A Retrospective Analysis of Primary Diagnosis,Comorbidities,Anticholinergic Load,and Other Factors on Treatment for Noisy Respiratory Secretions at the End of Life

Abstract

Background:

Noisy respiratory secretions (NRS) at the end of life have been described in 23%–92% of deaths in palliative care units. Despite limited evidence to support the efficacy of the antisecretory medications, hyoscine hydrobromide and glycopyrrolate, administration of these medications remains standard palliation. Predicting those at risk of NRS could allow early and targeted intervention.

Methods:

A retrospective analysis of 199 deaths in a palliative care unit was undertaken to identify if any patient, disease, or treatment-related factors might be predictive or protective of the symptom.

Results:

In the last 72 hours of life, 60% received antisecretory medication (mean, 2; range, 0–16 doses). Significant unadjusted associations were found between administration of antisecretory medications and survival as well as the anticholinergic drug load from other medications. Results obtained from logistic regressions revealed patients with a higher anticholinergic load from prescribed medications were more likely to require treatment for NRS (odds ratio [OR]=2.9, 95% confidence interval [CI]=1.4–5.7). There were no other factors significantly associated with developing the symptom.

Conclusion:

A high anticholinergic load from medications was not protective and instead predicted the need for treatment for NRS at the end of life.

Introduction

Observational studies report noisy respiratory secretions (NRS) at the end of life occur in 23%–92% of patients dying within palliative care units.^1,2 The rate of NRS in larger studies has been reported to be between 40%–60%.^3–5 These wide ranges reflect first, a lack of large prospective trials, and second, a difficulty in objectively measuring the symptom.

NSR is defined as noise produced by the oscillatory movements of secretions in the upper airways associated with the inspiratory and expiratory phases of respiration.⁶ This symptom usually occurs 17–57 hours prior to death⁷ and is often referred to as “death rattle.” Although the patients themselves are usually unconscious, there can be considerable distress to the family, caregivers, and health professional exposed to the respiratory noise. This can lead to marked distress at the time and in the bereavement period for families and caregivers^8,9 and to a negative impact on staff caring for this dying person.^10,11

The mechanism underlying the collection of respiratory secretions remains unclear. Bennett proposed two subgroups: type I, predominately due to salivary secretions and type II, due to bronchial secretions.³ A more recent article proposed a three-step mechanism of secretions pooling due to inability to cough or swallow, leading to a partial obstruction and resulting in further secretions being produced at the level of and in response to the obstruction.¹²

Despite good palliative care and use of antisecretory medications, the symptom is not resolved in 22%–50% of cases.⁷ Recent studies suggest that pharmacologic intervention commenced at low intensity NRS is more effective at resolving the symptom.¹³ It may follow that predicting which patients are at risk of developing NRS could allow focused observations and quicker initiation of interventions and perhaps more effective management of NRS. Previous work to identify people at risk of NRS is limited. There are reported associations between lung cancer, lung and brain metastases, pulmonary infection,^14,15 male gender and time in terminal stage.¹⁵ In a small audit, Bennett³ found no association with cardiac or respiratory comorbid diseases. To our knowledge this is the only previous trial to explore comorbid illness and NRS. There remains inclusive evidence and limited data in predicting which patients are most at risk of developing NRS.

There are numerous medications prescribed regularly in palliative care with anticholinergic effects, and the anticholinergic load of prescribed medications increases toward the end of life.¹⁶ A common adverse effect of these medications is a dry mouth, which occurs secondary to reduced saliva production.^16–18 It is this effect that underlies the prescription of anticholinergic medications such as glycopyrrolate and hyoscine hydrobromide⁷ to palliate NRS. Given that medications with similar effects are used with such frequency earlier in illness, it is possible to theorize that administration of higher anticholinergic loads prior to the terminal phase could protect people from developing NRS at the end of life.

The aim of this retrospective consecutive cohort analysis is to explore the association between different patient factors and the administration of antisecretory medications for NRS at the end of life. The particular patient factors include the anticholinergic load of prescribed medications prior to the terminal phase, the primary diagnosis, comorbid diseases, and use of intravenous fluids in the terminal phase.

Methods

Study setting

The Sacred Heart Palliative Care Unit forms part of the tertiary St. Vincent's Hospital Campus (Darlinghurst, New South Wales, Australia). This unit has approximately 750 admissions annually with approximately 400 deaths, 90% of which are caused by cancer.

Data collection

With the approval of the institutions' Human Research Office, all deaths over the period of January 1, 2009 to June 30, 2009 that occurred in the palliative care unit were reviewed. Data were extracted from patients' medical records and electronic databases by a single examiner using a pro forma designed for this study. Patient identification numbers were initially used to retrieve case files and database details and then removed so no personally identifiable data were collected.

Information was collected on patient's demographic data, dates of admission, phase changes, and dates of death using the electronic database. Survival was calculated from time of admission to hospice to death. Comorbidities and details of main diagnosis were collected using a template from the medical record. For the purposes of this audit, comorbidities were categorized into fluid overload states, respiratory diseases, pneumonia, and swallowing disorders. Administration of parenteral fluids within the last 72 hours of life was recorded.

Illness phase

Illness phase is a validated descriptive tool used to indicate how stable or otherwise a person is in their terminal illness.¹⁹ The four phases are stable, unstable, deteriorating, and terminal (Table 1). The date of phase changes to deteriorating and to terminal is recorded in the electronic data base. If this date was not recorded then, using the Palliative Care Outcomes Collaboration definitions,¹⁹ it was extrapolated by reviewing the medical notes.

Table 1.

Illness Phase and description

Phase type	Phase description
Stable	Symptoms are adequately controlled by established management. Further interventions to maintain symptom control and quality of life have been planned. The situation of the family/caregivers is relatively stable, with no new issues apparent.
Unstable	The patient experiences the development of a new unexpected problem or a rapid increase in the severity of existing problems, either of which requires an urgent change in management or emergency treatment. The family/caregivers experience a sudden change in their situation requiring urgent intervention by members of the multidisciplinary team
Deteriorating	The patient experiences a gradual worsening of existing symptoms or the development of new but expected problems. These require the application of specific plans of care and regular review but not urgent or emergency treatment. The family/caregivers experience gradually worsening distress and other difficulties, including social and practical difficulties, as a result of the illness of the person.
Terminal	Death is likely in a matter of days and no acute intervention is planned or required. Typically in this phase people are weak, essentially bed bound, drowsy for extended periods, disoriented for time and has a severely limited attention span, increasingly disinterested in food and drink, often finding it difficult to swallow medication. This requires the use of frequent, usually daily, interventions aimed at physical, emotional and spiritual issues.

Medications for NRS

Antisecretory medications data (hyoscine hydrobromide and glycopyrrolate administered in the last 72 hours of life) was obtained from an electronic prescribing database along with all medications administered on the date of each phase change. This outcome factor was then treated as a binary factor: those who did and those who did not receive antisecretory medications.

Anticholinergic load of prescribed medications

Medications were rated using the Clinician Rated Anticholinergic Scale (modified version).²⁰ Each medication is given a score of 0 (no anticholinergic effect) to 3 (marked anticholinergic effect). All medications administered on the date of phase change were assigned a score. The sum of these scores was then calculated and this numerical value was the total anticholinergic load. Each case had a load calculated for the beginning of the deteriorating phase and the beginning of the terminal phase.

Data analysis

Data were analyzed using the STATA V10.0 software (StataCorp, College Station, TX). Descriptive statistics were obtained on patients' characteristics and the use of antisecretory medication. Bivariate analyses were conducted to examine the unadjusted associations between these variables and antisecretory medications. Bivariate relationships between variables of interest and the outcome variable were tested with likelihood ratio tests. Significant variables identified in the bivariate analyses were selected to be included in further multiple logistic regression analyses. Odds ratios (OR) and their corresponding 95% confidence intervals (CI) were calculated for those significant variables retained in the final model. A significant level of 5% was used for all hypothesis testings.

Consideration was given to developing a model for multivariate analysis including all biologically plausible associations regardless of significance in bivariate model. Given the retrospective nature of this work, however, the above analysis plan was chosen.

Null hypothesis

A high anticholinergic load of medications is not protective against the need for antisecretory medications and there are no significant patient or disease related factors that predict development of NRS at the end of life.

Results

Descriptive data

There were 205 deaths over the 6-month period, with 199 complete records available for analyses. The demographic data and patients characteristics are summarized in Table 2. Malignancy was the primary diagnosis in 83% of cases and primary lung cancer for 40 (24%) cases. The illness phase change date was extrapolated in 69 (34%) cases for deteriorating phase and 101 (51%) terminal phase. The median time (days) from deteriorating phase change to death was 2 and from terminal phase to death was 1.

Table 2.

Patients' Characteristics (n=199)

Characteristics	Descriptive statistics
Age	Mean=70.5, SD=14.8, median=72.0, min.=29, max.=97
Gender
Male	106 (54%)
Female	89 (46%)
Survival time (days)	Mean=19.0, SD=33.6, median=9.0, min.=0, max.=368
Time from deteriorating phase change to death	median=2
Time from terminal phase change to death	median=1
Malignancy as primary diagnosis	166 (83%)
Primary tumor site
Lung	40 (24%)
Breast	17 (10%)
Colorectal	17 (10%)
Urologic	17 (10%)
Upper gastrointestinal	17 (10%)
Hematologic	16 (10%)
Other	41 (25%)
Lung metastases
Yes	79 (48%)
No	86 (52%)
Pleural effusion
Yes	20 (12%)
No	145 (88%)
Cardiac failure
Yes	42 (21%)
No	156 (79%)
Pneumonia
No	135 (68%)
Yes, no treatment	22 (11%)
Yes, treated oral or IV antibiotics	42 (41%)
Chronic airway disease
Yes	32 (16%)
No	166 (84%)
Parenteral Fluids
No	172 (86%)
PEG/Subcutaneous/Intravenous	27 (14%)
Anticholinergic load in deteriorating phase
<2	72 (36%)
2–3	56 (28%)
≥3	71 (36%)
Antisecretory medications
Yes	120 (60%)
No	79 (40%)
Total cases administered
Glycopyrrolate	82 (41%)
Hyoscine hydrobromide	65 (33%)
Glycopyrrolate and hyoscine hydrobromide	27 (14%)

SD, standard deviation.

One hundred twenty (60%) received an antisecretory medications in the last 72 hours of life. Six (3%) had treatment initiated in the last 96 and 120 hours of life, respectively. All had documentation of NRS in the patient's medical records. The median number of doses was 1. (Interquartile range 3 min.=0 max.=16.) Antisecretory medication was delivered via a syringe driver in only 2 (1%) cases.

Anticholinergic load is imparted from numerous medications (Table 3). At the beginning of the deteriorating phase 159 (80%) were prescribed medication with some anticholinergic effect. (Anticholinergic load ranged from 0–11, median 2, interquartile range, 3.) By the terminal phase, 174 (93%) of patients had anticholinergic load from medications (range, 0–12; median, 3; interquatile range, 3). Because of the wide ranges, the total anticholinergic load in the deteriorating phase was stratified into three categories. The number of patients in each category was 72 (36%), 56 (28%), and 71 (36%), respectively.

Table 3.

Common Medication with Anticholinergic Effects

Analgesic	Morphine, fentanyl, oxycodone
Benzodiazepines	Midazolam, lorazepam, temazepam, oxazepam
Corticosteroids	Dexamethasone, prednisone, fluticasone, hydrocortisone
Antiemetics	Cyclizine, levomepromazine, promethazine
Antisecretory	Hyoscine hydrobromide, glycopyrrolate
Adjuvant agents, cardiac medications, miscellaneous	TCA, sodium valproate, digoxin, frusemide, olanzapine, warfarin

Antisecretory medications contributed to the overall anticholinergic load at the beginning of the deteriorating phase in 20 (10%) of cases. By the beginning of terminal phase, the anticholinergic load was increased by antisecretory medications in 63 (31%).

Results of bivariate analysis are summarized in Table 4. Significant unadjusted associations were found between administration of antisecretory medications and survival, as well as anticholinergic drug load. There was no significant unadjusted association for age, gender, primary site of malignancy, site of metastases, or any of the comorbidities. Further bivariate analysis was performed after removing those 20 patients in whom antisecretory medications contributed to the anticholinergic load at the deteriorating phase, the association remained significant (p=0.002).

Table 4.

Unadjusted Associations between Patients' Characteristics and Study Outcome

Characteristics	Results
Age	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_1 = 0.09$$\end{document} , p=0.762
Gender	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_1 = 0.71$$\end{document} , p=0.401
Survival	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_1 = 3.97$$\end{document} , p=0.046
Lung metastases	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_1 = 0.02$$\end{document} , p=0.899
Brian metastases	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_2 = 0.00$$\end{document} , p=0.981
Pleural effusion	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_6 = 0.23$$\end{document} , p=0.628
Cardiac failure	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_1 = 0.07$$\end{document} , p=0.788
Pneumonia	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_2 = 3.22$$\end{document} , p=0.200
Chronic airway disease	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_2 = 0.49$$\end{document} , p=0.483
Parenteral fluids	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_1 = 1.37$$\end{document} , p=0.243
Anticholinergic load in deteriorating phase	\documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} $${\rm LR} \chi^2_2 = 9.25$$\end{document} , p=0.010

Results obtained from the multiple logistic regression indicated that of these variables, only anticholinergic load at the beginning of the deteriorating phase was significantly associated with the outcome (Table 5). The odds for needing antisecretory medications were about 3 times for those with a total anticholinergic load of greater than 3 when compared to those who did not receive any antisecretory medications (OR=2.9, 95% CI=1.4–5.7). The odds for antisecretory medications were about 2 times for patients who had an anti-cholinergic load of 2–3 compared to those who did not receive antisecretory medications (OR=1.9, 95% CI=0.9–3.8). However, this result was not statistically significant.

Table 5.

Results Obtained from the Logistic Regression of the Study Outcome on Anticholinergic Load in Deteriorating phase

Anticholinergic load at deteriorating phase	Total	Percentage	OR (95% CI)
>3	72	36%	2.9 (1.4–5.7)
2–3	56	28%	1.9 (0.9–3.8)

OR, odds ratio; CI, confidence interval.

Discussion

This retrospective study aimed to determine if the number of anticholinergic medications likely to reduce salivary flows impacted the development of NRS and whether disease- or treatment-related factors could predict who would develop NRS.

The role and effect of anticholinergic load from prescribed medications in the palliative population is an emerging research area. The association with constipation,²¹ delirium, and cognitive impairment^16,22 have been reported. It seemed reasonable to consider that if a patient was receiving medications giving them a high anticholinergic load resulting in a dry mouth, this would be protective against developing NRS. The findings of this study oppose this theory. There was a statistically significant association between a higher total calculated anticholinergic load as a patient deteriorates towards death and the use of antisecretory medications. There are a number of reasons why this might occur.

There are reports in the psychiatric literature of an anticholinergic discontinuation syndrome leading to cholinergic rebound. This phenomenon is associated with abrupt withdrawal of anticholinergic medications^23,24 and is characterized clinically by nausea, vomiting, abdominal cramping, sweating, and urinary urgency. The prevalence of cholinergic rebound after cessation of psychiatric medications remains unclear.^23,24 It is common practice to cease oral medications as people enter the terminal stages and this is likely to include medications contributing to the anticholinergic load. It is therefore possible to theorize prescribing earlier in life may contribute to the problem of NRS due to cholinergic rebound.

One could hypothesize that there may be anticholinergic tolerance. Prolonged and long-term exposure to anticholinergic medications cause an upregulation of muscarinic acetylcholine receptors, as has been demonstrated in rat models.²⁵ This upregulation may reduce anticholinergic side effects and result in loss of the “protective” effects of an anticholinergic load, perhaps resulting the need for treatment. Further to this, although not examined in this study, patients may go onto require higher doses of medications to achieve the same adverse effects of medications. As the aim of the antisecretory medications is to reduce salivary flow, patients who have received anticholinergic medications at higher doses before the terminal phase may continue to require higher doses to keep secretions minimized.

The results fail to identify any particular patient characteristics that help predict those who will develop NRS. There was no significant association between primary tumor type, metastases site or comorbid diagnoses and need for antisecretory medications. Therefore, this study did not support the findings from previous trials related to increased chance of NRS for male gender, primary lung cancer, or those patients with lung or brain metastases.^14,15 The unadjusted association with survival time seemed to support the findings of Kass et al.¹⁵ yet this was not significant following multiple logistic regression analyses. This study was of similar size but included a more diverse cohort including nonmalignant cases and all deaths including those who rapidly deteriorated. The prevalence of sites of metastatic disease was in keeping with those reported by Morita,¹⁴ yet the association with NRS was not significant in this study. Cardiorespiratory comorbidities have previously been examined and similarly no significant association was observed.¹⁸

It has previously been reported there was no significant relationship between level of hydration and respiratory tract secretions.^6,27 A prospective study examined biochemical abnormalities and patient reported symptoms of dehydration and reported no correlation with NRS.⁶ Of note, no patients in the cohort received parenteral hydration. In this study, 14% of patients received parenteral fluids (intravenous, subcutaneous or via percutaneous gastrostomy) yet there was no significant association with administration of antisecretory medications. A recent prospective observational study, limited to patients with intra-abdominal malignancies, also showed no association between artificial hydration and bronchial secretions.²⁶

Strengths of this study

To our knowledge this is the first study of anticholinergic load and the association with noisy respiratory secretions at the end of life. Given the intriguing result, this will be important in furthering our understanding of the symptom and its treatment.

This study included a higher percentage of nonmalignant cases, an emerging case load within palliative care units. The previously reported retrospective trial included only four nonmalignant cases¹⁵ and other studies have been exclusively in cancer patients.

Limitations of this study

This is a retrospective review and so relies on documentation in the medical admission and notes for comorbid diagnosis and extent of metastatic disease.

The incomplete recording of the phase changes meant that dates had to be extrapolated using the Palliative Care Outcomes collaboration definition. This relied on documentation by staff on consciousness level, time spent in bed and oral intake. Although this was easily obtained there may be some error.

Anticholinergic load was calculated using medications administered for one day only at two separate time points. This therefore is only a snap shot in time and may not reflect variations in anticholinergic load.

Using the administration of antisecretory medications as a surrogate marker for NRS will not be as representative as direct recording of the symptom, its severity and response to treatment. NRS is the only indication for these antisecretory medications at the unit. The number of doses given should reflect persistent NRS yet if the initial dose in ineffective, repeat doses may be abandoned and it is not possible to clarify this easily in a retrospective study. Patients with anxious, concerned care givers may have received more doses than those without family present. This is unable to be accurately accounted for. Therefore, for the purposes of evaluation, the use of antisecretory medication was treated as a binary outcome. This does not account for patients with tenacious secretions in whom antisecretory medications may dry secretions further and worsen the symptom. Antisecretory medications may be actively avoided in these cases. There was one case noted, with documentation to support this in the medical notes.

Implications for future research

The significant association with anticholinergic load in the later phases of life indicate that the use of anticholinergic medications in the palliation of this symptom needs further assessment. A recent randomized control trial found no significant difference between three anticholinergic agents¹³ and concluded treatment commenced at a lower noise intensity improved the response rate. What remains unclear is the natural history of the symptom. There is little placebo controlled data to aid our understanding of what happens to NRS if no treatment is initiated.

This study highlights the need for large and rigorously designed, placebo controlled trials to prove or disprove the efficacy of antisecretory medications in the treatment of NRS at the end of life.

Although it is fairly widely accepted that the patients themselves are not distressed as a result of the symptom, the intent of treatment is to minimize caregivers, family, and staff distress. Exploration into the role of education as a nonpharmacologic measure is also needed.

Footnotes

Acknowledgments

There was no specific funding grant for this research from government, commercial, or not-for-profit organizations. This research was supported by Sacred Heart Centre, Darlinghurst, and The Cunningham Centre for Palliative Care with statistical support from the University of Notre Dame.

The authors greatly acknowledge Professor David Currow and Dr. Meera Agar for review of the manuscript.

Author Disclosure Statement

No competing financial interests exist.

References

Wildiers

, Menten

. Death rattle: Prevalence, prevention and treatment. J Pain Symptom Manage, 2002; 23:310–317.

Hugel

, Ellershaw

, Gambles

. Respiratory tract secretions in the dying patient: A comparison between glycopyrronium and hyoscine hydrobromide. J Palliat Med, 2006; 9:279–284.

Bennett

. Death rattle: An audit of hyoscine (Scopolamine) use and review of management. J Pain Symptom Manage, 1996; 12:229–233.

Morita

, Hyodo

, Yoshimi

, Ikenaga

, Tamura

, Yoshizawa

et al. Incidence and underlying etiologies of bronchial secretion in terminally ill cancer patients: A multicenter, prospective, observational study. J Pain Symptom Manage, 2004; 27:533–539.

Lichter

, Hunt

. The last 48 hours of life. J Pall Care, 1990; 6:7–15.

Ellershaw

, Sutcliffe

, Saunders

. Dehydration and the dying patient. J Pain Symptom Manage, 1995; 10:192–197.

Wee

, Hillier

. Interventions for noisy breathing in patients near to death. Cochrane Database Syst Rev, 2008; 1:CD005177.

Wee

, Coleman

, Hillier

, Holgate

. The sound of death rattle I: Are relatives distressed by hearing this sound? Palliat Med, 2006; 20:171–175.

Wee

, Coleman

, Hillier

, Holgate

. The sound of death rattle II: How do relatives interpret the sound? Palliat Med, 2006; 20:177–181.

10.

Wee

, Coleman

, Hillier

, Holgate

. Death rattle: Its impact on staff and volunteers in palliative care. Palliat Med, 2008; 22:173–176.

11.

Watts

, Jenkins

, Back

. Problem and management of noisy rattling breathing in dying patients. Int J Palliat Nurs, 1997; 3:245–252.

12.

Clark

, Butler

. Noisy respiratory secretions at the end of life. Curr Opin Supp Palliat Care, 2009; 3:120–124.

13.

Wildiers

, Dhaenekint

, Demeulenaere

, Clement

PMJ

, Desmet

, Van Nuffelen

et al. Atropine, hyoscine butylbromide, or scopolamine are equally effective for the treatment of death rattle in terminal care. J Pain Symptom Manage, 2009; 38:124–133.

14.

Morita

, Tsunoda

, Inoue

, Chihara

. Risk factors for death rattle in terminally ill cancer patients: A prospective exploratory study. Palliat Med, 2000; 14:19–23.

15.

Kåss

, Ellershaw

. Respiratory tract secretions in the dying patient: A retrospective study. J Pain Symptom Manage, 2003; 26:897–902.

16.

Agar

, Currow

, Plummer

, Seidel

, Carnahan

, Abernethy

. Changes in anticholinergic load from regular prescribed medications in palliative care as death approaches. Palliat Med, 2009; 23:257–265.

17.

Bachrach

, Walter

, Trzcinski

. Use of glycopyrrolate and other anticholinergic medications for sialorrhea in children with cerebral palsy. Clin Pediatr, 1998; 37:485–490.

18.

Mier

, Bachrach

, Lakin

, Barker

, Childs

, Moran

. Treatment of sialorrhea with glycopyrrolate: A double-blind, dose-ranging study. Arch Pediatr Adolesc Med, 2000; 154:1214–1218.

19.

Eagar

, Green

, Gordon

. An Australian casemix classification for palliative care: Technical development and results. Palliat Med, 2004; 18:217–226.

20.

Carnahan

, Lund

, Perry

, Pollock

, Culp

. The Anticholinergic Drug Scale as a measure of drug-related anticholinergic burden: associations with serum anticholinergic activity. J Clin Pharma, 2006; 46:1481–1486.

21.

Clark

, Lam

, Agar

, Chye

, Currow

. The impact of opioids, anticholinergic medications and disease progression on the prescription of laxatives in hospitalized palliative care patients: A retrospective analysis. Palliat Med, 2010; 24:410–418.

22.

Rudolph

, Salow

, Angelini

, McGlinchey

. The Anticholinergic Risk Scale and anticholinergic adverse effects in older persons. Arch Intern Med, 2008; 168:508–513.

23.

Bhanji

, Chouinard

, Kolivakis

, Margolese

. Persistent tardive rebound panic disorder, rebound anxiety and insomnia following paroxetine withdrawal: a review of rebound-withdrawal phenomena. Can J Clin Pharma, 2006; 13:e69–e74.

24.

Howland

. Potential adverse effects of discontinuing psychotropic drugs: Part 1: adrenergic, cholinergic, and histamine drugs. J Psychosoc Nurs Mental Health Serv, 2010; 48:11–14.

25.

Majocha

, Baldessarini

. Tolerance to an anticholinergic agent is paralleled by increased binding to muscarinic receptors in rat brain and increased behavioral response to a centrally active cholinomimetic. Life Sciences, 1984; 35:2247–2255.

26.

Morita

, Hyodo

, Yoshimi

, Ikenaga

, Tamura

, Yoshizawa

et al. Artificial hydration therapy, laboratory findings, and fluid balance in terminally ill patients with abdominal malignancies. J Pain Symptom Manage, 2006; 31:130–139.

27.

Morita

, Hyodo

, Yoshimi

, Ikenaga

, Tamura

, Yoshizawa

et al. Association between hydration volume and symptoms in terminally ill cancer patients with abdominal malignancies. Ann Oncol, 2005; 16:640–647.