Coronavirus disease 2019-related risk factors for postoperative delirium: a secondary analysis of an observational cohort study Authors: Yi-Heng Liu, Chenghong Hu, Xia-Min Yang, Jun-Jie Zhang, Andreas W. Loepke, Meng Deng Published: 2026 ABSTRACT Background:Whether preoperative coronavirus disease 2019 (COVID-19) can affect postoperative neurological outcomes remains unknown. This study aimed to investigate the role of COVID-19-related risk factors (i.e., surgical timing following COVID-19 diagnosis and COVID-19 disease severity) on postoperative delirium (POD).Materials and methods:We performed a secondary analysis of an observational cohort study. Adult patients who underwent elective, noncardiac surgery under general anesthesia were included. Multivariable logistic regression was used to assess the impact of surgical timing relative to COVID-19 (605 individuals) and COVID-19 severity (573 individuals) on POD.Results:After adjustment for baseline characteristics, patients who underwent surgery <7 weeks after COVID-19 diagnosis [adjusted odds ratio (OR), 95% confidence interval (CI): 3.33, 1.43–7.74, P = 0.005], but not those who received surgery ≥7 weeks postinfection, had an increased risk of POD, compared with COVID-19-negative patients. Symptomatic infections (adjusted OR, 95% CI: 3.31, 1.46–7.48, P = 0.004), rather than asymptomatic infections, were associated with POD. These results remained consistent across several sensitivity analyses. In the subgroup analyses by age, recent COVID-19 (<7 weeks) preceding surgery and symptomatic infections were predictors of POD only among patients under 65 years of age. An interplay between COVID-19-related risk factors affected the COVID-19-induced risk of POD. The risk was only increased in symptomatic patients who received surgery <7 weeks after having COVID-19.Conclusion:Recent COVID-19 within 7 weeks preceding surgery and symptomatic infections are associated with POD. Complementary studies are needed to corroborate the impact of preoperative COVID-19 on postoperative neurological complications. Introduction Since 2020, the coronavirus disease 2019 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection has had a dramatic impact on healthcare services. Although the morbidity and mortality resulting from acute SARS-CoV-2 infection have significantly decreased, its long-term influence on multiple organ systems, especially the central nervous system (CNS), has drawn increasing attention, even for vaccinated individuals[1–4]. According to recent studies, up to 85% of COVID-19 patients suffer from neurological and psychiatric symptoms, such as headache, dysgeusia, anosmia, delirium, and cognitive impairment, in the acute and subacute phases[3,5,6]. However, the neurological burden imposed by COVID-19 during the perioperative period has not been extensively studied.HIGHLIGHTSTesting positive for COVID-19 within 7 weeks prior to surgery and symptomatic infections were associated with an increased risk of postoperative delirium.This risk was age-dependent and influenced by an interplay between surgical timing relative to COVID-19 diagnosis and COVID-19 disease severity.Our study indicates that preoperative COVID-19 could compromise perioperative brain health. Individualized assessment of postoperative neurological risks for COVID-19 survivors should be implemented in clinical practice. HIGHLIGHTS Testing positive for COVID-19 within 7 weeks prior to surgery and symptomatic infections were associated with an increased risk of postoperative delirium.This risk was age-dependent and influenced by an interplay between surgical timing relative to COVID-19 diagnosis and COVID-19 disease severity.Our study indicates that preoperative COVID-19 could compromise perioperative brain health. Individualized assessment of postoperative neurological risks for COVID-19 survivors should be implemented in clinical practice. Testing positive for COVID-19 within 7 weeks prior to surgery and symptomatic infections were associated with an increased risk of postoperative delirium. Testing positive for COVID-19 within 7 weeks prior to surgery and symptomatic infections were associated with an increased risk of postoperative delirium. This risk was age-dependent and influenced by an interplay between surgical timing relative to COVID-19 diagnosis and COVID-19 disease severity. This risk was age-dependent and influenced by an interplay between surgical timing relative to COVID-19 diagnosis and COVID-19 disease severity. Our study indicates that preoperative COVID-19 could compromise perioperative brain health. Individualized assessment of postoperative neurological risks for COVID-19 survivors should be implemented in clinical practice. Our study indicates that preoperative COVID-19 could compromise perioperative brain health. Individualized assessment of postoperative neurological risks for COVID-19 survivors should be implemented in clinical practice. Postoperative delirium (POD) is a common complication that can affect a wide range of patients[7]. It is characterized by an acute onset of fluctuating attention and mental status alterations from baseline in the early postoperative period, with an incidence of 5–50%, depending on several factors, such as the type of surgery and patient age[7,8]. Its occurrence is correlated with prolonged length of hospital stay, increased healthcare resource expenditures, and a greater likelihood of developing persistent cognitive decline and dementia[9,10]. To date, there are limited data elucidating the relationship between preoperative COVID-19 and POD. Using the requirement for antipsychotic drugs during postsurgical hospitalization as a surrogate of POD diagnosis, Cata and colleagues reported no association between a history of COVID-19 and POD[11]. However, the inaccurate diagnostic criteria applied in their study cannot completely reflect the postoperative neurological risk imposed by preoperative COVID-19. Additionally, the authors did not specifically investigate COVID-19-related risk factors, such as the timing of surgery relative to COVID-19 and the severity of COVID-19, which could influence the postoperative prognosis of COVID-19 survivors[12,13]. To explore the correlation between preoperative COVID-19 and POD more accurately, we performed a secondary analysis of a recent observational cohort study[14], utilizing data on the abovementioned COVID-19-related risk factors, as well as a POD assessment using the 3-min Diagnostic interview for Confusion Assessment Method (3D-CAM), a well-validated method of delirium diagnosis[15]. We hypothesized that recent COVID-19 preceding surgery (<7 weeks) and symptomatic infections were associated with an increased risk of POD. Materials and methods Study design and participants This study is a post hoc secondary analysis of an ambidirectional cohort study[14], which primarily assessed the association between preoperative COVID-19 and postoperative mortality. The original study was approved by the Institutional Review Board of Huashan Hospital of Fudan University (identifier: KY-2023-074, date of approval: February 28, 2023) and registered in the Chinese Clinical Trial Registry (www.chictr.org.cn, identifier: ChiCTR2300069380, date of registration: March 15, 2023). Informed consent was obtained from all participants. The original cohort consisted of adult patients (≥18 years of age) who underwent elective, noncardiac surgery under general anesthesia at Huashan Hospital of Fudan University from January to March 2023. In this secondary analysis, all patients with postoperative assessments of delirium were included. Patients were excluded if the POD assessment was incomplete or if information on preoperative COVID-19 was missing. The results are reported according to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines[16] and the Strengthening the Reporting of Cohort Studies in Surgery guidelines[17]. Outcome measures The primary outcome was POD, which was assessed by trained research personnel twice daily (between 7 and 10 am and between 5 and 7 pm) with 3D-CAM until hospital discharge or the fifth postprocedural day, whichever came first. The outcome assessors were blinded to the preoperative COVID-19 status. POD was identified if there was at least one positive 3D-CAM assessment. Characteristics of POD, such as duration and severity of POD, numbers of episodes, and time to onset, were all documented. Exposures of interest In this study, we aimed to assess the associations between POD and COVID-19-related risk factors, including the time interval from COVID-19 diagnosis to surgery and COVID-19 severity. Patients were classified as COVID-19-positive based on laboratory tests, radiological findings or clinical diagnosis[14]. On the basis of the timing of surgery relative to COVID-19 diagnosis, COVID-19-positive patients were categorized into <7 weeks and ≥7 weeks, as this time point has been identified as the margin between normal and increased risk of postoperative mortality and pulmonary complications[12,13]. On the basis of patients’ reports, COVID-19 severity was dichotomized to asymptomatic or symptomatic infections. In both scenarios, COVID-19-negative patients were designated as the reference group. Covariates Demographic and clinical data, including age, sex, body mass index (BMI), education level, American Society of Anesthesiologists (ASA) physical status, surgical specialty, and duration of surgery, were retrieved from electronic medical records. The subjective assessment of postoperative pain using the numerical rating scale and overall sleep quality using the Richards–Campbell Sleep Questionnaire, constructed as a visual analog scale (VAS) ranging from 0 (indicating poorest sleep quality) to 100 (indicating optimum sleep)[18], was obtained from postoperative follow-up records. COVID-19 vaccination status was classified as unvaccinated, fully vaccinated without a booster dose, or fully vaccinated with a booster dose. Statistical analysis Statistical analyses were performed using Python v3.10.12 software (CreateSpace, Scotts Valley, CA, USA) and SPSS v23.0 software (IBM, Armonk, NY, USA). In the primary analyses, we separately assessed the impact of surgical timing relative to COVID-19 diagnosis or COVID-19 severity on POD occurrence in two independent multivariable logistic regression models with the Broyden–Fletcher–Goldfarb–Shanno optimizer, adjusting for age, sex, ASA physical status, education level, surgical specialty, duration of surgery, COVID-19 vaccination status, and VAS scores of overall sleep quality. COVID-19 vaccination and sleep quality were added to the regression models because they have been demonstrated to be correlated with the development of neurological diseases[7,19,20]. We assessed model calibration using the Hosmer–Lemeshow goodness-of-fit test and evaluated model discrimination with the receiver operating characteristic (ROC) analysis. Several sensitivity analyses were performed. First, we assessed whether COVID-19-related factors were associated with POD occurring within three postoperative days, in accordance with the diagnostic criteria applied in previous studies[21]. Second, the analyses were restricted to patients with laboratory-confirmed COVID-19 to mitigate the influence of potential misdiagnosis caused by clinical or radiological diagnosis. Third, we introduced two confounding factors of POD, i.e., BMI[22] and postoperative pain[7,10], into the regression models, respectively. Finally, the E-value methodology was used to calculate the effect size of unmeasured confounding, which negates the exposure-outcome associations observed in the primary analyses[23]. To further assess the temporal association between preoperative COVID-19 and POD, surgical timing relative to COVID-19 was treated as a continuous variable in a multivariable regression model. To further determine the characteristics of COVID-19 survivors who are susceptible to POD, two additional analyses were carried out. Given that advanced age is considered a significant demography-related risk factor for POD[24], we conducted subgroup analyses by age (18–64 years; ≥65 years). Additionally, to explore the potential interplay between COVID-19-related risk factors, the impact of surgical timing relative to COVID-19 diagnosis on POD was separately assessed among patients with symptomatic and asymptomatic infections. Characteristics of POD were compared using the Kruskal–Wallis test. A P-value < 0.05 was considered statistically significant in all tests. A power analysis was performed on existing data using the logistic regression procedure of PASS v11.0 software (NCSS LLC, Kaysville, UT, USA). In this study, 13.2% of the COVID-19-negative patients had POD. The proportions of patients with recent and symptomatic infections were 16.2% and 16.6%, whereas their R-squared values with other covariates were 0.02 and 0.03, respectively. Sample sizes of 274 and 272 patients would provide 80% power to detect an OR of 3 or higher for recent and symptomatic infections, respectively, when α = 0.05. Results Timing from COVID-19 diagnosis to surgery and the risk of POD We identified 653 adult patients who underwent elective, noncardiac surgery under general anesthesia and had postoperative assessments of delirium during the study period. After excluding 48 patients who had incomplete assessments, 605 patients remained available for the analysis assessing the impact of surgical timing relative to COVID-19 diagnosis on POD. Among these patients, 129 (21.3%) did not have COVID-19 prior to surgery, whereas 98 (16.2%) and 378 (62.5%) patients were diagnosed with COVID-19 less than and more than 7 weeks before surgery, respectively (Supplemental Digital Content Figure S1, available at: http://links.lww.com/JS9/G469). Compared with the other two groups, patients who lacked a preoperative COVID-19 diagnosis were older, had higher BMI and ASA classifications, and were less likely to have received more education. Compared with those in the other two groups, a significantly lower proportion of patients who had COVID-19 7 weeks or more before surgery were male and unvaccinated (Table 1). Table 1Demographic and clinical characteristics stratified by timing of surgery relative to COVID-19 diagnosis.No preoperative COVID-19, n = 129Time interval from preoperative COVID-19 diagnosis to surgeryMaximum absolute standardized difference<7 weeks, n = 98≥7 weeks, n = 378Age, years60 [49–69]57 [47–64]56 [45–65]0.224Sex0.303 Male82 (63.6%)56 (57.1%)184 (48.7%) Female47 (36.4%)42 (42.9%)194 (51.3%)Body mass index, kg∙m−224.20 [21.88–26.23]23.76 [21.79–25.43]23.64 [21.72–25.89]0.205ASA physical status0.125 I or II110 (85.3%)87 (88.8%)338 (89.4%) ≥III19 (14.7%)11 (11.2%)40 (10.6%)Education level Primary school or lower31 (24.0%)9 (9.2%)67 (17.7%)0.407 High school70 (54.3%)58 (59.2%)199 (52.6%)0.132 College or higher28 (21.7%)31 (31.6%)112 (29.6%)0.226Surgical specialties0.144 High-risk surgerya48 (37.2%)31 (31.6%)115 (30.4%) Other81 (62.8%)67 (68.4%)263 (69.6%)Duration of surgery, min100 [80–150]105 [66–162]105 [70–170]0.066COVID-19 vaccination Unvaccinated18 (14.0%)14 (14.3%)36 (9.5%)0.147 Fully vaccinated without booster dose31 (24.0%)23 (23.5%)118 (31.2%)0.174 Received booster dose80 (62.0%)61 (62.2%)224 (59.3%)0.061COVID-19 disease severity Asymptomatic-66 (67.3%)283 (74.9%)0.168 Symptomatic-27 (27.6%)68 (18.0%)0.230 Missing-5 (5.1%)27 (7.1%)0.084COVID-19 diagnosis Laboratory confirmed-84 (85.7%)323 (85.4%)0.009 Clinical or radiological-11 (11.2%)42 (11.1%)0.003 Missing-3 (3.1%)13 (3.4%)0.017Data are expressed as n (%) or median [25th–75th percentiles] and are compared using absolute standardized difference, with an index <0.100 indicating a negligible difference between groups.aIncluding orthopedic and neurosurgery.COVID-19 indicates coronavirus disease 2019; ASA, American Society of Anesthesiologists. Table 1Demographic and clinical characteristics stratified by timing of surgery relative to COVID-19 diagnosis.No preoperative COVID-19, n = 129Time interval from preoperative COVID-19 diagnosis to surgeryMaximum absolute standardized difference<7 weeks, n = 98≥7 weeks, n = 378Age, years60 [49–69]57 [47–64]56 [45–65]0.224Sex0.303 Male82 (63.6%)56 (57.1%)184 (48.7%) Female47 (36.4%)42 (42.9%)194 (51.3%)Body mass index, kg∙m−224.20 [21.88–26.23]23.76 [21.79–25.43]23.64 [21.72–25.89]0.205ASA physical status0.125 I or II110 (85.3%)87 (88.8%)338 (89.4%) ≥III19 (14.7%)11 (11.2%)40 (10.6%)Education level Primary school or lower31 (24.0%)9 (9.2%)67 (17.7%)0.407 High school70 (54.3%)58 (59.2%)199 (52.6%)0.132 College or higher28 (21.7%)31 (31.6%)112 (29.6%)0.226Surgical specialties0.144 High-risk surgerya48 (37.2%)31 (31.6%)115 (30.4%) Other81 (62.8%)67 (68.4%)263 (69.6%)Duration of surgery, min100 [80–150]105 [66–162]105 [70–170]0.066COVID-19 vaccination Unvaccinated18 (14.0%)14 (14.3%)36 (9.5%)0.147 Fully vaccinated without booster dose31 (24.0%)23 (23.5%)118 (31.2%)0.174 Received booster dose80 (62.0%)61 (62.2%)224 (59.3%)0.061COVID-19 disease severity Asymptomatic-66 (67.3%)283 (74.9%)0.168 Symptomatic-27 (27.6%)68 (18.0%)0.230 Missing-5 (5.1%)27 (7.1%)0.084COVID-19 diagnosis Laboratory confirmed-84 (85.7%)323 (85.4%)0.009 Clinical or radiological-11 (11.2%)42 (11.1%)0.003 Missing-3 (3.1%)13 (3.4%)0.017Data are expressed as n (%) or median [25th–75th percentiles] and are compared using absolute standardized difference, with an index <0.100 indicating a negligible difference between groups.aIncluding orthopedic and neurosurgery.COVID-19 indicates coronavirus disease 2019; ASA, American Society of Anesthesiologists. Demographic and clinical characteristics stratified by timing of surgery relative to COVID-19 diagnosis. Data are expressed as n (%) or median [25th–75th percentiles] and are compared using absolute standardized difference, with an index <0.100 indicating a negligible difference between groups. Including orthopedic and neurosurgery. COVID-19 indicates coronavirus disease 2019; ASA, American Society of Anesthesiologists. The incidence of POD was 13.2% (17/129) among COVID-19-negative patients, and 18.4% (18/98) and 14.6% (55/378) among patients who underwent surgery <7 and ≥7 weeks after COVID-19 diagnosis, respectively. In the primary analysis, there was a significantly greater risk of POD in patients who had COVID-19 less than 7 weeks prior to surgery [adjusted odds ratio (OR), 95% confidence interval (CI): 3.33, 1.43–7.74, P = 0.005, Fig. 1A], compared with COVID-19-negative patients. The Hosmer–Lemeshow test indicated no evidence of lack of fit (P = 0.228). The area under the ROC curve was 0.820 (95% CI, 0.773–0.863). The E-values for the adjusted OR and its lower limit of CI were 6.12 and 2.21, respectively. Sensitivity analyses assessing POD within three postoperative days, restricted to laboratory-confirmed COVID-19 and adding BMI or postoperative pain as a distinct covariate also demonstrated an association between recent COVID-19 within 7 weeks prior to surgery and POD (Supplemental Digital Content Figure S2, available at: http://links.lww.com/JS9/G470). No significant risk of POD was detected in those who underwent surgery 7 weeks or later postinfection in the primary (adjusted OR, 95% CI: 1.49, 0.75-2.99, P = 0.257, Fig. 1A) and sensitivity analyses (Supplemental Digital Content Figure S2, available at: http://links.lww.com/JS9/G470). A lower education level, longer duration of surgery, and lower VAS score of postoperative sleep quality were also predictors of POD in the primary and sensitivity analyses, whereas COVID-19 vaccination did not reduce the risk of POD (Fig. 1A and Supplemental Digital Content Figure S2, available at: http://links.lww.com/JS9/G470). When the time from COVID-19 diagnosis to surgery was modeled as a continuous variable, the risk of POD decreased with each additional week (adjusted OR, 95% CI: 0.75, 0.63–0.89, P = 0.001, Supplemental Digital Content Figure S3, available at: http://links.lww.com/JS9/G471). Compared with that among patients under 65 years of age, the incidence of POD was greater among patients aged ≥65 years [10.6% (45/424) vs 24.9% (45/181), P < 0.001]. When stratifying the primary analysis by age, COVID-19 within 7 weeks preceding surgery remained as an independent risk factor for POD among patients aged 18–64 years, but not those ≥ 65 years of age (Fig. 1B-C). In contrast, better sleep quality was associated with a reduced risk of POD only among patients aged 65 years or above but not among younger patients (Fig. 1B-C). The duration, severity and episodes of POD, and the time to onset of delirium symptoms were not significantly different among the exposure groups (Table 2). Figure 1.Multivariable regression models assessing the associations between the timing from COVID-19 diagnosis to surgery and postoperative delirium: primary analysis (A) and subgroup analyses for patients aged 18–64 years (B) and ≥65 years (C). The graphs show adjusted ORs (boxes) with 95% CI (whiskers). ASA indicates American Society of Anesthesiologists; COVID-19, coronavirus disease 2019; OR, odds ratio; CI, confidence interval. Table 2Characteristics of postoperative delirium stratified by timing of surgery relative to COVID-19 diagnosis.No preoperative COVID-19Time interval from preoperative COVID-19 diagnosis to surgeryP-value<7 weeks≥7 weeksDuration of postoperative deliriuma, days1 [1–2]1 [1–1]1 [1–1]0.695Severity of postoperative deliriumb8 [6–9]5 [4–8]7 [5–8]0.104No. of postoperative delirium episodesc1 [1–1]1 [1–1]1 [1–1]0.851Time to onset of postoperative delirium, days1 [1–2]1 [1–2]2 [1–4]0.057Data are expressed as median [25th–75th percentiles] and are compared using the Kruskal–Wallis test.aDuration of delirium was defined as the period between the first and last delirium-positive day, even if there were nondelirious days in between.bSeverity of postoperative delirium was the maximum score on a 20-point scoring scale based on 3D-CAM assessments, with higher scores reflecting a worse condition[25,26].cPostoperative delirium episodes were defined as discrete delirium-positive calendar days. If there were discrete delirium-positive calendar days on day 2 and 4 postoperatively, this counted as two postoperative delirium episodes. If the delirium-positive calendar days were consecutive (e.g. days 1, 2, and 3), this was defined as one episode.COVID-19 indicates coronavirus disease 2019. Multivariable regression models assessing the associations between the timing from COVID-19 diagnosis to surgery and postoperative delirium: primary analysis (A) and subgroup analyses for patients aged 18–64 years (B) and ≥65 years (C). The graphs show adjusted ORs (boxes) with 95% CI (whiskers). ASA indicates American Society of Anesthesiologists; COVID-19, coronavirus disease 2019; OR, odds ratio; CI, confidence interval. Table 2Characteristics of postoperative delirium stratified by timing of surgery relative to COVID-19 diagnosis.No preoperative COVID-19Time interval from preoperative COVID-19 diagnosis to surgeryP-value<7 weeks≥7 weeksDuration of postoperative deliriuma, days1 [1–2]1 [1–1]1 [1–1]0.695Severity of postoperative deliriumb8 [6–9]5 [4–8]7 [5–8]0.104No. of postoperative delirium episodesc1 [1–1]1 [1–1]1 [1–1]0.851Time to onset of postoperative delirium, days1 [1–2]1 [1–2]2 [1–4]0.057Data are expressed as median [25th–75th percentiles] and are compared using the Kruskal–Wallis test.aDuration of delirium was defined as the period between the first and last delirium-positive day, even if there were nondelirious days in between.bSeverity of postoperative delirium was the maximum score on a 20-point scoring scale based on 3D-CAM assessments, with higher scores reflecting a worse condition[25,26].cPostoperative delirium episodes were defined as discrete delirium-positive calendar days. If there were discrete delirium-positive calendar days on day 2 and 4 postoperatively, this counted as two postoperative delirium episodes. If the delirium-positive calendar days were consecutive (e.g. days 1, 2, and 3), this was defined as one episode.COVID-19 indicates coronavirus disease 2019. Characteristics of postoperative delirium stratified by timing of surgery relative to COVID-19 diagnosis. Data are expressed as median [25th–75th percentiles] and are compared using the Kruskal–Wallis test. Duration of delirium was defined as the period between the first and last delirium-positive day, even if there were nondelirious days in between. Severity of postoperative delirium was the maximum score on a 20-point scoring scale based on 3D-CAM assessments, with higher scores reflecting a worse condition[25,26]. Postoperative delirium episodes were defined as discrete delirium-positive calendar days. If there were discrete delirium-positive calendar days on day 2 and 4 postoperatively, this counted as two postoperative delirium episodes. If the delirium-positive calendar days were consecutive (e.g. days 1, 2, and 3), this was defined as one episode. COVID-19 indicates coronavirus disease 2019. Severity of COVID-19 and the risk of POD A total of 653 patients were screened. Eighty patients were excluded because of missing values for COVID-19 severity (32 patients) or incomplete assessment of POD (48 patients). The data of the remaining 573 patients were used to explore the relationship between COVID-19 disease severity and POD. Of the eligible patients, 129 (22.5%) were COVID-19-negative, and 349 (60.9%) and 95 (16.6%) had asymptomatic and symptomatic infections, respectively (Supplemental Digital Content Figure S4, available at: http://links.lww.com/JS9/G472). Compared with COVID-19-positive patients, a greater proportion of patients without COVID-19 were male and underwent high-risk surgery. Compared with the other two groups, asymptomatic patients were younger, had lower ASA classifications, and received higher levels of education, whereas symptomatic patients were less likely to be unvaccinated (Table 3). Table 3Demographic and clinical characteristics stratified by COVID-19 disease severity.No preoperative COVID-19, n = 129COVID-19 disease severityMaximum absolute standardized differenceAsymptomatic infections, n = 349Symptomatic infections, n = 95Age, years60 [49–69]56 [45–65]58 [47–68]0.254Sex0.352 Male82 (63.6%)175 (50.1%)44 (46.3%) Female47 (36.4%)174 (49.9%)51 (53.7%)Body mass index, kg∙m−224.20 [21.88–26.23]23.67 [21.60–25.59]23.66 [22.03–26.84]0.087ASA physical status0.267 I or II110 (85.3%)318 (91.1%)78 (82.1%) ≥III19 (14.7%)31 (8.9%)17 (17.9%)Education level Primary school or lower31 (24.0%)49 (14.0%)19 (20.0%)0.257 High school70 (54.3%)189 (54.2%)47 (49.5%)0.096 College or higher28 (21.7%)111 (31.8%)29 (30.5%)0.230Surgical specialties0.259 High-risk surgerya48 (37.2%)109 (31.2%)24 (25.3%) Other81 (62.8%)240 (68.8%)71 (74.7%)Duration of surgery, min100 [80–150]100 [65–165]110 [70–160]0.071COVID-19 vaccination Unvaccinated18 (14.0%)42 (12.0%)4 (4.2%)0.344 Fully vaccinated without booster dose31 (24.0%)105 (30.1%)27 (28.4%)0.137 Received booster dose80 (62.0%)202 (57.9%)64 (67.4%)0.197Time interval from preoperative COVID-19 diagnosis to surgery0.225 < 7 weeks-66 (18.9%)27 (28.4%) ≥ 7 weeks-283 (81.1%)68 (71.6%)COVID-19 diagnosis Laboratory confirmed-311 (89.1%)84 (88.4%)0.022 Clinical or radiological-32 (9.2%)10 (10.5%)0.044 Missing-6 (1.7%)1 (1.1%)0.051Data are expressed as n (%) or median [25th-75th percentiles] and are compared using absolute standardized difference, with an index < 0.100 indicating a negligible difference between groups.aIncluding orthopedic and neurosurgery.COVID-19 indicates coronavirus disease 2019; ASA, American Society of Anesthesiologists. Table 3Demographic and clinical characteristics stratified by COVID-19 disease severity.No preoperative COVID-19, n = 129COVID-19 disease severityMaximum absolute standardized differenceAsymptomatic infections, n = 349Symptomatic infections, n = 95Age, years60 [49–69]56 [45–65]58 [47–68]0.254Sex0.352 Male82 (63.6%)175 (50.1%)44 (46.3%) Female47 (36.4%)174 (49.9%)51 (53.7%)Body mass index, kg∙m−224.20 [21.88–26.23]23.67 [21.60–25.59]23.66 [22.03–26.84]0.087ASA physical status0.267 I or II110 (85.3%)318 (91.1%)78 (82.1%) ≥III19 (14.7%)31 (8.9%)17 (17.9%)Education level Primary school or lower31 (24.0%)49 (14.0%)19 (20.0%)0.257 High school70 (54.3%)189 (54.2%)47 (49.5%)0.096 College or higher28 (21.7%)111 (31.8%)29 (30.5%)0.230Surgical specialties0.259 High-risk surgerya48 (37.2%)109 (31.2%)24 (25.3%) Other81 (62.8%)240 (68.8%)71 (74.7%)Duration of surgery, min100 [80–150]100 [65–165]110 [70–160]0.071COVID-19 vaccination Unvaccinated18 (14.0%)42 (12.0%)4 (4.2%)0.344 Fully vaccinated without booster dose31 (24.0%)105 (30.1%)27 (28.4%)0.137 Received booster dose80 (62.0%)202 (57.9%)64 (67.4%)0.197Time interval from preoperative COVID-19 diagnosis to surgery0.225 < 7 weeks-66 (18.9%)27 (28.4%) ≥ 7 weeks-283 (81.1%)68 (71.6%)COVID-19 diagnosis Laboratory confirmed-311 (89.1%)84 (88.4%)0.022 Clinical or radiological-32 (9.2%)10 (10.5%)0.044 Missing-6 (1.7%)1 (1.1%)0.051Data are expressed as n (%) or median [25th-75th percentiles] and are compared using absolute standardized difference, with an index < 0.100 indicating a negligible difference between groups.aIncluding orthopedic and neurosurgery.COVID-19 indicates coronavirus disease 2019; ASA, American Society of Anesthesiologists. Demographic and clinical characteristics stratified by COVID-19 disease severity. Data are expressed as n (%) or median [25th-75th percentiles] and are compared using absolute standardized difference, with an index < 0.100 indicating a negligible difference between groups. Including orthopedic and neurosurgery. COVID-19 indicates coronavirus disease 2019; ASA, American Society of Anesthesiologists. POD occurred in 13.2% (17/129) of COVID-19-negative patients, 11.5% (40/349) of asymptomatic patients, and 23.2% (22/95) of symptomatic patients. In the primary analysis, patients with symptomatic infections (adjusted OR, 95% CI: 3.31, 1.46–7.48, P = 0.004), but not asymptomatic infections (adjusted OR, 95% CI: 1.35, 0.67–2.74, P = 0.406, Fig. 2A), were at significantly greater risk of developing POD, compared with COVID-19-naïve patients. This model had no evidence of miss-fit (the Hosmer–Lemeshow test, P = 0.113) and had an area under the ROC curve of 0.808 (95% CI, 0.758–0.854). These results were consistent in the sensitivity analyses assessing POD within three postoperative days, restricted to laboratory-confirmed COVID-19 and adding BMI or postoperative pain as a distinct covariate (Supplemental Digital Content Figure S5, available at: http://links.lww.com/JS9/G473). The E-values for the adjusted OR and the lower CI bound for POD in symptomatic patients observed in the primary analysis were 6.08 and 2.28, respectively. In the primary and sensitivity analyses, education level, duration of surgery, and the VAS score of postoperative sleep quality were associated with the development of POD, whereas vaccination did not reduce the risk of POD (Fig. 2A and Supplemental Digital Content Figure S5, available at: http://links.lww.com/JS9/G473). Compared with younger patients, POD was more prevalent among patients ≥ 65 years of age [9.8% (39/400) vs 23.1% (40/173), P < 0.001]. In the subgroup analyses by age, a significant association between symptomatic infections and POD was observed only in patients under 65 years of age. Asymptomatic infections did not confer any added risk of POD in any age subgroup (Fig. 2B-C). In patients aged ≥ 65 years, but not younger patients, lower VAS scores of postoperative sleep quality were associated with POD (Fig. 2B-C). The duration, severity and episodes of POD, and the time to onset of delirium did not differ across the exposure groups (Table 4). Figure 2.Multivariable regression models assessing the associations between COVID-19 disease severity and postoperative delirium: primary analysis (A) and subgroup analyses for patients aged 18–64 years (B) and ≥65 years (C). The graphs show adjusted ORs (boxes) with 95% CIs (whiskers). ASA indicates American Society of Anesthesiologists; COVID-19, coronavirus disease 2019; OR, odds ratio; CI, confidence interval. Table 4Characteristics of postoperative delirium stratified by COVID-19 disease severity.No preoperative COVID-19COVID-19 disease severityP valueAsymptomatic infectionsSymptomatic infectionsDuration of postoperative deliriuma, days1 [1–2]1 [1–1]1 [1–1]0.699Severity of postoperative deliriumb8 [6–9]6 [5–8]7 [4–9]0.316No. of postoperative delirium episodesc1 [1–1]1 [1–1]1 [1–1]0.945Time to onset of postoperative delirium, d1 [1–2]2 [1–3]2 [1–3]0.414Data are expressed as median [25th–75th percentiles] and are compared using the Kruskal–Wallis test.aDuration of delirium was defined as the period between the first and last delirium-positive day, even if there were nondelirious days in between.bSeverity of postoperative delirium was the maximum score on a 20-point scoring scale based on 3D-CAM assessments, with higher scores reflecting a worse condition[25,26].cPostoperative delirium episodes were defined as discrete delirium-positive calendar days. If there were discrete delirium-positive calendar days on day 2 and 4 postoperatively, this counted as two postoperative delirium episodes. If the delirium-positive calendar days were consecutive (e.g. days 1, 2, and 3), this was defined as one episode.COVID-19 indicates coronavirus disease 2019. Multivariable regression models assessing the associations between COVID-19 disease severity and postoperative delirium: primary analysis (A) and subgroup analyses for patients aged 18–64 years (B) and ≥65 years (C). The graphs show adjusted ORs (boxes) with 95% CIs (whiskers). ASA indicates American Society of Anesthesiologists; COVID-19, coronavirus disease 2019; OR, odds ratio; CI, confidence interval. Table 4Characteristics of postoperative delirium stratified by COVID-19 disease severity.No preoperative COVID-19COVID-19 disease severityP valueAsymptomatic infectionsSymptomatic infectionsDuration of postoperative deliriuma, days1 [1–2]1 [1–1]1 [1–1]0.699Severity of postoperative deliriumb8 [6–9]6 [5–8]7 [4–9]0.316No. of postoperative delirium episodesc1 [1–1]1 [1–1]1 [1–1]0.945Time to onset of postoperative delirium, d1 [1–2]2 [1–3]2 [1–3]0.414Data are expressed as median [25th–75th percentiles] and are compared using the Kruskal–Wallis test.aDuration of delirium was defined as the period between the first and last delirium-positive day, even if there were nondelirious days in between.bSeverity of postoperative delirium was the maximum score on a 20-point scoring scale based on 3D-CAM assessments, with higher scores reflecting a worse condition[25,26].cPostoperative delirium episodes were defined as discrete delirium-positive calendar days. If there were discrete delirium-positive calendar days on day 2 and 4 postoperatively, this counted as two postoperative delirium episodes. If the delirium-positive calendar days were consecutive (e.g. days 1, 2, and 3), this was defined as one episode.COVID-19 indicates coronavirus disease 2019. Characteristics of postoperative delirium stratified by COVID-19 disease severity. Data are expressed as median [25th–75th percentiles] and are compared using the Kruskal–Wallis test. Duration of delirium was defined as the period between the first and last delirium-positive day, even if there were nondelirious days in between. Severity of postoperative delirium was the maximum score on a 20-point scoring scale based on 3D-CAM assessments, with higher scores reflecting a worse condition[25,26]. Postoperative delirium episodes were defined as discrete delirium-positive calendar days. If there were discrete delirium-positive calendar days on day 2 and 4 postoperatively, this counted as two postoperative delirium episodes. If the delirium-positive calendar days were consecutive (e.g. days 1, 2, and 3), this was defined as one episode. COVID-19 indicates coronavirus disease 2019. Interplay between timing of surgery and COVID-19 disease severity The influence of the time interval between COVID-19 diagnosis and surgery on POD among patients with asymptomatic and symptomatic infections was assessed. Among patients who had symptomatic disease, the risk of POD was increased only in those who underwent surgery less than 7 weeks after having COVID-19. Patients with asymptomatic infections did not demonstrate an increased risk, regardless of the timing of surgery (Supplemental Digital Content Figure S6, available at: http://links.lww.com/JS9/G474). Discussion The impact of preoperative COVID-19 on postoperative prognosis has been a topic of concern in perioperative medicine, with postoperative mortality and pulmonary complications being investigated the most[12,27,28]. However, to date, only limited information is available regarding the effects of preoperative COVID-19 on neurological morbidity following surgery and anesthesia. This issue is worthy of particular attention because the brain is among the organs most significantly affected by COVID-19[5,29], and postoperative neurological complications can negatively affect functional outcomes and long-term quality of life[8]. Herein, we assessed the correlation between preoperative COVID-19 and POD, which is considered a key component of perioperative brain health[7]. To our knowledge, this is the first study that explores the impact of temporal proximity from COVID-19 diagnosis to surgery and COVID-19 disease severity, as well as their interplay, on the risk of POD. Our analysis suggested that (1) COVID-19 within 7 weeks prior to surgery, as well as symptomatic infections, was associated with an increased risk of POD; (2) this risk was influenced by an interplay between surgical timing relative to COVID-19 diagnosis and COVID-19 disease severity; (3) as with previous studies, POD was more prevalent among aged patients; however, those under 65 years of age were more susceptible to COVID-19-induced POD; (4) each additional week between COVID-19 diagnosis and surgery was associated with a 25% reduction in the risk of POD; and (5) vaccination against COVID-19 was not a protective factor against developing POD. Notably, according to the Chinese Center for Disease Control and Prevention, the genome sequences from domestic cases were almost exclusively Omicron variants during the patient recruitment period[30]. Hence, our data primarily reflect the neurological burden imposed by the currently circulating variant of SARS-CoV-2[14,31] and indicate that although the Omicron variant is less virulent than its ancestral strains[32], it still compromises perioperative brain health. Our findings provide perioperative evidence that COVID-19 can affect the CNS. Indeed, in non-operative settings, the neurological symptoms and post-acute neurological sequelae of COVID-19 are common among infected patients, with a documented incidence of 30–85%[3,33,34]. These neurological problems are more prevalent among patients with severe infections[4,33] and can persist for 3–9 months or even longer[1–3,34,35]. Notably, the neurological manifestations of COVID-19 can gradually recover over time but might also recur during recovery[2,3]. Hence, COVID-19 survivors might have a greater likelihood of developing POD. This study is of clinical importance by demonstrating that symptomatic infections and 7 weeks postinfection are two important boundaries between increased and normal risk of delirium after elective surgery. Moreover, on the basis of multi-omics data, the pathomechanisms of COVID-19-related neurological abnormalities include neuroinflammation, blood–brain barrier (BBB) disruption, and dysregulated coagulation[36], which overlap with the pathophysiology of POD[10]. These common pathways could provide the pathophysiological basis for explaining how preoperative COVID-19 status can affect the occurrence of POD. Our findings not only emphasize the importance of recognizing the risk of POD among COVID-19 survivors but also encourage future research into the associations between preoperative COVID-19 and other forms of postoperative neurological complications. To facilitate an individualized assessment of postoperative neurological risk, we also identified the characteristics of at-risk patients. Consistent with previous evidence[8,10], we found that POD was more prevalent among aged patients; however, the impact of preoperative COVID-19 on POD was more pronounced in younger patients. Specifically, in the subgroup analyses by age, recent and symptomatic COVID-19 prior to surgery were predisposing factors for POD among patients under 65 years of age, but not those with advanced age, compared with their respective controls who were COVID-19-negative and in the same age subgroup. These findings are supported by recent observations that compared with elderly patients, young patients aged < 65 years are disproportionately affected by in-hospital delirium[37] and neurological sequelae subsequent to COVID-19[3,35,38]. These results indicate that in the context of the ongoing COVID-19 pandemic, maintaining perioperative brain health is indispensable not only for elderly patients, but also for younger patients who are more susceptible to COVID-19-induced POD. While the specific biological causes remain unknown, these findings might suggest the involvement of the immune system in the development of POD subsequent to COVID-19. During the process of aging, the immune system undergoes substantial changes in cell population and function, which ultimately contribute to immune decline[39–41]. Immune cells, including lymphocytes and microglia, contribute to crucial neuropathological changes, such as neuroinflammation, neuronal damage, and BBB breakdown, in both COVID-19 neuropathy[42–44] and POD[45,46]. Therefore, it is plausible that the age-related disparities in immune function may account for the divergent risk of POD imposed by preoperative COVID-19 in different age subgroups. Future studies are needed to elucidate the specific subsets of immune cells responsible for the age-related risk of POD in COVID-19 survivors, along with the underlying mechanisms of this intriguing phenomenon. It is also interesting that poor sleep quality was related to the prevalence of POD only among patients who were 65 years or older. Therefore, optimizing perioperative sleep might especially benefit this group of patients. In the assessment of the interplay between COVID-19-related factors, undergoing surgery less than 7 weeks after COVID-19 diagnosis increased the risk of POD in symptomatic rather than asymptomatic patients, indicating that the duration of COVID-19-induced risk of POD may be dependent on COVID-19 severity. Collectively, these results suggest that the implementation of multidisciplinary delirium care programs should be considered, especially for COVID-19 survivors between 18 and 64 years of age and those who have recent and symptomatic infections. After identifying the impact of previous COVID-19 on POD, it is crucial to explore risk-reducing interventions against the primary disease. Given the association between COVID-19 disease severity and POD observed in this study, COVID-19 vaccination could theoretically alleviate this risk because it has been proven effective in relieving disease severity[4,47]. Zhang and colleagues demonstrated that a single dose of the COVID-19 vaccine was effective in ameliorating POD in aged patients without COVID-19[48]. However, in this study, we did not find a protective role of complete COVID-19 vaccination or booster doses. Possible explanations are three-fold. First, compared with previous variants, the Omicron variant showed substantial neutralization escape[49–52]. As a result, the effectiveness of current vaccination protocols against the Omicron variant has significantly decreased[32]. Second, vaccination-induced immunity could be influenced by various factors, including the type of vaccine and the time interval from vaccination to surgery[32,53–56], which were not taken into consideration in this study or by Zhang and colleagues. Third, although vaccination can mitigate COVID-19-induced neuroinflammation, loss of neurogenesis, and subsequent memory deficits[57], the incidence of neurological complications following vaccination is also not insignificant[19,20]. Hence, the overall effect of COVID-19 vaccination on POD could be a consequence of both beneficial and detrimental effects. Given the complicated mechanisms by which vaccination interferes with COVID-19, human immunization, and the CNS, more elaborate and focused studies would be required. Although we did not provide direct evidence on how preoperative COVID-19 status could affect POD occurrence, this study indicated that advanced age and preoperative COVID-19 status were independent risk factors for POD. Additionally, the impact of preoperative COVID-19 on POD was age dependent. Young patients less than 65 years of age were more susceptible to COVID-19-induced POD. As SARS-CoV-2 is still resulting in tens of thousands of infections worldwide per week[31], there would be an urgent need to elucidate the underlying mechanisms. The results would be of great clinical significance not only for preoperative risk stratification, but also for the prevention and treatment of the neurological sequelae subsequent to COVID-19. There are several limitations to this study. First, owing to the observational nature of this study, the influence of residual confounding factors cannot be ruled out. However, an E-value analysis demonstrated that an extremely strong unmeasured confounder with an effect size exceeding 6 would be required to obviate the observed association. Given that this value is much greater than the adjusted ORs of measured covariates in this study, as well as those of unmeasured risk factors reported by previous research such as preoperative cognitive decline[58], it is unlikely that an unknown confounder would refute the impact of preoperative COVID-19-related risk factors on POD. Second, although symptomatic COVID-19 is associated with POD, we cannot determine which specific symptoms are responsible for the observed association between symptomatic infections and POD. Third, the statistical power for subgroup analyses and assessment of the interplay between COVID-19 risk factors might be limited. The implications of these analyses should be tested in future studies. Fourth, the single-center design may have limited the generalizability of the results. Fifth, we cannot further evaluate whether preoperative COVID-19 affects other characteristics of POD, such as the motoric subtypes, owing to the lack of relevant data. Conclusion Our study suggests that recent COVID-19 within 7 weeks prior to surgery, as well as symptomatic infections, increases the risk of POD. This risk might be more prominent in younger patients aged <65 years and symptomatic patients who underwent surgery <7 weeks after COVID-19 diagnosis. Our study underscores the importance of recognizing preoperative COVID-19 as a potential precipitating factor for postoperative neurological complications.