Association between peripheral serotonin levels and postoperative delirium in older adults undergoing gastrointestinal surgery: a single-centre, prospective cohort study protocol Authors: Shujie Wang, Yaqi Han, Kaiqiang Fan, Chuan Guo, Yiwei Lin, Jirun Wang, Yuhang Shi, Shaoqi Tian, Youzhuang Zhu Published: 2026 ABSTRACT Abstract AbstractIntroductionPostoperative delirium (POD) is a common and serious postoperative complication, with an incidence of approximately 30% among older adults undergoing gastrointestinal surgery. Serotonin has been implicated in a variety of neuropsychiatric disorders. Gastrointestinal surgery has been shown to increase serotonin synthesis and release from enterochromaffin cells; however, the causal relationship between serotonin and POD remains unclear. This study aims to investigate the causal relationship between peripheral serotonin levels and POD in older adults undergoing gastrointestinal surgery.Methods and analysisThis single-centre prospective cohort study will enrol 314 patients aged ≥60 years who are undergoing elective gastrointestinal surgery at the Affiliated Hospital of Qingdao University between December 2025 and December 2026. Peripheral serotonin levels will be measured prior to anaesthesia induction and prior to discharge from the post-anaesthesia care unit. The primary outcome will be the incidence of POD within 3 days. Secondary outcomes will include emergence agitation, number of POD episodes, POD onset time and duration, POD severity and subtype, pain scores, quality of recovery, gastrointestinal function and postoperative complications. A predefined directed acyclic graph will be used to guide covariate adjustment. Statistical analyses will include multivariate logistic regression, linear trend analysis, restricted cubic spline modelling, subgroup analyses, sensitivity analyses and Bayesian mediation analysis. Delirium assessors will undergo standardised training conducted by psychiatrists to ensure diagnostic reliability.Ethics and disseminationThe study protocol was approved by the Medical Ethics Committee of the Affiliated Hospital of Qingdao University (Approval No. QYFYEC2025-199) and was registered with the Chinese Clinical Trial Registry (Registration ChiCTR2500112152). Written informed consent will be obtained from all participants prior to enrolment. Data will be anonymised, securely stored on encrypted servers for a minimum of 10 years, and shared only with approval from the ethics committee. Results will be disseminated through publication in peer-reviewed journals and presentation at scientific conferences.Trial registration numberChiCTR2500112152.Keywords: Postoperative Delirium, Aged, Prospective studies, Cognitive dysfunction, GASTROENTEROLOGY STRENGTHS AND LIMITATIONS OF THIS STUDY Exposure and outcomes are rigorously defined using internationally accepted detection methods and validated assessment tools, enhancing diagnostic accuracy and methodological rigour. Exposure and outcomes are rigorously defined using internationally accepted detection methods and validated assessment tools, enhancing diagnostic accuracy and methodological rigour. The use of a directed acyclic graph to guide covariate selection clarifies the assumed causal structure between exposure, outcomes and potential confounders, thereby improving the transparency and reproducibility of causal inference. The use of a directed acyclic graph to guide covariate selection clarifies the assumed causal structure between exposure, outcomes and potential confounders, thereby improving the transparency and reproducibility of causal inference. As a single-centre study, variations in population characteristics, surgical techniques and perioperative management may limit the generalisability of the findings. As a single-centre study, variations in population characteristics, surgical techniques and perioperative management may limit the generalisability of the findings. The study does not directly assess central serotonin levels or blood-brain barrier permeability, which precludes full validation of the mechanistic pathway by which peripheral serotonin influences central nervous system function and warrants further investigation using animal and cellular models. The study does not directly assess central serotonin levels or blood-brain barrier permeability, which precludes full validation of the mechanistic pathway by which peripheral serotonin influences central nervous system function and warrants further investigation using animal and cellular models. Introduction Background and rationale Postoperative delirium Postoperative delirium (POD) is a common neuropsychiatric complication among geriatric surgical patients, characterised by acute and fluctuating disturbances in attention and consciousness, disorganised thinking, memory impairment and perceptual abnormalities.1 Epidemiological studies have demonstrated that POD occurs in approximately 15%–25% of patients following major elective surgery and in up to 50% of those undergoing cardiac surgery,2 while the incidence after abdominal surgery has been reported to reach 30%.3 POD has been shown to be significantly associated with prolonged hospitalisation, increased risk of nosocomial infections, frailty, elevated mortality and long-term impairments in cognitive function and activities of daily living.4 Serotonin synthesis and metabolism Serotonin, also known as 5-hydroxytryptamine (5-HT), is an inhibitory monoamine neurotransmitter. Approximately 90% of serotonin is synthesised in enterochromaffin cells (ECs) of the intestinal mucosa and stored in platelets.5 Serotonin synthesis uses tryptophan as a substrate, which is hydroxylated by tryptophan hydroxylase to form the intermediate 5-hydroxytryptophan (5-HTP). Subsequently, 5-HTP is decarboxylated by aromatic L-amino acid decarboxylase to produce serotonin. Serotonin plays a key role in regulating intestinal motility and contributes to the modulation of vascular tone and haemostatic function via platelets.6 Previous studies have suggested that serotonin cannot cross the blood-brain barrier (BBB) due to its polarity and the absence of a specific transporter.7 Central serotonergic activity depends on the active transport of L-tryptophan into the brain via neutral amino acid transporters, followed by its subsequent synthesis into serotonin.8 Within the central nervous system (CNS), serotonin is stored in synaptic vesicles and is released into the synaptic cleft on neuronal activation, where it binds to receptors and modulates neurotransmission. Following release, most serotonin undergoes reuptake into presynaptic neurons via the serotonin transporter (SERT), whereas the remaining fraction is metabolised by monoamine oxidase into 5-hydroxyindoleacetic acid for renal excretion. The central serotonergic system plays a crucial role in mood regulation, cognitive function and neuroimmune signal transduction.9 Serotonin and neurocognitive function Serotonin is widely distributed throughout the CNS and peripheral tissues, where it is involved in a wide range of physiological processes—including learning, memory, perception, reward, anger, aggression, appetite, sexual behaviour, attention and emotion—through multiple receptor subtypes.10 11 Although the BBB is generally impermeable to serotonin, emerging evidence suggests that peripheral serotonin may gain access to the CNS through distinct mechanisms. Surgical intervention, anaesthesia and the ageing process have been shown to compromise BBB integrity, thereby creating a pathway for peripheral serotonin to enter the CNS and subsequently contribute to cognitive dysfunction.12,14 Platelet-derived extracellular vesicles (EVs) have been shown to transport serotonin across the BBB; moreover, platelets themselves, which function as serotonin carriers, may directly penetrate the CNS parenchyma and interact with neuronal cells. Additionally, the SERT exhibits bidirectional transport properties, providing a potential mechanistic basis for serotonin entry into the CNS from the systemic circulation.15 Previous studies have demonstrated that abnormalities in central serotonin levels are closely associated with various psychiatric and neurological disorders.11 However, the relationship between serotonin and POD remains unclear. (1) Elevated serotonin levels may lead to cholinergic deficits and activation of the dopaminergic system via different receptor pathways, disrupting the balance of neurotransmitters and triggering POD.16 (2) Surgery and anaesthesia increase peripheral serotonin levels, which, in turn, interact with inflammatory mediators. This synergy leads to a breakdown of the BBB, inducing CNS inflammation and subsequently triggering POD.17 18 (3) Serotonin can activate the 5-HT₁A receptor, inhibiting the cAMP-PKA signalling pathway, thereby enhancing inhibitory neurotransmission and mitigating excitotoxicity and neuroinflammatory damage.19 Additionally, activation of the 5-HT₂A receptor can reduce the production of pro-inflammatory cytokines such as tumour necrosis factor-α by microglia and increase the expression of neurotrophic factors such as brain-derived neurotrophic factor (BDNF), promoting the shift of microglia to an anti-inflammatory and neuroprotective phenotype, thereby maintaining CNS homeostasis.9 20 (4) Serotonin is a key precursor in the synthesis of melatonin. A deficiency in serotonin disrupts the precise regulation of the serotonin-melatonin axis in the sleep-wake cycle, ultimately leading to neuronal dysfunction and the onset of POD.21 22 Gastrointestinal surgery, serotonin and POD Randomised controlled trials have demonstrated that serotonin inhibitors can prevent POD.23 24 Surgical trauma, inflammation and mechanical intestinal distension can directly stimulate ECs to synthesise and release endogenous serotonin.25 26 Gastric surgery may also indirectly stimulate ECs to synthesise and release endogenous serotonin by altering gastric acid secretion and the gastrointestinal hormonal milieu.27 28 Additionally, surgical procedures can activate the coagulation system, and during this process, platelets release serotonin, which promotes platelet aggregation and thrombus formation, thereby facilitating haemostasis.29 Peripheral serotonin can enter the central nervous system directly through the damaged BBB, or indirectly via platelets, EVs and the SERT. By affecting inflammatory factors and neurotransmitters, it induces neuronal apoptosis and impaired synaptic plasticity, ultimately contributing to the occurrence of POD30 31 (figure 1). Proposed mechanisms of the relationship between serotonin levels and postoperative delirium in older adults undergoing gastrointestinal surgery. Serotonin is synthesised from tryptophan via sequential catalysis by TPH and AADC, with 5-hydroxytryptophan as the intermediate. Surgical trauma, inflammation and intestinal mechanical distension promote the synthesis and release of endogenous serotonin from enterochromaffin cells. Peripheral serotonin gains access to the CNS through the impaired BBB, as well as via platelets, EVs and the SERT. It then triggers neuronal apoptosis and synaptic plasticity impairment by modulating inflammatory mediators and neurotransmitters, ultimately leading to POD. Created with MedPeer (medpeer. cn). AADC, aromatic L-amino acid decarboxylase; BBB, blood-brain barrier; CNS, central nervous system; EVs, extracellular vesicles; POD, postoperative delirium; SERT, serotonin transporter; TPH, tryptophan hydroxylase; 5-HT, 5-hydroxytryptamine. Objective This study aims to investigate the causal association between peripheral serotonin levels and POD in older adults undergoing gastrointestinal surgery, thereby providing a theoretical foundation for the identification of early biomarkers and intervention strategies for POD in geriatric patients. The null hypothesis posits no significant causal association between peripheral serotonin levels and POD in this patient cohort. Methods Study design and setting This study is a single-centre, prospective cohort study. The study will be conducted at the Affiliated Hospital of Qingdao University, a large capacity gastrointestinal surgery centre. Data collection is planned from December 2025 to December 2026. Each eligible subject will undergo 3 days of face-to-face follow-up (figure 2). The final results of this study protocol will be reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement. Study flow diagram. Ethical approval This prospective observational study will adhere to the Declaration of Helsinki and relevant regulations, and has been approved by the Medical Ethics Committee of the Affiliated Hospital of Qingdao University (Approval No. QYFYEC2025-199), and registered with the Chinese Clinical Trial Registry (Registration No. ChiCTR2500112152). Prior to commencement of the study, the research team obtained written informed consent from all patients to protect the legitimate rights and interests of the participants. Participants Inclusion criteria (1) Older adults aged ≥60 years; (2) scheduled for elective gastrointestinal surgery, including radical gastrectomy for gastric cancer/subtotal gastrectomy/total gastrectomy/gastrojejunostomy/radical resection for colorectal cancer/colorectal lesion resection/colostomy/small bowel lesion resection; (3) American Society of Anesthesiologists (ASA) physical status classification I-III; (4) provision of informed consent. Exclusion criteria (1) Patients with preoperative dementia (Mini-Mental State Examination (MMSE) score <18 for illiteracy, <20 for primary school education, <23 for secondary school education or <24 for higher education); (2) history of psychiatric disorders or long-term use of psychotropic medications; (3) preoperative use of selective serotonin reuptake inhibitors or serotonin-norepinephrine reuptake inhibitors; (4) severe hepatic or renal insufficiency (Child-Pugh class C or estimated glomerular filtration rate <30 mL/min); (5) emergency surgery; (6) patients with severe preoperative anaemia (defined as haemoglobin (Hb) <70 g/L) anticipated to require intraoperative transfusion; (7) patients with abnormal platelet counts (defined as platelets <100 or >300×10⁹/L). Exposure Measurement of exposure Serotonin will be collected from the patient’s peripheral arterial blood and tested using a serotonin kit. The specific procedures are as follows: 5 mL of arterial blood will be drawn at two predefined time points: 10 min before anaesthesia induction and 5 min after extubation in the post-anaesthesia care unit (PACU), collected in EDTA anticoagulant tubes. Gently invert and mix eight times immediately after collection to avoid vigorous shaking leading to platelet activation. The first centrifugation will be completed within ≤5 min of blood collection; the sample is kept at 4°C during the waiting period. First centrifuging: 2000×g, 15 min, 4°C, take supernatant. Second centrifugation: transfer supernatant to new tube, centrifuge again at 10 000×g for 10 min at 4°C and take supernatant as platelet-poor plasma. Platelet-poor plasma will be dispensed into low adsorption cryotubes (500 µL each) and frozen at −80°C. Each aliquot sample shall be frozen and thawed at most once; if it exceeds the limit, it shall be recorded. Commercial serotonin test kit will be used, and all reagents and samples will be equilibrated to the specified temperature according to the instructions for use. The concentration gradient of standard solution is prepared according to the instructions for use; the sample is diluted according to the recommended multiple, and if it exceeds the linear range, it is further diluted and retested. Readings, calculation methods and unit conversion are performed according to the kit instructions. At least duplicate wells shall be made for standard samples and samples. Blank wells, standard wells and quality control wells are set for each plate. Definition of exposure variables Given the absence of clear precedents for this research direction, to comprehensively investigate the association between serotonin levels and POD, this study defines and analyses the exposure factor across the following three dimensions: (1) postoperative serotonin levels will be treated as a continuous variable to explore its association with POD; (2) change from preoperative to postoperative: the difference between postoperative and preoperative serotonin levels will be calculated and analysed as a continuous variable to assess the impact of the magnitude of change in serotonin levels on POD; (3) binary categorical variable: a receiver operating characteristic curve will be plotted based on the delirium outcome and serotonin levels in this study. Using the optimal cut-off value derived from the receiver operating characteristic analysis, the study population will be divided into an exposed group and a non-exposed group to determine the association between abnormal postoperative serotonin levels and the POD. Outcomes Primary outcome The primary outcome is the risk of POD within 3 days postoperatively. For patients transferred to the general ward, delirium assessments will commence on postoperative day 1 and be performed two times a day for three consecutive days using the Chinese version of the 3-Minute Diagnostic Interview for Confusion Assessment Method-defined Delirium (3D-CAM), evaluations will be conducted within designated time windows (08:00–10:00 and 16:00–18:00), with a minimum interval of 6 hours between assessments. In addition, information from medical records over the preceding 24 hours and interviews with family caregivers will be incorporated into the assessment. In patients admitted to the intensive care unit (ICU), POD will be evaluated using the CAM for the ICU once an assessable level of sedation is attained, defined as a Richmond Agitation–Sedation Scale (RASS) score of −2 or higher, and will be conducted two times a day using the same time windows until ICU discharge. The 3D-CAM employs a diagnostic algorithm based on four core features of POD: (A) acute change or fluctuating mental status; (B) inattention; (C) disorganised thinking; and (D) altered level of consciousness. The diagnosis of POD requires the presence of both features A and B, plus either feature C or D (or both). The 3D-CAM demonstrates high sensitivity (range 84%–99%) and specificity (range 90%–97%).32 Should a participant experience hospital discharge or death within the 3-day postoperative period, the final POD assessment result will serve as a surrogate value for missing data.33 Secondary outcomes Incidence of postoperative emergence agitation: The RASS will be assessed immediately after extubation in the PACU, and at 15 min and 30 min post-extubation. Emergence agitation will be defined as an RASS score ≥+1. Incidence of postoperative emergence agitation: The RASS will be assessed immediately after extubation in the PACU, and at 15 min and 30 min post-extubation. Emergence agitation will be defined as an RASS score ≥+1. POD episodes: A single positive 3D-CAM assessment establishes a diagnosis of POD. The frequency of POD-positive status is defined as the number of discrete 3D-CAM-positive days; for example, if discrete CAM-positive days occur on postoperative days 1 and 3, this will be counted as two POD episodes. If CAM-positive days are consecutive (eg, days 1, 2 and 3), this will be defined as three episodes. POD episodes: A single positive 3D-CAM assessment establishes a diagnosis of POD. The frequency of POD-positive status is defined as the number of discrete 3D-CAM-positive days; for example, if discrete CAM-positive days occur on postoperative days 1 and 3, this will be counted as two POD episodes. If CAM-positive days are consecutive (eg, days 1, 2 and 3), this will be defined as three episodes. Onset and duration of POD: The onset time of POD is defined as the interval from patient transfer to the PACU until the first POD diagnosis. The duration of POD is defined as the time elapsed from POD diagnosis to two consecutive negative 3D-CAM assessments. For patients with multiple episodes, the longest POD duration will be recorded. Onset and duration of POD: The onset time of POD is defined as the interval from patient transfer to the PACU until the first POD diagnosis. The duration of POD is defined as the time elapsed from POD diagnosis to two consecutive negative 3D-CAM assessments. For patients with multiple episodes, the longest POD duration will be recorded. POD subtype: POD can be categorised into three motoric subtypes based on clinical presentation: hyperactive, hypoactive and mixed. The RASS will be used to classify delirium subtypes in patients with established delirium.34 Hyperactive delirium is defined as persistently positive RASS scores (+1 to +4); hypoactive delirium is defined as persistently neutral or negative RASS scores (−3 to 0); and mixed delirium is characterised by RASS scores that fluctuate between positive (+1 to +4) and neutral or negative (−3 to 0). POD subtype: POD can be categorised into three motoric subtypes based on clinical presentation: hyperactive, hypoactive and mixed. The RASS will be used to classify delirium subtypes in patients with established delirium.34 Hyperactive delirium is defined as persistently positive RASS scores (+1 to +4); hypoactive delirium is defined as persistently neutral or negative RASS scores (−3 to 0); and mixed delirium is characterised by RASS scores that fluctuate between positive (+1 to +4) and neutral or negative (−3 to 0). POD severity: POD severity will be assessed using the Confusion Assessment Method-Severity (CAM-S) instrument in patients with confirmed delirium, with the highest daily score recorded. The CAM-S instrument comprises four items with a total score of 7 points.35 Symptom severity can be rated as absent (0), mild (1) or marked (2). A total score of 0 indicates normal status, 1 indicates mild delirium, 3–5 indicates moderate delirium and 6–7 indicates severe delirium. POD severity: POD severity will be assessed using the Confusion Assessment Method-Severity (CAM-S) instrument in patients with confirmed delirium, with the highest daily score recorded. The CAM-S instrument comprises four items with a total score of 7 points.35 Symptom severity can be rated as absent (0), mild (1) or marked (2). A total score of 0 indicates normal status, 1 indicates mild delirium, 3–5 indicates moderate delirium and 6–7 indicates severe delirium. Pain scores on postoperative days 1–3: The Numerical Rating Scale (NRS) will be used to assess postoperative pain.36 The NRS consists of 11 numerical ratings from 0 to 10 to describe pain intensity: 0 indicates no pain, 1–3 indicates mild pain, 4–6 indicates moderate pain and 7–10 indicates severe pain. Pain scores on postoperative days 1–3: The Numerical Rating Scale (NRS) will be used to assess postoperative pain.36 The NRS consists of 11 numerical ratings from 0 to 10 to describe pain intensity: 0 indicates no pain, 1–3 indicates mild pain, 4–6 indicates moderate pain and 7–10 indicates severe pain. Quality of recovery scores on postoperative days 1–3: The Quality of Recovery-15 (QoR-15) scale will be used to evaluate postoperative recovery quality.37 The QoR-15 instrument contains 15 items, with higher scores representing better recovery quality. Quality of recovery scores on postoperative days 1–3: The Quality of Recovery-15 (QoR-15) scale will be used to evaluate postoperative recovery quality.37 The QoR-15 instrument contains 15 items, with higher scores representing better recovery quality. Gastrointestinal function recovery time: This refers to the time interval from the end of surgery until the recovery of intestinal motility and excretory function. It is defined as the time to first flatus after surgery. Gastrointestinal function recovery time: This refers to the time interval from the end of surgery until the recovery of intestinal motility and excretory function. It is defined as the time to first flatus after surgery. Postoperative complications, including nausea and vomiting, unplanned ICU transfer, postoperative cardiovascular events, postoperative cerebrovascular events, pulmonary complications and surgical site infections. Postoperative nausea and vomiting are defined as symptoms of nausea and/or vomiting occurring within 24 hours after surgery. Unplanned ICU transfer is defined as emergency admission to the ICU within 30 days postoperatively due to sudden clinical deterioration (eg, organ failure, septic shock). Postoperative cardiovascular events are defined as new-onset or acutely exacerbated arrhythmia, myocardial infarction or heart failure occurring within 30 days postoperatively. Postoperative cerebrovascular events are defined as acute intracerebral haemorrhage or cerebral infarction occurring within 30 days postoperatively. Pulmonary complications are defined as pulmonary infection, pulmonary oedema or respiratory failure occurring within 30 days postoperatively. Surgical site infection is defined as a surgical incision meeting at least one of the following criteria within 30 days postoperatively: (1) purulent drainage from the deep incision without communication with the organ/space component of the surgical site; (2) spontaneous dehiscence of the deep incision or deliberate opening by the surgeon due to wound infection with at least one of the following symptoms: fever (>38°C), localised pain or tenderness; (3) detection of an abscess or other evidence of infection involving the deep incision identified by direct examination, reoperation, histopathology or radiology; (4) diagnosis of deep incisional infection by the surgeon or attending physician. Postoperative complications, including nausea and vomiting, unplanned ICU transfer, postoperative cardiovascular events, postoperative cerebrovascular events, pulmonary complications and surgical site infections. Postoperative nausea and vomiting are defined as symptoms of nausea and/or vomiting occurring within 24 hours after surgery. Unplanned ICU transfer is defined as emergency admission to the ICU within 30 days postoperatively due to sudden clinical deterioration (eg, organ failure, septic shock). Postoperative cardiovascular events are defined as new-onset or acutely exacerbated arrhythmia, myocardial infarction or heart failure occurring within 30 days postoperatively. Postoperative cerebrovascular events are defined as acute intracerebral haemorrhage or cerebral infarction occurring within 30 days postoperatively. Pulmonary complications are defined as pulmonary infection, pulmonary oedema or respiratory failure occurring within 30 days postoperatively. Surgical site infection is defined as a surgical incision meeting at least one of the following criteria within 30 days postoperatively: (1) purulent drainage from the deep incision without communication with the organ/space component of the surgical site; (2) spontaneous dehiscence of the deep incision or deliberate opening by the surgeon due to wound infection with at least one of the following symptoms: fever (>38°C), localised pain or tenderness; (3) detection of an abscess or other evidence of infection involving the deep incision identified by direct examination, reoperation, histopathology or radiology; (4) diagnosis of deep incisional infection by the surgeon or attending physician. Length of hospital stay: Defined as the total number of days from hospital admission to hospital discharge. Length of hospital stay: Defined as the total number of days from hospital admission to hospital discharge. Total hospitalisation cost: Defined as the sum of all expenses incurred by the patient during hospitalisation, including those for diagnosis, treatment, medications, laboratory tests, procedures, nursing care and other related medical services. Total hospitalisation cost: Defined as the sum of all expenses incurred by the patient during hospitalisation, including those for diagnosis, treatment, medications, laboratory tests, procedures, nursing care and other related medical services. Covariates Guided by prior evidence and multidisciplinary expert input, we constructed a directed acyclic graph (DAG) to represent the assumed causal relations between serotonin, demographic characteristics, perioperative factors and postoperative delirium.38 39 On this basis, we prespecified a minimal sufficient adjustment set (MSAS) for confounding control before any outcome analyses (figure3 figure 4). The MSAS included lifestyle characteristics, age, sex, comorbidities, body mass index (BMI), surgery factors and anaesthesia factors.40,50 Unless otherwise indicated, all models were fully adjusted a priori for the DAG-derived MSAS. The detailed construction of confounding, mediating and instrumental variables in the DAG, as well as the corresponding code for generating the DAG in DAGitty (https://dagitty.net/), are provided in online supplemental 1. Timeline of variables involved in the study. aCCI,age-adjusted Charlson Comorbidity Index; BI,Barthel Index; BMI, body mass index; GDS-15,15-item Geriatric Depression Scale; MMSE, Mini-Mental State Examination; NRS, Numerical Rating Scale; POD, postoperative delirium; PSQI, Pittsburgh Sleep Quality Index; QoR-15, Quality of Recovery-15; 5-HT, 5-hydroxytryptamine. Directed acyclic graph illustrating the assumed causal structure between peripheral serotonin levels (including postoperative serotonin, delta value of serotonin and dichotomised serotonin) and POD. The exposure is represented by green nodes with a triangle symbol. The outcome, POD, is shown as a blue node with a thick black vertical line. Other variables are represented as follows: blue nodes depict ancestors of the outcome; pale grey nodes signify unobserved or latent variables; and darker grey nodes represent other variables. Confounders are denoted by white nodes and were adjusted for in the final model. aCCI, age-adjusted Charlson Comorbidity Index; BMI, body mass index; GDS-15,15-item Geriatric Depression Scale; MMSE, Mini-Mental State Examination; NRS, Numerical Rating Scale; POD, postoperative delirium; PSQI, Pittsburgh Sleep Quality Index; RASS, Richmond Agitation–Sedation Scale; 5-HT, 5-hydroxytryptamine. Training for postoperative delirium assessment To ensure the reliability, validity and standardised implementation of delirium assessments in this study, all participating assessors were required to undergo systematic and standardised training. Prior to study initiation, a training team comprising three psychiatrists provided standardised training to ensure that each assessor achieved proficiency in using delirium assessment instruments and could perform independent evaluations. The training programme consisted of three core modules. First, the theoretical module addressed the definition, clinical manifestations and classification systems of delirium and included a systematic introduction to the 3D-CAM and the CAM-S instruments, alongside standardised operational procedures and differential diagnosis. Second, during the simulated practical session, assessors observed three standardised interview videos and performed delirium assessments and severity grading; results were subsequently compared with predetermined ‘gold standard’ scores to quantify the assessors’ degree of mastery. Finally, during the evaluation phase, assessors evaluated simulated delirium and non-delirium cases alongside psychiatrists. A consensus on at least three cases—showing complete concordance with the psychiatrists’ judgements—was required to successfully complete the evaluation. Additionally, regular monthly refresher training and assessments were conducted, covering the latest protocols and practical experience summaries, to ensure that assessor reliability and scoring consistency were continuously monitored and enhanced. Anaesthesia protocol Following admission to the operating room, all patients will undergo radial artery catheterisation and receive routine monitoring including ECG, invasive arterial blood pressure, pulse oxygen saturation and nasopharyngeal temperature. For anaesthesia induction, patients will be positioned in a head-elevated supine position, receive 6 L/min oxygen via facemask and a slow intravenous bolus of sufentanil citrate 0.5–1 µg/kg. After 3 min, etomidate emulsion 0.2–0.3 mg/kg will be administered. Following loss of consciousness, cisatracurium 0.20 mg/kg will be administered and tracheal intubation will be performed with an appropriate endotracheal tube size. After tracheal intubation, patients will be connected to the mechanical ventilator. Ventilator parameters will be set to volume-controlled ventilation mode with a tidal volume of 6–8 mL/kg, maintaining end-tidal carbon dioxide at 35–40 mm Hg and partial pressure of oxygen at 120–150 mm Hg. Intraoperative anaesthesia maintenance will employ propofol 2 mg/kg/hour, cisatracurium 0.1 mg/kg/hour and remifentanil 0.15 µg/kg/min. To ensure haemodynamic stability, the anaesthesiologist will adjust vasoactive agents and fluid management based on patients’ vital signs to maintain adequate tissue perfusion (urine output ≥0.5 mL/(kg×hour)), avoid electrolyte disturbances and acid-base imbalance. Intraoperative monitoring of core temperature and depth of anaesthesia will be maintained, with core temperature kept at 36–37°C and bispectral index 40–60. On completion of surgery, patients will be transferred to the PACU for continued mechanical ventilation and routine monitoring. When patients meet the extubation criteria, tracheal extubation will be performed by an anaesthesia nurse. Following extubation, patients were continuously observed for respiratory stability, airway patency and haemodynamic status. Once patients meet the PACU discharge criteria (including stable vital signs, adequate pain control, no significant bleeding or respiratory depression, recovery of consciousness to or near baseline and a modified Aldrete score of ≥9), they will be evaluated by PACU medical staff and discharged to the general ward or another appropriate treatment unit. Following transfer to the general ward or ICU, a series of non-pharmacological measures will be implemented to reduce the risk of POD. For patients admitted to the ICU, excessive sedation will be avoided and efforts will be made to facilitate early recovery and expedite transfer out of the ICU when clinically appropriate. For patients in the general ward, early ambulation will be actively encouraged. In addition, orientation interventions will be provided for patients both in the ICU and general ward. Perioperative analgesia protocol Preoperatively, depending on the surgical site, regional anaesthesia techniques such as transversus abdominis plane block, rectus sheath block and local infiltration anaesthesia may be performed to reduce intraoperative opioid requirements and postoperative pain. Intraoperative analgesia is mainly remifentanil or sufentanil, titrated according to surgical stimulation and other drugs can be used to assist analgesia. Postoperative analgesia is mainly based on patient-controlled intravenous analgesia and non-steroidal anti-inflammatory drugs can be used as an adjunct or alone for analgesia. Patient-controlled intravenous analgesia will combine sufentanil (100 µg) and ondansetron (0.1 mg/kg). Parameter settings for patient-controlled intravenous anaesthesia: background dose of 2 mL/hour, bolus dose of 2 mL, lock-in time of 10 min, maximum dose per hour of 14 mL/hour. Data collection The following data will be collected through face-to-face interviews with participants or relatives, and extracted from electronic medical records systems (table 1). Table 1Schedule of data collection and measurement time pointsOutcome measureVisit 1Visit 2Visit 3Visit 4Visit 5Visit 6 Time Preoperative Intraoperative Postoperative day 1 Postoperative day 2 Postoperative day 3 Postoperative day 30 PreoperativeEligibility screening × Informed consent × Demographic characteristics × Baseline measures × Preoperative comorbidities × Laboratory indices × Preop 5-HT × IntraoperativeSurgery-related variables × Anaesthesia-related variables × Laboratory indices × PostoperativePostop 5-HT × POD × × × POD-related indicators × × × Laboratory indices × Numerical Rating Scale score × × × Quality of Recovery-15 score × × × Gastrointestinal function recovery time × × × Length of hospital stay × × × Total hospitalisation cost × × × Postoperative adverse events × × × × Preop 5-HT, preoperative 5-hydroxytryptamine; Postop 5-HT, postoperative 5-hydroxytryptamine; POD, postoperative delirium. Schedule of data collection and measurement time points Preop 5-HT, preoperative 5-hydroxytryptamine; Postop 5-HT, postoperative 5-hydroxytryptamine; POD, postoperative delirium. Preoperative data will include patients’ demographic characteristics, including age (years), sex (male/female), BMI (kg/m2), smoking history (defined as continuous or cumulative smoking for >6 months, ≥1 cigarette daily and currently still smoking), alcohol consumption history (defined as ≥500 mL beer or 50 mL spirits daily for men; ≥400 mL beer or 40 mL spirits daily for women), education level (illiterate, primary school, secondary school or higher education (higher vocational education, college or university)), baseline MMSE score, ASA physical status classification, hearing impairment (yes/no), visual impairment (yes/no). Age-adjusted Charlson Comorbidity Index, Geriatric Depression Scale, Frailty score, Pittsburgh Sleep Quality Index will be used to assess sleep quality 1 month before surgery and Barthel Index. Preoperative comorbidities: coronary heart disease (yes/no), hypertension (yes/no), diabetes mellitus (yes/no), atrial fibrillation (yes/no), chronic obstructive pulmonary disease (yes/no), fragile brain function (referring to the presence of underlying cerebrovascular or neurological lesions not in an acute exacerbation state), assessment of preoperative nutritional status using the Nutritional Risk Screening, chronic pain history (defined as pain persisting or recurring for >3 months), tumour staging. Medication use during the month prior to surgery: use of beta-blockers (eg, metoprolol, atenolol); use of calcium channel blockers (eg, amlodipine, verapamil); use of antiplatelet agents (eg, clopidogrel, aspirin). Preoperative laboratory and functional indicators: pre-anaesthetic serotonin level (ng/mL), electrolyte abnormalities (K+ <3.5 mmol/L or >5.5 mmol/L, Na+ <135 mmol/L or >145 mmol/L, Ca2+ <1.12 mmol/L or >1.23 mmol/L, Mg2+ <0.75 mmol/L or >1.02 mmol/L), low Hb (male Hb <120 g/L, female Hb <110 g/L), low albumin (ALB <40 g/L), platelet count (100–300×10⁹/L), elevated C-reactive protein (CRP >5 mg/L). Surgery-related factors: type of surgery (gastric surgery/bowel surgery), surgical approach (laparoscopic/open), anaesthesia duration (min), surgery duration (min), proportion of time with intraoperative hypotension (defined as cumulative time with mean arterial pressure <60 mm Hg divided by total anaesthesia time), duration of hypothermia (defined as the cumulative duration of core temperature below 36°C), intraoperative blood loss (mL), type and volume of transfusion, urine output (mL), fluid infusion volume (mL). Intraoperative medication dosages: opioids (converted to oral morphine equivalents), benzodiazepines, dexmedetomidine, magnesium sulfate, esketamine, vasopressors, propofol, sevoflurane, non-steroidal anti-inflammatory drugs, serotonin receptor antagonists (eg, ondansetron). Postoperative parameters: unplanned secondary surgery, postoperative analgesic regimen, postoperative rescue analgesic dosage, postoperative serotonin level (ng/mL) and laboratory parameters on the first postoperative day (electrolytes, Hb, ALB, platelet count, CRP), incidence of postoperative emergence agitation, incidence of POD, POD episodes, onset and duration of POD, POD subtype, POD severity, pain score, quality of recovery scores, gastrointestinal function recovery time, postoperative complications, length of hospital stay and total hospitalisation cost. Data management Data entry Research data will be entered into the study database using a double independent entry method. After data entry, the system will automatically generate logic check reports to examine numerical ranges, missing values and logical inconsistencies. If discrepancies are identified, investigators will review original medical records or laboratory reports for verification and correction. Data storage and security FREE Electronic Data Capture System (V.2.0; Beijing Free Clinical Medical Technology, Beijing, China). Only the principal investigator and data manager will have full database access, while other research personnel will access data according to their assigned permissions. All electronic data will be stored on encrypted servers with regular backups. Paper case report forms (CRFs) will be kept by designated personnel in locked cabinets. Personally identifiable information will be converted to unique study codes during entry to ensure privacy. Data quality control Prior to study initiation, all research personnel will undergo standardised training covering data collection methods, standard definitions and completion protocols. Regular on-site or remote monitoring will be conducted, with random case sampling and comparison against original medical records to verify accuracy. Key indicators (eg, serotonin levels, POD) will undergo double verification, independently confirmed by two investigators. Data locking and export On completion of data cleaning and verification, the database will be locked after joint confirmation of accuracy by the principal investigator and data manager. The locked dataset will serve as the final analysis database and must not be modified arbitrarily. Should revisions be necessary, a formal request must be submitted with a complete audit trail maintained throughout the process. Statistical analysts will use only the locked, exported and anonymised data. Confidentiality and preservation All research data, case report forms and monitoring records will be preserved for a minimum of 5 years following study completion. If data publication or sharing is required, ethics committee approval must be obtained and anonymisation completed prior to use. Sample size calculation Sample size was calculated using PASS V.2020 software (NCSS LLC, Kaysville, Utah, USA). Due to the limited evidence regarding the association between peripheral serotonin levels and POD, this study assumes a POD incidence of 30% in the non-exposed group (serotonin below cut-off) based on previously published meta-analyses.3 An OR of 2.0 for POD between the exposed (serotonin above cut-off) and non-exposed groups is assumed. With a two-sided α level of 0.05 and a statistical power of 80%, the required sample size was calculated to be 141 participants per group. Allowing for an anticipated 10% postoperative loss to follow-up, the final required sample size was increased to 157 participants per group, resulting in a total sample size of 314. Statistical methods Normally distributed data will be expressed as mean±SD, non-normally distributed data as median and IQR and categorical variables as frequency (n) and percentage (%). Baseline data will be compared between groups using two-sample t-tests, Mann-Whitney U tests or χ2 tests as appropriate. Univariate logistic regression analysis will be used to assess the association between serotonin levels, covariates and POD, with calculation of ORs and 95% CIs. Multivariable logistic regression analysis will be performed to examine the association between serotonin and POD, with adjustment for confounders defined as the minimal adjustment set according to the DAG, while mediators lying on the causal pathway will not be included in the adjustment model, results will be reported as adjusted ORs with corresponding 95% CIs. The median value of tertiles will be employed as a quasi-continuous variable in the model to analyse linear trends between serotonin and POD. To explore the non-linear relationship between intraoperative serotonin and POD, restricted cubic spline regression will be used. Threshold analysis will be performed using piecewise regression, and the goodness-of-fit between linear and non-linear models will be compared through likelihood ratio tests to determine the optimal model. Based on the identified threshold, an adjusted piecewise logistic regression model will be constructed to estimate the OR and 95% CI before and after the threshold. Subgroup analyses will be conducted based on prespecified variables, including age, sex, BMI, educational level, ASA grading, preoperative cognitive status, surgical type, surgical approach and potential interactions will be assessed. Sensitivity analyses excluding patients aged 60–65 years and examining alternative definitions of the exposure variables for the primary outcome will be conducted to assess the robustness of the findings. Bayesian mediation analysis is used to investigate biochemical indexes, pain scores, quality of recovery scores in the association between serotonin and POD. All statistical tests in this study will be two-sided, with p values<0.05 considered statistically significant. Data processing and analysis will be performed using R V.4.3.0, along with Zstats V.1.0(www.zstats.net). Handling of missing data For covariates (eg, demographic and baseline clinical characteristics), imputation was not performed if the proportion of missing was low (<5%). Multiple imputation will be used if the proportion of missing key covariates or primary outcome. Assuming data are missing at random, missing values will be addressed using multiple imputation methods prior to statistical analysis. Imputation will be performed five times with a maximum of 50 iterations. Predictive mean matching models will be employed for continuous variables, while logistic regression models will be used for imputation of categorical variables. Patient and public involvement Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research. Discussion This single-centre, prospective cohort study investigated the association between peripheral serotonin levels and POD in older adults undergoing gastrointestinal surgery. Drawing on established pathophysiological mechanisms and identified clinical research gaps, a rigorous study design was employed to explore the potential causal relationship between peripheral serotonin and POD, providing evidence-based support for early prediction and targeted intervention strategies. Previous studies have demonstrated associations between POD and the biomarker such as inflammatory factors, amyloid-β42, neurofilament light chain, S100 calcium binding protein β, microtubule-associated protein tau, neuron-specific enolase, prostaglandin E2, neutrophil-lymphocyte ratio and neopterin.51,54 In addition, cortisol is a principal stress hormone reflecting activation of the hypothalamic-pituitary-adrenal axis and has been linked to both the occurrence of delirium and subsequent long-term cognitive impairment in critically ill patients,55 suggesting that dysregulation of neuroendocrine function may also contribute to the development of POD. However, the relationship between peripheral serotonin levels and POD, particularly within this specific cohort of older gastrointestinal surgery patients, has yet to be systematically investigated. The core value of this study lies in filling this gap. Most previous studies on POD have enrolled patients aged ≥60 years, and confirmed a higher incidence of POD in this population.56,61 This differs from the WHO definition of older adults as those aged ≥65 years,62 and may be explained by the early emergence of age-related brain frailty at approximately 60 years old, resulting in significantly increased vulnerability to perioperative stressors including anaesthesia and surgical trauma, ultimately predisposing these patients to POD. Accordingly, this study focuses on patients aged ≥60 years as a clinically relevant high-risk population, particularly those undergoing gastrointestinal surgery. Gastrointestinal surgery directly promotes serotonin release from ECs via mechanical distension and inflammatory stimulation, thereby enhancing the aetiological specificity of research in this population.25 26 Furthermore, determining whether peripheral serotonin is an independent risk factor for POD can provide clinicians with an accessible, quantifiable biomarker to address the current paucity of objective indicators for early warning—a critical need for risk stratification in high-risk older adults with comorbidities. Post-intensive care syndrome, defined as persistent physical, psychological and cognitive impairments following critical illness due to both the primary disease and ICU-related care, has cognitive dysfunction as a core component.63 Notably, the occurrence and duration of ICU delirium are independent risk factors for long-term cognitive decline, underscoring the importance of delirium prevention. The expected results of this study hold significant potential for clinical translation. If elevated peripheral serotonin is confirmed as an independent risk factor, serotonin testing could be integrated into preoperative assessment protocols for older gastrointestinal surgery patients to enable targeted interventions. Potential strategies include refining anaesthetic protocols to mitigate BBB injury, using serotonin receptor antagonists to modulate neurotransmitter balance or administering melatonin to improve sleep-wake cycles. Additionally, the correlation data regarding delirium subtypes and severity can serve as a foundation for developing individualised, precision-based intervention strategies. This study has certain limitations. Central serotonin levels and BBB permeability were not directly measured; thus, the translocation of peripheral serotonin into the CNS could not be definitively confirmed. This mechanism requires further validation through animal models or cerebrospinal fluid analysis. Furthermore, as patients were followed only up to 3 days postoperatively, the association between serotonin and long-term cognitive outcomes, such as delayed neurocognitive disorder, requires further elucidation through extended follow-up periods. Future research should expand on these findings in several ways. First, multicentre cohort studies should be conducted to increase sample sizes across diverse regions and surgical types, validating the universality of serotonin as a biomarker. Second, follow-up should be extended to 12 months postoperatively to assess the prognostic value of serotonin regarding long-term cognitive function and quality of life. Third, the integration of metabolomics and genomics could identify synergistic biomarkers, facilitating the construction of multi-indicator warning models. Finally, interventional trials are warranted to verify whether the targeted regulation of serotonin levels can effectively reduce POD incidence, providing high-level evidence for clinical practice. Supplementary material Footnotes Funding: This prospective observational study will be supported by the Youth Research Fund of Qingdao University (QDFYQN2023226), the Shandong Provincial Medical and Health Science and Technology Guidance Project (202418000774) and Shandong Provincial Natural Science Foundation (ZR2024LMB001). The funders will have no role in the design of the study, the collection of the data, the statistical analysis or the writing of the study report. Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2026-119278 ). Provenance and peer review: Not commissioned; externally peer reviewed. Patient consent for publication: Not applicable. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. References Supplementary Materials