Clinical practice guidelines have been developed to improve screening, prevention and management of delirium.
Clinical practice guidelines have been developed to improve screening, prevention and management of delirium.
To implement delirium guidelines in general medical patients to reduce incidence and duration of delirium and improve outcomes in delirious patients.
Implementation was led by a multidisciplinary team of clinicians and project staff on one medical ward. Evaluation was undertaken as a controlled trial in patients aged 65 years or older with/at risk of delirium, compared with a control medical ward. Interventions included risk screening, delirium detection, multidisciplinary education, ward modifications including a four-bed delirium bay, behaviour and medication protocols, and use of nursing assistant and volunteers. Primary outcome measures were incidence and duration of delirium; secondary outcomes were length of stay, mortality, falls and discharge destination in delirious subgroup. Process measures included ward moves, use of neuroleptics, allied health review and delirium bay use.
Of 206 consenting older medical patients, 22% were delirious at admission and 44% were at risk. No incident cases of delirium were identified. In the delirious subgroup, significantly fewer intervention participants were discharged with persistent delirium (32% vs 71%, P = 0.016), with trends to reduced inpatient mortality (0% vs 18.5%, P = 0.07) and falls (11% vs 22%, P = 0.16), at the expense of a longer medical ward stay (16 days vs 8 days, P = 0.01).
Low incidence of new delirium may reflect the established interdisciplinary care environment. Improved outcomes in the delirious group are encouraging although implementation was costly, including increased length of acute ward stay.
Delirium is an acute disturbance of attention and cognition accompanying medical illness and complicates 15–37% of acute general medical hospitalisations in people aged 65 years and older.[2-6] Delirium independently predicts poorer hospital outcomes, including prolonged hospital stays, higher mortality and greater discharge to residential aged care, and long-term outcomes, such as mortality, dementia and institutional care use.[4, 7-9]
Delirium prevention and management have been proposed as indicators of high-quality care of the older hospitalised patient. Several experimental,[5, 6, 10-14] quasi-experimental[15, 16] and observational[17-19] studies have reported interventions designed to reduce the incidence, duration and/or severity of delirium in hospitalised patients. Interventions have included screening, geriatric consultation, education, changes to the physical environment (including use of a delirium room), pharmacological protocols, nursing protocols targeting risk factors and behaviour management, and use of additional staff members or volunteers. Recommendations based on these studies have been synthesised into expert guidelines.[20, 21] However, the evidence base remains limited, particularly for management of established delirium, with highly heterogeneous interventions, variable study quality and inconsistent findings. Many guideline recommendations are based on expert opinion or low-grade evidence, and the practical implementation of guidelines has not been widely reported.
The Royal Brisbane and Women's Hospital (RBWH) general medical wards introduced consistent, well-resourced interdisciplinary teams in 2003, providing the opportunity to improve team-based care of vulnerable older patients. The general medical wards care for over 5000 acute medical emergency admissions per annum, with two thirds aged over 65 years. A prospective cohort study performed in 2006 identified delirium in 27% of general medical patients aged 65 years and older. Publication of the Australian Clinical Practice Guidelines for the Management of Delirium in Older People (hereafter called the Guidelines) provided the impetus for this quality improvement project.
The aim was to implement the Guidelines in a general medical service to improve processes for screening, prevention and management of delirium. We hypothesised that this would reduce the incidence of new delirium, reduce the duration of prevalent delirium and improve secondary outcomes, including reductions in length of stay, falls and hospital mortality.
The project was undertaken in the general medical wards of RBWH, a large metropolitan teaching hospital in Australia. The model of care and patient mix has been described previously. General medical admissions are almost all unplanned admissions and are assigned to one of five general interdisciplinary medical teams based on an administratively determined, cyclical rotating roster at the time of admission from the emergency department. The general medical teams occupy four medical wards, including a Medical Assessment and Planning Unit and three home wards. High occupancy rates mean that ward admission may be delayed, resulting in ‘boarding’ in the emergency department and other transit or outlier wards. This environment offered a number of challenges for implementation and evaluation. Screening for delirium risk factors, and assessment for delirium in patients at risk, needed to be undertaken soon after admission, often before ward allocation was determined. Team-based clinical practice strategies, such as medication review, managing causes of delirium, attention to nutrition and regular reorientation could be commenced as soon as the intervention team was assigned. Environmental strategies, such as a delirium bay, additional trained nursing assistants and changes to lighting and noise levels needed to be implemented on a specific ward.
The intervention was planned and overseen by a multidisciplinary steering committee led by a consultant physician from the intervention ward. The project team met regularly to prioritise activities, plan specific strategies, identify and address barriers to implementation and assess progress of the evaluation. Strategies were based on Guideline recommendations for both prevention and management of delirium, and are detailed in Table 1. Strategies were implemented and revised over the 1-year project period. Regular feedback was sought from ward staff by surveys, case study forums, a communication book and weekly one-on-one reviews with the nurse unit manager, volunteers and nursing assistant. Evaluation was commenced once strategies were embedded.
|Screening for risk factors|| |
Screening for all patients aged 65 years and older within 48 h of admission by project staff (screening 5 days per week)
|Delirium detection|| |
|Education and training|| |
|Ward-based strategies|| |
|Team strategies|| |
|Patient/carer information|| |
The evaluation was planned as a concurrent controlled trial (Fig. 1). Participants were patients aged 65 years or older admitted to the intervention team (medical unit 2) or control teams (medical unit 4 or 5) with an anticipated length of stay of 3 days or more. Previous studies in this clinical environment demonstrated an even distribution of patient casemix between teams and wards, facilitating comparative evaluations.[22, 23] The intervention ward was chosen because of a consistent occupancy by a single medical team (reducing the risk of contamination) and identifiable medical and nursing champions. The control ward was chosen because it was in close physical proximity and had similar staffing and policies. Patients were ineligible if they were palliative, unconscious, critically ill or had previously documented severe dementia (mini-mental status examination <10), psychiatric or intellectual disability, or dysphasia which would preclude adequate assessment. Patients were also excluded if they did not have adequate English knowledge to participate in assessment and consent processes, as the project did not allow provision of additional interpreters.
Descriptions of intervention fidelity, including rate and timing of delirium screening, education sessions delivered, number of patients admitted to the delirium bay and use of information pamphlets were documented prospectively by project staff. Processes of care reflecting compliance with the Guidelines included number of bed/ward moves, timely allied health review (within 72 h), one-on-one nursing, use of restraints and use of neuroleptic medications, obtained by clinical record review.
Primary outcomes were incidence of new delirium in patients at risk, and duration of delirium episodes, assessed by the screening process described in Table 1. Secondary outcomes were obtained by clinical record review and included falls, functional decline at discharge compared with premorbid function (any increase in the number of activities of daily living (ADL) – bathing, dressing, toileting, eating, transfers and mobilising – requiring assistance, per nursing documentation), length of stay in the general medical ward, length of overall hospital stay (including subacute ward care) and inpatient mortality. Process and outcome measures were collected and entered into the study database by project staff who were aware of group assignment. Sample size for this time-limited project was determined by available time and resources within the total project budget.
Analyses were confined to participants who were delirious or had two or more risk factors for delirium on initial screening. Processes and outcomes were summarised using standard summary statistics. Categorical variables were compared between the intervention and control group using contingency tables and chi-square testing (or Fisher's exact test if expected cell counts <5), while continuous variables were compared using t-test (normally distributed variables) or Mann–Whitney U-test. Because intervention strategies targeted both prevention and management of delirium, analyses were undertaken in all at-risk participants and in the delirious subgroup. A P-value of <0.05 was considered significant in all analyses.
The study was approved by the Human Research Ethics Committee. Delirium studies are challenging because of difficulties obtaining consent. While this was a quality improvement initiative which did not pose risks to participants, individual data collection was required for robust analysis. Study staff assessed capacity to consent to data collection for each patient and sought proxy consent if necessary from a statutory health attorney (usually a family member). When such an authority was not available, patients with delirium were commenced in the study, and consent for data inclusion was sought if capacity to consent returned. If consent for data collection was denied, or capacity did not return, the patient was excluded from the evaluation.
The implementation occurred in three phases: a 3-month establishment and planning phase (including establishing project leadership and governance, recruitment of project staff, development of screening tools and education programmes, and ethics application); a 4-month implementation phase with iterative development and implementation of project strategies; and a 4-month evaluation phase. A series of five 30-min weekly education sessions based on the Guidelines was delivered by the project team to nursing staff and interdisciplinary team members during the implementation phase, with each topic delivered three times during the week to optimise reach among shift workers. Overall, topic attendance was 40–73% of all ward staff; written materials were mailed to staff unable to attend. Monthly casebased nursing forums facilitated by one of the intervention unit physicians were held for the duration of the intervention, and 200 patient/carer brochures were utilised.
Participant enrolment is shown in Figure 1. Of 206 older patients who underwent admission screening, 46 (22%) were delirious and a further 90 (44%) were considered at risk, providing 136 participants for analysis. Participant characteristics were similar between the intervention and control groups (Table 2).
|n = 74||n = 62|
|Age, mean (SD) (years)||82.3 (7.7)||79.6 (8.2)||0.36|
|Male, n (%)||38 (51.4)||30 (48.4)||0.73|
|Risk factors, n (%)|
|Impaired cognition||42 (56.8)||32 (51.6)||0.55|
|Impaired vision or hearing||73 (98.6)||62 (100)||1.0|
|Dehydration||37 (50.0)||37 (59.7)||0.26|
|Hyponatraemia||12 (16.2)||13 (21.0)||0.48|
|From residential aged care, n (%)||15 (20.3)||6 (9.8)||0.10|
|Independent in ADL pre-admission, n (%)||34 (45.9)||32 (51.6)||0.51|
|CAM screening within 48 h, n (%)||62 (83.8)||48 (77.4)||0.35|
|Prevalent delirium, n (%)||27 (36.4)||19 (30.6)||0.47|
Twenty-eight (45%) of intervention participants were managed for all or part of their stay in the four-bed delirium bay: 12/19 (63%) delirious participants and 16/43 (37%) of those at risk. Ward moves were significantly reduced in the intervention group (34% vs 51% had three or more moves during admission (P = 0.05)). Patterns of early allied health review (by day 3 of admission) were similar (Fig. 2a).
No at-risk participant developed incident delirium during the acute admission. There was a trend to longer medical ward stay in the intervention group (median 11 days vs 8 days, P = 0.07). Mortality, falls, functional decline and new residential care placement were similar between groups (Table 3).
|All participants||Participants with delirium|
|n = 74||n = 62||n = 27||n = 19|
|Length of acute stay, median (IQR)||8 (5–14)||11 (6–19)||0.07||8 (4–20)||16 (12–20)||0.01|
|Length of hospital stay, median (IQR)||9 (5–20)||14 (7–24)||0.08||10 (4–24)||16 (13–26)||0.11|
|Inpatient mortality, n (%)||5 (6.8)||2 (3.2)||0.35||5 (18.5)||0 (0)||0.07|
|Falls, n (%)||6 (8.1)||4 (6.5)||0.92||6 (22.2)||2 (10.5)||0.16|
|Functional decline, n (%)†||37/66 (56.1)||28/59 (47.5)||0.34||15/24 (68.2)||9/16 (56.3)||0.45|
|Discharged back to community, n (%)‡||46/59 (78.0)||42/56 (75.0)||0.77||12/20 (60.0)||11/16 (68.8)||0.59|
|Persistent delirium at ward discharge, n (%)§||NA||NA||NA||17/24 (70.8)||6/19 (31.6)||0.02|
Compared with participants at risk, those with delirium were older (mean age 83.1 vs 80.0, P = 0.02) and more likely to be dependent in one or more ADL prior to the acute illness (67% vs 43%, P = 0.008). There was no reduction in the requirements for additional one-on-one nursing care (required in six (32%) intervention and six (22%) control participants; median duration 5 days for intervention vs 4 days for control). Intervention participants were more likely to receive psychogeriatric consultation (32% vs 11%, P = 0.04) and there was a trend to less use of neuroleptic drugs (26% vs 41%, P = 0.46; median 1.5 days for intervention vs 3 days for control). No participant in either group was managed using physical restraints. Early allied health review is documented in Figure 2b. Less occupational therapy review in the intervention group reflects cognitive testing by project and intervention medical staff, a task often delegated to occupational therapists in our teams. There was a non-significant increase in early review by physiotherapy, nutrition therapy and speech pathology.
The median duration of delirium episodes during the acute stay was 7 days, but this underestimates true duration because 23 participants (50%) were discharged from the study wards with evidence of persisting delirium (five to high-level residential aged care, three to hostel care, eight to other wards and seven home to community). Participants were significantly less likely to be discharged from the acute ward with persisting delirium in the intervention group (32% vs 71%, P = 0.016). In the control group, 16/20 (80%) delirious participants who had been admitted from the community and 5/7 (71%) from residential care had persisting delirium; in the intervention, only 4/15 (27%) from the community had persistent delirium at discharge (with none discharged home delirious), while three of four (75%) from residential care had persisting delirium. As a consequence, length of acute stay was longer in the delirious subgroup in the intervention unit (median 16 days vs 8 days, P = 0.01). However, there was a trend to reduced mortality (0% vs 18.5%, P = 0.07) and falls (11% vs 22%, P = 0.11) in the intervention group (Table 3).
This quality improvement intervention implemented many of the recommendations of the Guidelines on a general medical ward, including screening, education, interdisciplinary clinical practice protocols and environmental changes within the ward and four-bed delirium bay. The implementation had the intensive support of a project team and additional nursing assistant and volunteer time to support practical interventions. However, there were only small improvements in targeted processes of care, such as reduced ward moves, a trend to increased early review by some allied health disciplines and a trend to reduced neuroleptic use. One third of the intervention cohort still had three or more ward transfers, demonstrating the difficulty of implementing this strategy effectively. Multiple ward moves may have delayed the review by allied health team members.
Delirium was common in this sample of older medical patients, affecting 22% of the screened participants at admission, similar to previous intervention studies.[2, 3, 5, 6, 12] Exclusion of patients with terminal illness and those unable to consent means this may underestimate the true prevalence. No new case of delirium was identified during in-hospital follow up. Recent Australian and New Zealand cohort studies found a delirium prevalence rate of 18–23% and incidence rate of 2–6%,[2, 3] while previous intervention studies report incidence of new cases of 0–3%.[5, 6] In contrast, some delirium prevention trials have reported the incidence of new delirium during hospitalisation of 10–18%.[11, 13, 14] There are several potential reasons for our observed low incidence of new delirium. Project staff were employed only 5 days per week, leading to potential misclassification of early incident cases as prevalent. However, 81% were assessed within 48 h of admission, and there was no significant difference in prevalence between groups (Table 3), so it is unlikely that the intervention had an unrecognised effect on early incident cases. Rescreening was only feasible twice weekly, so brief episodes may have been missed, although project staff referred to nursing and medical notes to seek evidence of fluctuating consciousness or behaviour. The practical and ethical constraints on study inclusion may have excluded a subgroup of patients at increased risk for incident delirium. Our service had already implemented interdisciplinary team care with a focus on strategies, such as early mobilisation, adequate nutrition and hydration and recognition of delirium risk factors, which may reduce delirium incidence. This is supported by the non-use of restraints and relatively high rates of timely allied health assessments in both groups.
Delirium duration could not be adequately assessed because of the number of participants discharged with persisting delirium. The intervention significantly reduced persistent delirium, reflecting better recognition by the care team. A recent review confirmed the high prevalence of persistent delirium, with estimates of 45% at discharge and 21% at 6 months. Persistent delirium is associated with poorer outcomes, but it is unclear whether continuing care in an acute care environment until delirium resolves mitigates this risk. One study reported that home care reduced incident delirium compared with hospital care in older rehabilitation patients, although there is no evidence that early discharge results in earlier resolution of established delirium. The safety, healthcare costs and carer stress of caring for patients with persistent delirium after hospital discharge have not been explored.
We demonstrated encouraging trends in reduction in mortality and falls. Previous studies of delirious medical patients report in-hospital mortality rates of 7–36%,[2, 4, 5, 9, 14] similar to our control rate, while models using multi-component interventions and a delirium bay have been reported as 1–8%.[17-19] The control group rate of patients who fell was similar to the 15–27% reported previously in delirious medical cohorts,[2, 9, 19] and only half this rate fell in the intervention group. However, this trend to better outcomes and lower rates of delirium at medical ward discharge was achieved at the cost of a longer acute stay in delirious participants in the intervention cohort. This provides a significant barrier to replicating and sustaining this programme. Only one previous intervention study has shown reduced length of stay in delirious patients, and it did not explicitly report persistent delirium.
Our intervention included a delirium bay for patients with or at risk of delirium. No previous controlled studies have reported use of a delirium bay, although observational studies report similar outcomes to our intervention group.[17-19] Suggested benefits of such an area include ability to provide a more patient-friendly environment, greater efficiency of visual observations and targeted nursing assistant time, and raising delirium awareness. We could provide limited customisation of an existing ward bay, and some physical changes (e.g. removing bedside lockers to reduce clutter, providing a table in the room) were poorly acceptable to nursing staff as they were perceived to reduce efficiency of core nursing tasks. Cohorting delirious patients was unpopular due to discrepancies in nursing workload which were not overcome by the additional nursing assistant, who was only available for limited shifts. Some ‘at risk’ participants were also managed in the delirium bay, leading to tension regarding the optimal role as prevention or management. Use of the bay may have increased bed moves in the intervention group, offsetting the benefits of direct ward admission. In the context of a multi-component intervention, it is not possible to determine whether the delirium bay contributed to the outcomes. However, our experience suggests that this strategy requires strong nursing leadership and advocacy, and it has not been sustained in our ward.
We recognise that this study is limited by its small size, increasing the risk of type 2 error, and short-term outcomes collected by project staff who were not blinded to the study purpose. It is from a single institution, but adds to existing literature because it was implemented in a general medical ward rather than a specialised physical or clinical environment, albeit one with a well-established interdisciplinary care model. It is impractical to undertake individual patient randomisation for complex wardbased and/or team-based interventions due to logistic considerations and the risk of contamination. However, our intervention and control groups were well matched, with similar models of care and staff resourcing. It is possible that strategies to increase delirium awareness among staff may have produced some contamination in the other medical units, although cross-staffing of the intervention and control wards is very uncommon. Limitations with patient exclusions and consent requirements have been discussed.
It has been difficult to sustain many of the intervention processes in clinical practice. We hoped to demonstrate less incident delirium and therefore a reduced length of stay, which would then provide argument for additional resources, but our unexpected finding of a longer stay and an underpowered evaluation to unequivocally demonstrate improved outcomes has made advocacy difficult. The medical and nursing champions of this project are no longer working on the implementation ward, leading to loss of ownership of the initiatives and clinician disengagement as other clinical priorities arise. High hospital occupancy has precluded effective strategies to prioritise high-risk patients for early ward transfers, and cohorting high-risk/delirious patients is not acceptable for nursing staff without dedicated additional staffing. We have continued a programme of interdisciplinary education across all four medical wards, which includes other acute aged care topics in addition to delirium, and have introduced routine risk assessment as part of an integrated risk assessment and care planning tool. However, without a project officer or nursing assistant, it has not been possible to coordinate ongoing volunteer input or reliably provide cognitive stimulation activities or patient/carer leaflets. We continue to engage with the hospital quality and safety unit to explore novel ways to embed these processes.
By implementing clinical practice guidelines for delirium on a single medical ward, we found a marked reduction in discharge of persistently delirious patients in the intervention group, but this resulted in a significantlylonger hospital stay, and there was no reduction seen in one-on-one nursing use, so the intervention was costly. The trends to improved in-hospital mortality and falls were encouraging but must be interpreted with caution given the small sample size. The low incidence of new delirium, low mortality in the delirious cohort and limited evidence of process improvements may reflect the effectiveness of our existing interdisciplinary model of care rather than ineffective Guideline recommendations or poor implementation.
We are grateful for the support of staff and patients in the Department of Internal Medicine and Aged Care, RBWH. The study was presented in preliminary form at the World Congress of Internal Medicine, Melbourne, Australia on March 2010. The study was registered ACTRN 12608000204347.