
2023 Impact Factor
Dementia is characterized by cognitive and functional decline and poses a significant economic and societal burden. Characterized by the loss of cognitive and emotional abilities, behavioral changes, and the ability to perform daily activities, dementia commonly occurs in later life due to neurodegenerative and cerebrovascular processes [1]. Dementia significantly burdens both patients and their caregivers, stemming not only from cognitive decline but also from its behavioral and psychological symptoms [2]. Due to the current lack of disease-modifying treatments and its substantial impact as a public health concern, dementia is projected to incur greater costs in the future [3,4]. Thus, identifying risk factors and preventing or delaying the clinical onset of dementia are becoming increasingly important. Several potentially modifiable risk factors have been identified, including diabetes, hypertension, midlife obesity, excessive alcohol consumption, smoking, hearing loss, traumatic brain injury, lack of physical activity, social isolation, and psychiatric disorder such as depression [5,6].
Recently, anesthesia administered during surgical or interventional procedures has gained increasing attention as a potentially modifiable risk factor for dementia. Post-operative cognitive decline (POCD), which is cognitive decline lasting hours, days, or months after major surgery, has been reported in the literature [7]. However, recent studies have reported varying results regarding whether POCD leads to dementia [8].
Anesthesia is categorized into general, regional, and local anesthesia, depending on the extent and depth of anesthesia. General anesthesia is further divided into intravenous and inhalation anesthesia [9]. Some studies have shown that inhalation anesthesia is associated with a higher risk of dementia compared to intravenous general anesthesia or regional anesthesia [10], while others have not shown such a difference [11-13], resulting in mixed results. A possible reason for these mixed results may be the use of different anesthetic agents, even with the same type of anesthesia.
Among general anesthetics, propofol, midazolam, etomidate, and ketamine are widely used for intravenous anesthesia [14-16], and desflurane, isoflurane, and sevoflurane are currently used for inhalation anesthesia [17]. Studies investigating the cognitive effects of individual anesthetic agents have been conducted primarily in animal models [18-22], with some studies measuring cognitive changes occurring from hours to days following surgery. Several studies have explored the association between general anesthesia and the development of dementia in clinical populations [11-13,23,24]; however, few studies have explored the association between individual anesthetic agents used in general anesthesia and dementia incidence.
In this study, we aim to compare the incidence of dementia according to (1) the mode of anesthesia and (2) individual anesthetic agents used for anesthesia, using nationwide cohort sample data to investigate the correlation between anesthesia and subsequent dementia diagnosis.
The Health Insurance Review and Assessment (HIRA) Service (https://opendata.hira.or.kr/home.do) was used to access the claims data from the national healthcare system. HIRA contains claims data for approximately 98% of the South Korean population and has been used in various epidemiological studies [25]. This study was approved by the Institutional Review Board of Korea University Guro Hospital (IRB number 2020GR0508). The need for written consent was waived because we used the South Korean HIRA dataset, which consists of de-identified secondary data for research purposes.
Initially, a total of 192,695 individuals aged 40 years or older at the time of inclusion were recruited for the retrospective cohort from the HIRA database. The aim was to achieve a sufficient sample size suitable for statistical analysis, enabling comparison between the groups exposed and not exposed to anesthesia in the HIRA data-base. Furthermore, the inclusion criteria for the entire cohort sample were primarily determined based on insurance claim codes with minimal impact on the cognitive function of individuals. The insurance claim code ‘Rash’ (R21) was chosen as it met the aforementioned criteria. Consequently, individuals who submitted insurance claims with the R21 code between 2007 and 2020 were primarily included in the cohort sample.
The exclusion criteria included a washout period wherein individuals diagnosed with dementia within 5 years prior (2007−2011) to the follow-up assessment period of dementia development and those with access to care under the diagnoses of Parkinson’s disease (G20 codes), cerebral infarction (I63 codes), cerebral hemorrhage (I60, I62, S06 codes), which may have a direct impact on dementia development and cognitive function, and a history of brain surgery in this period (S4615, S4616, S4621, S4622, S4625, S4634-7, S4633, S4641, S4642, S4653-8, S4661, S4662, S4670, S4671, S4681-5, S4711-4713, S4721-4724, S4731-7, S0471-6, S4730, S4741-4, S4756-8, S4760, S4771-2, S4780, S4796, S4792-4, S4797-9, S0479, S4801-3, S4805 codes) were also excluded. Additionally, we excluded participants with a cancer diagnosis (C-code) from the final analysis.
Within the chosen study samples, our investigation focused on examining the progression of dementia following anesthesia. Specifically, we explored whether individuals who underwent general anesthesia, local, or regional anesthesia, or a combination of general and local/regional anesthesia for the first time between 2012 and 2013 subsequently sought medical care with a diagnosis of dementia up to the year 2020. We identified indivi-duals who received anesthesia by recognizing the HIRA claim codes for intravenous (L0101, L0102, L0103, and L0104), inhalation (L1211, L1212, L1221, L1222, L6010-L6970, L7010-L7970), and neuraxial/local anesthesia (L0201, L0410, L0402, L0403, L0404, L0405, L0406, L0407, L0411, L0412, L0413, L0414, L0415, L1210, L1213, L1214, L1215, L1223, L1224, L1225). The control group initially included only patients who had never received anesthesia during the study period (2012−2020).
For the individual anesthetic agent-based analysis, we confirmed the administered anesthesia agents in the sample using the drug codes of the HIRA claims data: propofol (2198xxxxx), midazolam (1952xxxxx), ketamine (1795xxxxx), etomidate (1570xxxxx), sevoflurane (2274xxxxx), desflurane (1413xxxxx), isoflurane (1780xxxxx), lidocaine (1839, 3140, 3141, 3143, 3145, 6498, 1838, 4546xxxxx), levobupivacaine (4306xxxxx), bupivacaine (1201xxxxx), ropivacaine (2250xxxxx), and mepivacaine (1904xxxxx). Subgroups were formed according to the individual agents used for anesthesia. Individuals who were unsure of their anesthetics were excluded from the analyses based on individual anesthetic agents used. A schematic of the cohort study design is shown in Figure 1.
Within these subject groups, we evaluated the incidence of individuals diagnosed with dementia after their first administration of anesthesia, defined as those who 1) had received the following dementia diagnosis codes (F00, F01, F02, F03, G30, G31) in at least two separate visits to a medical doctor and 2) were prescribed cognitive enhancers (donepezil, rivastigmine, galantamine, memantine).
The presence of diabetes, hypertension, hyperlipide-mia, and other coexisting conditions among participants during the study period, which was used to calculate the Charlson comorbidity index (CCI) [26], was also investi-gated. The CCI categorizes a patient’s comorbidities based on the International Classification of Diseases (ICD) diagnosis codes found in administrative data, such as hospital abstract data. Each comorbidity category had an associated weight (1−6) based on the adjusted risk of mortality or resource use. The sum of all weights resulted in a single comorbidity score for each patient. A higher score indicates a greater likelihood of the predicted outcome resulting in death or greater resource use [26]. Age, sex, and CCI were used as covariates in multivariate models to analyze the effect of anesthesia on the development of dementia.
General characteristics were compared between those who received anesthesia with 1) different modes of anesthesia and 2) different individual anesthetics using analysis of variance (ANOVA) for continuous variables and the chi-squared test for categorical variables. After performing ANOVA in the comparison by anesthesia mode, additional post-hoc analysis was conducted, incorporating Bonferroni correction to address the issue of multiple comparisons. A pvalue less than 0.05 was considered statistically significant. Nevertheless, in the comparison of individual anesthetic drugs, the threshold for determining statistical significance was adjusted to 0.05/9 = 0.005 through Bonferroni correction, mitigating errors arising from multiple comparisons. For multivariable analysis evaluating the effect of different types of anesthetics on the incidence of dementia, the Cox proportional hazard model was used to derive hazard ratios (HRs) for 1) different modes of anesthesia and 2) individual anesthetic agents on dementia incidence. A pvalue less than 0.05 was considered significant. All analyses were conducted using the Statistical Analysis System Software (version 9.4; SAS Institute).
The final analysis included a total of 62,541 participants. At the start of data collection in 2012, these participants had no history of anesthesia or diagnosis of dementia within the previous 5 years. Among them, 15,857 participants received anesthesia between 2012 and 2013; of these, 5,124 received both regional/local and general anesthesia, 8,581 received regional or local anesthesia only, and 2,152 received general anesthesia only, and they were included in the final analysis regarding the different modes of anesthesia. In total, 46,684 controls did not receive anesthesia during the study period. For the analysis of the risk of developing dementia according to the individual anesthetic agent, 2,732 people were excluded due to unclear information regarding the anesthetic agent used. Additionally, during the analyses of individual anesthesia, samples with a small number of instances using a particular agent, where confirming statistical significance proved challenging, were excluded from the analysis; isoflurane: 32; etomidate: 7; ketamine: 16; levobupivacaine: 16. The number of individuals included in the analysis by the agent is shown in Figure 1.
Individuals who exclusively received general anesthesia were younger compared to those who underwent other modes of anesthesia or no anesthesia. Compared to the individuals in the no anesthesia group, those who received any type of anesthesia were less likely to be male, had a higher prevalence of comorbidities, such as hypertension, diabetes, and hyperlipidemia, and had a higher mean CCI score (Table 1). The baseline demographics and comorbidities of the individual subgroups categorized by the individual agents used in anesthesia are presented in Supplementary Table 1 (available online).
After receiving anesthesia in 2012−2013, the number of patients diagnosed with dementia who accessed follow-up care through 2020 was 850 out of 15,857 (5.36%). Among these, 570 out of 8,581 (6.64%) received regional/local anesthesia only, 90 out of 2,152 (4.18%) received general anesthesia only, and 190 out of 5,124 (3.71%) received both general and regional/local anesthesia. In the group that did not receive anesthesia during the study period, the number of patients diagnosed with dementia and accessed care was 1,201 out of 46,684 (2.57%) (Tables 1, 2).
The number of patients diagnosed with dementia who accessed follow-up care by 2020, depending on the anesthesia agent administered, is presented in Table 3. Among the agents used for inhalant anesthesia, the desflurane group (5.57%, 24/431) displayed the highest incidence of dementia during the study period, followed by sevoflurane (4.08%). The midazolam group reported an incidence of dementia of 5.26%, whereas the incidence in the propofol group was 3.00%. Regarding individual agents used in regional anesthesia, the various incidence of dementia were as follows; lidocaine (5.31%, 376/7,082), bupivacaine (6.80%, 106/1,558), ropivacaine (4.80%, 18/375), mepivacaine (6.98%, 49/702) (Table 3).
The risk of developing dementia over the 8-year period after anesthesia, categorized by anesthesia mode, is presented in Table 2 and Figure 2. When the Cox proportional hazards models were calculated for each mode of anesthesia the risk of developing dementia was higher in the regional anesthesia-only group and the general anesthesia only group compared to the no anesthesia group, and this remained statistically significant after adjusting for the effects of age, sex, and CCI. For individuals who received both general anesthesia and local/regional anesthesia, the HR showed a significant increase (1.200; 95% CI, 1.026−1.404) until adjustments were made for age and gender. However, after further adjustment for the CCI, the heightened HR (1.097; 95% CI, 0.937−1.284) lost statistical significance.
When Cox proportional hazards models were calculated for each anesthetic agent, the intravenous anesthetic midazolam and inhalational anesthetic desflurane had significantly higher HRs compared to the no anesthesia group, and this trend remained after adjusting for age, sex, and comorbidities. However, inhalational anesthetic sevoflurane and intravenous anesthetic propofol did not have a significantly higher or lower HR for the development of dementia. The anesthetics used for regional/local anesthesia (lidocaine, bupivacaine, ropivacaine, and mepivacaine) were associated with a significantly higher incidence of dementia compared to the control group, even after adjusting for the confounding effects of age, sex, and comorbidities (Table 2, Fig. 3).
This study evaluated the risk of dementia development based on the mode of anesthesia and individual anesthetic agents in a nationwide cohort. After eight years of follow-up, the risk of developing dementia increased in individuals under either general or regional anesthesia compared to individuals without anesthesia experience. Nonetheless, the risk of dementia did not exhibit a significant increase when individuals received both general anesthesia and regional/local anesthesia, even after adjustments were made for sex, age, and comorbidities. In addition, the risk of dementia differed for each anesthetic agent used. After adjusting for sex, age, and comorbidities, anesthesia with desflurane for inhalation and midazolam, lidocaine, bupivacaine, ropivacaine, and mepivacaine for intravenous anesthesia were associated with an increased risk of developing dementia during the 8-year follow-up.
In this study, the risk of dementia increased with all types of anesthesia administered, which is consistent with trends observed in previous studies that have shown an increased risk of dementia regardless of the type of anesthesia [11-13]. However, whether inhalation or intravenous anesthesia is more likely to cause dementia remains controversial. Some studies have reported that inhalation anesthesia is associated with a higher risk of dementia than intravenous or regional anesthesia. For instance, patients aged 55 years and older who underwent coronary artery bypass grafting (CABG) under general anesthesia or percutaneous transluminal coronary angioplasty (PTCA) under intravenous anesthesia were studied to assess the incidence of Alzheimer’s dementia (AD) after anesthesia. Over 5 years postoperatively, the CABG group showed a 1.7-fold increased risk of dementia compared with the PTCA group; however, no comparison was made with individuals without anesthesia [10]. Additionally, in female patients who underwent hysterectomies 14 years apart, general anesthesia had a 2.68-fold increased HR for developing dementia compared to regional anesthesia [27]. However, other studies have not found that inhalation anesthesia is associated with a higher risk of dementia than other intravenous or regional anesthetic agents. In a retrospective cohort study from Taiwan [11], patients undergoing anesthesia for the first time had a higher incidence of dementia in all cases of general, regional, and intravenous/muscular anesthesia compared to un-anesthetized controls at 3 to 7 years of follow-up, with a shorter mean time to dementia diagnosis from the date of surgery [11]. Similarly, a retrospective study from a registry in Virginia, USA, compared patients who developed Alzheimer’s disease after spinal surgery with those who did not and found no significant differences in the choice of inhalational anesthetic agent versus propofol anesthesia or duration of anesthesia between the two groups [12]. Another retrospective cohort study using a registry in Ontario, Canada, compared the risk of dementia between regional and general anesthesia and found no significant difference between both groups [13].
Prospective studies have followed postoperative cognitive decline for months to years. In a study comparing propofol and lidocaine to sevoflurane in patients undergoing spine surgery, the incidence of Alzheimer’s disease and cognitive decline did not differ among the three groups; however, the incidence of mild amnestic cognitive impairment was higher in the sevoflurane group [28]. A study examining cognitive function 3 months postoperatively compared to 1 week preoperatively demonstrated a postoperative cognitive decline associated with increasing age and duration of anesthesia [29]. In a study that prospectively followed cognitive function after CABG under general anesthesia for up to 7.5 years, the rate of dementia-level cognitive decline after CABG was higher than the incidence of dementia in the general population derived from other population-based studies [30]. However, no prospective studies have directly compared the incidence of dementia between anesthesia and no anesthesia groups.
There are several possible explanations for the mechanisms by which general inhalation anesthesia may increase the risk of dementia. Several animal studies have reported that inhaled and intravenous anesthetic agents may worsen the neurohistological findings of dementia associated with degenerative brain diseases, such as Alzheimer’s disease [21,31,32]. Inhalational anesthetics, particularly isoflurane, and sevoflurane, have also been reported to promote amyloid beta 42 accumulation and induce cell death in animal models [33,34]. Human studies have also reported an increase in tau protein levels in the cerebrospinal fluid during surgery [35].
In the current study, among the inhalation anesthetics, desflurane displayed a significantly higher dementia incidence compared to the controls, even after adjustment for age, sex, and comorbidities. While sevoflurane initially resulted in an increased incidence of dementia compared with the control group, it failed to maintain statistical significance after adjusting for possible confounding factors. Previous human studies have focused on differences in the form of anesthesia; however, no studies have examined the effects of individual anesthetics. The same inhaled anesthetic agent may have different effects on the body and physiological markers, such as oxygen saturation and minimum alveolar concentration, during surgery, and therefore, may have different effects on the brain and the development of dementia. The rapid administration of desflurane causes hypertension and tachycardia. Additionally, desflurane and isoflurane are known to cause airway irritation and bronchospasm in patients with asthma, and sevoflurane is used to induce general anesthesia in patients with asthma [17]. These different adverse effect profiles between individual anesthetic agents may have affected the varying effects on dementia incidence depending on the agent used for anesthesia.
Propofol is a commonly used drug for intravenous anesthesia, and although its mechanism of action remains unclear, it is believed to induce sedation through GABA-mediated chloride channels [14]. Propofol has been associated with a lower risk of mild cognitive impairment in human studies compared to inhaled general anesthesia [28]. However, other studies have shown no difference in the risk of dementia compared to inhaled anesthesia [12]. In terms of Alzheimer’s pathology, postoperative plasma concentrations of blood biomarkers, including Aβ 1-40 and S-100β, were lower in the propofol group compared with the inhalation anesthesia group [36]. In animal studies, propofol has been shown to cause cognitive decline, induce cell death [18], and promote tau phosphorylation in the hippocampus [37] in association with AD pathology. However, in vitro studies have shown a protective effect by inhibiting the oligomerization of amyloid peptides [21]. This combination of the negative and protective effects of propofol on dementia development may have contributed to a similar level of dementia development in the non-anesthetized group in this study.
Midazolam is an ultra-short-acting benzodiazepine widely used as a rapid intra-venous anesthetic. It is primarily used for preoperative sedation and the induction of general anesthesia. Midazolam acts on GABA-A receptors to decrease neuronal excitability, resulting in the impairment of multidomain cognitive function immediately after dosing [15]. Although a recent meta-analysis has shown an association between long-term benzodiazepine use and cognitive decline [38], few studies have examined the long-term cognitive effects of a single dose of midazolam. Furthermore, it is essential to consider the practice of administering midazolam for anxiety symptom control before surgery and general anesthesia when assessing the potential impact of midazolam on dementia [39]. Future prospective studies are needed to confirm the increased risk of dementia after midazolam anesthesia seen in this study.
In contrast, few studies have evaluated the effects of regional and local anesthetic agents on dementia. Retro-spective cohort studies have shown an increased risk of dementia with regional anesthesia, similar to that with general anesthesia [11-13]. The increased risk of dementia in this study, even after adjusting for age, sex, and comorbidities, was observed with regional and local anesthesia, suggesting that factors other than comorbidities and age contribute to the increased risk of dementia with regional anesthesia. Other factors not investigated in this study, such as the invasiveness of surgery, duration of surgery, duration of anesthesia, cessation of anesthesia, postoperative recovery time, and preoperative cognitive level, should also be evaluated to determine their impact on the risk of developing dementia.
There are few studies on the risk of developing dementia due to the neurotoxicity of the anesthetic agents used for regional and local anesthesia. There are reports on animal models of neurotoxicity of local and regional anesthetics used in this study, including lidocaine, and rare reports of central nervous system toxicity, such as seizures, following the use of local anesthetics in humans [40]. However, studies on cognitive decline caused by neurotoxicity are lacking. Future studies are needed to evaluate the neurotoxicity of regional anesthetics and their effects on cognitive function.
Notably, this investigation revealed a lack of significant escalation in dementia risk among patients subjected to both general anesthesia and regional/local anesthesia during extended follow-up periods. There has been a proposal for employing a combined protocol of regional/local anesthesia alongside general anesthesia in diverse clinical scenarios, with documented instances of heightened survival rates [39]. The augmented survival associated with combined anesthesia, as opposed to exclusive employment of general anesthesia, may be attributed to diminished inflammation and stress responses [41,42], along with reduced utilization of inhaled anesthetics [43,44] and opioids [43,45] during concurrent anesthesia administration. The inflammation and stress response [46] and the use of inhaled anesthetics [10,27] and opioids [47] have also been posited as potential contributors to the development of dementia. Earlier research has indicated that the utilization of combined anesthesia is linked to a lower incidence of cognitive decline [48] and delirium [49] compared to exclusive reliance on general anesthesia. While it remains unverified whether the cohort receiving both general anesthesia and regional/local anesthesia in this study underwent combined anesthesia for a singular surgical procedure, considering the contemporary inclination toward such combined approaches, it is plausible that these individuals experienced combined anesthesia for a singular intervention. Consequently, a more nuanced investigation is imperative to comprehensively assess the risk of dementia development associated with combined regional/local anesthesia and general anesthesia.
This study analyzed nationwide insurance claims data, and patients included in the study may have had different medical conditions, resulting in differences in the assessment tests and diagnostic methods used. The diagnosis of dementia may include patients who have undergone a systematic examination using biological markers and patients with dementia who were diagnosed based on clinical suspicion without adequate diagnostic evaluation. To address this variability, we included patients with dementia who had at least two visits with a diagnosis of dementia and a history of cognitive enhancer prescriptions. Additionally, this study did not categorize dementia diagnoses by underlying diseases, such as Alzheimer’s, Lewy body, vascular, or Parkinson’s disease, limiting our ability to explain the biological pathogenesis of dementia caused by specific diseases. Moreover, this study did not investigate the type of anesthesia, number of individual anesthesia doses, duration of anesthesia, depth of anesthesia, or type of surgery, which could have influenced the results. As the study did not explore repeated anesthesia exposure beyond the initial exposure period, it was not possible to assess the cumulative effects. Psychiatric comorbidities, educational level, socioeconomic status, and problematic substance use data were not available in the HIRA dataset, thereby limiting adjustments for these variables. We also lacked information on preoperative cognitive function, making it difficult to exclude the influence of baseline cognitive function.
To the best of our knowledge, this is the first study to examine the risk of developing dementia based on the type of anesthesia and individual anesthetic agents used over an extended period. In summary, the results of this study suggest that the risk of dementia varies depending on the type of anesthesia and the specific anesthetic agent used. Clinically, this underscores the importance of monitoring cognitive decline post-anesthesia, regardless of the type of anesthesia and suggests that long-term follow-up may facilitate appropriate cognitive function assessment and intervention. Future prospective studies should explore whether specific types of anesthesia or anesthetic agents increase the risk of dementia.
This study was supported by the Korea Health Technology R&D Project (HI22C1453) through the Korea Health Industry Development Institute (KHIDI) that were funded by the Ministry of Health & Welfare, Republic of Korea.
No potential conflict of interest relevant to this article was reported.
Conceptualization: Hyun-Ghang Jeong, Seung-Hoon Lee. Data acquisition: Seung-Hoon Lee, Won Seok William Hyung, Surin Seo, Hyun-Ghang Jeong. Formal analysis: Seung-Hoon Lee, Won Seok William Hyung. Funding: Hyun-Ghang Jeong. Supervision: Hyung-Ghang Jeong, Changsu Han, Junhyung Kim, Kwang-Yeon Choi, HyunChul Youn. Writing—original draft: Seung-Hoon Lee. Writing—review & editing: Surin Seo, Won Seok William Hyung.
![]() |
![]() |