Abstract
Background:
Delirium is a preventable and reversible complication for intensive care unit (ICU) patients, which can be linked to negative outcomes. Early intervention to cope with the risk factors of delirium is necessary. Yet no specific description of the Artificial Intelligence Assisted Prevention and Management for Delirium (AI-AntiDelirium) following the Template for Intervention Description and Replication (TIDieR) checklist was reported. This is the first study to describe a detailed process for the development of an evidence-based delirium intervention.
Aims:
To describe an individualised delirium intervention which is delivered by an artificial intelligence-assisted system in the ICU for critically ill patients.
Methods and results:
The TIDieR checklist improved the description of ICU delirium interventions, including several key features for improved implementation of the intervention. This descriptive research describes the AI-assisted ICU delirium interventions for improving cognitive load and adherence of nurses and reducing ICU delirium incidence. Following the TIDieR checklist, we standardised the flow chart of ICU delirium assessment tools; formed an evaluation sheet of ICU delirium risk factors; and translated the evidence-based ABCDEF bundle intervention into practice. Therefore, nurses and researchers would benefit from replicating the interventions for clinical use or experimental research.
Conclusions:
The TIDieR checklist provided a systematic approach for reporting the complex ICU delirium interventions delivered in a clinical interventional trial, which contributes to the nursing practice policy for the standardisation of interventions.
Introduction
Incorporation of evidence-based interventions into clinical settings requires a comprehensive set of descriptions, which not only serve to offer reliable implementation of interventions but also allow replication in other studies (Hoffmann et al., 2014). However, numerous reports of randomised controlled trials (RCT) lack a sufficient description of the interventions due to limited space in an article (Hoffmann et al., 2014; Hoffmann and Walker, 2015), therefore, a better way to report the interventions is in a separate paper. The Template for Intervention Description and Replication (TIDieR) checklist, was based on the Consolidated Standards of Reporting Trials (CONSORT) 2010 statement (Schulz et al., 2010) and the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) 2013 statement (Chan et al., 2013). TiDieR is a guide to report the detailed interventions in a trial and ensure the replicability of interventions.
This checklist can be useful in reporting complex interventions, for example, intensive care unit (ICU) delirium, which is triggered by multiple risk factors (Ying Guo, 2016). The individualised delirium intervention in the intensive care unit is aimed to assist ICU nurses to implement evidence-based delirium interventions. To report the trial results for replication and further studies, the details of the interventions are reported here. A complex set of delirium interventions contains multiple detailed procedures. These procedures include (1) the use of the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) or Intensive Care Delirium Screening Checklist (ICDSC) for ICU delirium assessment; (2) risk factors evaluation and use of an ICU delirium prediction model to classify patients into different risk levels; (3) early prevention or management interventions for ICU delirium (Kobayashi et al., 2013), use of the ABCDEF (A stands for Assessment and measures on managing pain (which is a crucial risk factor of ICU delirium); B represents Both conducting spontaneous awakening trials (SATs) and spontaneous breathing trials (SBTs), deep sedation and mechanical ventilation are crucial risk factors of ICU delirium; C stands for Choice of analgesics and sedatives in maintaining light sedation; D denotes Delirium prevention and management using non-pharmacological intervention; E stands for Exercise or mobility (immobility is a crucial risk factor of ICU delirium); and F represents Family participation (restrictive ICU visit is a crucial risk factor of ICU delirium) bundle, which is recommended by the Pain, Agitation/Sedation, Delirium, Immobility, and Sleep Disruption (PADIS) Guidelines in adult patients in the ICU as a way to reduce the incidence of ICU delirium or a management intervention once ICU delirium occurs (Hsieh et al., 2019; Trogrlic et al., 2019).
However, adherence to ABCDEF bundle intervention may be sub-optimal in routine clinical care (Zhang et al., 2021). Previous studies demonstrate that various barriers may hinder adherence in implementing the delirium intervention, for example, unfamiliarity with assessment tools (Alhaidari and Allen-Narker, 2017), and increased work burden (Devlin et al., 2008). The most important reason for lower adherence is that healthcare providers are unclear about why, how, who, and when to implement the effective intervention. The above reasons could result in slowing down of receiving and information processing, and the decreased capacity of cognitive resources (Wu et al., 2017). When the cognitive resources are less than the requirement of cognitive tasks, there may be increased cognitive load and a lower level of performance adherence in nurses (Wu et al., 2017).
To overcome those barriers, the Clinical Decision Support System (CDSS), an artificial intelligence (AI)-assisted system was designed to aid clinical decision-making and reduce nurses’ cognitive load (Lyell et al., 2018; Pickering et al., 2010). Wu et al. (Wu et al., 2017) and Dal Sasso et al. (Dal Sasso and Barra, 2015) found that medical staff had less clinical information (e.g. diseases assessment results, medications record, interventions) to remember using the CDSS, which reduced the cognitive resources and led to a significant reduction in cognitive load and improvement in adherence. Therefore, we developed an
This is the first study to report a detailed process for the development of the AI-AntiDelirium database following the TIDieR checklist (Hoffmann et al., 2014). To allow replication of each component of the intervention, the description of interventions should involve sufficient details such as procedures, materials, number of times, duration, mode of delivery, and how and when to administer essential processes.
Methods
As shown in Figure 1, all methods were performed following the Template for Intervention Description and Replication (TIDieR) checklist (Hoffmann et al., 2014). The Ethics Committee of the University approved this study (Z2019SY21). Informed consent was obtained from all subjects prior to study participation. All study data were anonymised and treated confidentially, and will not be disclosed to anyone other than the research group. For the duration of the study, all paper documents of each participant were locked up and stored in cabinets, and electronic data were input into a password-protected database.
Flowchart of the TIDieR checklist.
There were three steps to developing the AI-AntiDelirium database to support successful implementation by ICU nurses, including: Step 1. Needs assessment; Step 2. Development of intervention; Step 3. Formulation of the AI-AntiDelirium database.
Step1: Qualitative study for needs assessment
The AI-AntiDelirium database aims to reduce cognitive load and enhance adherence to implementing the methods by ICU nurses and minimise the negative effects of ICU delirium. As ICU nurses were the end users, the first step was to investigate ICU nurses’ attitudes toward delirium assessment, prevention and management, and collect the needs requirement for specific functions they needed including the content of the intervention and the mode of presenting information (such as pictures, texts and sounds), and procedure of the interventions (e.g. who carries out, how to do, duration and frequency). The needs of nurses regarding their understanding of delirium interventions were carefully assessed.
Our research team members conducted a qualitative study with one-to-one semi-structured in-depth interviews among ICU nurses to identify their experiences of caring for delirium patients, and specify needs in preventing and managing ICU delirium for future nursing care. During the interview, participants were asked a series of open-ended questions, such as, ‘Can you describe a delirious patient who impressed you most?’, ‘How do you identify delirium?’, ‘What are interventions you do after detecting the delirium?’, ‘What problems are encountered in delirium assessment and intervention implementation?’, ‘If there was a system to assist in delirium assessment and intervention implementation, what function would you like it to have?’. ICU nurses illuminated their desired system during answering these questions. The obtained data were analysed according to Colaizzi’s (1978) phenomenological procedure.
Step 2: Development of the intervention
We developed the AI-AntiDelirium database according to the results of the needs assessment. We performed a comprehensive search to identify guidelines and articles related to delirium.
Step 3: Formulation of the AI-AntiDelirium database
Expert discussion meeting
The AI-AntiDelirium database protocol contains delirium assessment, risk factor recognition, prevention, and treatment strategies. To make sure the protocol is appropriate for clinical application, expert discussion meetings were held, and six well-known clinical experts (each with more than 10 years of experience in nursing care of neurological or critically ill patients) were invited to discuss and revise the draft version of the protocol. After the expert discussion meeting, the expert opinions were collected and integrated to refine the protocol into its final version.
Translating evidence-based interventions into practical applications
After approval by the experts, the evidence-based interventions were translated into practical applications. We used the computer logic programme method to tailor individualised interventions. The following are required in this method: (1) a database containing all the interventions; (2) a series of ‘IF-THEN’ trigger rules; (3) a channel that forms tailored interventions for each patient based on his/her condition; and (4) a data source. Additionally, detailed content and delivery of each intervention were clearly described, including the content of the interventions, the persons to deliver or receive the interventions, and the time, dose, and frequency of the interventions.
Theory for the designing of the AI-AntiDelirium database
In the 1980s, Sweller proposed the CLT based on research about the limitation of cognitive resources in the field of psychology (Sweller, 1988). The theory assumes that human cognitive resources are limited, and a certain number of resources are needed to process and maintain information. Based on the theory, the cognitive load of ICU nurses that arises from ICU delirium assessment, risk factor evaluation, and preventive and management interventions can be reduced by optimised modes of presentation.
Results
Item 1. Brief name
AI-assisted delirium intervention.
Item 2. Why: Rationale, theory, or goal of the elements essential to the intervention
Step1: Needs assessment
Eleven ICU nurses (seven senior nurses, two supervisor nurses, two co-chief nurses) were included in the qualitative study, average age was 35.5 (SD 3.6), mean ICU working experience was 12.4 years (SD 4.7), 54.55% had bachelor’s degrees or above. Four main problems were extracted from the interview, including lack of awareness and knowledge on delirium assessment instrument, unfamiliar with risk factors of delirium, no existing decision-making support system for making nursing care plan, and unfamiliar with the procedure of delirium interventions. These problems are presented as themes below.
Theme 1: Lack of awareness and knowledge on delirium assessment instruments
During the interview, all the ICU nurses stated that delirium was not routinely assessed and monitored in clinical practice: Our department policy does not require us to assess delirium daily, so we rarely use delirium assessment tools. [supervisor nurse]
In the process of using delirium assessment tools, many errors were prone to occur, such as a wrong memory of the result, error in total calculation score, therefore, nurses were reluctant to use delirium assessment tools: We need to remember the results of each item, and finally calculate the total score to judge whether the patient is with delirium or not, this process needs us to spend time memorising each item. [co-chief nurse]
Nurses expressed that they needed a time-saving and easy-to-use assessment tool or a system to provide delirium assessment tools that automatically calculates results.
Theme 2: Unfamiliar with risk factors of delirium
All the ICU nurses indicated they were not comprehensively assessing the delirium risk factors. Seven ICU nurses mentioned that they were unclear about risk factors and four nurses expressed that there was challenge in collecting numerous risk factors through multiple channels: We know that many risk factors contributed to delirium, but we are not willing to assess delirium risk factors due to the complexity of the assessment pathway for risk factors, for example, we need to check the results of laboratory tests to determine whether there are electrolyte disturbances, metabolic acidosis; check the prescribed medication to determine whether to use psychoactive drugs; communicate with the patient to determine whether there are sleep disturbances and cognitive dysfunction. [senior nurse]
Nurses expressed that they needed an informational hand-out about delirium risk factors, and an ICU delirium risk prediction model which can dynamically assess patients’ risk levels in developing delirium, or a system to automatically collect delirium risk factors or provide an easy way to assist in collecting risk factors.
Theme 3: No existing decision-making support system for making nursing care plan
Ten ICU nurses mentioned that they were not able to make a risk-factors-targeted nursing-care plan even though they knew the risk factors of the patient. Seven ICU nurses expressed that they had a lack of knowledge about delirium interventions, and therefore hoped that an evidence-based intervention that can be easily tailored by the nurse, or a system that could automatically provide nursing-care plan based on patients’ existing risk factors would be desirable.
Theme 4: Unfamiliar with the procedure of delirium interventions
Six ICU nurses expressed that they were unfamiliar with the procedure of the interventions (e.g. who carries out, how to do, duration and frequency). Therefore, they hoped the system could provide relevant knowledge: We are not familiar with the content of delirium guidelines, and never receive training in terms of delirium interventions, only occasionally heard from doctors talking about delirium, so we do not know how and when to implement delirium interventions to prevent or manage delirium.[supervisor nurse]
In summary, the results of the ICU nurses’ needs requirement culminates in a delirium system where they would like to receive help in areas including ICU delirium assessment tools, risk factors assessment and targeting interventions. If they can quickly check the patient data (e.g. medication, medical or nursing records, surgical information and laboratory test results) on one system and receive the feedback about delirium risk factors and targeting interventions, it would be very helpful to promote adherence to delirium interventions.
Step 2: Development of the intervention
Flow chart of ICU delirium assessment tools
The PADIS guidelines recommend the use of CAM-ICU or ICDSC to recognise ICU delirium (Devlin et al., 2018). A flow chart of assessment tool procedures was designed based on the features of each tool. The patient’s risk for developing delirium is based on the nurse’s assessment of each item using the rules of each assessment tool.
Formulation of risk factors assessment sheet
We performed a comprehensive search to identify the guidelines related to ICU delirium. ICU delirium is caused or exacerbated by the combination of multiple risk factors (Fan et al., 2019); three types of risk factors were summarised in our study, patient-related factors (e.g. hearing and visual impairment), disease-related factors (e.g. pain, infection), environmental or iatrogenic factors (e.g. physical restraints, mechanical ventilation).
Translating evidence-based ABCDEF bundle intervention into practical applications
The PADIS Guidelines have recommended using a bundle approach (Devlin et al., 2018), namely ‘ABCDEF bundle’ which includes measures on managing pain, conducting spontaneous awakening trials and spontaneous breathing trials, maintaining light sedation, and encouraging early mobility, family participation and use of non-pharmacological interventions to prevent and manage ICU delirium. The ABCDEF bundle is targeted at eliminating various modifiable risk factors of ICU delirium, such as pain, mechanical ventilation, and immobility, and can be used as either a preventive intervention to reduce the incidence of ICU delirium or as a management intervention once ICU delirium occurs. The appropriate subset of interventions from the ABCDEF bundle should be tailored to the patient’s specific risk factors (see Supplemental Appendix Table S1).
Step 3: Formulation of the AI-AntiDelirium database
We established an interdisciplinary team of clinical nurses, physicians, and researchers, to determine what modifications were necessary to streamline the delivery of the protocol. This team kept in touch online daily and met weekly throughout the project, and discussed a set of change requests based on CLT and design requirements, including modifying document templates, creating new viewing fields, and revising the algorithm between delirium risk factors and interventions. The modification requests were an iterative, ongoing input and cooperative process between clinical staff and the research team to optimise the delirium intervention. Finally, a set of delirium interventions to reduce the ICU delirium incidence was formulated (see Supplemental Appendix Table S1).
Item 3. Materials: Physical and informational materials used in the intervention, including those provided to participants or used in intervention delivery or training of intervention providers
The manual of the AI-AntiDelirium was developed to ensure that ICU nurses understand the use of these delirium interventions, it consists of three main parts (1) assessment tools, the rule for assisting nurses to judge if the patient is having delirium or not; (2) risk factors evaluation sheet, the definition/criteria of each risk factor has been predefined, for example, sleep quality was assessed by the Richards-Campell Sleep Questionnaire (RCSQ), and a total RCSQ score of ⩽25 was defined as sleep disorder; and (3) risk factor targeting interventions, detailed nursing interventions with duration or frequency were included in the instructions. These interventions are described as clearly as possible, so nurses can learn them with minimal training (Supplemental Table S1).
Item 4. Procedures, activities and/or processes used in the intervention
Intervention group
Nurses in the intervention group provide delirium intervention based on the AI-AntiDelirium. Firstly, an educational programme is delivered by researchers, eligible nurses are trained on how to operate the AI-AntiDelirium. The AI-AntiDelirium consists of four main modules: assessment tools, risk factor assessment, nursing care plan, and care activity checklist. Briefly, the introduction of the steps to use the AI-AntiDelirium: (1) logging into the system on a mobile phone; (2) screening ICU delirium: nurses choose one instrument (CAM-ICU or ICDSC) and complete the items according to the prompts, the AI-AntiDelirium automatically presents whether the patient has delirium or not; (3) assessing risk factors: the AI-AntiDelirium automatically retrieves risk factors of ICU delirium from the hospital information system, nursing information system and laboratory information system. Risk factors that cannot be obtained from other systems are evaluated by the nurse (e.g. family members visit, hearing or visual impairment), and the nurse manually enters this information into the AI-AntiDelirium. Additionally, the AI-AntiDelirium automatically calculates a predictive risk value for developing ICU delirium based on the prediction model (Fan et al., 2019); (4) viewing the nursing care plan: the AI-AntiDelirium can automatically tailor personalised ICU delirium prevention or management interventions based on the risk factors identified; (5) viewing care activity checklist: nurses view the care activity checklist which includes frequency of the intervention and carry out the interventions for each patient and record the reasons why patients did not receive these interventions.
Control group
Nurses in the control group provide delirium intervention based on the paper version of assessment tools (CAM-ICU, ICDSC), delirium risk factors and ABCDEF bundle interventions. Before the study, an educational programme related to ICU delirium is delivered by researchers to train nurses on how to use paper-based materials. During the study, nurses use the paper version of CAM-ICU or ICDSC to assess ICU delirium, use an evaluation list to assess risk factors, and use the appropriate subset of interventions from the ABCDEF bundle targeted to patients’ specific risk factors.
Item 5. Description of the expertise, background and specific training given to intervention providers
ICU nurses are positioned best to provide early delirium detection and early management intervention for critically ill patients when necessary because they are always at the patient’s bedside and monitoring the changes in the patient’s condition. The AI-AntiDelirium has been used by ICU nurses who are qualified registered nurses, with a minimum of 1 year of experience in intensive care.
ICU nurses in the intervention group attended a one-hour training session according to a standardised manual that describes how to operate the AI-AntiDelirium, including logging on to the system, using the delirium assessment tools, recording risk factors, and providing individualised interventions. Additionally, the ICU nurses and the researchers were required to attend regular meetings to discuss and troubleshoot the use of the interventions. The purpose of these procedures is to ensure minimal deviation from the manual. Nurses in the control group provide ICU delirium-related nursing care based on the paper version of assessment tools, risk factors, and the ABCDEF bundle. Before the study, an educational programme related to the use of paper-based ICU delirium materials was delivered by researchers with the same content delivered to the intervention group.
Item 6. Mode of delivery
The AI-AntiDelirium automatically provides personalised interventions based on the patient’s delirium assessment and risk factor assessment. These interventions are delivered primarily by nurses face-to-face; in addition, a few interventions (e.g. encourage family members to visit patients and help patients with orientation training) are implemented by family members during visiting hours.
Item 7. Type(s) of location(s) where the intervention occurred, including any necessary infrastructure or relevant features
AI-AntiDelirium was delivered in the ICU.
Item 8. Number of times the intervention was delivered and over what period of time including the number of sessions, their schedule and their duration, intensity or dose
AI-AntiDelirium is delivered 7 days a week, until the patient’s discharge. Nursing care activities varied with a specific frequency, duration, intensity, or dose. For example, assessment of delirium is recommended at least one time per shift. If the patient can exercise, nurses will guide the patient to do the active range-of-motion exercises, 10 times for each joint, three times a day. ICU nurses provide as much intervention as the patient could tolerate based on the patient’s condition. Some factors may result in a lowered dose of the intervention including patients’ illness, patients’ family members, and staff shortages. The actual implementation of interventions for each patient was recorded on a separate report form.
Item 9. Tailoring of the intervention
The AI-AntiDelirium is designed for adult patients with an expected ICU stay of at least 24 hours. On the first day of admission, the nurse assesses the patient’s s basic characteristics, vital signs, disease history, diagnosis, test results, medications, etc. Each day, different nursing interventions are provided for patients based on their daily assessment. In the manual, interventions for different delirium risk factors are tailored into detailed nursing interventions respectively (see Supplemental Table S1).
Item 10. Modifications of the intervention during the study
Before the AI-AntiDelirium trial, a pilot study was conducted to examine the feasibility of the intervention, and modifications were made according to providers’ feedback. No modifications were made to the current intervention protocol during the AI-AntiDelirium trial.
Item 11. Planned procedures for how adherence or fidelity was assessed, describe how and by whom, and if any strategies were used to maintain or improve fidelity, describe them
The detail of the intervention providers’ training is presented in item 5. In this section, adherence to AI-AntiDelirium by study nurses was described. The head nurse for each study unit, who received training in the intervention, serves the role of supervisor for the nurses’ fidelity to the manual. In addition, the principal investigator of this project acts as an observer during the implementation of the study in each centre and does not provide any feedback to ICU nurses, because we record the authentic adherence without any interference.
Item 11.1. Planned procedures to assess feasibility in the ongoing trial
The feasibility of AI-AntiDelirium is to be assessed in a pilot study during the 2-month intervention. Registered nurses are eligible for the study if they (1) have a minimum of 1year intensive care experience; (2) currently work full-time in ICU; and (3) consent to participate in this study. A two-arm, cluster randomisation controlled pragmatic trial design will be used with the cluster at the unit level. Four ICUs in a tertiary hospital will be included, and randomised into either the intervention group (AI-AntiDelirium) or the control group. This cluster RCT aims to improve guideline adherence among ICU nurses by using the AI-AntiDelirium compared to the control group. Based on a previous similar study that aimed to improve guideline adherence (Trogrlic et al., 2019), we expect that the intervention adherence among ICU nurses in the intervention group would be 80%, and 50% in the control group, with an intra-cluster (within-unit) correlation of 0.00001, a p-value set at 0.05, and power set at 80% following two-sided significant testing. Our target sample is 18 ICU nurses in each ICU. Currently, 24 nurses (mean (SD) age, 33.33 (5.57) years; 19 (79.2%) female) are included in the intervention group, and 23 nurses (mean (SD) age, 32.65 (3.46) years; 22 (95.7%) female) are recruited in the control group. ICU nurses in the two groups did not differ significantly in terms of age, gender, ICU experience, marital status, education level and professional title.
Item 12. Actual adherence or fidelity
The fidelity of the AI-AntiDelirium protocol in the trial is measured by the intervention recording form, which was used in the pilot study to record the frequency, duration, intensity, or dose of intervention. Data from the RCT will be submitted for publication on completion.
Discussion
To the best of our knowledge, this is the first study to report a detailed process for the development of theory-based and evidence-based AI-AntiDelirium, which aims at reducing cognitive load and improving the adherence of ICU nurses by AI technology, as well as reducing ICU delirium incidence and other negative outcomes in ICU patients. Following the CLT, interventions are formed in this study which integrates the nurses’ needs, evidence-based measures and experts’ opinions. ICU delirium interventions are described in detail following the TIDieR checklist. Therefore, it will be easy for the clinical staff and researchers to replicate the delirium interventions. However, implementation of the interventions described in the manual should be delayed until the full study results are published.
The major advantage of this study is that the TIDieR checklist, which has rarely been applied to delirium trials before, was adopted to describe evidence-based delirium interventions. The TIDieR checklist has been developed to provide a systematic way to report key items of the AI-AntiDelirium used in clinical trials, including the following main headings: why (rationale and underlying theory), what (materials and procedures), when (how often and how much), how (mode of delivery, tailoring, and modifications), where (location and infrastructure), and who (providers). Several points are worthy of discussion about the feasibility of the TIDieR checklist to describe a complex delirium intervention. First, the TIDieR checklist emphasises the facilitators and obstacles that promote or compromise the delivery of interventions (Yamato et al., 2016). Second, TIDieR requires a detailed description of the intervention nurses such as their background, competencies, and training received, which would benefit other hospital staff to replicate the interventions. Third, the TIDieR checklist serves as a universal norm for describing interventions, making it useful at the stage of study design (Alvarez et al., 2016). Above all, the TIDieR checklist guides the authors to report their interventions more effectively, whereby clinical staff can apply the interventions and help researchers to replicate the evidence in further studies (Yamato et al., 2016).
It is also noteworthy that all the interventions used in this study are evidence-based bundle interventions, which increases the reliability and feasibility of interventions to improve delirium-related outcomes. The evidence-based bundle intervention is aimed at facilitating nurses’ early recognition of ICU patients at high risk for delirium and addresses multiple delirium risk factors in either prevention and/or management interventions, which make a potential difference to the clinical outcomes of ICU patients. Information on the impact of bundle intervention will offer knowledge to clinical staff for providing individualised measures to critically ill patients.
Another strength of this study is that an AI-assisted system is adapted to support the translation of evidence-based interventions into practical clinical applications. Currently, applications of AI in healthcare include screening for disease, calculating the risk of disease, and providing treatment and nursing intervention for patients based on predicted outcomes. With the development of technology, AI can be better integrated into clinical nursing services to optimise nursing practice (Yang et al., 2021; von Gerich et al., 2022). Several studies have shown that AI-assisted systems can reduce the cognitive load of nursing staff (Lyell et al., 2018; Pickering et al., 2010). ICU nurses are always in a state of workload pressure and dealing with complex decision-making, which leads to psychological distress and mental fatigue (Wu et al., 2017). PoLiang and colleagues (Wu et al., 2017) found that medical staff needed more cognitive resources to complete activities, such as recalling information and collecting information from multi-channel and real-time monitoring, which when lacking resulted in poor adherence to clinical practice guidelines. Therefore, a CDSS which assists clinical staff to decrease cognitive load and enhance adherence to ICU delirium interventions may be beneficial.
Several limitations should also be noted. One is that there is still a need for a randomised controlled trial to verify the effectiveness of the intervention. The other limitation is that the needs assessment was limited to only a literature review and interviews. Multi-aspect needs assessment methods that integrate individual interviews, focus group discussions and literature should be applied in future studies to identify more problems regarding the implementation of delirium interventions. Finally, the bundle interventions only focused on particular important delirium risk factors (e.g. pain, mechanical ventilation, immobility and family members’ absence) rather than targeting other risk factors such as hypotension and metabolic acidosis. Further studies are needed that focus on expanded delirium risk factors and related interventions. Despite these limitations, this paper is the current best description of an evidence-based ICU delirium intervention which followed the TIDieR checklist.
Conclusion
The AI-AntiDelirium is described according to the TIDieR checklist, which contributes to the nursing practice policy for the standardisation of interventions. The results from this study also serve as a theoretical and methodological basis for exploring the application of the CLT in designing delirium interventions delivered by ICU nurses for critically ill patients. The needs of nurses must be taken into account when improving adherence and cognitive load. This study encourages hospital administrators to adopt AI-assisted system in complex activities, such as nursing evaluation and nursing intervention. Future well-designed RCTs are needed to examine the effectiveness of the AI-AntiDelirium in reducing ICU delirium incidence and duration.
Key points for policy, practice and/or research
This paper is the current best description of an ICU delirium intervention which followed the TIDieR checklist.
All the interventions mentioned in this study are evidence-based bundle interventions, which increases the reliability and feasibility of interventions to improve delirium-related outcomes.
An artificial intelligence-assisted system was adapted to support the translation of evidence-based interventions for delirium into practical clinical application.
To allow replication of each component of the delirium intervention, the delirium intervention is described according to the TIDieR checklist, which contributes to clinical practice for the standardisation of interventions.
This information may allow hospital administrators to rationally allocate nursing staff toward implementing delirium intervention.
Supplemental Material
sj-docx-1-jrn-10.1177_17449871231219124 – Supplemental material for Description of an individualised delirium intervention in intensive care units for critically ill patients delivered by an artificial intelligence-assisted system: using the TIDieR checklist
Supplemental material, sj-docx-1-jrn-10.1177_17449871231219124 for Description of an individualised delirium intervention in intensive care units for critically ill patients delivered by an artificial intelligence-assisted system: using the TIDieR checklist by Shan Zhang, Wei Cui, Ying Wu and Meihua Ji in Journal of Research in Nursing
Supplemental Material
sj-docx-2-jrn-10.1177_17449871231219124 – Supplemental material for Description of an individualised delirium intervention in intensive care units for critically ill patients delivered by an artificial intelligence-assisted system: using the TIDieR checklist
Supplemental material, sj-docx-2-jrn-10.1177_17449871231219124 for Description of an individualised delirium intervention in intensive care units for critically ill patients delivered by an artificial intelligence-assisted system: using the TIDieR checklist by Shan Zhang, Wei Cui, Ying Wu and Meihua Ji in Journal of Research in Nursing
Footnotes
Acknowledgements
We would like to thank all the nursing staff at the University Teaching Hospital of Beijing who took part in the study for their active participation and confidence in the researchers.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research of this article: This research was supported by Grant 72304196 from the National Natural Science Foundation of China. The funders had and will not have a role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Ethical approval
The Ethics Committee of Capital Medical University approved this study (#Z2019SY021) on 23 April 2019.
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
