Abstract
delirium is a common condition associated with hospital admission. Detection and diagnosis is important to identify the underlying precipitating cause and implement effective management and treatment. Quality improvement (QI) methodology has been applied in limited publications. There are even fewer publications of the role of development of the electronic health record (EHR) to enhance implementation.
we used QI methodology to improve delirium detection in the emergency department (ED). Plan Do Study Act (PDSA) cycles could be broadly categorised into technology, training and education and leadership. As part of the technology PDSA an electronic delirium pathway was developed as part of an NHS England digital systems improvement initiative (NHS England Global Digital Exemplar). The electronic pathway incorporated the 4AT screening tool, the Confusion Assessment Method, the TIME delirium management bundle, investigation order sets and automated coding of delirium as a health issue.
development of the EHR combined with education initiatives had benefit in terms of the number of people assessed for delirium on admission to the ED and the total number of people diagnosed with delirium across the organisation. The implementation of a delirium pathway as part of the EHR improved the use of 4AT in those 65 years and over from baseline of 3% completion in October 2017 to 43% in January 2018.
we showed that enhancement of the digital record can improve delirium assessment and diagnosis. Furthermore, the implementation of a delirium pathway is enhanced by staff education.
Key points
Delirium is a common reason for hospital admission.
Quality Improvement methodology can result in significant improvement in delirium detection in the acute hospital setting.
Detection of delirium can be improved by the use of a bespoke electronic health record and education.
Background
Delirium is a clinical syndrome which is acute in onset and involves abnormalities of attention, thought and level of consciousness [1]. Recognition and rate of diagnosis are often poor. Delirium is common affecting one in eight hospital inpatients, though prevalence varies dependent on ward type [2]. Risk factors for delirium include people with dementia, older age and post-surgery [3]. Outcomes for people with delirium are universally poor including increased mortality, increased length of hospital stay, higher risk of readmission and higher rates of institutionalisation [4]. The consequences of not adhering to appropriate screening procedures for delirium, in addition to impact on patient and staff safety, are distressing and costly. Akune et al. [5] calculated that each episode of inpatient delirium costs approximately £13 k.
How to encourage staff to assess for delirium in those at risk on admission to hospital has remained elusive to many. Though delirium detection is increasingly recognised as important timely assessment, diagnosis and management has been difficult to introduce into everyday clinical practise [6]. Quality improvement (QI) methodology provides an elegant framework to improve clinical safety and standards and can provide the how as well as the why a set of interventions are effective. Within the acute hospital general ward setting (as opposed to other environments such as Intensive Care or hospice), there have been a limited number of studies using QI methodology. Baurenfreund et al. [7] implemented the 4AT assessment tool and the triggers/investigation/management/explanation (TIME) management bundle on an acute medical unit. The 4AT is a well described tool to screen for delirium and the TIME bundle a management bundle, both recommended in published SIGN guidelines [8]. Through a combination of education for staff, management checklists, online delirium order set and use of a bedside laminated orientation tool use of the 4AT improved from 40 to 61% and comprehensive assessment for the causes/precipitants of delirium increased from 73 to 94% of cases. Using an assessment sticker and education, Bearn et al. [9] improved assessment of delirium in an acute medical unit from 0 to 64%. McCleary et al. [10] improved the use of the Single Question in Delirium (SQuID) on an acute surgical unit and found use of information cards and posters and incorporation into nursing care round forms to increase use of the SQuID to 50%. Dormandy et al. [11] undertook QI project in a small population of hip fracture patients. Comparing 40 patients one year to 45 the next they showed that a three stage process including the SQuiD, an electronic version of the 4AT and the assessment “PINCH ME” tool improved the use of the 4AT but had no impact on secondary measures including length of hospital stay. Evaluation with a focus on digital solutions to enhance delirium management is limited to a single paper which evaluated the use of modification of the electronic health record (EHR) in the ICU setting showing this to be an effective intervention [12].
Project rationale and aims
Organisational context
The project was completed at Salford Royal Hospital, an integrated care provider of hospital, community, primary and social care services in the North of England. It is a tertiary and level 1 major trauma centre and regional stroke centre consisting of 35 wards and 728 beds.
Baseline delirium assessment data
The hospital diagnosis and management of delirium guideline was updated in February 2017, recommending use of the 4AT. It was decided to focus on improving delirium assessment on admission for those at risk. The 4AT had been available in the emergency department (ED) clinical assessment since 2016 but was not easily visible and embedded in a safeguarding section of the proforma. This allowed baseline data collection which showed the 4AT delirium screening tool was poorly completed in the ED (around 3% for those at risk, that is aged 65 years and over).
Project rationale
The Royal College for Emergency Medicine Audit [13] for Salford in 2014 showed 19% of those 75 years and over to have a cognitive screen, with only 11% of these having a 4AT. Evidence of the poor outcomes for delirium and high prevalence from the literature provided, alongside recognition of low rates of assessment and development of hospital pathways defining standards for delirium assessment and care, the rationale for the project to embed screening for delirium for those at risk presenting to the ED.
Project team and aim
A project team was convened including the following staff groups: ED medical and nursing, lead medical and nursing, informatics and operational management and started working together in April 2017. The team had a senior sponsor in the director of nursing. The project aim and driver diagram was developed by the team as shown in Figure 1. The project aim was to screen 65% of those admissions aged 65 years and over from the A&E department for delirium by March 2018.
Project Driver diagram.
Design and strategy
The team designed Plan Do Study Act (PDSA) cycles based on the primary and secondary drivers in the diagram. The themes for PDSA cycles were based on primary drivers of technology, training and education and leadership. These are listed and detailed in Figure 2.
PDSA cycle summary.EAU = emergency assessment unit; ED = emergency department; GDE = Global Digital Exemplar; EHR = electronic health record; ANP = advanced nurse practitioner.
Technology
Salford Royal hospital has had an EHR since 1999 and uses Allscripts Sunrise Clinical Manager. In 2016, it was awarded Global Digital Exemplar (GDE) status, an improvement initiative led by NHS Digital in the United Kingdom offering the opportunity to redesign the EHR to enhance clinical care. One of the projects was the GDE delirium and dementia project [14]. In brief, bespoke electronic documentation for the assessment and management of delirium were developed and introduced into the EHR in September 2017. The aims included to improve delirium assessment on admission to ED and in any inpatient with new confusion, improve delirium management and to improve coding for delirium in the health issues section of the record. We gained permissions to publish the 4AT, Confusion Assessment Method (CAM) and TIME bundle, Royal College of Psychiatrist’s information sheet link, from relevant publication owners [15–18]. Details of the EHR reconfiguration and some of the technical prerequisites are detailed in Appendix 1.
The documentation was designed with the EHR development team and clinicians. To complement the launch of the new EHR documentation training was completed in both a training room environment and with trainers visiting the ward to show clinicians (medical and nursing staff) the signed off document. The GDE project was supported by a communication lead supporting the EHR change launch with screensaver, posters and brochures. The launch was also supported by:
organisation newsletters
email via clinical and nursing leads
nurse educator training
training guides including hints and tips made available via the hospital intranet and circulated by email
Training and education
A survey was designed to assess the attitudes towards and knowledge of delirium amongst staff in the ED. The survey was undertaken in April 2017 and repeated in April/May 2018.
A short presentation on delirium screening and importance was produced and delivered to nurses, doctors and health care assistants in the week commencing 28 May 2018 by a medical student, and presented during a teaching session for advanced nurse practitioners.
Training sessions were also delivered by a Consultant Geriatrician to Senior ED staff at the start of the project and at induction of junior medical staff during the project.
Leadership
A monthly “delirium champion” was introduced in May 2018. Clinicians had their rates of delirium screening of the over 65 s monitored throughout the month. The clinician shown to have recorded the highest rate of screening in those over 65 was recognised as a “delirium champion” and given a certificate for their training record. An email was then sent to doctors and nurses working in the ED congratulating the individual and giving a delirium screening rate for the department for the preceding month to give staff members an idea of the amount of screening occurring.
Measurement
The percentage of people aged 65 years and over admitted through ED to hospital was measured on a weekly basis. The percentage of patients admitted through ED screened for delirium was collected on a daily basis, before being collated into weekly totals and plotted on a statistical process control (SPC) chart. Knowledge of delirium amongst ED staff was measured before and after education PDSA.
During the implementation of the EHR Delirium and Dementia assessment tool, working with the GDE project clinical lead, data from the wider hospital was collected on delirium diagnostic rates and outcomes. This additional data were provided by the business intelligence team and anonymised data were analysed by a statistician using Stata version 14.
Results and study of intervention
4AT completion in ED
Prior to any changes being tested baseline data were collected between 1 April 2017 and 14 June 2017, the baseline mean was established to be 3%. Between June and August 2017 several changes focussing on raising awareness of 4AT were tested and during this period the mean increased to 7%.
In September 2017 the changes to EHR as part of the GDE Programme went live, this was immediately followed by several tests of change raising awareness of the new functionality. By the beginning of November 2017, the mean percentage of patients screened had increased to 39%. Tests of change continued into 2018, including sharing achievement posters and presenting at an organisation wide team brief. This resulted in a mean of 43% that was sustained from January to May 2018 which marked the end of the active measurement period (Figure 3).
SPC run chart of percentage of patients admitted through ED screened for delirium.
Survey staff knowledge
The initial survey had 23 responses and the second 46. The baseline survey found that only 39% of respondents could suggest a tool for delirium screening, 17% named the 4AT and only 35% were aware of that a tool could be accessed via the EHR system. More than half of respondents stated that they did not use a screening tool. In the follow up survey, 54% clinicians correctly identified the 4AT as the departmental tool used to assess delirium and 76% knew a tool was available to them in EHR. Reassurance was required for clinical staff about “burden” of additional documentation and referring to consequences of not implementing these changes.
In the original survey, 52% of clinicians stated that they did not use the tool and 13% said that they would only use it some of the time. A major theme identified was that clinicians felt they could rely on clinical judgement for diagnosis or absence of delirium without the use of a screening tool such as the 4AT or CAM. In the follow up survey, only 2% of clinicians said they did not use the tool and 17% said they used it some of the time. Many clinicians still felt that clinical judgement was sufficient to make a diagnosis so only used the tool on occasions when they deemed it appropriate and useful to do so.
Wider hospital data
We looked at other outcome for those with a confirmed diagnosis of delirium as part of GDE project benefits analysis (Table 1). This showed that between 2016/17 and 2017/18 the number of 4AT completed in people 65 years and over increased from 595 to 4,285. The number of people diagnosed with a delirium rose from 1,899 to 2,546, which as a proportion of non-elective admissions aged 65 and over (11– 15%) was statistically significant (P < 0.001). The length of stay for people diagnosed with delirium reduced by 2 days (median 13 versus 11 days, P < 0.001). Length of stay for all non-elective 65 years and over was median and interquartile range 4 (1–11) for both years, analysed with Mann Whitney test P = 0.007 and using an alternative test the Moods test χ2 = 2.7562 and P = 0.097. Two statistical tests were performed for Length of stay as the best test was uncertain. Length of stay for both all over 65 years and the delirium only group were both significant, slightly more so in the delirium group as may be expected based on the difference in medians compared to the over 65s group. Readmissions for delirium patients reduced from 15 to 14%, though this was not statistically significant.
Whole hospital outcomes data
| Measure | 16 October/17 September | 17 October/18 September | Statistical test and significance level |
|---|---|---|---|
| Total non-elective admissions to hospital | 40,206 | 42,050 | – |
| Total non- elective admissions to hospital age 65 years and over (% of total) | 16,551 (41%) | 16,614 (40%) | χ2 = 23.4069, P < 0.001 |
| 4AT completed in people 65 years and over | 595 | 4285 | a |
| Total number of non-elective admissions with delirium diagnosed during admission (% of age 65 years and over) including ED | 1899 (11%) | 2546 (15%) | χ2 = 105.9312, P < 0.001 |
| Length of stay for people with delirium (median and IQR) | 13 (6–30) | 11 (5–25) | Mann–Whitney U test, P < 0.001 |
| Readmissions at 30 days for people with delirium (% of admission) | 285 (15%) | 346 (14%) | χ2 = 1.7954, P = 0.180 |
| TIME bundles completed | 0 | 94 | – |
| Measure | 16 October/17 September | 17 October/18 September | Statistical test and significance level |
|---|---|---|---|
| Total non-elective admissions to hospital | 40,206 | 42,050 | – |
| Total non- elective admissions to hospital age 65 years and over (% of total) | 16,551 (41%) | 16,614 (40%) | χ2 = 23.4069, P < 0.001 |
| 4AT completed in people 65 years and over | 595 | 4285 | a |
| Total number of non-elective admissions with delirium diagnosed during admission (% of age 65 years and over) including ED | 1899 (11%) | 2546 (15%) | χ2 = 105.9312, P < 0.001 |
| Length of stay for people with delirium (median and IQR) | 13 (6–30) | 11 (5–25) | Mann–Whitney U test, P < 0.001 |
| Readmissions at 30 days for people with delirium (% of admission) | 285 (15%) | 346 (14%) | χ2 = 1.7954, P = 0.180 |
| TIME bundles completed | 0 | 94 | – |
IQR = interquartile range.
Time periods referred to are 1 October to 30 September for each year. Other than where specified all figures are totals.
a4AT may be repeated more than one per individual hence statistical comparison not performed.
Whole hospital outcomes data
| Measure | 16 October/17 September | 17 October/18 September | Statistical test and significance level |
|---|---|---|---|
| Total non-elective admissions to hospital | 40,206 | 42,050 | – |
| Total non- elective admissions to hospital age 65 years and over (% of total) | 16,551 (41%) | 16,614 (40%) | χ2 = 23.4069, P < 0.001 |
| 4AT completed in people 65 years and over | 595 | 4285 | a |
| Total number of non-elective admissions with delirium diagnosed during admission (% of age 65 years and over) including ED | 1899 (11%) | 2546 (15%) | χ2 = 105.9312, P < 0.001 |
| Length of stay for people with delirium (median and IQR) | 13 (6–30) | 11 (5–25) | Mann–Whitney U test, P < 0.001 |
| Readmissions at 30 days for people with delirium (% of admission) | 285 (15%) | 346 (14%) | χ2 = 1.7954, P = 0.180 |
| TIME bundles completed | 0 | 94 | – |
| Measure | 16 October/17 September | 17 October/18 September | Statistical test and significance level |
|---|---|---|---|
| Total non-elective admissions to hospital | 40,206 | 42,050 | – |
| Total non- elective admissions to hospital age 65 years and over (% of total) | 16,551 (41%) | 16,614 (40%) | χ2 = 23.4069, P < 0.001 |
| 4AT completed in people 65 years and over | 595 | 4285 | a |
| Total number of non-elective admissions with delirium diagnosed during admission (% of age 65 years and over) including ED | 1899 (11%) | 2546 (15%) | χ2 = 105.9312, P < 0.001 |
| Length of stay for people with delirium (median and IQR) | 13 (6–30) | 11 (5–25) | Mann–Whitney U test, P < 0.001 |
| Readmissions at 30 days for people with delirium (% of admission) | 285 (15%) | 346 (14%) | χ2 = 1.7954, P = 0.180 |
| TIME bundles completed | 0 | 94 | – |
IQR = interquartile range.
Time periods referred to are 1 October to 30 September for each year. Other than where specified all figures are totals.
a4AT may be repeated more than one per individual hence statistical comparison not performed.
Lessons and limitations
Through PDSA cycles, this project showed that changes in the EHR had an immediate impact on the use of screening tools for delirium in those at risk in a busy ED. The SPC run chart also indicates that education sessions for junior doctors were effective. Education on the clinical aspects of delirium as well as how to use the new digital assessment on EHR was important. The delivery of education was time consuming, and we learned that a bigger project team would be beneficial for timely delivery and for maintaining staff knowledge.
Staff surveys helped to identify gaps in knowledge around delirium screening. It was interesting that staff cited clinical judgement as sufficient to make a diagnosis of delirium, despite published literature to the contrary [19]. A further barrier to screening amongst clinical staff was the attitudes of clinicians about the process. Some expressed a feeling of futility towards screening due to a lack of actions being put in place within the ED before the TIME bundle package was implemented. It was identified there was some debate amongst clinical staff as to whether delirium assessment was a medical or nursing role. Both groups of clinicians are expected to fill the assessment out; the task is not designated to one particular group of staff which may be a reason for some staff not screening.
Though the project resulted in some improvement in delirium assessment using the 4AT the mean completion for patients 65 years and over in ED was 43%. Some of the barriers in relation to staff behaviour are described earlier. We hypothesise additional contributing factors include: frequency of staff education sessions, poor use of the TIME bundle and also the relatively short time frame of the project. The TIME bundle was used only 94 times between 1 October 2017 and 30 September 2018. Case note audit as part of the national dementia audit for the hospital showed 31/50 (77%) people with dementia had a full clinical assessment after delirium was identified [20]. The focus of the project was implementation of screening for delirium using the 4AT. The project did not specifically focus on implementation of the TIME bundle, other than to enhance availability within the EHR. Further, PDSA cycles aimed at encouraging TIME bundle usage might in turn incentivise staff to complete the 4AT.
We have summarised the impact of the EHR changes across the organisation. However, due to scale we did not use QI methodology to assess the contribution of the EHR changes and bespoke departmental educational sessions resulted in the improvements realised, or downstream impact of ED assessment in individual clinical inpatient areas. Hence changes in length of stay cannot be causally related to the EHR and education interventions. It may be that less severe cases of delirium were detected earlier changing the case mix with resultant difference in length of stay.
Conclusions
To conclude we have shown that delirium screening can be improved through development of the EHR and adoption of QI methodology to develop and test changes in the areas of technology, improving teaching and education and leadership. Though we are aware of other centres that have incorporated delirium assessment into EHR, this is the first large study to publish the impact of re-designing the EHR to incorporate a suite of options to assess and guide management of delirium. This is particularly timely with implementation of the National Early Warning Score 2 [21] across England and Wales, with detection of new confusion a new parameter, and thus finding effective ways to embed of pathways and assessment for delirium will be vital. This project illustrates improvements from almost no delirium screening in those at risk to near screening of 50% with existing workforce. Further, PDSA cycles might include making the assessment mandatory or support of an assistance workforce as described by Hullick et al. [22]. Further, the involvement of patients and carers in education sessions was not fully explored but would be interesting to incorporate in the future.
Acknowledgements
We would like to thank Tony Holmes and Sarah Monks who contributed to the quality improvement project, Fraser Brooks who completed the initial emergency department staff knowledge survey and Calvin Heal of the University of Manchester who assisted in the statistical analysis.
Declaration of Conflicts of Interest
None.
Declaration of Sources of Funding
The development of the Electronic health record was funded by the Global Digital Exemplar funding through NHS England.
Comments