Nursing perspectives on the confusion assessment method: a qualitative focus group study

Abstract

Background

the Confusion Assessment Method (CAM) is commonly used to detect delirium. Although accurate when administered by trained researchers, its sensitivity is low when performed by nurses in clinical practice. We aimed to understand the perspectives of nurses who used the CAM on orthopaedic wards.

Design

qualitative focus group study.

Setting

two academic hospitals in Hamilton, Ontario, Canada.

Participants

forty-three nurses who worked on orthopaedic inpatient units and used the CAM daily participated in one of eight focus group sessions.

Measurements

structured focus groups explored nurses’ perception of delirium and the use of the CAM. Each transcript was coded and sampling continued until theme saturation.

Results

the participants (84% female, mean age 40 years, mean years in practice 12.8) had mixed feelings about the CAM. Some nurses praised its simplicity, while others preferred a narrative description of the delirium episode. Only 35% recalled receiving training to administer the CAM. Across the groups, disorientation was inappropriately used to evaluate level of consciousness and inattention. Objective testing was reportedly rarely used for assessing inattention. Most nurses retrospectively completed the CAM at the end of their shift by extrapolating from earlier observations rather than formally administering the tool. Reported challenges included differentiating delirium from dementia, assessing non-verbal patients and those with language barriers, time constraints, discrepancy with physicians’ assessments and pressure to diagnose delirium.

Conclusion

despite its widespread use, the CAM was poorly understood by orthopaedic nurses at two academic institutions. The CAM may be difficult to implement in practice.

Introduction

Delirium is an acute decline in cognitive function featuring inattention, fluctuating course, disorganised thinking and motor disturbance [1]. It is common in postoperative hip fracture patients with a prevalence of 30–50% [2–4]. Delirium is associated with poor outcomes, such as increased length of stay, accelerated cognitive decline, decreased ambulatory function and increased need for long-term care [1].

The Confusion Assessment Method (CAM) is a validated tool to detect delirium based on four cardinal features: (i) acute onset and fluctuating course, (ii) inattention, (iii) disorganised thinking and (iv) altered level of consciousness [1]. When conducted by trained researchers, the sensitivity and specificity of the CAM are 94% and 89%, respectively [5]. It requires both subjective and objective testing for proper assessment [6]. Subjective testing is based on clinical observation and collateral history from family and others. For inattention, the CAM asks the subjective question: ‘Is the patient easily distracted or inattentive?’ The CAM also requires objective testing of inattention (e.g. digit span).

Studies have shown that nurses often under recognise delirium when compared to trained researchers using the CAM [7–11]. Qualitative research points to the following reasons for missed diagnosis: misconceptions of acute versus chronic confusion, assumption of age-related cognitive decline and misattribution of confusion as a normal response to disease or treatment [12]. Studies in palliative care have shown similar findings [13, 14]. The vast majority of studies focused on the conceptual misunderstanding of delirium as the root cause for its under recognition. As a result, proposed interventions have largely focused on educational programs as potential solutions [9]. However, CAM operationalization (i.e. how the assessment is carried out) has not been formally explored as a potential contributing factor. For example, delirium may go unrecognised if nurses do not perform objective mental status testing in a consistent or reliable manner.

Related to operationalization is the overall nursing experience with the CAM. There has been little published in this area [15–17]. Understanding the nursing experience with the CAM, including attitudes, ease of use and barriers, may help guide quality assurance initiatives to improve delirium detection.

At two academic hospitals in Hamilton, Canada, orthopaedic nurses are required to administer the CAM for each patient with hip fracture at least once during a 12-h shift. Nurses enter responses to the individual CAM components into an electronic documentation system or on a paper flowsheet, which is available to physicians and allied health staff. This study aimed to: (1) characterise the current use of the CAM tool by orthopaedic nurses at two academic tertiary care hospitals, (2) identify both facilitators and barriers to correct CAM administration and delirium assessment and (3) explore the current CAM training process and identify any potential areas of improvement.

Methods

We invited nurses on the orthopaedic floors of both hospitals to participate in a focus group session. Each group was scheduled at a pre-arranged time at the end of a day or night shift and lasted approximately 60 min. Registered nurses (RNs) and registered practical nurses (RPNs) were included. In Ontario, Canada, RNs undertake longer university-based training, while RPNs complete shorter training through a community college, developing focused skills for an area of patient care (e.g. orthopaedic ward). In the focus groups, nurse managers, educators and clinical nurse specialists were excluded. Nurses were enrolled into focus groups, which continued until saturation of themes was reached. At each focus group, after written consent, participants completed a short demographic survey and a meal was provided prior to the session. Four to eight nurses participated in each focus group, which was hosted by a geriatrician moderator (C.P.) and assistant moderators (E.W., J.L.). All sessions were audio recorded and later transcribed. Participants were assigned a numeric code and any identifying information was removed. The protocol was approved by the Hamilton Integrated Research Ethics Board (15–060). The nurse managers approved this study at both hospitals and encouraged their staff to participate in the focus groups.

Interview guide

The moderator followed an interview guide for all sessions. The guide outlined the roles of the investigators and participants. It directed the moderator to introduce the study and focus group sessions in a structured manner. The moderator facilitated the discussion with the following questions:

• Opening question: How do you go about detecting delirium in your patients?

• What is your experience with the CAM?

  • Is it helpful and easy to use? Why or why not?

  • How do you assess the individual components of the tool?

  • What are the barriers to its use in your practice?

• What training did you receive?

  • By whom?

  • What feedback did you receive?

• How is delirium interpreted in patients with dementia?

  • How do you complete the CAM in the context of dementia?

Analysis

Data on the demographic survey were analyzed using descriptive statistics. A transcription of each session was produced from the audio recording. The transcript was checked for accuracy by the principal investigator. Themes were organised using the qualitative research software NVivo 11 (QSR International Pty Ltd, Australia). The first transcript was coded independently by two investigators, and consensus was reached on thematic coding. The remaining transcripts were coded by the principal investigator. A table of emerging themes was created and revised by two investigators. Analysis was iterative with transcripts being reviewed again as new themes became apparent. A qualitative research expert guided the analysis (K.N.). As a quality improvement initiative, the results were anonymously disclosed to the participants, nurse managers, nurse educators, relevant physicians and hospital administrators in a timely manner.

Results

Forty-three nurses were enrolled into eight focus groups (Table 1). The mean age of participants was 40.7 years, with a mean of 12.8 years of clinical experience. The majority (84%) were female. RNs and RPNs accounted for 58% and 42% of the participants, respectively. At each site, three groups were held after day shift and one group after night shift. Only 35% of participants recalled having had any formal CAM training, and only 9% recalled any previous education on hypoactive delirium. The participants estimated the delirium prevalence on their ward to be 33%. Both sites were similar in baseline characteristics except for (i) younger (mean age 35.3 vs. 46.6 years) and less experienced (9.3 vs. 16.8 mean years in practice) nurses at site 2, and (ii) more participants recalling CAM training (50% vs. 22%) at site 1.

Major themes were grouped into the following six categories (Table 2): views on delirium, attitudes toward the CAM, operationalization of the CAM components, barriers to use, ways to improve delirium care and unexpected findings. Consistent ideas across themes emerged, including (i) difficulty distinguishing delirium and dementia, (ii) need for improved communication about delirium, (iii) shortfall in understanding the CAM and (iv) perceived issues with the CAM. This data is presented below.

Difficulty distinguishing delirium from dementia

The participants openly discussed their experience caring for patients in delirium. Numerous examples of delirium were described, in particular the change in behaviour, agitation and fluctuations between day and night. Difficulty in distinguishing delirium from dementia was a common theme among all groups. Determining the cognitive baseline for patients was particularly challenging, especially for patients without a visiting caregiver. Without an accurate baseline, the nurses found Criterion 1 (acute change and fluctuating course) to be the most difficult part of the CAM to evaluate. Even if a diagnosis of dementia was documented, some participants found it difficult to distinguish sundowning with delirium fluctuations. Some participants noted that severe dementia was more difficult to separate from delirium, especially when the patient was aphasic. There was a high degree of agreement between participants on this issue. For example, the following description illustrates the challenge in separating delirium from dementia:

‘It appears like nonsense to us but to them it’s not. There was a lady yesterday and she kept saying, ‘I want to go home to Howard,’ but her husband’s name actually was not Howard. I asked who Howard was, she said ‘that’s my husband!’ Well, he’s not. Someone else told me your husband has a different name. And actually, Howard turned out to be her older brother. And she just lives 5 houses down from him. It wasn’t so much that she was confused, but he lived in the house that she was born in, so for her, that’s where she wanted to be home. For me, that wasn’t so much confusion. Just that’s the place she’s familiar with.’ [Session 3, Participant 6, Q2]

Need for improved communication about delirium

Although the majority of groups expressed a positive attitude toward the CAM tool, there were some negative thoughts as well. The CAM was praised for its simplicity and convenience of documentation. However, some participants at both sites wanted space for a narrative description. They found the yes/no responses to be limiting, preferring to write observations so that the nurse on the next shift could reference them. There appeared to be a difference of opinion between younger and older nurses, with younger nurses preferring narrative documentation and older nurses preferring simple checklists. At both sites, a description of mental status was not routinely transferred at shift change, which made baseline cognitive status difficult to determine, particularly for night shift nurses.

As a method of communicating the presence of delirium, nurses found that physicians generally responded to calls about a positive CAM. Physicians typically complete a delirium order set or consult geriatric medicine for help. However, some participants disagreed, citing situations when physicians questioned the nurses’ assessment. Some participants were frustrated that physicians would make a determination of normal mental status when delirium appeared to have cleared by morning. Across the groups, there were contrasting accounts of physicians being resistant to completing a delirium order set. In one group, participants thought that the CAM was never reviewed by physicians:

Participant 6: Sometimes I just feel like the CAM is done and nobody even looks at it. You do it… yeah, there’s fluctuation, they’re delirious, but does geriatrics look at that and say, ‘hey, maybe they’re like this because we cut back all [the opioid] medications?’ I don’t think the tool… we do fill it out if it’s ordered, but I don’t think any physician goes back to look at it.

Moderator: So even if you do complete the [CAM] form, you don’t think people take any notice of it?

Participant 6: I don’t… I really don’t.

Participant 5: It sits in the chart and that’s it… [Session 6, Q2/a]

Resident physicians were sometimes unfamiliar with the CAM, which led to a breakdown of communication in the care team. At Site 2, several groups brought up issues with carrying out medical orders when a hospitalist (internist or family physician), orthopaedic surgeon, geriatrician and psychiatrist had competing opinions about management. This multidisciplinary team approach is unique to Site 2. Participants generally recommended better communication between physicians and nurses on the CAM and delirium management.

Shortfalls in understanding the CAM

Despite better recall of ever having CAM training at Site 1, participants at both sites had limited understanding of the tool’s proper use. The first criterion of acute onset and fluctuating course was well understood but difficult to determine. The second criterion of inattention was assessed by regular conversation during routine assessment. None of the participants used objective testing to measure inattention, even though many of them were aware of available tools. Many participants inappropriately used disorientation to place and time as a measure of inattention. The third criterion of disorganized thinking appeared to be best understood and tested. Nurses readily identified examples of disorganized thinking such as disorientation, hallucinations, or inappropriate behaviour. For the fourth criterion of altered level of consciousness, participants often used disorientation to determine its presence or absence. Most participants were unaware of the motor subtypes of delirium.

In every focus group, participants were unable to recall the four CAM components even though they complete it on a daily basis. Each group asked the moderator to see a copy of the tool when asked about the components. When we further explored the reason for poor recall, we found that nurses generally complete the CAM at the end of a shift by retrospectively recalling evidence of fluctuation, inattention, disorganized thinking and altered level of consciousness. This suggests that the CAM was not being used as a prospective bedside tool, but rather a method of retrospective documentation of delirium.

Perceived issues with the CAM

Knowledge about delirium as a concept was sometimes lacking, in addition to a lack of formal CAM training. Younger nurses tended to have more awareness of the CAM, perhaps as a result of more recent training. Time constraints and language barriers were seen as issues with the CAM in multiple groups. Some participants disliked the CAM because it did not help reveal the underlying cause of delirium, which reflects both a misunderstanding of its purpose but also a focus on problem solving and management rather than diagnosis. Along the same theme, some participants thought that the CAM was redundant as it did not help them recognise delirium any better than by routine assessment.

Moderator: Do you pick up more delirium using CAM or you think you did pretty well before?

Participant: I don’t think there is any difference, to tell you the truth. Anything abnormal would stand out and you would still tell the doctor, physiotherapist, and charge nurse. I don’t know if it helps at all. [Session 4, Participant 3]

In one group, a participant was uncomfortable ‘diagnosing’ delirium, which the CAM facilitates because of its high specificity [18]. These nurses were hesitant to make diagnoses, as this was seen as a physician’s responsibility.

‘The one question that I don’t like either: ‘Do they have delirium according to CAM?’ I feel like I’m diagnosing and I don’t like to answer that question.’ [Session 4, Participant 2, Q2/c]

Across all the groups, nurses wanted more training on delirium recognition and the CAM tool. None of the nurses received feedback on the accuracy of their CAM scores. Despite this, nurses generally thought their role in delirium care was important.

Comparison between the two sites

The themes were generally consistent among participants at both sites, including attitudes and CAM knowledge. Site 1 used an electronic documentation system to track CAM scores, which participants found easy to use. Site 2 used paper charting, recording the CAM on a separate paper form. Participants at that site thought the CAM was inconvenient to complete due to the perception of extra paperwork. Nurses at Site 2 were more proactive at finding a cause for any change in mental status and often bypassed CAM documentation to focus on management. Nurses at Site 1 completed the CAM more consistently but were often dependent on physicians to assess and manage delirium.

Undertreated pain in patients with delirium

The open discussion at these focus groups revealed an expected finding. At both sites, nurses voiced concerns about inadequate treatment of postoperative pain in patients with delirium because of pressure from both staff and patients to minimise opioid use. This often led to abrupt discontinuation of opioid analgesics. Nurses at Site 1 were hesitant to reveal where this pressure to avoid opioids came from. Participants at Site 2 felt that opioid analgesia was often under- and overused in orthopaedic patients, with contrasting stories of when discontinuation led to delirium either resolving or worsening. Across groups, participants used a verbal numeric pain scale, while acknowledging its inaccuracy in patients with delirium.

Discussion

Our focus groups used facilitated discussion to investigate the current state of delirium detection on our orthopaedic wards. By listening to nurses’ experiences and perspectives, we learned how delirium is detected (or goes undetected) and some of the challenges faced in delirium care. We heard about the very human experiences that nurses encounter on a daily basis when caring for delirious patients.

The CAM tool was often improperly used, perhaps as a reflection that only 35% of our participants recalled any formal training. This is the first study to show that the CAM is not used as a bedside instrument, but rather as a way of charting delirium episodes retrospectively. The components were scored by recalling how the patient behaved during the shift, instead of being used for prospective assessment. This deviates from the purpose for which CAM was intended [6].

In each focus group, there was a consistent shortfall in knowledge about the individual components of the CAM, especially inattention and level of consciousness. Disorientation is frequently used to assess multiple CAM components, without objective testing. Studies show that subtle degrees of inattention can be easily missed without objective testing [19]. The danger of inaccurate assessment of consciousness level and inattention is that hypoactive delirium, which accounts for approximately 1/3 of delirium cases [3], is easily missed [20]. Hypoactive delirium is associated with worse clinical outcomes [21], and its under recognition may prevent patients from receiving timely treatment. Nurses are generally proficient at identifying hyperactive delirium [7], so perhaps a properly used CAM would be most helpful for recognising the hypoactive subtype.

This study identifies some of the issues with the CAM and offers justification for operationalized tools like the 3D-CAM [22] or 4AT [23], where assessment of individual criteria are more explicit and objective. The Delirium Observation Screening (DOS) Scale and Nursing Delirium Screening Scale (Nu-DESC) are delirium screening tools designed and validated specifically for use by nurses [24–26]. Individual items are easier to interpret, but training is still required for accurate use. These tools may be more appropriate for training novice learners because they help the user focus on symptoms that reflect underlying delirium. In our focus groups, nurses mostly used the CAM tool to chart delirium and not necessarily to help improve its recognition. The perception of added workload without obvious benefit signifies an important barrier to use.

Delirium features a fluctuating course. Although the CAM is documented once every 12-h shift at our centres, it is meant to capture any delirium that happened during the shift. If the CAM was performed prospectively once during each shift, it may not detect a fluctuating delirium. Some participants preferred a narrative description of delirium rather than the CAM, as a description would presumably better capture any incident delirium during the shift. A delirium screening tool should expose delirium which is not detected by routine clinical observation. The best way to improve delirium detection is through a knowledgeable and attentive observer. If the tool does not enhance understanding of the disease, clinical staff may not invest the time to learn it.

Communication is often a barrier to delirium care. Nurse-to-nurse communication is important in relaying information about baseline cognitive status during previous shifts. The transfer of accountability at shift change should include a description of mental status to establish a baseline for comparison. The CAM did not appear to improve transfer of accountability on the two orthopaedic units. Equally important is communication between nurses and physicians. All medical staff working on orthopaedic wards should be aware of the CAM. Although we did not investigate physician knowledge of delirium in our study, a prior study showed that physicians are often unfamiliar with the CAM [27]. Optimal delirium care requires a collaborative effort among physicians, nurses, allied health staff and family members.

Finally, an unexpected finding emerged from our focus groups. There was a concern about inadequately treating postoperative pain in hip fracture patients who are delirious. Both pain and opioid medications are associated with increased delirium risk [1]. The nurses interviewed sometimes felt pressure to discontinue opioid medications when a postoperative patient became delirious. Sometimes the change led to resolution of delirium, but it often caused a perception of inadequate pain control. Most studies suggest that adequately treated pain does not increase delirium risk [28, 29]. Education about pain management in delirium is needed.

The strengths of this study include conducting groups with nurses at two academic hospitals with differences in baseline characteristics. We reached saturation of themes and included RNs and RPNs from both day and night shifts. This study is limited in generalisability as it involved only orthopaedic nurses within two academic centres in Hamilton, Ontario, Canada. Night shift nurses were less well represented because of difficulty in recruitment. We interviewed about 50% of the entire orthopaedic nursing complement at each site.

Conclusion

This study reinforces known shortfalls in delirium detection and provides direction to improving the quality of delirium care in our orthopaedic patients. Operational issues with the CAM were explored, and specific areas identified to improve training. Administrators should not rely on a single tool for delirium detection. Astute observation and good clinical judgement are key. A culture of ongoing education, openness to feedback and collaborative interdisciplinary communication is critical to enhancing the current state of delirium recognition. While this study is limited to two local academic hospitals, other centres should consider conducting internal audits to ensure timely detection and care of delirious patients.

Acknowledgements

We thank all of the orthopaedic nurses who took time after busy shifts to share their experiences with us. We also thank the nurse managers, educators and clinical nurse specialists at both hospitals for their support. Finally, we thank Xena Li and Natalie Munn for their help with the transcription of the focus group sessions, and Dr Sharon Straus for reviewing the manuscript.

Funding

Department of Medicine, Division of Geriatric Medicine, McMaster University and the St. Peter’s/McMaster Chair of Aging.

Presented at the 36th Annual Meeting of the Canadian Geriatrics Society on 15 April 2016.

Conflict of interest

None.

References