The American Delirium Society (ADS) has hosted a delirium year-in-review session since 2015 to present that year’s most impactful papers. Over the years, the processes for manuscript selection have been standardised to ensure methodological rigour.
We present the methods of selecting articles for the 2022 year-in-review, including the processes for nomination and scoring. Next, we summarise the works presented during the year-in-review webinar on December 12, 2022.
Since its inception, the ADS Research Committee (RC) has coordinated the ADS year-in review session as a standing charge from the ADS Board of Directors. The RC is a multidisciplinary group with an interest in delirium research and its dissemination. It includes patient advocates as well as professionals of all career stages from the fields of neurology, critical care, psychiatry, geriatrics, anesthesiolgy, nursing, and physicial therapy. Each ADS RC member was invited to nominate up to six primary research publications they considered the most influential for the delirium field." Articles were eligible if their digital object identifier (DOI) was assigned between September 1, 2021, and July 31, 2022. We chose DOI assignment date to standardise the publication time frame as the time from online publication to inclusion in a journal issue varies across journals. September 1, 2021, was consistent with the preceding ADS annual meeting, and we advanced the time by a month to provide adequate time to prepare the session. The eligibility period for future year-in-review sessions will be August 1 to July 31.
Only primary studies on delirium were eligible; we did not consider case reports or systematic reviews. A total of 23 studies were nominated, divided into three groups: medication intervention trials, non-medication trials and implementation studies, and delirium detection/basic science studies. This balanced submissions among the three categories, with 7 studies in the first group and 8 in each of the latter groups. All submitted articles were included for review. Each RC member was assigned to a group based on area of expertise, and they reviewed and rated all studies in their category. To minimize conflict of interest, where an RC member was also a study author, they were recused from scoring and contributing to RC discussions. Reviewers were instructed to score papers from 0 to 10 on scientific rigour (Suppmementary Table 1) and impact (Suppmementary Table 2). We could not identify a reliable scale that adequately evaluates both intervention and observational studies, which is why we created these scoring rubrics. Both scales were added to generate an overall score (total possible score 20). The a priori decision was to select the three studies with the highest average overall scores. Publications with equal scores were discussed by virtual meeting. We calculated z-scores for each reviewer’s overall scores to standardise scores across reviewers, and these informed the consensus decision.
Each study was scored by 5 to 8 RC member reviewers. Mean scores (SD) for each group were: medication intervention trial group =12.8 (1.1); non-medication intervention trial/outcomes group =13.1 (1.1); delirium detection and basic science group =12.6 (1.0).
Medical Intervention Trials
1. Effect of Regional vs General Anesthesia on Incidence of Postoperative Delirium in Older Patients Undergoing Hip Fracture Surgery: The RAGA Randomized Trial
The Regional Anesthesia versus General Anesthesia (RAGA) randomised trial1 tested the primary hypothesis that regional anaesthesia would reduce the incidence of postoperative delirium in older adults undergoing surgical repair for hip fracture. The trial had an allocation-concealed, open-label, multi-centre design. Patients were randomised to receive regional anaesthesia without sedation, or general anaesthesia followed by daily clinical delirium assessments via the Confusion Assessment Method (CAM)2 the first week after surgery. Patients aged ≥ 65 years, with or without preexisting dementia, and a fragility hip fracture requiring surgical repair, were recruited from nine university teaching hospitals in Southeastern China.
Among the 950 patients enrolled and randomised, 941 were evaluable for the primary outcome of postoperative delirium incidence. Postoperative delirium occurred in 29 (6.2%) in the regional anaesthesia group versus 24 (5.1%) in the general anaesthesia group (unadjusted risk difference [RD], 1.1%; 95%CI, -1.7% to 3.8%; p = 0.48; unadjusted relative risk [RR], 1.2 [95%CI, 0.7 to 2.0]; p = 0.57]). Secondary outcomes, including mean delirium severity score, motoric subtypes of postoperative delirium, median pain score, median length of hospitalisation, and 30-day all-cause mortality were not different between groups.
The RAGA randomised trial concluded that regional anaesthesia without sedation does not significantly reduce the incidence of postoperative delirium during the first week after surgery in older patients with hip fracture. The rates of delirium were much lower than reported in other studies of hip fracture; this may have been due to the lower levels of frailty in the study population and low sensitivity of the delirium detection process as operationalised in the study. These findings add to a growing body of evidence suggesting that general anaesthetics alone do not cause postoperative delirium in older adults.
2. Spinal Anaesthesia or General Anaesthesia for Hip Surgery in Older Adults
The primary hypothesis of the Regional versus General Anesthesia for Promoting Independence after Hip Fracture (REGAIN) trial3,4 was that patients assigned to receive spinal, as compared with general, anaesthesia would be more likely to be alive and walking independently 60 days after surgery. REGAIN was a multi-centre, pragmatic, randomised superiority trial in which patients aged ≥ 50 years were randomised in a 1:1 ratio at 46 North American hospitals. The primary outcome was a composite of death or an inability to walk independently or with a cane or walker 60 days after randomisation. Secondary outcomes included postoperative delirium incidence, death within 60 days, time to discharge, and ambulation at 60 days after surgery. Postoperative delirium was assessed daily via the 3-Minute Diagnostic Interview for CAM (3D-CAM) for the first three days after surgery.5
Of the 1600 enrolled patients, 666 randomised to the spinal anaesthesia group and 769 randomised to the general anaesthesia group received their intended treatment. The primary outcome occurred similarly in each group: 18.5% with spinal anaesthesia and 18% with general anaesthesia (RR 1.03; 95% CI, 0.84 to 1.27; p = 0.83). No differences were found in secondary outcomes. Postoperative delirium occurred in 20.5% with spinal anaesthesia versus 19.7% with general anaesthesia.
REGAIN addressed the pragmatic question of whether anaesthesia type affected patient-centred outcomes in a hip fracture surgery cohort. For postoperative delirium, both groups received anaesthetic or sedative agents (albeit to a lesser extent with spinal anaesthesia), which is common practice in the U.S. and Canada. Nevertheless, the REGAIN findings for postoperative delirium are consistent with the RAGA trial.1
3. Prophylactic Melatonin for Delirium in Intensive Care (Pro-MEDIC): A Randomized Controlled Trial
In the Pro-MEDIC trial,6 the hypothesis was that evening melatonin would reduce delriium prevalence in critically ill patients if administered early in their hospital course. This multi-centre, randomised, placebo-controlled, double-blind trial was conducted across 12 Australian ICUs over three years. 847 patients, aged ≥ 18 years and with an expected ICU LOS > 72 hours, were enrolled within 48 hours of ICU admission. Patients were randomised to melatonin 4 mg or placebo at 21:00. Delirium was assessed twice daily using the CAM-ICU7 for 14 consecutive days or until ICU discharge. The primary outcome was the proportion of delirium-free assessments as a measure of delirium prevalence within 14 days or before ICU discharge.
There was no difference in the average proportion of delirium-free assessments per patient between the melatonin and placebo groups (79.2% versus 80% respectively, p = 0.55). This finding was consistent across subgroups, including diagnostic category, age, delirium risk, and secondary outcomes, including delirium severity, length of stay, mechanical ventilation, and mortality. This trial does not support the routine use of melatonin in the critically ill to prevent or treat delirium.
Non-Medical Intervention Trials and Implementation Studies
1. Outcomes of a Delirium Prevention Program in Older Persons After Elective Surgery: A Stepped-Wedge Cluster Randomized Clinical Trial
PAWEL (German acronym for “patient safety, cost-effectiveness, and quality of life”)8 was a stepped-wedge, cluster randomised trial comparing a tailored, multifaceted intervention to standard care across 5 German tertiary medical centres. The aim was to reduce postoperative delirium incidence and duration. The AKTIVER (German acronym for “everyday skills and cognition training and interdisciplinarity improves outcome and mitigates risk”) intervention included cognitive, motor, and sensory stimulation, meal companionship, stress relaxation, sleep promotion, and diagnostic test and operation room accompaniment. Patients aged ≥ 70 years undergoing elective orthopaedic, general, or cardiac surgery were eligible. The primary outcome was delirium incidence, as assessed with the I-CAM (short-form CAM plus an additional item for psychomotor activity as required per ICD-10 criteria),9 performed daily through postoperative day 7 by psychogeriatric research nurses. Delirium was assessed after discharge using a validated medical record review method.
Of the 1470 enrolled participants, postoperative delirium occurred in 147 (19.9%) of the intervention group and 171 (23.4%) of controls (RR, 0.85; 95% CI, 0.70–1.03; p = 0.10). Among cardiac surgery patients, there was no difference in delirium incidence between groups (intervention 35.2% vs control 36.5%; RR, 0.97; 95% CI, 0.77–1.22; p = 0.79). However, among non-cardiac surgery patients, delirium incidence was significantly lower in the intervention group (intervention 10.9% vs control 16.3%; RR, 0.67; 95% CI, 0.48–0.93; p = 0.008).
Adding to evidence for the efficacy of non-pharmacological interventions to prevent delirium, the AKTIVER intervention significantly reduced postoperative delirium risk in patients ≥ 70 years undergoing elective general and orthopaedic but not cardiac surgery. This was an effective intervention, but its reliance on volunteers may be challenging to implement in some centres.
2. Association Between Implementation of a Geriatric Trauma Clinical Pathway and Changes in Rates of Delirium in Older Adults with Traumatic Injury
This retrospective case-control study of electronic health records of patients admitted to a single level-one trauma centre10 compared cohorts before and after implementation of the Stanford Geriatric Trauma Care Pathway. This programme is based on geriatric best practices, which include order sets, guidelines, automated consultations, and escalation pathways delivered by a multidisciplinary team. The intervention was hypothesised to reduce delirium incidence (primary outcome) and hospital length of stay (secondary outcome). Delirium was measured using the CAM as performed by nurses each shift during routine care for the duration of hospitalisation and also if there was a diagnosis by the clinical team.
This study included 712 patients aged ≥ 65 treated for traumatic injury not undergoing operation. Delirium incidence was significantly lower in the post-implementation group compared with the baseline group (18.5% vs 28.3%; p = 0.002), even after accounting for patient demographics and injury characteristics (OR, 0.54; 95% CI 0.37–0.80; p < .001). Notably, the primary analysis did not adjust for frailty or dementia. Hospital length of stay was not different between post-implementation and baseline (mean [SD], 4.3 [2.9] days vs 4.3 [3.8] days; p = 0.20). The study provides evidence for the effectiveness of formalised, integrated, multidisciplinary systems to prevent delirium in trauma centres and warrants replication in randomised trials.
3. Pre-operative Breathing Training Based on Video Learning Reduces Emergence Delirium in Preschool Children: A Randomised Clinical Trial
This was a single-centre, double-blinded, randomised control trial to determine whether breathing training could reduce the incidence of emergence delirium in preschool children after otorhinolaryngologic surgery performed with general anaesthesia.11 The sample was 154 children aged 3 to 7 years with an ASA physical status I or II undergoing elective tonsillectomy and/or adenoidectomy under general anesthesia. The intervention involved practising controlled breathing through a straw to simulate conditions during anaesthesia emergence. Participants watched a standardised training video the day before surgery and performed breathing training sessions 5 minutes and 30 minutes after watching the video, then again before bed. The primary outcome was emergence delirium incidence, including hypoactive and hyperactive subtypes. Delirium incidence and severity were assessed with the Pediatric Emergence Delirium (PAED) scale12 administered by post-anaesthesia care unit (PACU) nurses at 0, 5, 15, and 25 minutes after awakening.
For the primary outcome, emergence delirium was lower in the group receiving breathing training than the control group (10.4% vs. 35.1%, p < 0.001). The maximum PAED score (delirium severity) was lower in the training group (5.0 ± 5.0) compared with the control group (7.0 ± 7.0) (p = 0.001). There was no difference between groups in extubation time (intervention 35.0 ± 16.0 min vs control 30 ± 20.0 min, p = 0.8) or PACU stay time (65.0 ± 20.0 min vs 65.0 ± 20.0 min, p = 0.6).
Video-based breathing training during the preoperative visit in children aged 3 to 7 undergoing otorhinolaryngologic surgery led to reduced incidence and severity of emergence delirium, deserving further investigation other settings.
Delirium Detection and Basic Science Studies
1. Comparative Implementation of a Brief App-Directed Protocol for Delirium Identification by Hospitalists, Nurses, and Nursing Assistants: A Cohort Study
This prospective cohort study13 evaluated the real-world effectiveness of implementing a smart tablet app-based delirium identification protocol, based on clinician profession. Clinicians included hospitalists, nurses, and certified nursing assistants (CNAs). Patients were assessed daily for two consecutive days by: (1) trained research personnel using a rigorous research reference standard delirium assessment based on Mini-Mental Status Examination, long-form CAM, and supplemental items to assess attention and delirium symptom severity; (2) CNAs using an app-based ultra-brief 2-item screener (UB-2); and (3) physicians and nurses using an app-based 2-step protocol consisting of the UB-2, followed by the 3-Minute Diagnostic Interview for Confusion Assessment Method (3D-CAM) in UB-2 positive patients. The authors hypothesised that the app-based protocol would be feasible (≥ 95% completion); brief (< 1 min for the UB-2; < 2 min for the 2-step protocol); and achieve an accuracy > 85%, sensitivity > 80% and specificity > 90%.
The study enrolled 527 older adult inpatients and involved 399 clinicians from two hospitals. Of the 924 delirium assessments completed, 154 (17%) of 114 patients (22%) were positive for delirium. The overall completion rate of the app-based protocol exceeded 97%. On average, CNAs completed the UB-2 in 62 seconds, while nurses and physicians completed the 2-step protocol in 104 seconds and 106 seconds, respectively. When the UB-2 was administered by physicians, nurses, and CNAs, the corresponding sensitivities were 88%, 87%, and 82%; specificities, 64%, 70%, and 66%; and accuracy, 69%, 73%, and 70%. For physicians and nurses, the 2-step protocol had sensitivities of 65% and 63%; specificities of 93% and 91%; and accuracy of 89% and 87%, respectively.
These findings demonstrate that the app-based delirium detection protocol was feasible and quick when administered in real-world practice by hospitalists, nurses, and CNAs alike. Many patients were excluded due to inability to effectively communicate in English, necessating further study in other languages.
2. The Electroencephalographic Confusion Assessment Method Severity Score (E-CAM-S)
This was a retrospective cohort study14 that aimed to develop an automated, neurophysiologically derived measure of delirium and its severity: the Electroencephalographic Confusion Assessment Method Severity Score (E-CAM-S). The E-CAM-S was calculated by applying a learning-to-rank machine learning model to frontal electroencephalography (EEG) signals from 373 adult patients. The short-form Confusion Assessment Method Severity (CAM-S) was the clinical reference standard.
For the primary outcome, E-CAM-S correlated with CAM-S (r = 0.68; p < 0.0001); 53 EEG features were predictive of delirium severity in the whole sample and 23 in the non-comatose subsample. In terms of clinical outcomes, E-CAM-S and CAM-S had similar strengths of association with hospital length of stay (Spearman’s rho: 0.33 vs 0.41, p = 0.08) and in-hospital mortality (area under the curve: 0.77 vs 0.81; p = 0.3).
These findings suggest that the E-CAM-S derived from 4 frontal EEG leads objectively quantified delirium severity and predicted clinical outcomes similar to conventional interview-based clinical assessment. The E-CAM-S shows promise for automatic detection and routine screening of delirium severity. However, future research is needed to understand the wide range of EEG features associated with delirium severity and to prospectively assess the clinical application of the features selected by the machine learning model.
3. Protective Effects of Omega-3 Fatty Acids in a Blood-Brain Barrier-on-Chip Model and on Postoperative Delirium-Like Behaviour in Mice
A microfluidic-assisted blood-brain barrier (BBB) device used primary human cells to examine the effect of omega-3 fatty acids on interleukin (IL)-115 Omega-3 fatty acids protected BBB homeostasis and inhibited IL-1 -induced inflammation in BBB-associated endothelial cells and pericytes, including vascular cell adhesion molecule 1 overexpression, intercellular gaps between human brain microvascular endothelial cells, and disruption of the human astrocytic network density.-induced inflammation.
There were effects of omega-3 fatty acid on postoperative BBB disruption, microglial activation, and delirium-like behaviour (inattention) in a mouse model. Male and female mice 12–16 weeks old were randomised to three groups: control, surgery, and surgery with application of omega-3 fatty acids. After surgery, the hippocampus was isolated to evaluate gene expression of neuroinflammatory markers; a five-choice serial reaction time task was used to assess inattention.
Omega-3 fatty acids contributed to the maintenance of BBB integrity and protected against surgery-induced microglial activation and attention deficits in mice. An additional novel finding was that surgery-induced pericyte loss and perivascular macrophage activation was effectively rescued by omega-3 fatty acid pretreatment.
This researcher-developed micro-physiological BBB device provides evidence in support of the BBB-protective effect of omega-3 fatty acids in a mouse model of delirium. It also identified pericytes and perivascular macrophages as potential new intervention targets to prevent postoperative delirium.
The nine studies presented for the 2022 ADS year-in-review webinar illustrates the breadth of the delirium field, and the growing body of clinical, basic, and translational research.
The first two trials addressed questions regarding anaesthesia type in relation to delirium and suggested that general anaesthesia does not confer substantially more risk of delirium than regional anesthesia. The third medication trial, Pro-MEDIC, added to research investigating sleep and circadian interventions as delirium targets. It concluded that evening melatonin alone was insufficient for reducing the prevalence of delirium in a mixed critically ill population.
In the next section, the first two publications studied the clinical implementation of delirium pathways. The first found that a non-pharmacological delirium bundle reduced delirium incidence among non-cardiac surgery patients but not among cardiac patients. The second study, in a cohort of older adults on a trauma service, found that a multidisciplinary team intervention was associated with reduced delirium incidence. Questions about the reliability of delirium assessment during routine care aside, both studies support the effectiveness of non-pharmacological bundles to prevent delirium, though these may not be effective in every clinical population. The third study, which focused on children undergoing ENT surgery with anaesthesia, found that a novel breathing-training intervention reduced delirium over the 30 minutes after emerging from anaesthesia. It demonstrates the importance of tailoring interventions to the population and highlights the relevance of delirium across the age spectrum.
The final three studies were diverse. A smart tablet app was a feasible and effective means of detecting delirium quickly and reliably when used by CNAs, nurses, and physicians. In the next study, investigators demonstrated a physiology-based delirium severity measure using frontal EEG and identified novel EEG signatures of delirium using machine learning deserving further investigation. Finally, basic science and animal models suggested maintained BBB integrity through omega-3 supplementation may be a novel means to prevent delirium.
Strengths of the year-in-review process described include a systematic, consensus-based approach. This approach differs from conventional narrative reviews, which can lack transparency in article selection. In contrast to systematic reviews, which typically answer a narrow research question, the year-in-review process allows the choice of top-quality studies across diverse topics and designs. Furthermore, the year-in-review process includes a rigorous quality appraisal process, in contrast with scoping reviews that map all existing research activities of a given topic, typically include no quality appraisal, and are often time-intensive to complete. Our exclusion of articles published in non-English languages is a limitation. There is also potential that our approach could inadvertently exclude high-quality articles due to lack of nomination. The involvement of a diverse group of delirium research experts and reviewers mitigates this risk.
The 2022 ADS year-in-review session highlighted the growing scientific interest in delirium and the diversity of study designs in delirium research. The highlighted publications address fundamental questions regarding anaesthesia and sleep, the complexity and importance of implementation science, and the diversity of approaches available for detecting and preventing delirium. This diversity speaks to the importance of improving delirium management within the clinical and scientific community and provides the rationale for journals devoted explicitly to delirium.
Sources of Support
This work received no direct support. CAA is supported by the National Institute on Aging of the National Institutes of Health under Award Number R21 AG080331. BJAP is supported by the National Institute on Aging of the National Institutes of Health under Award Number R01 AG057901. SKS is supported by an International Anesthesia Research Society Mentored Research Training Grant. SK is supported by National Institute on Aging under Award Number K76 AG074925. HL is supported by the National Institute on Aging of the National Institutes of Health under Award Number K23 AG076662. MO is supported by the National Institute on Aging of the National Institutes of Health under Award Number K23 AG072383.
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
CAA and BJAP serve on the Editorial Board for Delirium Communications. All other authors report no conflicts of interest.