Volume 121, Issue 6 p. 960-967
Original Article
Free Access

Bedside clinical signs associated with impending death in patients with advanced cancer: Preliminary findings of a prospective, longitudinal cohort study

David Hui MD, MSc

Corresponding Author

David Hui MD, MSc

Department of Palliative Care and Rehabilitation Medicine, University of Texas MD Anderson Cancer Center, Houston, Texas

Corresponding author: David Hui, MD, MSc, Department of Palliative Care and Rehabilitation Medicine, University of Texas MD Anderson Cancer Center, Unit 1414, 1515 Holcombe Boulevard, Houston, TX 77030; Fax: (713) 792-6092; [email protected]Search for more papers by this author
Renata dos Santos MD

Renata dos Santos MD

Department of Palliative Care, Barretos Cancer Hospital, Barretos, Brazil

Search for more papers by this author
Gary Chisholm MS

Gary Chisholm MS

Department of Biostatistics, University of Texas MD Anderson Cancer Center, Houston, Texas

Search for more papers by this author
Swati Bansal MPH

Swati Bansal MPH

Department of Palliative Care and Rehabilitation Medicine, University of Texas MD Anderson Cancer Center, Houston, Texas

Search for more papers by this author
Camila Souza Crovador RN

Camila Souza Crovador RN

Department of Palliative Care, Barretos Cancer Hospital, Barretos, Brazil

Search for more papers by this author
Eduardo Bruera MD

Eduardo Bruera MD

Department of Palliative Care and Rehabilitation Medicine, University of Texas MD Anderson Cancer Center, Houston, Texas

Search for more papers by this author
First published: 09 February 2015
Citations: 85

Abstract

BACKGROUND

Five highly specific physical signs associated with death within 3 days among cancer patients were recently reported that may aid in the diagnosis of impending death. In this study, the frequency and onset of another 52 bedside physical signs and their diagnostic performance for impending death were examined.

METHODS

Three hundred fifty-seven consecutive patients with advanced cancer who had been admitted to acute palliative care units at 2 tertiary care cancer centers were enrolled. Fifty-two physical signs were systematically documented every 12 hours from admission to death or discharge. The frequency and median time of onset of each sign from death backwards were examined, and the likelihood ratios (LRs) associated with death within 3 days were calculated.

RESULTS

Two hundred three of the 357 patients (57%) died at the end of the admission. Eight physical signs that were highly diagnostic of impending death were identified. These signs occurred in 5% to 78% of the patients within the last 3 days of life, had a late onset, and had a high specificity (>95%) and a high positive LR for death within 3 days. They included nonreactive pupils (positive LR, 16.7; 95% confidence interval [CI], 14.9-18.6), a decreased response to verbal stimuli (positive LR, 8.3; 95% CI, 7.7-9), a decreased response to visual stimuli (positive LR, 6.7; 95% CI, 6.3-7.1), an inability to close eyelids (positive LR, 13.6; 95% CI, 11.7-15.5), drooping of the nasolabial fold (positive LR, 8.3; 95% CI, 7.7-8.9), hyperextension of the neck (positive LR, 7.3; 95% CI, 6.7-8), grunting of vocal cords (positive LR, 11.8; 95% CI, 10.3-13.4), and upper gastrointestinal bleeding (positive LR, 10.3; 95% CI, 9.5-11.1).

CONCLUSIONS

Eight highly specific physical signs associated with death within 3 days among cancer patients were identified. These signs may inform the diagnosis of impending death. Cancer 2015;121:960–967. © 2014 American Cancer Society.

INTRODUCTION

As patients approach the last days of their lives, they experience a multitude of physiological changes affecting their neurocognitive, cardiovascular, respiratory, and muscular function.1 These bodily changes may be observed at the bedside and may assist clinicians in establishing the diagnosis of impending death (ie, death within days). The ability to make the diagnosis of impending death with confidence is of great importance to clinicians who attend to patients at the end of life because it could affect their communication with patients and families and inform complex decision making such as the discontinuation of investigations and aggressive treatments, discharge planning, and enrollment into clinical care pathways.2, 3

There has been a paucity of studies examining diagnostic signs of impending death. A majority of the studies on this topic started monitoring patients when they were recognized as actively dying, and this potentially could have resulted in a biased population and an overestimation of the frequency of these physical signs.4, 5 Recognizing this limitation, we recently conducted the Investigating the Process of Dying Study, a prospective, longitudinal, observational study that systematically documented an array of clinical signs every 12 hours in consecutive patients from the time of admission to an acute palliative care unit (APCU).6 Among the 10 target signs, we identified 5 signs (ie, pulselessness of the radial artery, decreased urine output, Cheyne-Stokes breathing, respiration with mandibular movement, and death rattle) that occurred only in the last days of life and were highly predictive of an impending death within 3 days. In this study, we report the frequency and onset of an additional 52 bedside physical signs and their diagnostic performance for impending death.

MATERIALS AND METHODS

Study Setting and Criteria

This study is a planned secondary analysis of the Investigating the Process of Dying Study designed to identify physical signs of impending death. The methods have been reported in detail previously.6 Briefly, we enrolled consecutive patients with a diagnosis of advanced cancer who were ≥18 years of age and were admitted to the APCUs at the MD Anderson Cancer Center in the United States between April 5, 2010 and July 6, 2010 and at Barretos Cancer Hospital in Brazil between January 27, 2011 and June 1, 2011. The institutional review boards at both institutions approved this study. A waiver of consent for patient participation was endorsed to minimize distress during the consent process and to ensure that we could collect data on consecutive patients. All nurses who participated in this study signed the informed consent before enrollment.

The APCU was the setting of choice for this study because of (1) the relatively high mortality rate, (2) the presence of clinical staff 24 hours a day, and, importantly, (3) the experience of APCU nurses in providing care to patients in the last days of life and their commitment to complete the study assessments every shift. Patients with advanced cancer were admitted to APCUs for symptom control and/or transition of care at the end of life.7 Both participating APCUs are situated within tertiary care cancer centers and provide interprofessional symptom management and psychosocial care, active management of acute complications, and discharge planning.8

Data Collection

We selected a list of clinical signs to be documented every 12 hours from admission until death or discharge; this list was based on a literature review of published studies, review articles, and educational materials.1, 4, 9-11 The final list included an array of 62 signs that systematically cover the neurological, cardiovascular, respiratory, gastrointestinal, genitourinary, gynecological, muscular, and integumentary systems. Ten of these signs were selected as target signs on the basis of their prevalence in the literature (ie, apnea periods, Cheyne-Stokes breathing, death rattle, dysphagia of liquids, decreased level of consciousness, palliative performance scale ≤ 20%, peripheral cyanosis, pulselessness of the radial artery, respiration with mandibular movement, and decreased urine output), and they had been reported previously.6 This study focused on the remaining 52 physical signs.

We collected baseline patient demographics on admission. All nurses who participated in this study worked full time in palliative care and were experienced in providing care at the end of life. All nurses attended an orientation session to review the study objectives and data collection forms, with a particular emphasis on physical signs not commonly described, such as grunting of vocal cords and drooping of the nasolabial fold. Moreover, the principal investigators and charge nurses provided longitudinal support during the study by reviewing the forms on a daily basis to ensure that they were complete and accurate, and they provided education to the nurses on an as-needed basis. The 2 study sites had a weekly video conference to ensure that data were collected systematically and accurately. The study forms were translated into Portuguese to facilitate data collection in Brazil and were back-translated to ensure the accuracy of the translation. Clinical nurses completed standardized data collection forms independently of prior assessments. The 12-hour period was chosen on the basis of the duration of the nursing shift.

Table 1 includes a detailed description and coding for the 52 clinical signs. A majority of the signs were marked as either absent or present during the past 12 hours on the basis of the nurses' clinical observations (eg, drooping of the nasolabial fold and respiration with mandibular movement). A grading system was used for the remaining signs (eg, myoclonus). Delirium was documented with the Memorial Delirium Assessment Scale every 12 hours. The Memorial Delirium Assessment Scale has been validated for delirium screening in cancer patients. It comprises 10 items, each with a score between 0 (normal) and 3 (worst). A total score > 13/30 suggests delirium.12 We also documented vital signs, including the heart rate, respiratory rate, blood pressure, temperature, and oxygen saturation. The values were routinely collected for all admitted patients unless the attending physician ordered otherwise.

Table 1. Definitions of Clinical Signs
Variable (Definition) Criteria for Negative Sign Criteria for Positive Sign
Vitals
Heart rate (bpm) ≤100 >100
Respiratory rate (breaths/min) ≤20 >20
Systolic blood pressure (mm Hg) ≥100 <100
Diastolic blood pressure (mm Hg) ≥60 <60
Temperature (°C) ≥36 <36
Oxygen saturation (%) ≥90 <90
Supplemental oxygen use Absent Present
Nervous system
Memorial Delirium Assessment Scale 0-13 ≥13
Agitation/purposeless movements None; rarely Obvious; severe
Decreased level of consciousness Absent Present
Nonreactive pupils Absent Present
Decreased response to verbal stimuli Absent Present
Decreased response to visual stimuli Absent Present
Decreased speech Absent Present
Motor abnormality: tremors Absent Present
Motor abnormality: weakness Absent Present
Myoclonus None Mild fasciculations involving face/distal extremities; marked movements of face/limbs; severe movements involving limbs and trunk
Seizures Absent Present
Head and neck
Inability to close eyelids Absent Present
Drooping of nasolabial fold (decrease in prominence/visibility of nasolabial fold) Absent Present
Hyperextension of neck Absent Present
Epistaxis None Streaks; teaspoon; tablespoon
Thrush Absent Present
Cardiovascular
Mottling (stasis, purplish discoloration of extremities) None Toes; feet; up to knees
Peripheral edema None Up to ankle; up to knee; anasarca
Capillary refill > 3 s Absent Present
Decreased heart sounds Absent Present
Respiratory
Grunting of vocal cords (sound produced predominantly on expiration, related to vibrations of vocal cords) Absent Present
Irregular breathing pattern Absent Present
Decreased breath sounds Absent Present
Tachypnea ≤20 21-30; 31-40; >40
Use of accessory muscles Absent Present
Hemoptysis None Streaks; teaspoon; tablespoon
Inability to clear secretions Absent Present
Gastrointestinal
Upper gastrointestinal bleed None ≤20 mL; >20 mL; hemodynamic Δ
Lower gastrointestinal bleed (hematochezia, melena) None ≤20 mL; >20 mL; hemodynamic Δ
Dysphagia of solids Absent Present
Meals left over None; <half >Half; entire meal
Ascites Absent Present
Bowel sounds Within normal limits; hyperactive Hypoactive; absent
Fecal incontinence Absent Present
Genitourinary
Voiding difficulties Normal Abnormal; diaper; urinary catheter
Urinary incontinence Absent Present
Vaginal bleed Absent Present
Vaginal discharge Absent Present
Hematuria None Hematuria but no clots; blood clots
Integumentary
Skin turgor Good, <1-s recoil Fair, 1- to 2-s recoil; poor, >2-s recoil
Cool/cold extremities None Toes; feet; up to knees
Skin moist or diaphoretic Absent Present
Pressure ulcer Absent Present
Cool skin temperature Absent Present
Sweating None Limited; diffuse; drenching

Survival from the time of APCU admission was collected from institutional databases and electronic health records.

Statistical Analysis

We analyzed the data with the same methodology used in previous reports.6, 13 Our preplanned sample size was a combined total of 200 deaths at the 2 study sites as stated previously.6 This analysis was planned on the basis of the combined data a priori. Briefly, we dichotomized all the variables into absent or present (see Table 1). Predefined cutoffs were used for continuous variables (eg, vital signs). For the patients who died in the APCUs, we calculated the frequency of each sign from the time of death backwards. We also determined the median duration between the first documentation of each sign and death with the Kaplan-Meier method; this was conditional on the observation of that particular sign. In the time-to-event analysis, patients were left censored if they entered the APCU with the sign already present.

To determine the diagnostic utility of each sign, we computed the sensitivity, specificity, positive likelihood ratio (LR), and negative LR for death within 3 days in 357 patients.6, 13, 14 Positive LRs indicate how many times the sign of interest appeared in patients who died within 3 days in comparison with patients who did not die within 3 days. Positive LRs > 5 and > 10 suggest good and excellent discriminatory test performance, respectively.15 The last 3 days was chosen as the cutoff for impending death because our previous study showed the emergence of many of the signs of impending death during this period, and knowing a patient is within the last 3 days of life could affect many medical decisions such as hospital discharge, discontinuation of prescription medications, artificial nutrition, and use of life-support measures. To account for the multiple observations for each patient, we randomly sampled our data with 1 observation per patient to construct a data set with independent samples, and we then calculated the sensitivity, specificity, positive LR, and negative LR for the resampled data set. We repeated this resampling algorithm 100 times to obtain the average and 95% confidence interval for each statistic.

We also examined the univariate odds ratios (ORs) with a similar resampling methodology. ORs are complementary to LRs because they provide an estimate of the strength of association between impending death and the physical signs. We also determined the prevalence (ie, pretest probability) of death within 3 days. We selected the 7 neurological signs with a positive LR > 5 for inclusion in a multivariate logistic regression model with backward selection using a similar resampling strategy with 500 repetitions. Upper gastrointestinal bleeding was excluded from the model because of its low frequency. We examined the goodness of fit of the model with the Hosmer-Lemeshow statistic, for which a P value > .05 indicates a good fit.16, 17

The Statistical Analysis System (SAS version 9.2, SAS Institute, Cary, NA) was used for the statistical analysis. A P value < .05 was considered significant.

RESULTS

Patient Characteristics

The patient demographics have been reported in a previous study.6 In brief, the average age was 58 years (range, 18-88 years), 195 (55%) were female, 233 (65%) were of Hispanic origin, and 101 (28%) had a diagnosis of gastrointestinal cancer. The median duration of the APCU admission was 6 days (interquartile range, 4-9 days).

Frequency and Onset of Clinical Signs

The frequency and median onset of the 52 physical signs are shown in Table 2. Figure 1 shows the frequency over the last 14 days of life for 7 neurological signs selected on the basis of their high diagnostic performance. These 7 signs had a late median onset within the last 3 days of life (Fig. 2).

Table 2. Frequencies, Onsets, Odds Ratios, Sensitivities, Specificities, and Negative and Positive LRs for 52 Physical Signs in 3 Daysa
Sign Prevalence of Sign in Last 3 Days of Life, n (%)b Time From Onset to Death (d), Median (95% CI) Univariate Odds Ratio (95% CI) Sensitivity (95% CI)c Specificity (95% CI)c Negative LR (95% CI)c Positive LR (95% CI)c
Vitals
Diastolic blood pressure < 60 mm Hg 76 (62) 2.5 (2.0-3.5) 4.5 (2.6-10.2) 19.3 (18.7-19.9) 94.9 (94.6-95.2) 0.85 (0.84-0.86) 4.1 (3.8-4.4)
Heart rate > 100/min 130 (92) 8.5 (6.0-11.5) 2.1 (1.4-2.9) 53.1 (52.4-53.8) 65 (64.5-65.5) 0.72 (0.71-0.74) 1.5 (1.5-1.6)
Oxygen saturation < 90% 107 (90) 5.5 (4.0-7.0) 3.9 (2.6-6.1) 40.5 (39.7-41.2) 86 (85.5-86.4) 0.69 (0.68-0.7) 3 (2.8-3.1)
Respiratory rate > 20/min 102 (72) 4.5 (3.5-5.5) 2.5 (1.7-4.3) 24.1 (23.5-24.7) 88.6 (88.2-89) 0.86 (0.85-0.87) 2.2 (2.1-2.3)
Supplemental oxygen use 170 (89) 6.5 (4.5-10.5) 6.8 (0.8-27) 1.9 (1.6-2.2) 99.1 (99-99.3) 0.99 (0.99-0.99) 2.9 (2.5-3.4)
Systolic blood pressure < 100 mm Hg 112 (82) 4.0 (3.0-6.5) 4.1 (2.7-6.7) 34.9 (34.3-35.6) 88.2 (87.8-88.6) 0.74 (0.73-0.75) 3.1 (2.9-3.2)
Temperature < 36°C 81 (60) 5.5 (3.5-7.5) 1.2 (0.7-2.2) 18.2 (17.6-18.8) 85.2 (84.7-85.6) 0.96 (0.95-0.97) 1.3 (1.2-1.3)
Nervous System
Agitation/purposeless movements 53 (36) 3.5 (2.5-4.0) 2.2 (1.4-4.6) 10.9 (10.5-11.2) 94.9 (94.6-95.2) 0.94 (0.93-0.94) 2.4 (2.2-2.5)
Decreased level of consciousness 186 (97) 7.0 (5.5-8.5) 5.6 (4.2-7.5) 74 (73.4-74.5) 66.1 (65.7-66.6) 0.39 (0.39-0.4) 2.2 (2.2-2.2)
Decreased response to verbal stimuli 118 (69) 2.0 (1.5-4.0) 10 (5.2-23.8) 30 (29.4-30.5) 96 (95.8-96.3) 0.73 (0.72-0.74) 8.3 (7.7-9)
Decreased response to visual stimuli 121 (70) 3.0 (2.0-4.0) 7.6 (5-16.1) 31.9 (31.4-32.4) 94.9 (94.6-95.1) 0.72 (0.71-0.72) 6.7 (6.3-7.1)
Decreased speech 143 (79) 4.0 (3.0-6.0) 6.9 (4.8-10.9) 42.6 (42-43.2) 90.2 (89.8-90.5) 0.64 (0.63-0.64) 4.4 (4.3-4.6)
Memorial Delirium Rating Scale > 13 105 (89) 12 (11.0-13.0) 7.7 (4.8-12) 66.3 (65.6-67) 80.5 (79.9-81.1) 0.42 (0.41-0.43) 3.5 (3.4-3.6)
Motor abnormality: tremors 47 (32) 7.0 (4.0-9.0) 2.6 (1.3-5.9) 9 (8.7-9.4) 96.8 (96.5-97) 0.94 (0.94-0.94) 3.2 (2.9-3.5)
Motor abnormality: weakness 168 (89) 31.0 (11.0-50.5) 1 (0.7-1.3) 65.6 (65-66.2) 34.6 (34.1-35.1) 1 (0.98-1.03) 1 (1-1)
Myoclonus 63 (42) 4.0 (3.5-6.5) 2.4 (1.3-5.9) 14.4 (13.9-14.9) 93.2 (92.9-93.5) 0.92 (0.91-0.93) 2.3 (2.1-2.5)
Nonreactive pupils 53 (38) 2.0 (1.5-3.0) 13.7 (6.1-89.9) 15.3 (14.9-15.7) 99 (98.8-99.1) 0.86 (0.85-0.86) 16.7 (14.9-18.6)
Seizures 5 (4) 2.5 (0.5-28.0) 4.4 (0.5-8.2) 0.7 (0.6-0.8) 99.7 (99.6-99.7) 1 (1-1) 2.6 (2.2-3)
Head and neck
Drooping of nasolabial fold 137 (78) 2.5 (1.5-3.0) 9.4 (6-21.1) 33.7 (33.2-34.3) 95.5 (95.3-95.8) 0.69 (0.69-0.7) 8.3 (7.7-8.9)
Epistaxis 7 (5) 2.8 (0.5-4.5) 1.5 (0.3-6.2) 0.8 (0.7-0.9) 99.4 (99.3-99.5) 1 (1-1) 1.9 (1.5-2.4)
Hyperextension of neck 73 (46) 2.5 (2.0-3.0) 7.6 (4.4-16.9) 21.2 (20.6-21.7) 96.7 (96.5-96.9) 0.82 (0.81-0.82) 7.3 (6.7-8)
Inability to close eyelids 93 (57) 1.5 (1.0-1.5) 11.3 (5.5-36.5) 21.4 (20.9-21.8) 97.9 (97.7-98.1) 0.8 (0.8-0.81) 13.6 (11.7-15.5)
Thrush 29 (21) 4.0 (3.0-7.0) 0.9 (0.5-2.1) 6.6 (6.3-6.9) 93.6 (93.3-93.9) 1 (0.99-1) 1.1 (1-1.2)
Cardiovascular
Capillary refill > 3 s 155 (87) 7.0 (4.5-14.0) 4.1 (2.7-5.9) 47.3 (46.6-48) 82.4 (81.9-82.8) 0.64 (0.63-0.65) 2.7 (2.6-2.8)
Decreased heart sounds 76 (48) 2.5 (1.0-3.5) 3.6 (2-6.8) 16.2 (15.7-16.7) 94.7 (94.4-95) 0.89 (0.88-0.89) 3.4 (3.1-3.8)
Mottling 71 (46) 2.5 (1.5-3.5) 4.6 (2.9-12.1) 17.1 (16.6-17.6) 95.8 (95.5-96) 0.87 (0.86-0.87) 4.4 (4.1-4.6)
Peripheral edema 178 (94) 13.5 (8.5-30.0) 2 (1.5-3) 68.8 (68.2-69.4) 47.9 (47.4-48.4) 0.65 (0.64-0.67) 1.3 (1.3-1.3)
Respiratory
Decreased breath sounds 100 (61) 8.5 (7.0-10.5) 1.3 (1-1.7) 40.8 (40.3-41.3) 65.5 (65-66) 0.91 (0.89-0.92) 1.2 (1.2-1.2)
Grunting of vocal cords 86 (54) 1.5 (1.0-2.0) 10.7 (4.6-32.5) 19.5 (19-19.9) 97.9 (97.7-98.1) 0.82 (0.82-0.83) 11.8 (10.3-13.4)
Hemoptysis 10 (8) 4.0 (1.5-31.5) 3.1 (1-16.3) 2.8 (2.6-3) 99.1 (99-99.2) 0.98 (0.98-0.98) 4.7 (3.9-5.5)
Inability to clear secretions 155 (87) 3.5 (2.5-4.5) 4.8 (2.9-6.9) 46.1 (45.6-46.7) 84.9 (84.5-85.3) 0.64 (0.63-0.64) 3.1 (3-3.2)
Irregular breathing pattern 109 (65) 4.5 (3.0-7.0) 4.7 (3.1-8.2) 38 (37.4-38.5) 88.6 (88.2-89) 0.7 (0.69-0.71) 3.4 (3.3-3.6)
Tachypnea 129 (76) 4.5 (3.5-6.0) 1.8 (1.2-2.7) 26.8 (26.2-27.4) 83.1 (82.7-83.5) 0.88 (0.87-0.89) 1.6 (1.6-1.7)
Use of accessory muscles 135 (77) 3.5 (2.5-5.0) 4.7 (3.1-6.9) 41.4 (40.7-42) 86.7 (86.3-87.1) 0.68 (0.67-0.68) 3.2 (3.1-3.3)
Gastrointestinal
Ascites 50 (35) 4.5 (3.0-7.0) 1.2 (0.7-1.9) 14.7 (14.3-15.1) 87.5 (87.1-87.8) 0.98 (0.97-0.98) 1.2 (1.1-1.3)
Decreased or absent bowel sounds 49 (36) 3.0 (2.0-3.0) 2 (1.3-4.2) 12.4 (12-12.7) 93.5 (93.3-93.8) 0.94 (0.93-0.94) 2 (1.9-2.2)
Dysphagia of solids 118 (96) 11.0 (6.0-18.5) 2.5 (1.9-3.6) 50.1 (49.4-50.8) 71.9 (71.5-72.4) 0.7 (0.68-0.71) 1.8 (1.8-1.9)
Fecal incontinence 84 (57) 4.5 (3.0-6.5) 2.4 (1.5-4.1) 28.2 (27.7-28.7) 86.1 (85.7-86.4) 0.83 (0.83-0.84) 2.1 (2-2.1)
Lower gastrointestinal bleed 8 (6) 12.0 (2.5-46.5) 1.5 (0.7-3.9) 2.9 (2.7-3) 98.1 (97.9-98.2) 0.99 (0.99-0.99) 1.8 (1.5-2)
Meals: >half left over 183 (95) 7.5 (5.0-9.5) 3.5 (2.6-5.3) 81.1 (80.6-81.5) 45.7 (45.2-46.2) 0.41 (0.4-0.42) 1.5 (1.5-1.5)
Upper gastrointestinal bleed 6 (5) 5.5 (0.5-17.0) 10.9 (2.6-18.5) 2.9 (2.8-3) 99.7 (99.6-99.7) 0.97 (0.97-0.98) 10.3 (9.5-11.1)
Genitourinary
Hematuria 25 (19) 3.5 (2.0-4.5) 2.9 (1.1-9.6) 5.9 (5.6-6.1) 97.8 (97.6-97.9) 0.96 (0.96-0.97) 3.4 (2.7-4.1)
Urinary incontinence 124 (72) 6.0 (4.5-10.5) 1.9 (1.4-2.4) 47.9 (47.4-48.5) 67.2 (66.8-67.6) 0.78 (0.77-0.79) 1.5 (1.4-1.5)
Vaginal bleed 5 (7) 13.5 (3.0-NR) 1.9 (0.2-6) 3.2 (2.8-3.5) 97.8 (97.6-98) 0.99 (0.99-1) 2.4 (1.8-3)
Vaginal discharge 8 (11) 3.0 (0.5-34.0) 2 (0.6-6.5) 4.1 (3.7-4.4) 97.7 (97.5-97.9) 0.98 (0.98-0.99) 2.5 (2-2.9)
Voiding difficulties 190 (97) 11.5 (9.0-31.0) 5.2 (3.8-7.6) 80.5 (80-80.9) 56.7 (56.3-57.1) 0.35 (0.34-0.35) 1.9 (1.8-1.9)
Integumentary
Cool skin temperature 59 (40) 3.0 (2.0-6.0) 4.8 (2.2-13.5) 13.6 (13.2-14) 96.8 (96.6-97) 0.89 (0.89-0.9) 4.9 (4.4-5.3)
Cool/cold extremities 151 (84) 7.5 (5.5-9.0) 2.5 (1.7-3.2) 47.9 (47.4-48.5) 71.4 (70.8-72) 0.73 (0.72-0.74) 1.7 (1.7-1.7)
Pressure ulcer 65 (47) 7.5 (4.5-13.0) 1.3 (0.9-1.9) 16.6 (16.2-17) 86.5 (86.1-86.9) 0.97 (0.96-0.97) 1.3 (1.2-1.3)
Skin moist or diaphoretic 9 (8) 3.5 (3.0-6.0) 1.1 (0.2-10.8) 1.3 (1.1-1.4) 99.1 (99-99.2) 1 (0.99-1) 2.3 (1.9-2.7)
Skin turgor recoil ≥ 1 s 198 (99) 40.0 (21.0-NR) 1.8 (1.2-2.6) 85.6 (85.2-86) 23.8 (23.3-24.3) 0.61 (0.59-0.63) 1.1 (1.1-1.1)
Sweating 110 (66) 3.5 (2.5-5.5) 2.4 (1.2-3.6) 20.9 (20.3-21.5) 89.6 (89.2-89.9) 0.88 (0.88-0.89) 2.1 (2-2.2)
  • Abbreviations: CI, confidence interval; LR, likelihood ratio; NR, not reached.
  • a Entries with positive LRs > 5 are highlighted in bold.
  • b Any occurrence of the sign of interest within the last 3 days of life among the 203 patients who died in the APCU.
  • c The sensitivity, specificity, positive LR, and negative LR were computed for each sign of death within 3 days with data from all 357 patients. To account for the multiple observations for each patient, we resampled our data by randomly selecting 1 observation from each subject to construct a data set with independent samples, and we then calculated the sensitivity, specificity, positive LR, and negative LR for the resampled data set. We repeated this resampling algorithm 100 times to obtain the average and 95% CI for each statistic.
Details are in the caption following the image

Frequency of 7 neurological signs of impending death among 203 patients with advanced cancer who died in an acute palliative care unit. The frequency of these signs increased over the last few days.

Details are in the caption following the image

Onset of 7 neurological signs of impending death. The median onset (95% confidence interval) was ≤3 days before death for all 7 signs.

LRs for Death in 3 Days

The prevalence (ie, pretest probability) of death within 3 days was 38% (95% confidence interval, 19%-57%), for our cohort. Table 2 shows the sensitivity, specificity, positive LR, and negative LR values associated with death within 3 days for the 52 physical signs. Eight of these signs, including nonreactive pupils, decreased response to verbal stimuli, decreased response to visual stimuli, inability to close eyelids, drooping of the nasolabial fold, hyperextension of the neck, grunting of vocal cords, and upper gastrointestinal bleeding, had a high specificity (95% or higher) and a high positive LR > 5. The multivariate analysis revealed that a decreased response to verbal stimuli and drooping of the nasolabial fold were significantly associated with death within 3 days (Table 3). The median P value for the Hosmer-Lemeshow statistic was .54, which indicated good calibration.

Table 3. Multivariate Logistic Regression for Death Within 3 Days
Sign Odds Ratio (95% CI) P
Decreased response to verbal stimuli 6.61 (2.49-23.17) .0004
Drooping of nasal labial fold 6.51 (2.67-20.07) .0001
  • Abbreviation: CI, confidence interval.

DISCUSSION

We previously reported 5 physical signs that were highly diagnostic of impending death. This study systematically examined a comprehensive list of signs that represent multiple organ systems, and it identified 8 additional highly specific physical signs for impending death within 3 days (7 neurological and 1 bleeding complication). The presence of these signs was associated with a high likelihood of death within 3 days, although their absence cannot rule out that the patient will die shortly because of their low sensitivity. The use of these bedside physical signs individually or in combination may assist clinicians in making the diagnosis of impending death.

With the exception of upper gastrointestinal bleeding, all the signs identified in this study were related to deterioration in neurocognitive (ie, nonreactive pupils, decreased response to verbal stimuli, and decreased response to visual stimuli) and neuromuscular function (ie, inability to close eyelids, drooping of the nasolabial fold, hyperextension of the neck, and grunting of vocal cords). In addition, 3 of the 5 physical signs identified in our previous study (ie, Cheyne-Stokes breathing, respiration with mandibular movement, and death rattle) were also related to an altered neurological status. Importantly, all 10 of these highly specific neurological signs occurred relatively late, with a median onset of 3 days before death. The high specificity suggests that few patients who did not die within 3 days were observed to have these signs. These signs were commonly observed in the last 3 days of life with a frequency in patients between 38% and 78%. Our findings highlight that the progressive decline in neurological function is associated with the dying process.

Upper gastrointestinal bleeding in patients admitted to APCUs also had a high positive LR for impending death. Understandably, hematemesis is a life-threatening complication, and it can lead to hemodynamic instability and result in death acutely. Because the frequency of upper gastrointestinal bleeding was low (5%) in the last days of life, further studies are needed to confirm the diagnostic utility of this sign.

A multivariate analysis identified 2 physical signs (ie, decreased response to verbal stimuli and drooping of the nasolabial fold) that were independently significant. However, the other physical signs with high positive LRs, when present, could still be helpful in the diagnosis of impending death. This is particularly true because patients often do not present with all the physical signs at the same time.

The positive LRs for the 8 diagnostic signs ranged from 6.7 to 16.7. The positive LR can be easily applied in the clinical setting to facilitate the diagnosis of impending death with either a formula or nomogram. For instance, the pretest probability for dying within 3 days of admission to our APCUs was 38%. The presence of grunting of vocal cords (positive LR, 11.8) in a patient results in a posttest probability of 88% of death within the next 3 days. Given the high likelihood of impending death, the clinician may recommend discontinuation of blood work and select medications and hold hospital discharge after a discussion with the patient and/or his or her family.

Although some of the signs identified here have been described anecdotally in review articles and books,1, 4, 10, 11 this is the first study to systematically characterize their frequencies, onset, LRs, and ORs and thus allow clinicians to differentiate their relative importance and utility for the diagnosis of impending death. For instance, the positive LRs for a diastolic blood pressure < 60 mm Hg, mottling, hemoptysis, and a cool skin temperature were between 4 and 5 (Table 2). Although these signs may be associated with impending death, they had lower diagnostic utility for impending death in comparison with the 8 signs identified here. Interestingly, many other signs such as delirium, dysphagia, and incontinence that occur commonly at the end of life were not diagnostic, likely because they had an earlier onset, and thus they were less useful for informing us of imminent death. We also identified many signs such as epistaxis, myoclonus, and thrush that were not associated with impending death, and this provided good internal control.

This study has several limitations. First, we included only cancer patients admitted to APCUs, who often have severe symptoms.18 The frequency, onset, and utility of the signs of impending death may differ in other health care settings and patient populations and need to be further examined. Second, the mortality rates of the 2 APCUs differed significantly. This is related to health care system differences: patients in Brazil could stay in the hospital for longer periods of time in comparison with those in the United States. Importantly, the 2 APCUs have similar palliative care practices, and the investigators from each site have also visited each other. Despite the different mortality rates in the 2 APCUs, we found comparable specificities and sensitivities for the signs between the 2 participating institutions, and this further strengthens our results and their generalizability. Third, the inter-rater reliability of these signs will also need to be assessed in future studies. We documented the signs only every 12 hours, and this limits the resolution of data for identifying the median onset. Fourth, we examined a large number of clinical signs in a relatively small number of patients. Our findings should thus be considered preliminary until they are validated in future studies. The physical signs reported here, in conjunction with those reported previously,6 may pave the way toward the development and validation of a bedside diagnostic tool for impending death.

In summary, this study identified 8 physical signs with high specificity and positive LRs for impending death within 3 days. Upon further validation, the presence of these telltale signs would suggest that patients have are actively dying.19 Taken together with the 5 physical signs identified earlier, these objective bedside signs may assist clinicians, family members, and researchers in recognizing when the patient has entered the final days of life.

FUNDING SUPPORT

This research was supported in part by a University of Texas MD Anderson Cancer Center support grant (CA 016672), which provided the funds for data collection at both study sites. Eduardo Bruera is supported in part by the National Institutes of Health (grants R01NR010162-01A1, R01CA122292-01, and R01CA124481-01).

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.