Personalized Sedation Goal for Agitated Delirium in Patients with Cancer: Balancing Comfort and Communication

Abstract

Background:

Agitated delirium is common and highly distressing. Medications are often needed to reduce agitation but it is unclear what is the desired level of sedation. We assessed personalized sedation goal (PSG) and its predictors for patients on a delirium clinical trial and in clinical vignettes.

Methods:

This is a pre-planned secondary analysis of a double-blind randomized clinical trial examining the sedative effect of chlorpromazine and/or haloperidol in patients with agitated delirium. At baseline, caregivers and nurses were independently asked to select the personalized sedation goal for the trial patient based on 5 choices corresponding to Richmond Agitation Sedation Scale scores of 0 (no sedation), −1 to −2, −3, −4 and −5 (deep sedation). Respondents also selected PSG for 6 vignettes that differed by the level of agitation, ability to communicate and survival.

Results:

42 caregivers and 40 nurses answered questions regarding PSG. For the study patient, caregivers preferred RASS −1 to −2 most often (36%) while nurses preferred RASS −3 most often (50%). Caregivers were significantly more likely than nurses to choose lighter sedation (OR=4.8, P=0.01) despite reporting greater delirium-related distress (P=0.0006). Patients were under-sedated 33–53% of the time and over-sedated 0–15% of the time according to PSG response criterion. In the case vignettes, deeper sedation was preferred by nurses (P<0.0001) and in patients who were unable to communicate (OR=3.1–4.4, P<0.0001) and had a shorter life expectancy (OR=1.7, P=0.002).

Conclusions:

Caregivers often preferred lighter sedation than nurses. Many patients were under-sedated relative to caregivers’ PSG, highlighting room for improvement.

Lay summary:

In the last days of life, many patients with cancer develop delirium and become restless/agitated which can be highly distressing. Caregivers and doctors alike are often concerned about use of sedatives for agitated delirium, trying to find a balance between maximizing comfort while maintaining communication. We examined the concept of Personalized Sedation Goal to set an individualized target for level of sedation. We found caregivers often preferred lighter sedation than nurses. Many patients were under-sedated relative to caregivers stated goal, highlighting room for improvement.

Precis:

This prospective study examined personalized delirium sedation goal using 6 clinical vignettes and as a clinical response criteria in patients enrolled onto a randomized clinical trial. Personalized sedation goal varied widely and was associated with the level of agitation, distress, communication, survival and respondent type, highlighting the need to tailor sedation goals for each individual.

Introduction

In the last days of life, a vast majority of cancer patients develop delirium and a large proportion develop restlessness/agitation,^{1, 2} which can be highly distressing to patients, caregivers, and clinicians alike.^3–5 In a 2017 integrative review, Finucane et al. reported that caregivers of patients with delirium in palliative care settings experienced high levels of distress, which was related to both the irreversibility nature of delirium and the uncertainty regarding patient’ ability to communicate.⁶ Neuroleptics, benzodiazepines and other sedatives may be offered to reduce restlessness and agitation; however, clinicians are often confronted by the clinical dilemma of how much sedation is appropriate.^7–10 On the one hand, inadequate sedation may lead to more breakthrough restlessness episodes which could be disturbing.¹¹ On the other hand, too much sedation may take away the patient’s ability to communicate with their family.¹² Although patients with terminal delirium often already have impaired communication, delirium is fluctuating by nature and there may be moments of alertness and/or lucidity.¹³ For patients with days or hours of life expectancy, simple verbal responses such as sounds and words and even non-verbal communication such as eye contact and nodding may be considered meaningful.

In clinical practice, the level of sedation is often actively titrated based on family’s preference.¹⁴ Clinicians often find it challenging to determine the appropriate amount of sedatives to maximize comfort while maintaining communication for each patient. In clinical trials, the ideal level of sedation has not been established. Several delirium studies in the intensive care setting have set a target range of sedation for all patients; others have allowed the clinician to set their own target. Studies in the terminal agitation setting found that a 4-point reduction in the Richmond Agitation Sedation Scale (RASS) score was necessary for patients to be perceived to be comfortable.¹⁵ However, these targets have not taken into account family caregivers’ preferences, individual variability in survival and ability to communicate. Similar to pain and other symptoms,^16–18 it may be possible to establish a personalized response criteria by asking caregivers to set a personalized sedation goal (PSG) that they would consider to be acceptable.^{19, 20} To our knowledge, no studies have ever examined the concept of PSG. A better understanding of this concept may facilitate its use as a novel treatment response criterion. We assessed PSG and its predictors for patients on a delirium clinical trial and in 6 clinical vignettes.

Patients and Methods

Study design

This is a pre-planned secondary analysis of a single-center, double-blind, double-dummy parallel group randomized clinical trial on terminal agitated delirium (clinicaltrials.gov NCT03021486). Details of the study has been reported elsewhere.²¹ Briefly, patients admitted to the acute palliative care unit at the University of Texas MD Anderson Cancer Center with refractory agitation despite low dose haloperidol were randomly assigned in a 1:1:1 ratio to (1) haloperidol dose escalation, (2) neuroleptic rotation to chlorpromazine, or (3) combination therapy with haloperidol and chlorpromazine until death, discharge or withdrawal. Key inclusion criteria were diagnosis of advanced cancer, age 18 years or older, a diagnosis of delirium by DSM-V criteria, a history of agitation with Richmond Agitation Sedation Scale (RASS) ≥+1 over the past 24 hours despite being on scheduled haloperidol of 1–8 mg/day or receiving ≥4 mg/day of rescue haloperidol. Exclusion criteria included scheduled chlorpromazine within the past 48 hours of study enrollment and contraindications to neuroleptics (i.e. Parkinson’s disease, Alzheimer’s dementia, myasthenia gravis, acute narrow angle glaucoma, neuroleptic malignant syndrome, active seizure disorder, documented QTc prolongation, hypersensitivity). This protocol was approved by the Institutional Review Board at MD Anderson Cancer Center (protocol number 2016–0687). Patients’ medical power of attorney or legal representative provided written surrogate consent. Study enrollment occurred from 7/5/2017 to 7/1/2019.

Study Procedures

The 3 study groups were haloperidol dose escalation, chlorpromazine rotation or combination of haloperidol and chlorpromazine.²¹ The study medication regimen consisted of a scheduled dose of haloperidol 2 mg, chlorpromazine 25 mg, or haloperidol 1 mg plus chlorpromazine 12.5 mg every 4 hours intravenously. The same medication dose was available as rescue every 1 hour intravenously as needed. Doses could be titrated up according to protocol if patients remained agitated or adjusted down if patients became too sedated. Treatment continued until discharge, death or withdrawal from study. This study used a double-dummy design to ensure physicians, nurses, research staff conducting assessments, patients and caregivers were blinded to study assignment. Only the investigational pharmacists were aware of treatment assignment.

Study Outcomes

The primary outcome was the Richmond Agitation Sedation Scale (RASS) at 24 hours after initiation of blinded study medication. RASS is a validated 10-point numeric rating scale that ranges from −5 to +4 at 8 hours.^22–24 Negative numbers indicate deeper levels of sedation (−1=drowsy, −2=light sedation, −3=moderate sedation, −4=deep sedation, and −5=unarousable), 0 indicates that the patient was alert and calm, and positive numbers indicate increasing levels of agitation (+1=restless, +2=agitated, +3=very agitated and +4=combative). The bedside nurse assessed RASS at fixed times: immediately prior to study medication administration (time 0), 0.5, 1, 2 hours, then every 2 hours during the first 24 hours and then every 4 hours until death, discharge or withdrawal.

We assessed the caregivers’ perception of the patient’s (1) ability to communicate coherently, (2) ability to communicate meaningfully and (3) the frequency of restlessness/agitation over the past 2 days. Responses to these 3 questions were “All the time”, “Most of the time”, “Sometimes”, “Rarely”, and “Never”. We then asked caregivers how distressed they were from the agitation using a numeric rating scale from 0 (not distress) to 10 (worst possible). We also assessed how much time they spent with the patient in the hospital over the last 2 days (answers >90%, 70–90%, 40–60%, 10–30%, <10%).

We assessed PSG at enrollment by asking the caregivers who were all legally designated surrogate decision makers to state their preference for the study patient: “From your perspective as the patient’s caregiver, what would be your preferred level of alertness to ensure that he/she is comfortable?” The choices were descriptive in nature and designed to correspond with RASS scores. However, the RASS score numbers were intentionally omitted from the answers.

• Not give any medicines that would affect her level of alertness, even when she is restless or agitated. She may continue to be awake most of the day and can be quite restless/agitated at times. (i.e. RASS 0 or higher)

• Use medicines to help her sleep a bit more. When asleep, she may open her eyes when you call on her and has eye contact. She may not be able to communicate with you. She may still have long periods of awakening with some moments of restlessness/agitation. (i.e. RASS −1 or −2)

• Use medicines to help her sleep a majority of the time. When asleep, she may open eyes when you call on her but without eye contact and unable to communicate with you. When awake (10–20% of the time), there may be moments of restlessness/agitation. (i.e. RASS −3)

• Use medicines to help her sleep almost all the time. When asleep (90–99% of the time), she will not respond to voice but may have some movement or eye opening when you touch her. She will have almost no restlessness/agitation. (i.e. RASS −4)

• Use medicines to help her sleep all the time. When asleep (100% of the time), she will be in deep sleep and may rarely be able to or not at all be able to be awakened. (i.e. RASS −5)

To control for the variability in patient characteristics, we also provided 3 hypothetical scenarios involving patients with terminal delirium (see Supplement 1).

• The first patient had significant restlessness and agitation (awake most of the day, sometimes quite agitated, pull on urinary catheter, climb out of bed much of the night) and no meaningful communication for days.

• The second patient had restlessness and agitation at times (sleeping about half of the time, often awake at night, caregiver awakened 3 times because of her restlessness) and no meaningful communication for days.

• The third patient had restlessness and agitation at times (sleeping about half of the time, often awake at night, caregiver awakened 3 times because of her restlessness) and had a few brief but meaningful conversations.

For each case vignette, we asked the caregiver to choose the PSG using the same choices as above, assuming the patient had “days to live”. We also posed the question if the patient had “weeks to live”.

Statistical Analysis

Sample size calculation was reported previously.²¹ Briefly, we calculated that 15 patients per group would provide 90% power to detect an effect size of 0.2 for within-group change in RASS with an alpha of 2.5% assuming an intra-patient correlation of 0.7.

We used descriptive statistics including means, standard deviations, medians, interquartile ranges and 95% confidence intervals to summarize the data. We compared between caregivers and nurses using the Wilcoxon rank sum test for continuous and ordinal variables and Fisher exact test for categorical variables. Multi-level cumulative logits models were applied to examine the association between PSG for the study patient and various factors, including level of agitation, level of distress in respondent, ability to communicate, and respondent type (nurse vs. caregiver). Achievement of PSG response was defined as RASS score at each time point ±1 category of PSG for each patient. For the case vignettes, we also used multi-level cumulative logits models to assess the association between PSG and various factors, including level of agitation and ability to communicate, survival time and respondent type (nurse vs. caregiver).

The Statistical Analysis System (SAS version 9·4, SAS Institute, Cary, North Carolina) was used for statistical analysis.

Results

Characteristics of Caregivers

Among 116 fully eligible patients, 68 enrolled and 45 proceeded to receive the blinded study intervention.²¹ A total of 42 caregivers responded to PSG at baseline. The mean age was 56 (range 19–86), 31 (74%) were female, and 31 (74%) were White. 27 (64%) were spouses and 11 (26%) were children. All were designated surrogate decision makers on medical power of attorney.

Table 1 shows the patient’s level of agitation and ability to communicate from the perspective of caregivers and nurses. Caregivers were more likely than nurses to report having spent more time with the patient (P<0.0001). Although both groups observed similar frequency of restlessness/agitation, caregivers reported significantly more delirium-related distress than nurses (median 6 vs. 2, P=0.0006). Caregivers were also more likely than nurses to report that the patient was able to communicate coherently (P=0.0498) and meaningfully (P=0.02) with them.

Table 1.

Delirium experience from the perspective of caregivers and bedside nurses

Item	Response	Caregiver	Bedside Nurse	P-Value
Over the last 2 days, has the patient been able to communicate coherently with you?	Yes, all the time	1 (2)	0 (0)	0.0498
	Yes, most of the time	15 (36)	5 (13)
	Yes, sometimes	11 (26)	16 (41)
	Yes, but only rarely	9 (21)	10 (26)
	Never	6 (14)	8 (21)
Over the last 2 days, has the patient been able to communicate meaningfully with you?	Yes, all the time	2 (5)	0 (0)	0.02
	Yes, most of the time	15 (36)	5 (13)
	Yes, sometimes	8 (19)	11 (28)
	Yes, but only rarely	10 (24)	12 (31)
	Never	7 (17)	11 (28)
Over the last 2 days, have you noticed any restlessness or agitation in the patient?	All the time	0 (0)	0 (0)	0.31
	Most of the time	12 (29)	8 (21)
	Sometimes	27 (64)	26 (67)
	Only rarely	2 (5)	4 (10)
	Never	1 (2)	1 (3)
If you have noticed any restless or agitation, how distressed did you feel?	N	42	38	0.0006
	Mean (SD)	5.5 (3.0)	2.8 (3.0)
	Median (IQR)	6.0 (4, 8)	2.0 (0, 4)
Over the last 2 days of this admission, how much time have you spent with the patient (i.e. in the hospital)?	>90%	23 (55)	3 (8)	<0.0001
	70–90%	11 (26)	1 (3)
	40–60%	7 (17)	12 (32)
	10–30%	1 (2)	22 (58)
	<10%	0 (0)	0 (0)

Abbreviations: IQR, interquartile range; SD, standard deviation

PSG for Study Patients

For study patients, PSG corresponding to RASS 0 or higher, −1 to −2, −3, −4 and −5 were chosen by 1 (2%), 15 (36%), 14 (33%), 10 (24%) and 2 (5%) of caregivers, and 0 (0%), 6 (15%), 20 (51%), 13 (33%) and 0 (0%) of nurses, respectively (Figure 1a–b).

Figure 1. Distribution of Personalized Symptom Sedation Goal.

(A) Caregivers’ choice for the study patient; (B) Nurses’ choice for the study patient; (C) Caregivers’ choice in case vignettes; (D) Nurses’ choice in case vignettes.

A deeper level of sedation was preferred among respondents who reported a higher level of delirium-related distress (OR=4.4 for severe distress, 95% CI 1.1–17.2; P=0.03). Nurses were also more likely than caregivers to prefer a deeper level of sedation (OR 4.8, 95% CI 1.4, 16.2; P=0.01) (Table 2).

Table 2.

Factors Associated with Personalized Sedation Goal in Study Patients^a

Model	Predictor	Odds ratio	95% CI	P-value
Delirium related distress	Delirium-related distress in respondent (Ref: mild 0–3)			0.10
	Moderate 4–6	2.21	0.54 – 9.07	0.26
	Severe 7–10	4.39	1.12 – 17.17	0.03
	Bedside Nurse (Ref: Caregiver)	4.78	1.41 – 16.21	0.01
Meaningful Communication	Able to meaningfully communicate (Ref: “All or most of the time”)			0.26
	Sometimes	2.63	0.76 – 9.15	0.12
	Rarely or Never	2.12	0.70 – 6.35	0.18
	Bedside Nurse (Ref: Caregiver)	2.09	0.85 – 5.13	0.11
Amount of time with patient	Amount of time with patient (Ref: >60%)			0.87
	40–60%	1.03	0.24 – 4.45	0.97
	30% or less	0.73	0.14 – 3.80	0.70
	Bedside Nurse (Ref: Caregiver)	3.40	0.79 – 14.58	0.10

^aThe multi-level cumulative logits model was used to examine the association between personalized sedation goal for the actual study patient (ordinal scale) and independent variables. A higher odds ratio indicates preference for deeper sedation.

Personalized Treatment Response

The proportion of patients who achieved a personalized response (i.e. ±1 point of caregiver’s PSG) is shown in Figure 2. Between hour 2 and 24, 40–61% of patients achieved RASS scores that were within, with no significant difference detected among treatment groups (P=0.45). During this timeframe, 33–53% of patients were considered to be under-sedated and 0–15% of patients were over-sedated.

Figure 2. Treatment Response by PSG.

PSG was graded over 5 categories, corresponding to RASS scores of 0 or higher, −1 to −2, −3, −4 and −5). A PSG response was defined as actual RASS score within ±1 category of the stated PSG; under-sedation was defined as actual RASS score greater than 1 category above PSG; over-sedation was defined as actual RASS score greater than 1 category below PSG. Twelve of the 42 patients who had vignette and goal data collected were omitted from the RASS analyses because they left the study prior to randomization.

PSG Distribution in Case Vignettes

The distribution of PSG for the case vignettes is shown for caregivers (Figure 1c) and nurses (Figure 1d). Overall, caregivers most preferred PSG corresponding to RASS of −1 to −2 (36%) and −3 (38%) while nurses preferred RASS of −3 (38%) and −4 (40%).

Table 3 shows how the choice of PSG varied by scenarios and respondents. Respondents were significantly more likely to prefer a deeper level of sedation if patients in the scenarios were not able to communicate (OR 3.1–4.4, P<0.001) and if the expected survival was in terms of days instead of weeks (odds ratio 1.7, 95% CI 1.2, 2.5, P=0.002). Nurses were also significantly more likely than caregivers to prefer a deeper level of sedation (odds ratio 2.5, 95% CI 1.8, 3.6; P<0.0001).

Table 3.

Factors Associated with Personalized Sedation Goal in Case Vignettes

	Odds ratioa	95% CI	P-Value
Vignette			<0.0001
1 (Ref: 3)	3.09	1.99 – 4.78	<0.0001
2 (Ref: 3)	4.39	2.82 – 6.83	<0.0001
Bedside Nurse (Ref: Caregiver)	2.50	1.75 – 3.57	<0.0001
Survival in Days (Ref: in Weeks)	1.73	1.22 – 2.45	0.002

^aThe multi-level cumulative logits model was used to examine the association between personalized sedation goal for each case vignette (ordinal scale) and independent variables. A higher odds ratio indicates preference for deeper sedation.

Discussion

By examining PSG in both study patients with agitated delirium and case vignettes, we found caregivers preferred lighter sedation than nurses. Other determinants of PSG identified in this study included delirium-related distress, patients’ ability to communicate and life expectancy. The use of PSG as a response criterion also highlighted that many patients were under-sedated. Findings from this study may have implications for both clinical care and research involving patients with agitated delirium.

To our knowledge, this is the first study to examine the concept of PSG in the research setting. In clinical practice, the principles of PSGs are often deployed by palliative care teams where clinicians often have to discuss and sometimes negotiate sedation levels with surrogate decision makers and caregivers. The PSG approach described in this study provides a useful standardized framework with which to approach this clinical practice and also to systematically identify discrepancies between clinicians’ and families’ expectations. We found that a vast majority of respondents preferred PSG corresponding to RASS scores −1 to −4 for the study patient and in the case vignettes. No sedation (RASS 0 or higher) and deep sedation (RASS −5) were rarely chosen. Even in case vignettes in which the key variables were standardized, we found that there was much variation in the preference for level of sedation, highlighting the need to establish a personalized sedation goal and to actively titrate medications.^{25, 26}

Caregivers consistently preferred less sedation than nurses in both case vignettes and study patient, despite reporting more delirium-related distress compared to nurses. Interestingly, caregivers were also more likely than nurses to perceive that the patient was able to communicate meaningfully with them. Our findings provide insights into how much caregivers value retaining consciousness and communication abilities in patients with delirium.²⁷ As patients approach the final days of life, families often desire opportunities to express their love and to say goodbye.²⁶ The ability for a patient to be able to hear, understand, and respond is of paramount importance.

Respondents who reported delirium-related distress were more likely to indicate a preference for deeper sedation. This is consistent our previous finding that higher doses of neuroleptics were associated with greater delirium-related distress among clinicians.¹¹ Indeed, witnessing terminal restlessness and agitation can be traumatizing and unbearable for the caregivers.^3–6 Patients frequently move their body, pull on tubings and/or clothing, and/or make distressing noises. These behaviors may be interpreted as manifestations of pain, emotional anguish, and/or spiritual suffering,⁶ which may contribute to caregivers being more willing to keep the patient sedated to ease the suffering. Effective interventions for terminal restlessness may alleviate delirium-related distress in caregivers.

Our study also found that deeper sedation was preferred in patients with shorter life expectancy, highlighting the importance of prognostication in the last days of life in decision making.^28–30 This choice likely reflects the delicate balance between maximizing comfort while not overly sedating if communication (or improvement) is possible in patients with longer life expectancy.

To assess treatment response in clinical trials in agitated delirium, we previously reported that the minimal clinically important difference (MCID) for RASS was −4 points.¹⁵ However, this concept has several limitations. First, it can only be applied to average group difference rather than individuals. Second, MCIDs should be specific for the patient population and time frame. For example, a MCID value for change in RASS over 8 hours may or may not be applicable as RASS over 24 hours.²⁰ The use of PSG may overcome these challenges. We found that 40–60% of patients met the PSG response. Although over-sedation is a commonly mentioned concern with sedatives and we used relatively high doses of neuroleptics in this trial,²¹ only 0–15% of study patients were over-sedated based on study criteria. In contrast, the use of PSG as a response criterion showed that 33–53% of study patients were under-sedated. This finding may explain why only 60–70% of caregivers perceived the patient to be more comfortable.²¹ Future studies should examine higher doses, more active dose titration and/or novel interventions to better achieve the desired level of sedation.

The concept of PSG has implications for both clinical care and research. Clinically, sedatives are often used in palliative care setting, although there is much debate on the choice of medications, the doses and the level of sedation.^{31, 32} We observed much variability in PSG in the case vignettes even when key variables were standardized, suggesting the need to individualize sedation goals. The PSG choices in this study may provide a framework for discussion with caregivers. Of interest, our study highlights that caregivers rarely preferred continuous deep sedation (RASS −5) even in patients with persistent agitated delirium. We consider the use of scheduled neuroleptics in this randomized trial to be a step before palliative sedation.²⁰ Further studies are needed to examine when caregivers would desire continuous deep sedation. From the research perspective, our study demonstrated the potential use of PSG to tailor treatment response assessment. This is similar to the concepts of personalized pain goal and personalized dyspnea goal to tailor treatment response.^16–18 Upon further refinement and validation, PSG may be incorporated as a key outcome in clinical trials in agitated delirium.

This study has several limitations. First, this is a single center study occurring at an acute Palliative and Supportive Care Unit at a tertiary care cancer center. The findings may not be generalizable to other settings. Second, all participants were enrolled onto a clinical trial to treat terminal agitated delirium. The preferences for PSG may differ in caregivers who declined to participate. Third, the sample size was small and may lead to false negative findings. The associations found here should be considered exploratory in nature for hypothesis-generating purposes. Fourth, as the first study to examine the concept of PSG, the vignettes and choices for PSG have not undergone full psychometric evaluation. Fifth, we only examined PSG at baseline and did not examine its stability over time. Caregivers’ preferences may shift with time and this area requires further investigation.

In summary, we examined a novel concept to define the level of sedation for patients with terminal agitated delirium, which could be used to assess treatment response. Our research highlights the PSG is a delicate balance between agitation control, communication capacity and prognosis, and may vary significantly by respondent. Further research is needed to examine the use of PSG to help personalize the management of delirium and improve goal concordant care.