. Berger, a National Institutes of Health (Bethesda, Maryland) grant (No. R03-AG050918) to Dr. Berger, and support from the Department of Anesthesiology, Duke University (Durham, North Carolina), to Dr. Berger. Dr. Berger also acknowledges support from the National Institutes of Health (grant No. T32-GM08600); a Jahnigen Scholars Fellowship award from the American Geriatrics Society (New York, New York) and the Foundation for Anesthesia Education and Research (Schaumburg, Illinois), and additional support from the National Institutes of Health (grant Nos. P30-AG028716 and K76-AG057022). Dr. Browndyke acknowledges support from the National Institutes of Health (grant Nos. U01-HL088942, R01-HL130443, and R01-AG042599). Dr. Whitson acknowledges support from the National Institutes of Health (grant Nos. UH2-AG056925-02 and UL1TR002553-02). Dr. Mathew acknowledges support from the National Institutes of Health (grant No. R01-HL130443). Dr. Reese acknowledges support from the Duke Aging Center Postdoctoral Research Training Grant (Durham, North Carolina; grant No. NIA T32 AG000029).

Competing Interests

Dr. Berger has received material support (i.e., electroencephalogram monitor loan) for a postoperative recovery study in older adults from Masimo (Irvine, California) and has received private legal consulting fees related to perioperative neurocognitive disorders. Dr. Browndyke has received funding from Claret Medical, Inc (Santa Rosa, California). Dr. Habib declares the following: Haisco USA (Bridgewater, New Jersey); Pacira Pharmaceuticals (Tampa, Florida); Heron Therapeutics (San Diego, California); advisory board: Merck (Rahway, New Jersey), Heron Therapeutics, MDoloris (Loos, France), Vertex (Boston, Massachusetts). Dr. Shaw declares consulting for Biogen (Cambridge, Massachusetts) and Roche (Indianapolis, Indiana). Dr. Whitson declares the following: Uptodate, Inc. (Wellesley, Massachusetts); American Geriatrics Society (New York, New York); served on external advisory boards for Yale (New Haven, Connecticut) and Wake Forest (Winston-Salem, North Carolina). Dr. Welsh-Bohmer declares the following: Genentech/Roche (San Fransisco, California); Senaptec (Beaverton, Oregon); JIGSAWDIO (Raleigh, North Carolina); External Advisory Board to the Center for Biomedical Research Excellence of the University of Nevada, Las Vegas Center for Neurodegeneration and Translational Neuroscience (Paradise, Nevada). Dr. Mathew declares the following: Clinetic (Durham, North Carolina), Scientific Advisory Board; Morpheus Consortium (Durham, North Carolina), Board of Directors; Sapphire Health Innovation (Madison, Wisconsin), owner. The other authors declare no competing interests.

Supplemental descriptions, tables, and figures: https://links.lww.com/ALN/D457

1.
Evered
LA
,
Silbert
BS
:
Postoperative cognitive dysfunction and noncardiac surgery.
Anesth Analg
2018
;
127
:
496
505
2.
Berger
M
,
Terrando
N
,
Smith
SK
,
Browndyke
JN
,
Newman
MF
,
Mathew
JP
:
Neurocognitive function after cardiac surgery: From phenotypes to mechanisms.
Anesthesiology
2018
;
129
:
829
51
3.
Evered
L
,
Silbert
B
,
Knopman
DS
et al
;
Nomenclature Consensus Working Group
:
Recommendations for the nomenclature of cognitive change associated with anaesthesia and surgery-2018.
Br J Anaesth
2018
;
121
:
1005
12
4.
Monk
TG
,
Weldon
BC
,
Garvan
CW
et al
:
Predictors of cognitive dysfunction after major noncardiac surgery.
Anesthesiology
2008
;
108
:
18
30
5.
Berger
M
,
Nadler
JW
,
Browndyke
J
et al
:
Postoperative cognitive dysfunction: Minding the gaps in our knowledge of a common postoperative complication in the elderly.
Anesthesiol Clin
2015
;
33
:
517
50
6.
Eckenhoff
RG
,
Johansson
JS
,
Wei
H
et al
:
Inhaled anesthetic enhancement of amyloid-beta oligomerization and cytotoxicity.
Anesthesiology
2004
;
101
:
703
9
7.
Planel
E
,
Bretteville
A
,
Liu
L
et al
:
Acceleration and persistence of neurofibrillary pathology in a mouse model of tauopathy following anesthesia.
FASEB J
2009
;
23
:
2595
604
8.
Anckarsäter
R
,
Anckarsäter
H
,
Bromander
S
,
Blennow
K
,
Wass
C
,
Zetterberg
H
:
Non-neurological surgery and cerebrospinal fluid biomarkers for neuronal and astroglial integrity.
J Neural Transm (Vienna)
2014
;
121
:
649
53
9.
Berger
M
,
Nadler
JW
,
Friedman
A
et al
;
MAD-PIA Trial Team
:
The effect of propofol versus isoflurane anesthesia on human cerebrospinal fluid markers of Alzheimer’s disease: Results of a randomized trial.
J Alzheimers Dis
2016
;
52
:
1299
310
10.
Palotas
A
,
Reis
HJ
,
Bogats
G
et al
:
Coronary artery bypass surgery provokes Alzheimer’s disease-like changes in the cerebrospinal fluid.
J Alzheimers Dis
2010
;
21
:
1153
64
11.
Reinsfelt
B
,
Ricksten
SE
,
Zetterberg
H
,
Blennow
K
,
Freden-Lindqvist
J
,
Westerlind
A
:
Cerebrospinal fluid markers of brain injury, inflammation, and blood-brain barrier dysfunction in cardiac surgery.
Ann Thorac Surg
2012
;
94
:
549
55
12.
Reinsfelt
B
,
Westerlind
A
,
Blennow
K
,
Zetterberg
H
,
Ricksten
SE
:
Open-heart surgery increases cerebrospinal fluid levels of Alzheimer-associated amyloid beta.
Acta Anaesthesiol Scand
2013
;
57
:
82
8
13.
Tang
JX
,
Baranov
D
,
Hammond
M
,
Shaw
LM
,
Eckenhoff
MF
,
Eckenhoff
RG
:
Human Alzheimer and inflammation biomarkers after anesthesia and surgery.
Anesthesiology
2011
;
115
:
727
32
14.
Olsson
M
,
Ärlig
J
,
Hedner
J
,
Blennow
K
,
Zetterberg
H
:
Repeated lumbar punctures within 3 days may affect CSF biomarker levels.
Fluids Barriers CNS
2019
;
16
:
1
5
15.
Boehnke
SE
,
Robertson
EL
,
Armitage-Brown
B
et al
:
The effect of lumbar puncture on the neurodegeneration biomarker neurofilament light in macaque monkeys.
Alzheimers Dement (Amst)
2020
;
12
:
e12069
16.
Berger
M
,
Browndyke
JN
,
Cooter Wright
M
et al
;
MADCO-PC Investigators
:
Postoperative changes in cognition and cerebrospinal fluid neurodegenerative disease biomarkers.
Ann Clin Transl Neurol
2022
;
9
:
155
70
17.
McDonagh
DL
,
Mathew
JP
,
White
WD
et al
;
Neurologic Outcome Research Group
:
Cognitive function after major noncardiac surgery, apolipoprotein E4 genotype, and biomarkers of brain injury.
Anesthesiology
2010
;
112
:
852
9
18.
Newman
MF
,
Kirchner
JL
,
Phillips-Bute
B
et al
;
Neurological Outcome Research Group and the Cardiothoracic Anesthesiology Research Endeavors Investigators
:
Longitudinal assessment of neurocognitive function after coronary-artery bypass surgery.
N Engl J Med
2001
;
344
:
395
402
19.
Newman
MF
,
Kirchner
JL
,
Phillips-Bute
B
et al
;
Neurological Outcome Research Group and the Cardiothoracic Anesthesiology Research Endeavors Investigators
:
Longitudinal assessment of neurocognitive function after coronary-artery bypass surgery.
N Engl J Med
2001
;
344
:
395
402
20.
Rossi
PD
,
Damanti
S
,
Nani
C
et al
:
Repeated cerebrospinal fluid removal procedure in older patients with idiopathic normal pressure hydrocephalus ineligible for surgical treatment.
J Am Med Dir Assoc
2019
;
20
:
373
376.e3
21.
R Core Team
:
R: A Language and Environment for Statistical Computing
.
Vienna, Austria
,
R Foundation for Statistical Computing
,
2022
22.
Glance
LG
,
Dick
AW
:
In response.
Anesth Analg
2016
;
122
:
1722
7
23.
Shaw
LM
,
Vanderstichele
H
,
Knapik-Czajka
M
et al
:
Cerebrospinal fluid biomarker signature in Alzheimer’s disease neuroimaging initiative subjects.
Ann Neurol
2009
;
65
:
403
13
24.
Shaw
LM
,
Waligorska
T
,
Fields
L
et al
:
Derivation of cutoffs for the Elecsys((R)) amyloid beta (1-42) assay in Alzheimer’s disease.
Alzheimers Dement (Amst)
2018
;
10
:
698
705
25.
Kaufer
DI
,
Williams
CS
,
Braaten
AJ
,
Gill
K
,
Zimmerman
S
,
Sloane
PD
:
Cognitive screening for dementia and mild cognitive impairment in assisted living: Comparison of 3 tests.
J Am Med Dir Assoc
2008
;
9
:
586
93
26.
Scott
JA
,
Tosun
D
,
Braskie
MN
et al
;
ADNI
:
Independent value added by diffusion MRI for prediction of cognitive function in older adults.
Neuroimage Clin
2017
;
14
:
166
73
27.
McDonald
WM
:
Overview of neurocognitive disorders.
Focus (Am Psychiatr Publ)
2017
;
15
:
4
12
28.
Zhang
Y
,
Xu
Z
,
Wang
H
et al
:
Anesthetics isoflurane and desflurane differently affect mitochondrial function, learning, and memory.
Ann Neurol
2012
;
71
:
687
98
29.
Hughes
CG
,
Patel
MB
,
Jackson
JC
et al
;
MIND-ICU, BRAIN-ICU Investigators
:
Surgery and anesthesia exposure is not a risk factor for cognitive impairment after major noncardiac surgery and critical illness.
Ann Surg
2017
;
265
:
1126
33
30.
Nadelson
MR
,
Sanders
RD
,
Avidan
MS
:
Perioperative cognitive trajectory in adults.
Br J Anaesth
2014
;
112
:
440
51
31.
Handley
JD
,
Williams
DM
,
Caplin
S
,
Stephens
JW
,
Barry
J
:
Changes in cognitive function following bariatric surgery: A systematic review.
Obes Surg
2016
;
26
:
2530
7
32.
Buchman
AS
,
Yu
L
,
Wilson
RS
et al
:
Physical activity, common brain pathologies, and cognition in community-dwelling older adults.
Neurology
2019
;
92
:
e811
22
33.
Wang
M
,
Wang
J
,
Li
X
,
Xu
X
,
Zhao
Q
,
Li
Y
:
A predictive model for postoperative cognitive dysfunction in elderly patients with gastric cancer: A retrospective study.
Am J Transl Res
2022
;
14
:
679
86
34.
Arora
RC
,
Djaiani
G
,
Rudolph
JL
:
Detection, prevention, and management of delirium in the critically ill cardiac patient and patients who undergo cardiac procedures.
Can J Cardiol
2017
;
33
:
80
7
35.
Guay
J
,
Kopp
S
:
Peripheral nerve blocks for hip fractures in adults.
Cochrane Database Syst Rev
2020
;
2021
:
CD001159
36.
Hughes
CG
,
Boncyk
CS
,
Culley
DJ
et al
;
Perioperative Quality Initiative (POQI) 6 Workgroup
:
American Society for Enhanced Recovery and Perioperative Quality Initiative joint consensus statement on postoperative delirium prevention.
Anesth Analg
2020
;
130
:
1572
90
37.
Neuman
MD
,
Ellenberg
SS
,
Sieber
FE
,
Magaziner
JS
,
Feng
R
,
Carson
JL
;
REGAIN Investigators
:
Regional versus General Anesthesia for Promoting Independence after Hip Fracture (REGAIN): Protocol for a pragmatic, international multicentre trial.
BMJ Open
2016
;
6
:
e013473
38.
Li
T
,
Li
J
,
Yuan
L
et al
;
RAGA Study Investigators
:
Effect of regional vs general anesthesia on incidence of postoperative delirium in older patients undergoing hip fracture surgery: The RAGA randomized trial.
JAMA
2022
;
327
:
50
8
39.
Kepp
KP
,
Robakis
NK
,
Hoilund-Carlsen
PF
,
Sensi
SL
,
Vissel
B
:
The amyloid cascade hypothesis: an updated critical review.
Brain
2023
;
146
:
3969
90
40.
Lewis
MS
,
Maruff
P
,
Silbert
BS
,
Evered
LA
,
Scott
DA
:
The influence of different error estimates in the detection of post-operative cognitive dysfunction using reliable change indices with correction for practice effects.
Arch Clin Neuropsychol
2006
;
21
:
421
7
41.
Atkins
D
,
Best
D
,
Briss
PA
et al
;
GRADE Working Group
:
Grading quality of evidence and strength of recommendations.
BMJ
2004
;
328
:
1490
42.
Evered
L
,
Silbert
B
,
Scott
DA
,
Ames
D
,
Maruff
P
,
Blennow
K
:
Cerebrospinal fluid biomarker for Alzheimer disease predicts postoperative cognitive dysfunction.
Anesthesiology
2016
;
124
:
353
61
43.
Moller
JT
,
Cluitmans
P
,
Rasmussen
LS
et al
:
Long-term postoperative cognitive dysfunction in the elderly ISPOCD1 study. ISPOCD Investigators. International Study of Post-Operative Cognitive Dysfunction.
Lancet
1998
;
351
:
857
61
44.
Zimpfer
D
,
Czerny
M
,
Vogt
F
et al
:
Neurocognitive deficit following coronary artery bypass grafting: A prospective study of surgical patients and nonsurgical controls.
Ann Thorac Surg
2004
;
78
:
513
8
45.
Johnson
T
,
Monk
T
,
Rasmussen
LS
et al
;
ISPOCD2 Investigators
:
Postoperative cognitive dysfunction in middle-aged patients.
Anesthesiology
2002
;
96
:
1351
7
46.
Goldstein
MZ
,
Fogel
BS
,
Young
BL
:
Effect of elective surgery under general anesthesia on mental status variables in elderly women and men: 10-Month follow-up.
Int Psychogeriatr
1996
;
8
:
135
49
47.
Gilberstadt
H
,
Aberwald
R
,
Crosbie
S
,
Schuell
H
,
Jimenez
E
:
Effect of surgery on psychological and social functioning in elderly patients.
Arch Intern Med
1968
;
122
:
109
15
48.
Avidan
MS
,
Searleman
AC
,
Storandt
M
et al
:
Long-term cognitive decline in older subjects was not attributable to noncardiac surgery or major illness.
Anesthesiology
2009
;
111
:
964
70
49.
Andrew
MJ
,
Baker
RA
,
Bennetts
J
,
Kneebone
AC
,
Knight
JL
:
A comparison of neuropsychologic deficits after extracardiac and intracardiac surgery.
J Cardiothorac Vasc Anesth
2001
;
15
:
9
14
50.
Ancelin
ML
,
de Roquefeuil
G
,
Scali
J
et al
:
Long-term post-operative cognitive decline in the elderly: The effects of anesthesia type, apolipoprotein E genotype, and clinical antecedents.
J Alzheimers Dis
2010
;
22
:
105
13
51.
Abildstrom
H
,
Rasmussen
LS
,
Rentowl
P
et al
:
Cognitive dysfunction 1-2 years after non-cardiac surgery in the elderly. ISPOCD Group. International Study of Post-Operative Cognitive Dysfunction.
Acta Anaesthesiol Scand
2000
;
44
:
1246
51
52.
Relander
K
,
Hietanen
M
,
Nuotio
K
et al
:
Cognitive dysfunction and mortality after carotid endarterectomy.
Front Neurol
2021
;
11
:
593719
53.
Browndyke
JN
,
Berger
M
,
Smith
PJ
et al
;
Duke Neurologic Outcomes Research Group (NORG)
:
Task-related changes in degree centrality and local coherence of the posterior cingulate cortex after major cardiac surgery in older adults.
Hum Brain Mapp
2018
;
39
:
985
1003
54.
Avidan
MS
,
Evers
AS
:
Review of clinical evidence for persistent cognitive decline or incident dementia attributable to surgery or general anesthesia.
J Alzheimers Dis
2011
;
24
:
201
16
55.
Newman
S
,
Stygall
J
,
Hirani
S
,
Shaefi
S
,
Maze
M
:
Postoperative cognitive dysfunction after noncardiac surgery: A systematic review.
Anesthesiology
2007
;
106
:
572
90
56.
Shoair
OA
,
Grasso Ii
MP
,
Lahaye
LA
,
Daniel
R
,
Biddle
CJ
,
Slattum
PW
:
Incidence and risk factors for postoperative cognitive dysfunction in older adults undergoing major noncardiac surgery: A prospective study.
J Anaesthesiol Clin Pharmacol
2015
;
31
:
30
6
57.
Krause
BM
,
Sabia
S
,
Manning
HJ
,
Singh-Manoux
A
,
Sanders
RD
:
Association between major surgical admissions and the cognitive trajectory: 19 Year follow-up of Whitehall II cohort study.
BMJ
2019
;
366
:
l4466
58.
Whitlock
EL
,
Diaz-Ramirez
LG
,
Smith
AK
et al
:
Association of coronary artery bypass grafting vs percutaneous coronary intervention with memory decline in older adults undergoing coronary revascularization.
JAMA
2021
;
325
:
1955
64
59.
Xie
Z
,
Xu
Z
:
General anesthetics and beta-amyloid protein.
Prog Neuropsychopharmacol Biol Psychiatry
2013
;
47
:
140
6
60.
Vasunilashorn
SM
,
Ngo
LH
,
Inouye
SK
et al
:
Apolipoprotein E genotype and the association between C-reactive protein and postoperative delirium: Importance of gene-protein interactions.
Alzheimers Dement
2020
;
16
:
572
80
61.
Schenning
KJ
,
Murchison
CF
,
Mattek
NC
,
Silbert
LC
,
Kaye
JA
,
Quinn
JF
:
Surgery is associated with ventricular enlargement as well as cognitive and functional decline.
Alzheimers Dement
2016
;
12
:
590
7
62.
Bartels
K
,
Li
YJ
,
Li
YW
et al
:
Apolipoprotein epsilon 4 genotype is associated with less improvement in cognitive function five years after cardiac surgery: A retrospective cohort study.
Can J Anaesth
2015
;
62
:
618
26
63.
Inouye
SK
,
Marcantonio
ER
,
Kosar
CM
et al
:
The short-term and long-term relationship between delirium and cognitive trajectory in older surgical patients.
Alzheimers Dement
2016
;
12
:
766
75
64.
Kim
DH
,
Lee
SB
,
Park
CM
et al
:
Comparative safety analysis of oral antipsychotics for in-hospital adverse clinical events in older adults after major surgery: A nationwide cohort study.
Ann Intern Med
2023
;
176
:
1153
62
65.
Saczynski
JS
,
Marcantonio
ER
,
Quach
L
et al
:
Cognitive trajectories after postoperative delirium.
N Engl J Med
2012
;
367
:
30
9
66.
Brown
CH
,
Probert
J
,
Healy
R
et al
:
Cognitive decline after delirium in patients undergoing cardiac surgery.
Anesthesiology
2018
;
129
:
406
16
67.
Pandharipande
P
,
Cotton
BA
,
Shintani
A
et al
:
Prevalence and risk factors for development of delirium in surgical and trauma intensive care unit patients.
J Trauma
2008
;
65
:
34
41

Appendix

The Markers of Alzheimer’s Disease and neuroCognitive Outcomes after Perioperative Care (MADCO-PC) Investigators also include C. L. Amundsen, S. Bengali, E. Bennett, M. F. Berry, D. G. Blazer, M. P. Bolognesi, R. Brassard, B. E. Brigman, M. Bullock, J. Carter, J. Chapman, B. Colin, T. A. D’Amico, J. K. DeOrio, D. Erdmann, R. M. Esclamado, M. Ferrandino, B. Funk, J. Gadsden, J. Gardner, G. Garrigues, C. Giattino, D. T. Gold, S. Grant, J. Guercio, D. K. Gupta, A. Habib, D. H. Harpole, S. M. Harris, M. G. Hartwig, S. T. Hollenbeck, J. Hu, E. Iboaya, B. A. Inman, D. W. Jang, J. Kaisen, A. Khan, S. Lagoo-Deenadayalan, D. T. Laskowitz, P. S. Lee, W. T. Lee, J. Lemm, H. Levinson, M. E. Lipkin, C. R. Mantyh, D. L. McDonagh, J. Migaly, S. K. Mithani, P. Mosca, J. Moul, M. F. Newman, K. Ni, B. Ohlendorf, M. W. Onaitis, T. N. Pappas, A. N. Perez, A. C. Peterson, T. J. Polascik, A. Podgoreanu, G. M. Preminger, Q. Quinones, E. N. Rampersaud, A. Ray, K. Roberts, C. N. Robertson, S. A. Roman, S. Runyon, A. Sandler, F. Sbahi, C. D. Scales, R. P. Scheri, S. K. Smith, L. Talbot, J. K. M. Thacker, J. Thomas, B. C. Tong, Y. Toulgoat-Dubois, A. Tu, S. N. Vaslef, J. Whittle, M. Woldorff, N. Waldron, D. S. Warner, X. Wang, S. S. Wellman, T. Wickenheisser, C. Young, and S. Zani.

This is an open-access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.