All 120 employed nurses in three post-acute care facilities were invited to participate in an assessment which involved completing(1) a demographic instrument reporting years of experience as a nurse, age, ethnicity, and recent formal delirium education (2) a ten-question multiple-choice delirium knowledge test (3) the scoring of two videos of delirious patients using a validated delirium assessment tool, The Confusion Assessment Method (CAM) short form,¹⁸ and (4) a five-point Likert scale questionnaire indicating confidence in screening for delirium and notifying a provider of a positive delirium result (see supplemental materials). This assessment was designed by study team members, reviewed by two delirium content experts and pilot-tested with several nurses to improve content validity. The involvement of experts and practising nurses with advanced knowledge in delirium helped identify essential topics to include in assessing knowledge and confidence surrounding delirium care.

A retrospective review of paper and electronic health records of 99 randomly chosen discharged patients was conducted in three post-acute care environments (n = 33 records per facility). This number of records was based on power calculations (effect size of 0.34 using 2 degrees of freedom Chi-Square test with a significance level of 0.025) for a future study to detect a change in nurses’ delirium detection rates from baseline following an educational intervention.

Eligibility criteria included records of any discharged patients who spent at least 72 hours on the participating post-acute unit within the past six months. The records were picked randomly by a medical records specialist at each facility. A group of research interns conducted the initial record review for each of the 99 charts. This included the collection of basic demographics, admission diagnosis, discharge destination, history of cognitive impairment (dementia or delirium) and complications (falls, skin breakdown, aspiration pneumonia, emergency department visit or hospital re-admission) occurring during the post-acute stay.

The investigators next determined which patients in the sample experienced delirium during their post-acute stay using a validated chart-based delirium identification instrument known as CHART-DEL.¹⁹ This tool uses Confusion Assessment Method criteria and has 79% sensitivity, 83% specificity and a likelihood ratio of 4.4 for identifying delirium. Any record found to contain trigger words or phrases that raise the suspicion of delirium, such as mental status change, disorientation, hallucinations, agitation, confusion, lethargy was referred to two delirium experts for further review using the CHART-DEL Instrument. In cases of disagreement between the delirium experts regarding the presence or absence of evidence supporting a delirium diagnosis, the adjudication process detailed in the CHART-DEL manual was followed to establish consensus.

Any record identified as positive for delirium was further reviewed to collect all terminology documented by interprofessional team members to describe the patient’s mental state or behaviour and determine if the clinical staff correctly identified delirium. Correct identification of delirium by clinical staff consisted of using the term delirium or other terminology to describe and recognise an acute change in the patient’s baseline cognitive function. An example of a correct identification is a patient with baseline dementia whose admission documentation states, disoriented to situation and day 3 documentation states Patient is more confused than usual. An incorrect or missed identification of delirium in a baseline cognitively intact patient may be documentation stating Patient is confused without recognising this as a change from baseline. The discipline to first mention the qualifying terminology for delirium in their note was labelled the discipline responsible for correct identification.

Actions taken by nurses in response to a recognised delirium were categorised as appropriate vs inappropriate. Appropriate actions were consistent with published best practices to address the aetiology of delirium or mitigate the adverse outcomes associated with delirium. Inappropriate actions included a failure or a delay to act consistent with published delirium care guidelines.²⁰

Twenty-two (22%) of the 99 records reviewed via CHART-DEL methods met criteria for delirium. The delirium prevalence for the three post-acute facilities ranged from 12-39%. Eleven (50%) of the 22 records contained documentation supporting a delirium diagnosis within 24 hours of admission to the post-acute unit. Six (54%) of those eleven patients were noted on the hospital discharge summary to have experienced delirium during the hospitalisation. Of the remaining eleven patients who were determined to have experienced incident delirium later than 24 hours after admission to the post-acute facility, eight developed it within the first week and three during weeks 2-4.

In this study, the age range of delirious patients was 65-98 with the average age of 85.4 years. Nineteen of the 22 (86%) records identified as delirium-positive represented Caucasians and 12 of the 22 (55%) were women. Nine of the 22 (41%) had a diagnosis of dementia, seven of the 22 (31%) had a history of delirium, and four of the 22 (18%) had both dementia and a history of delirium.

Figure 1 compares pre-existing cognitive conditions (prior delirium and dementia) and new-onset negative outcomes (falls, skin breakdown, re-hospitalisation and death) between delirious and delirium-free patients. Dementia was significantly more common (p<.001) in patients who experienced delirium during the post-acute stay compared to those who remained delirium-free. Delirious patients were re-hospitalised more often (p<.001) than those without delirium. Although not reaching statistical significance, trends showed a prior episode of delirium to be associated with a post-acute delirium occurrence. Falls, skin breakdown and death were more prevalent among patients experiencing delirium during their post-acute stay than among those patients who remained delirium-free.

114 of the 120 invited nurses from the three post-acute units participated in the delirium knowledge survey. The nurses were primarily Caucasian women with reported ages spanning five decades [ages 21-30 (19%); ages 31-40 (20%); ages 41-50: (25%); ages 51-60 (25%); ages 61+ (11%)]. A slight majority (53%) graduated from an associate degree program, while the others held bachelor’s (35%) or graduate degrees (12%) in nursing. Few (5%) reported receiving any formal education on delirium outside the work setting within the past three months, such as attending a conference or reading a publication. The average score on the knowledge test was 75% correct. Errors were primarily related to the definition and features of delirium, and the identification of delirium superimposed upon dementia. Video #1 (hypoactive delirium with visual hallucinations) was scored correctly as CAM-positive by 89% of the nurses. Video #2: (hypoactive delirium superimposed upon mild cognitive impairment) was scored correctly as CAM positive by 49% of the nurses, with 38% incorrectly scoring the CAM as negative and 13% scoring the CAM as unable to assess. (supplemental materials)

Most (85%) of the nurse participants indicated a total or substantial lack of confidence in screening for delirium. They reported the least confidence in assessing for the acute onset/change from baseline and inattention features of delirium and 56% indicated little or no confidence in reporting abnormal findings to the provider. (supplemental materials)

The most common term found in the nursing documentation associated with delirium was confusion, followed by restless and forgetful (Table 1). The term delirium was found in only two records, documented by a geriatrician and a psychiatrist but never by the nurse. In six of the 15 (40%) delirium-positive records, nurses cited families as the first party to raise concerns about the patient’s confusion, forgetfulness, memory issues and changing or worsening mental status. Families used descriptors such as not himself or different from normal to describe their delirious loved ones.

Fewer than half (45%) of the 22 patients whose CHART-DEL review suggested behaviours consistent with delirium were identified correctly as delirious by any clinical staff. Table 1 further describes correct and missed delirium identification by discipline. Nursing documentation was the source of delirium determination by experts using CHART-DEL in 15 of the 22 cases of positive delirium. Nurses correctly identified 6/15 (40%) but missed 9/15 (60%) of these delirious cases. Providers, which included physicians and advanced practice providers, documented terminology consistent with delirium in three records, but only identified two of the three patients (67%) as experiencing delirium. Similarly, the rehabilitation professionals, physical and occupational therapists, provided the initial documentation of delirium in three records, but only identified two of the three patients (67%) as experiencing delirium. A social worker provided the documentation in one case but did not recognise it as delirium.

Of the six cases in which nurses identified a new delirium, some action was taken by the nurse in five of the six instances (83%) (Table 2). As a result, one patient was sent to the hospital (who did die) and another had a potentially deliriogenic medication discontinued. In four instances, protocols for urinary assessment and management were implemented. In the one case involving an inappropriate response by the nurse, more than 24 hours passed before the provider was notified. This delayed the ordering of diagnostics to determine the aetiology of delirium. In addition, a potentially inappropriate medication (lorazepam) was administered to the patient.

Conceptualisation and methodology CW, JC, TN, SG

Formal Analysis JK, JC

Interpretation JK,JC,TN,SG,CW

Writing- original draft preparation JK, CW

Revising and editing JK,JC,TN,SG,CW

This study protocol was reviewed and approved by the Hartford Healthcare IRB.

None of the authors declare any conflicts of interest