Causal relationships between delirium and Alzheimer's disease: a bidirectional two-sample Mendelian randomization study

Delirium is an acute cognitive function statues changing in a short time among hospitalized older adults, which can lead to serious outcomes and increase the medical costs [1, 2]. Delirium is often associated with longer hospital stays, serious complications, and mortality [3‐5]. In addition, there is growing evidence suggesting that delirium may also increase the risk of developing dementia. Delirium has been reported to be a strong risk factor for dementia in the patients aged over 85 years and can increase the rate of cognitive deterioration in Alzheimer’s disease (AD) patients [6‐8]. However, dementia itself is a significant predisposing factor for delirium, and nearly half of dementia cases go undiagnosed, whether delirium is related to new-onset dementia remains controversial [9‐11]. AD is the most common cause of dementia, which accounts of 60–80% of all cases [12]. Although previous observational studies have identified a connection between dementia and delirium, the limitations of those observational studies (i.e., the inability to completely eliminate potential confounding factors, selection bias and small sample size) and the expensive and time-consuming nature of randomized controlled trails, make it challenging to determine a causal association between delirium and AD.

Mendelian randomization (MR) is a valuable tool for exploring causality between exposures and outcomes. By using single nucleotide polymorphisms (SNPs) that linked with exposures, MR can avoid the distribution of confounding bias and reverse causality that are present in observational studies [13, 14]. Therefore, we aimed to investigate the relationship between delirium and AD using a bidirectional two-sample MR in this study.

We conducted a bidirectional two-sample MR approach to evaluate the causal association between delirium and AD. Results from IVW and other MR methods, showed that there was no bidirectional causality between delirium and AD. Specifically, we found that patients with AD may have a higher risk of delirium, but delirium was not significant in relation to AD. To the best of our knowledge, the present study firstly explored a causal relationship between delirium and AD using MR. Our findings provide valuable insights into the potential causal relationship between these two conditions.

Previous studies have found a correlation between delirium and AD. Delirium has been identified as a risk factor that increases the likelihood of developing or exacerbating dementia, and it can accelerate cognitive decline in AD patients [6, 22, 23]. However, it is important to note that in the present study, we only included patients with or without AD, and our research focused on assessing the causal relationship between delirium and AD. The process and development of AD go beyond the scope of this specific study. Although we did not find that delirium can increase the risk for developing AD, it is not contradictory to the conclusion that delirium can accelerate cognitive decline in AD patients. Inflammatory response is believed to be a primary driver of acute brain dysfunction or delirium, in addition, critical illness with acute inflammatory insult was also a risk factor for AD [24, 25]. A meta-analysis including 28 observational studies found that biomarkers associated with dementia, such as elevated blood-based levels of IL-6, C-reactive protein, S100B, and NfL, were also associated with delirium [26]. The fact that delirium and AD share similar biomarkers suggests that they may share similar pathologies in the development of disease. These biomarkers associated delirium may also cause harm to the process of AD and speed up cognitive decline instead of delirium itself.

There are common risk factors, such as low vitamin D concentrations, that may contribute to both delirium and AD. Two MR analysis studies have found an association between low vitamin D concentrations and the development of AD and delirium [27, 28]. Witlox J et al. concluded that patients with delirium have a higher incidence of developing AD in a meta-analysis. This meta-analysis included two prospective follow-up studies that were conducted over a period of 5 and 3 years [5, 29, 30]. However, causality cannot be established because of the small sample size and the limitation of observational studies. In this study, we used five MR analysis methods and found consistent results showing that delirium is not an independent risk factor for AD. The results were robust, as we did not observe any pleiotropy or heterogeneity in the sensitivity analysis. For all that, we cannot completely deny the possibility that delirium may influence the process of AD through similar pathological mechanisms, such as abnormal nerve transmission and neuroinflammation. Future studies including lager sample sizes should be done, to confirm whether delirium has an effect on the process of AD and to explore the true underlying factors that contribute to accelerating the process of AD.

It is well known that AD is the risk factors for delirium. In our study, the effect estimate of IVW (OR 1.32, 95% CI 1.13–1.54) demonstrates that AD is indeed a risk factor for delirium, which aligns with previous research demonstrating that dementia and recognition decline are independent risk factors for delirium [31, 32]. Although the exact mechanisms of delirium are not completely understood, it is believed to be a multifactorial process that can result in brain dysfunction, often involving abnormal nerve transmission and neuroinflammation [10, 33, 34]. Patients with dementia and pre-existing neurodegeneration may be more prone to neuroinflammatory cytokines, and trigger greater oxidative stress during surgery, pain or other stressful situations, potentially increasing the risk of delirium. Furthermore, patients with delirium superimposed on dementia have poorer outcomes, such as higher mortality rates, greater dependence on walking and increased healthcare expenses [35‐37]. Therefore, recognition and prevention of the occurrence of delirium at an early stage is crucial for improving patient outcomes.

Delirium is currently classified into major subtypes: hyperactive type (which is more apparent), hypoactive type (which is more prevalent but often overlooked) and a combined hyperactive and hypoactive types [38]. Research has found that staff and volunteers tend to focus more on treating hyperactive symptoms, while may miss the prevention and early recognition of hypoactive delirium [39]. Clinicians and nurses may overlook patients with hypoactive delirium because they appear to be performing normally, even though they actually unwell. Prevention is always better than treatment, and there are two types of risk factors for delirium: predisposing and precipitating factors. The predisposing factors are associated with the patients’ internal state, such as advanced age, functional deficiency, and comorbidities. Precipitating factors, on the other hand, refer to external disturbances, such a trauma, sedative drugs, sepsis, surgery and anesthesia. Patients with more predisposing factors more likely to develop delirium with fewer precipitating factors [2, 40]. Post-operative delirium is a common post-surgical complication, particularly in older adults [41]. Although it may be difficult to make changes in predisposing factors, it possible to mitigate the risk of precipitating factors. Clinicians can avoid the use of benzodiazepines or anticholinergic drugs, manage pain, encourage early post-operative mobilization and regular ambulation to reduce the risk of post-operative delirium. Implementation nonpharmacologic prevention strategies can also significantly improve outcomes for older patients [42]. Therefore, the clinicians and nurses should pay close attention to patients with AD, and the presence of predisposing risk factors for delirium. By taking early preventative measures, clinicians can improve the chances of early recognition and effective prevention of delirium.

Several limitations of our study should be noted. Firstly, our findings may not be applicable to populations with different ethnicities or geographic locations as our study only included individuals of European ancestry. Further data collection and analysis is needed to confirm the generalizability of our findings. Secondly, we only included the delirium phenotype as a binary variable, and did not consider the three subtypes or delirium or the severity of delirium. Therefore, we could not assess the association between AD and delirium subtypes or severity. It is worth noting that previous study has shown that patients with persistent delirium have worse outcomes than those with transient delirium [43]. Thirdly, the Cochran’s Q test revealed heterogeneity in the effect of AD on delirium. Various GWAS cohort studies might have differences in gene annotation analysis platforms, inclusion and exclusion criteria for cases, which could contribute to the heterogeneity in our study [44]. Additionally, the GWAS data for AD are obtained from several consortia, may contribute to the heterogeneity. However, this heterogeneity did not impact the main analysis, as the consistent results from other MR methods, such as MR Egger and weighted median model, are robust to genetic heterogeneity and can provide valid estimates [45‐47]. Moreover, the other three MR methods (simple mode, weighted median and weighted mode) showed similar result (p < 0.05), and the p value was also close to statistical significance using MR Egger method (P = 5.70E−02). In addition, no directional pleiotropy was observed in the MR Egger intercept analysis. The leave-one-out sensitivity analysis was also performed, and it showed that even when individual SNPs were removed one by one, the overall results remained stable (the overall error line did not fluctuate significantly). These findings strongly suggest that the results of our MR analysis are robust.

Our bidirectional two-sample MR analysis showed that although AD increase the risk for delirium, while delirium is not associated with AD. These findings may suggest that clinicians should identify and prevent the occurrence of delirium superimposed on dementia in an early stage. Future study should investigate the mechanisms underlying the effect of AD on delirium, and researchers could include larger sample sizes to confirm the effect of delirium on AD.