Predictive modeling for post operative delirium in elderly

Abstract

Delirium, a complex neurocognitive syndrome, frequently emerges following surgery, presenting diverse manifestations and considerable obstacles, especially among the elderly. This editorial delves into the intricate phenomenon of postoperative delirium (POD), shedding light on a study that explores POD in elderly individuals undergoing abdominal malignancy surgery. The study examines pathophysiology and predictive determinants, offering valuable insights into this challenging clinical scenario. Employing the synthetic minority oversampling technique, a predictive model is developed, incorporating critical risk factors such as comorbidity index, anesthesia grade, and surgical duration. There is an urgent need for accurate risk factor identification to mitigate POD incidence. While specific to elderly patients with abdominal malignancies, the findings contribute significantly to understanding delirium pathophysiology and prediction. Further research is warranted to establish standardized predictive for enhanced generalizability.

Key Words: Post-operative delirium; Elderly delirium; Neurocognitive syndrome; Neurotransmitters; Abdominal malignancy; Predictive model; Synthetic minority oversampling technique

Core Tip: Postoperative delirium (POD) presents significant challenges in elderly patients, with no current gold standard for prevention. This editorial sheds light on a study that introduces a predictive model utilizing synthetic minority oversampling technique (SMOTE) to identify high-risk patients. Key risk factors include comorbidity index, anesthesia grade, and surgical duration. The editorial discusses that standardizing predictive models across surgical subspecialties is crucial for effective POD management. Further advancements in SMOTE algorithms offer promising avenues for handling unbalanced datasets prevalent in research.

INTRODUCTION

Delirium, a neurocognitive syndrome, arises from reversible neuronal disruption due to systemic perturbations and serves as an acute marker of brain dysfunction. The Diagnostic and Statistical Manual of Mental Disorders (DSM-5) offers diagnostic criteria for delirium, emphasizing disturbances in attention, cognition, and/or awareness that deviate from baseline function. Manifesting in hyperactive, hypoactive, or mixed forms, delirium can occur postoperatively, ranging from 10 minutes after anesthesia to 7 days post-op, with hypoactive delirium being particularly prevalent and identifiable in the Post-Anesthesia Care Unit. Postoperative delirium (POD) in elderly poses significant challenges due to its high incidence and associated adverse outcomes.

PREDICTING POST OPERATIVE DELIRIUM IN PATIENTS

Like many other major bodily systems, the gastrointestinal and the central nervous system (CNS) are intricately linked, with disturbances in one system impacting the other's homeostasis and functionality[1]. The pathophysiology of delirium in the elderly post-operative period involves neurotransmitter imbalances, particularly acetylcholine, serotonin, dopamine, and gamma-aminobutyric acid[2-4]. Age-related declines in acetylcholine synthesis and serotonin levels, alongside dopamine elevation, contribute to delirium symptoms[3,5]. Additionally, inflammatory markers like C-reactive protein and interleukin 6, exacerbated by stress events such as surgery, heighten delirium risk[4]. There are indications that these stress-related immunologic changes affect plasma amino acid concentrations, such as tryptophan, and multiple cerebral neurotransmitters, especially serotonin[2,5]. Age-related changes in blood-brain barrier integrity further increase vulnerability to post-operative delirium[5].

POD stands out as one of the most common neurological consequences following surgery, with prevalence rates ranging from 5% to as high as 87.0%[6-8]. This highlights the importance of preoperative consideration and prevention strategies. The high prevalence of this condition along with the lack of a gold standard for pharmaceutical POD prevention constitutes an urgent need for research development[9]. Non-pharmaceutical interventions have focused on identifying high-risk factors to implement targeted interventions aimed at reducing POD incidence. Given the absence of standardized pharmaceutical interventions, accurate prediction and prevention through risk factor identification becomes paramount. The study titled “Predictive modeling for POD in elderly patients with abdominal malignancies using synthetic minority oversampling technique” seeks to formulate an accurate model for predicting POD in elderly abdominal cancer patients[9].

The methodology employed in developing a delirium prediction model in this article utilizes the application of a synthetic minority oversampling technique (SMOTE), to improve predictive accuracy and aid in the early identification and mitigation of POD. The validation of this model relies on a selection of pivotal risk factors proven to be significant predictors of delirium. These factors encompass the Charlson comorbidity index, anesthesia grade, history of cerebrovascular disease, surgical duration, perioperative transfusion, and postoperative pain. This research study aims to address a crucial gap in the field by providing a more reliable method for predicting and potentially preventing POD in elderly patients with abdominal malignancies.

One evident consideration when analyzing this article is the specificity of the patient population recruited. All participants included in this study were elderly (65-75 years old) and diagnosed with colorectal cancer. While this specificity helps eliminate confounding variables and promotes a potentially effective intervention for this demographic, it may limit the generalizability of the predictive model to gastrointestinal surgical interventions addressing different pathologies or within various surgical subspecialties. Additionally, since the United States Preventive Services Task Force recommends colon cancer screening starting at the age of 45, the results of this predictive model may not be entirely applicable to all individuals with colon cancer[10]. This is particularly relevant considering that advanced age is identified as a risk factor for POD in other studies[6,7,11].

Further scrutiny of the study involves an examination of the diagnostic criteria used to identify delirium. Typically, DSM-5 serves as the "gold standard" for delirium diagnosis, yet it was not employed in this investigation[11]. Instead, the diagnostic framework relied on the Ambiguity Assessment Scale and the acute onset of specific symptoms such as difficulty concentrating, confused thinking, and altered consciousness. While the absence of the DSM-5 criteria may raise concerns about the diagnostic validity, it's essential to acknowledge that the Ambiguity Assessment Scale is a validated tool for delirium screening. Moreover, the criteria utilized in this study have been extensively elucidated, demonstrating high sensitivity and specificity levels of 94%-100% and 90%-95%, respectively. Furthermore, the diagnostic evaluation was conducted daily for 7 days post-surgery, aligning with the timeframe known for the peak onset of POD, as evidenced in previous literature[11]. Notably, in this study, these criteria yielded a POD incidence rate of 22.91%, falling within the predictive range observed in prior studies (5%-51%)[6,7].

This study highlights the importance of identifying key risk factors of POD. Utilizing Multivariate logistic regression analysis, key factors such as Charlson comorbidity index, American Society of Anesthesiologists classification, history of cerebrovascular diseases, surgical duration, and perioperative blood transfusion were identified as significant contributors to POD development. In contrast, other predictive models like intensive care unit (ICU)-POD utilize different risk factors such as Physiological and Operative Severity Score for Mortality and morbidity, acid–base disturbances, and medical history[8]. Comparing performance, both Multivariate logistic regression and SMOTE logistic models showed comparable results with an AUC of 0.862 and 0.856, respectively, similar to the ICU-POD model (0.852)[8]. Furthermore, other studies have found other risk factors like age, pre-existing dementia, Barthel Index score, preoperative infection, preoperative hematocrit, and preoperative opioid use to also significantly impact the development of POD[11]. These findings highlight the uncertainty surrounding the exact etiology of POD and its primary contributing factors.

Future meta-analyses are pivotal for identifying preventable risk factors that have the most significant impact on POD across various predictive models. Insights from these analyses could inform comprehensive pre- and post-operative protocols aimed at mitigating POD risk, thereby benefiting surgical, anesthesia, and geriatric/internal medicine teams. Furthermore, subsequent studies could explore variations in POD predictive models across surgical subspecialties to develop more standardized guidelines while also facilitating the design of more effective pharmaceutical and non-pharmaceutical interventions.

With numerous global studies proving the significance of differing risk factors on the development of POD, standardizing datasets can be quite challenging. The authors propose an alternative approach to address this issue by advocating for further advancements and refinements in SMOTE algorithms. The article demonstrates SMOTE's effectiveness in addressing the inherent complexity of unbalanced datasets prevalent in bariatric oncology research. SMOTE achieves this by generating synthetic minority samples from existing data, thus addressing the imbalance and streamlining the process of acquiring high-quality and accurate predictive models.

CONCLUSION

In conclusion, the intricate dynamics of delirium, particularly in elderly patients undergoing abdominal malignancy surgery, stem from the complex interplay between the gastrointestinal and CNSs. We have underscored the significance of neurotransmitter imbalances, inflammatory markers, and age-related changes in blood-brain barrier integrity in delirium development. The utilization of SMOTE has notably improved predictive accuracy, enabling the identification of crucial risk factors such as comorbidity index, anesthesia grade, and surgical duration. However, it is essential to acknowledge limitations, including the specificity of this patient population and deviation from DSM-5 diagnostic criteria. Looking ahead, standardizing predictive models and interventions across surgical subspecialties is paramount for effectively addressing POD. Given the ongoing global research efforts generating valuable data, there is an urgent need for methods to standardize scattered data scales. This standardization will facilitate the creation of comprehensive predictive models, ensuring that emerging research findings are leveraged efficiently for the development of robust predictive models in the field.