Abstract
Background
In France, a significant proportion of end-of-life (EOL) decisions occur in emergency departments (EDs), where time constraints and limited resources may hinder optimal care. Although withholding or withdrawing life-sustaining treatments (WHWD) is common in this context, the impact of timing on patient outcomes remains unclear. The aim of this study was to assess the effect of WHWD decision timing on short- and long-term outcomes among adult emergency department patients.
Methods
We conducted a retrospective, observational study at Beaujon Hospital’s ED (Paris, France) between January 2020 and December 2021. We included all adult patients admitted to the observation unit with a WHWD decision. Patients were categorized into early (eWHWD) and late (lWHWD) groups based on the median time to decision. The primary endpoint was 28-day all-cause mortality. Cox regression was used for survival analysis, adjusting for age, sex, comorbidities, functional status, and severity of illness.
Results
WHWD was decided in 354 patients. Patients in the eWHWD group (n = 176) were older, more functionally impaired, and had higher illness severity than those in the lWHWD group (n = 178). Early decisions occurred within a median of 2 h versus 7 h in the late group. Short-term mortality was significantly higher in the eWHWD group at 7 days (58% vs. 44%) and 28 days (76% vs. 65%). After adjustment, late WHWD remained independently associated with better survival at 7-days, 28-days, 90-days and 1-year. Respectively adjusted Hazard Ratios were equal to 0.71; 95% confidence interval (CI) (0.66 to 0.76), 0.73; 95% CI (0.68 to 0.78), 0.76; 95% CI (0.72 to 0.81) and 0.86; 95% CI (0.81 to 0.91).
Conclusion
Early WHWD decisions reflect more severe patient conditions and are linked to poorer short and long-term outcomes. Enhancing advance care planning and integrating palliative care into ED practices may improve decision quality and align care with patient values.
Keywords: Emergency department, End of life, Withholding or withdrawing life-sustaining treatments
Introduction
Each year, approximately 650,000 deaths occur in France [1]. Despite most patients expressing a preference to die at home, 58% of these deaths still took place in hospitals, as reported by the French General inspection of Social Affairs in 2009 [2]. This was even though patients at the end of life (EOL) generally wanted to die at home [3, 4]. This discrepancy between patients’ end-of-life wishes and the reality is likely stemmed from a lack of anticipation and insufficient resources to support home care in the final days. Notably, 16% of hospital deaths occurred within the first 24 h of admission, often in emergency departments (ED) [2]. While deaths accounted for only about 0.2% of all ED visits [5], this still represented approximately 40,000 deaths out of 20.3 million emergency visits in France in 2021. Caring for EOL patients in such settings demands significant time, material, and human resources, placing a significant emotional burden on families and healthcare teams. Importantly, between 79% and 87% of these deaths are preceded by decisions to withhold or withdraw life-sustaining treatments [6, 7]. The primary mission of emergency medical services is to provide care for all individuals in emergency situations, including psychiatric emergencies, 24/7, without selection, especially in distress or life-threatening cases [8]. This mission necessitated a specific organisational structure designed to meet various constraints. However, these constraints often create an “emergency environment” which is not conducive to make withholding or withdrawing (WHWD) decisions or offering appropriate EOL care. In this fast-paced setting, the need to make and implement decisions swiftly, yet carefully, is critical. Additionally, the organisation of care heavily relies on discharge rates, which hinge on the availability of beds in other units. Frequent bed shortages contribute to overcrowding in EDs, forcing patients to wait in suboptimal conditions, with limited privacy and calm. In response to these challenges, the French Intensive Care Society and the French Society of Emergency Medicine published a joint position paper in 2018 addressing WHWD decisions in emergency care [9]. The authors acknowledged the tension between regulatory requirements and everyday realities in emergency departments. The position paper emphasised the critical importance of understanding the daily realities in EDs and respecting the patient’s wishes when making WHWD decisions. It highlighted the need to review patients’ medical records and assess their autonomy and cognitive state before deciding. The position paper also covered the use of deep sedation in emergency settings, stressing that the procedure should be thoroughly documented in the medical record and that physicians must ensure the patient’s relatives fully understand the process. In 2020, the SFMU, in collaboration with the French Society of Palliative Care (SFAP), published a paper-based WHWD decision aid specifically designed for emergency settings [10]. However, data on the total number of WHWD implementations remained scarce. Most studies have focused on patients who died in emergency departments following a WHWD decision. However, there is limited information on patients for whom a WHWD was initiated but who survived beyond their ED stay.
The aim of this study was to assess the effect of WHWD decision timing on short- and long-term outcomes among adult emergency department patients who survived long enough to be admitted to the emergency department observation unit after an initial ED assessment.
Methods
This retrospective, observational, single-centre study was conducted within the Emergency Department of Beaujon Hospital (Assistance Publique des hôpitaux de Paris (AP-HP)). The data collection period spanned from January 1, 2020, to December 31, 2021.
Demographic, clinical, discharge, transfer, WHWD and death data were extracted from the Orbis® medical data processing software (healthcare software, Dedalus, Florence, Italy) or by contacting the hospital if the department does not use Orbis®. Autonomy for activities of daily living was assessed using the KNAUS classification [11]. The Knaus classification was assigned retrospectively from the patients’ electronic medical records using the information available before the acute illness. When the available information was insufficient to reliably determine the patient’s pre-admission functional status, no assumption was made regarding the level of autonomy and the Knaus classification was recorded as missing. This classification system describes the level of care required by older and disabled individuals, and ranges from A (completely autonomous) to D (most dependent). Patients classified as KNAUS A or B represent individuals who do not experience problems with autonomy in their daily tasks. The APACHE II score [12], the presence of legal protection guardianship, and the presence of daily nurse visits or home care were determined from the patient’s medical record.
To be included, adult patients hospitalised after their ED visit must be admitted to the observation unit (OU) in association with a mention of a WHWD of the life-sustaining treatment in the ED medical record. Life-sustaining treatments included cardiopulmonary resuscitation, invasive mechanical ventilation, non-invasive ventilation, high-flow oxygen therapy, vasopressor or catecholamine infusion, renal replacement therapy, emergency surgery or endoscopy performed as life-saving procedures, and fluid resuscitation. A WHWD decision was considered only when an explicit medical decision to withhold or withdraw one or more of these treatments was documented in the patient’s electronic medical record. Treatments that were not initiated because they were not clinically indicated or because of logistical or resource-related reasons were not considered WHWD decisions. The time between arrival and the WHWD decision was extracted from Orbis®, and the comparison of patients was done according to the median time of this decision. Patients were classified into early and late WHWD groups according to the median time from ED arrival to WHWD decision. The terms “early” and “late” therefore refer to the relative timing within this cohort and should not be interpreted as absolute clinical definitions of early or delayed end-of-life decision-making. After WHWD, a follow-up was conducted first, for 7 and 28 days, up to the point of death or discharge from the hospital. And a follow-up was extended to 90 days and one year. The primary outcome was 28-day all-cause mortality after the WHWD decision.
According to French legislation and national recommendations, WHWD decisions should follow a collegial decision-making process whenever possible. This process involves the treating emergency physician together with other physicians involved in the patient’s care (such as intensivists or disease-specific specialists), while considering the patient’s wishes, advance directives, the opinion of the trusted person when designated, and the views of relatives whenever available. Intensivists were consulted whenever admission to the intensive care unit was considered or when additional expertise regarding the appropriateness of life-sustaining treatments was required. Their participation contributed to the collegial decision-making process.
Ethics
The work is conformed to the Declaration of Helsinki. The study was approved by the Scientific and Ethical Committee of Assistance Publique – Hopitaux de Paris (AP-HP) clinical data warehouse (IRB00011591). The database was authorised by the National Freedom and Informatics Commission (CNIL Number: 1980120). Assistance Publique – Hopitaux de Paris (AP-HP) clinical data warehouse initiative ensures patients’ information and consent regarding the approved studies through a transparency portal following European Regulation on data protection and authorisation (number 1980120) from the National Freedom and Informatics Commission. The need for informed consent was waived by the Scientific and Ethical Committee of Assistance Publique – Hopitaux de Paris (AP-HP) clinical data warehouse, because of the retrospective nature of the study. The study was retrospectively registered at ClinicalTrials.gov on 4 August 2025 (NCT07115719).
Statistical analysis
Continuous variables are expressed as median [interquartile range (IQR)]. Categorical variables are presented as a number (percentage). The normality of all variables was checked using the Shapiro-Wilk test. Group characteristics were compared using the t-test for normally distributed continuous variables, the Mann-Whitney test for non-normally distributed continuous variables, or the χ² test for categorical variables. Hazard ratio (HR) and 95% confidence interval (95% CI) for 7, 28, 90-day and 1-year all-cause mortality were calculated using a Cox proportional hazards regression model. Variables with a p < 0.10 on univariate analysis were entered into the Cox regression model to describe how the factors jointly impact survival. We also kept comorbidity as a key confounder. The optimal model selection involved finding the best compromise between the number of factors to keep and the accuracy of the Cox regression analysis. The model was adjusted with age, sexe, dementia, active cancer, KNAUS scale, APACHE II score and home care. The proportional hazards assumption was checked using statistical tests based on the scaled Schoenfeld residuals. Missing data were not imputed, and analyses were conducted using complete-case analysis. The number of available observations for each variable is reported in the tables. The variables included in the multivariable Cox regression model had less than 2% missing data, limiting the risk of bias related to missing values. Analyses were performed using R® software (R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria). A two-sided p-value of less than 0.05 was considered statistically significant.
Sample size
The estimated population is 140 patients, considering the 0.2% mortality rate in the emergency department, and the 35,000 annual visits at the Beaujon emergency department over two years.
Results
Between January 1, 2020, and December 31, 2021, there were 64,239 emergency department (ED) visits at Beaujon Hospital. Among these, 7,707 patients were hospitalised, and 354 cases involving a decision to limit or withdraw life-sustaining treatments (WHWD) were included in the study (Fig. 1).
Fig. 1.
Flow chart
The median age of included patients was 83 years (IQR 72–90), and 53% were women (Table 1). The median time between ED arrival and WHWD decision was 242 min (IQR 119–426). The median time to decision was approximately 2 h in the eWHWD (early WHWD) group and 7 h in the lWHWD (late WHWD) group. Sixty-one per cent of patients lived at home, while 39% were institutionalised. According to the Knaus classification, 7% of patients were class A or B, 46% class C, and 46% class D. Patients classified as Knaus D were more frequent in the eWHWD group (56% vs. 37%; p = 0.0005), while class C was more common in the lWHWD group (56% vs. 35%; p < 0.001).
Table 1.
Baseline characteristics
| Variable | Available data N (%) | Total | EWHWD (N = 176) |
LWHWD (N = 178) |
P-value | ||
|---|---|---|---|---|---|---|---|
| Age Median (Q1 - Q3) | 354 (100) | 83 (72–90) | 83 (75–92) | 81 (70–89) | 0.074 | ||
| Time between Arrival and WHWD decision | 354 (100) | 242 (119–426) | 117 (51–181) | 414 (311–696) | < 0.001 | ||
| Sexe | Men | 354 (100) | 165 (47) | 72 (41) | 93 (52) | 0.033 | |
| Women | 189 (53) | 104 (59) | 85 (48) | ||||
| Living place | Home | 354 (100) | 216 (61) | 101 (57) | 114 (64) | 0.20 | |
| Nursing Home | 138 (39) | 75 (43) | 63 (34) | 0.066 | |||
| Social isolation | 354 (100) | 55 (16) | 27 (15) | 28 (16) | 0.90 | ||
| Legal protection guardianship | 354 (100) | 59 (17) | 28 (16) | 31 (17) | 0.70 | ||
| Home care | 352 (99) | 250 (71) | 141 (81) | 109 (62) | < 0.001 | ||
| Knaus scale | A | 347 (98) | 1 (0) | 1(1) | 0 (0) | 0.50 | |
| B | 26 (7) | 14 (8) | 12 (7) | 0.65 | |||
| C | 159 (46) | 61 (35) | 98 (56) | < 0.001 | |||
| D | 161 (46) | 96 (56) | 65 (37) | < 0.001 | |||
| Medical history | Psychiatric disorders | 354 (100) | 19 (5) | 8 (5) | 11 (6) | 0.50 | |
| Dementia | 178 (50) | 100 (57) | 78 (44) | 0.014 | |||
| Neurodegenerative disorders | 51 (14) | 23 (13) | 28 (16) | 0.48 | |||
| Stroke | 50 (14) | 29 (16) | 21 (12) | 0.21 | |||
| Heart disease | 140 (40) | 75 (43) | 65 (37) | 0.22 | |||
| Respiratory disease | 57 (16) | 30 (17) | 27 (15) | 0.63 | |||
| Chronicle kidney disease | 44 (12) | 26 (15) | 18 (10) | 0.18 | |||
| Liver disease | 11 (3) | 4 (2) | 7 (4) | 0.37 | |||
| Diabetes | 89 (25) | 46 (26) | 43 (24) | 0.67 | |||
| Immunodefiency | 3 (1) | 1 (1) | 2 (1) | 1.00 | |||
| Active cancer | 93 (26) | 38 (22) | 55 (31) | 0.047 | |||
| ED presentation | Ambulance | 354 (100) | 282 (80) | 124 (70) | 158(89) | < 0.001 | |
| EMS | 69 (19) | 51 (29) | 18 (10) | < 0.001 | |||
| Self-presented | 3 (1) | 1 (1) | 2 (1) | 1.00 | |||
| Hospitalisation during last 6 months | 354 (100) | 203 (57) | 103 (59) | 100 (56) | 0.66 | ||
| Main complaint | Dyspnea | 354 (100) | 217 (61) | 118 (67) | 99 (56) | 0.027 | |
| Hypotension | 94 (27) | 66 (38) | 28 (16) | < 0.001 | |||
| Neurologic | 164 (46) | 103 (59) | 61 (34) | < 0.001 | |||
| Abdomen | 61 (17) | 18 (10) | 43 (24) | < 0.001 | |||
| Metabolic | 23 (6) | 11 (6) | 12 (7) | 0.85 | |||
| Traumatology | 20 (6) | 10 (6) | 10 (6) | 0.98 | |||
| Haemorrhage | 27 (8) | 9 (5) | 18 (10) | 0.079 | |||
| Acute pain | 40 (11) | 13 (7) | 27 (15) | 0.021 | |||
| Infection | 187 (53) | 95 (54) | 92 (52) | 0.67 | |||
| Toxicology | 5 (1) | 2 (2) | 3 (2) | 0.62 | |||
| Asthenia | 97 (27) | 31 (18) | 66 (37) | < 0.001 | |||
| Multiple | 310 (88) | 153 (87) | 157 (88) | 0.72 | |||
| Clinical parameters | Temperature | 350 (99) | 36.6 (36.1–37.2) | 36.6 (36–37.3) | 36.6 (36.2–37.2) | 0.83 | |
| Mean blood pressure | 83.5 (64–100) | 79 (60–97) | 85.5 (70–103) | 0.016 | |||
| Heart rate (bpm) | 94 (80–108) | 98 (81.5–110) | 93 (78–107) | 0.18 | |||
| Glasgow coma scale | 14 (9–15) | 12 (7–14) | 14 (13–15) | < 0.001 | |||
| Oxygen need (L/min) | 4 (0–9) | 5 (0–12) | 3 (0–6) | 0.002 | |||
| Laboratory tests | PaO2 (mmHg) | 162 (46) | 82 (67–119) | 91 (70–130) | 77 (66–108) | 0.002 | |
| pH | 162 (46) | 7.36 (7.28–7.43) | 7.34 (7.26–7.41) | 7.37 (7.3–7.44) | 0.094 | ||
| Sodium (mmol/L) | 334 (94) | 138 (134–142) | 139 (134–142) | 138 (134–142) | 0.32 | ||
| Potassium (mmol/L) | 334 (94) | 4.4 (3.9–5.2) | 4.5 (3.9–5.25) | 4.4 (3.9–5) | 0.81 | ||
| Creatinine (µmol/L) | 334 (94) | 115 (72–199) | 131 (74–209) | 105.5 (71–176) | 0.74 | ||
| Leucocytes (/mm3) | 330 (93) | 12.6 (8.6–18.5) | 13.8 (9.3–20) | 11.9 (8.3–17.6) | 0.079 | ||
| Haemoglobin (g/dl) | 330 (93) | 11.7 (9.8–13.3) | 11.9 (10.1–13.4) | 11.7 (9.7–13.2) | 0.24 | ||
| APACHE II Median (Q1 - Q3) | 334 (94) | 16 (13–21) | 18 (14.5–22.5) | 14.5 (12–18) | < 0.001 | ||
APACHE I: Acute Physiologic Assessment and Chronic Health Evaluation Scoring System II
EMS: emergency medical service
GIR: groupe iso-ressource
WHWD: Withholding/Withdrawing
Regarding medical history, dementia was present in 50% of patients, heart disease in 40%, and active cancer in 26%. Dementia was more prevalent in the eWHWD group (57% vs. 44%; p = 0.014), while active cancers were more frequent in the lWHWD group (19% vs. 10%; p = 0.026). Seventy-seven per cent of patients were transported by ambulance, and 19% by Emergency medical service (EMS). Medicalised transport was more common in the eWHWD group (29% vs. 10%; p < 0.001). The most frequent reasons for ED admission were respiratory (61%) and neurological (46%). Patients in the eWHWD group more often presented with respiratory (56% vs. 40%; p = 0.004), circulatory (39% vs. 23%; p = 0.001), and neurological failures (31% vs. 15%; p = 0.0003). Clinically, early WHWD patients had lower mean arterial pressure and Glasgow scores and required more oxygen. The median APACHE II score for all patients was 16. The score was significantly higher in eWHWD patients (median 18 vs. 14.5; p < 0.001).
The principle of collegiality in WHWD decisions was followed in 84% of cases, with no difference between groups (Table 2). Intensivists were the most frequent medical partners (45%). Because 77% of patients were considered unable to express their wishes, WHWD decisions were predominantly made by physicians within the legally required collegial process, while taking into account advance directives, the opinion of the trusted person, relatives, and other healthcare professionals whenever available. Detailed information regarding the participants involved in the decision-making process is presented in Table 2. The eWHWD group had a lower proportion of patients considered capable of expressing their wishes (14% vs. 33%; p < 0.001), but overall patient involvement was low (7%) and not significantly different between groups. Advance directives were recorded in only 3% of cases, and a trusted person was identified in 13%. There was no significant difference between groups regarding advance directive presence, identification of a trusted person, or physician involvement. The main reasons cited for WHWD were disproportionate care (86%), poor anticipated quality of life (44%), and terminal prognosis (43%). Poor quality of life was oftentimes cited in eWHWD decisions (50% vs. 37%; p = 0.014). Supportive care, including oxygen therapy and fluid resuscitation, was more frequently administered in the eWHWD group. These therapies and non-invasive ventilation were also more often limited in that group. WHWD involved limitation of treatment in 88% of cases and withdrawal in 12%. Only 7% of patients were evaluated by the hospital palliative care team, with no significant difference between groups. Pain assessment was recorded in 81% of patients, with no group differences. Comfort care measures included sedation with Midazolam in 21% of patients and analgesia with Morphine in 29%. Both were used more frequently in the eWHWD group: Midazolam (27% vs. 15%; p = 0.0009) and Morphine (38% vs. 21%; p = 0.008).
Table 2.
End of life decision process and treatment
| Variable | Available data N (%) | Total | EWHWD (N = 176) |
LWHWD (N = 178) |
P-value | |
|---|---|---|---|---|---|---|
| Collegial WHWD process | 354 (100) | 297 (84) | 151 (86) | 146 (82) | 0.33 | |
| Participant | None | 354 (100) | 54 (15) | 23 (13) | 31 (17) | 0.26 |
| Many | 46 (13) | 25 (14) | 21 (12) | 0.50 | ||
| Intensivist | 160 (45) | 80 (45) | 80 (45) | 0.92 | ||
| Gerontologist | 18 (5) | 12 (7) | 6 (3) | 0.14 | ||
| Oncologist | 25 (7) | 8 (5) | 17 (10) | 0.066 | ||
| Surgeon | 13 (4) | 4 (2) | 9 (5) | 0.16 | ||
| Orther ED physician | 29 (8) | 21 (12) | 8 (4) | 0.011 | ||
| Other | 12 (3) | 4 (2) | 8 (4) | 0.25 | ||
| Patient able to decide | 354 (100) | 83 (23) | 25 (14) | 58 (33) | < 0.001 | |
| Criteria for assessing patient incapacity to make a decision | Neurocognitive disorders | 354 (100) | 61 (17) | 31 (18) | 30 (17) | 0.85 |
| Confusion ou coma | 89 (25) | 48 (27) | 41 (23) | 0.36 | ||
| Both | 119 (34) | 69 (39) | 50 (28) | 0.027 | ||
| Patient directly involved in making decisions | 354 (100) | 25 (7) | 11 (6) | 14 (8) | 0.55 | |
| Advanced health care directives | 354 (100) | 12 (3) | 7 (4) | 5 (3) | 0.54 | |
| Known surrogate | 354 (100) | 45 (13) | 23 (13) | 22 (12) | 0.84 | |
| Involvement of the primary care physician in decision-making | 354 (100) | 7 (2) | 3 (2) | 4 (2) | 1.00 | |
| Involvement of the specialist in decision-making* | 354 (100) | 55 (16) | 22 (13) | 33 (19) | 0.12 | |
| Involvement of the paramedical team in decision-making | 354 (100) | 9 (3) | 4 (2) | 5 (3) | 1.00 | |
| Assessment of autonomy and quality of life | 354 (100) | 332 (94) | 168 (95) | 164 (92) | 0.20 | |
| Reason of WHWD | Predictable short-term mortality linked to an incurable chronic disease | 354 (100) | 151 (43) | 66 (38) | 85 (48) | 0.051 |
| Irreversibility of the acute condition. | 94 (27) | 43 (24) | 51 (29) | 0.37 | ||
| Current or future quality of life | 154 (44) | 88 (50) | 66 (37) | 0.014 | ||
| Care considered too invasive given the risk-benefit ratio | 303 (86) | 148 (84) | 155 (87) | 0.42 | ||
| Family request | 17 (5) | 12 (7) | 5 (3) | 0.42 | ||
| none | 44 (12) | 25 (14) | 19 (11) | 0.31 | ||
| Life-sustaining treatments initiated before WHWD | CPR | 354 (100) | 9 (3) | 4 (2) | 5 (3) | 1.00 |
| Orotracheal intubation and mechanical ventilation | 4 (1) | 2 (1) | 2 (1) | 1.00 | ||
| Non-invasive ventilation | 28 (8) | 11 (6) | 17 (10) | 0.25 | ||
| High-flow oxygen therapy | 115 (32) | 66 (38) | 49 (28) | 0.045 | ||
| Catecholamines infusion | 4 (1) | 2 (1) | 2 (1) | 1.00 | ||
| Fluid resuscitation | 105 (30) | 62 (35) | 43 (24) | 0.023 | ||
| Renal replacement therapy | 0 (0) | 0 (0) | 0 (0) | NA | ||
| Emergency surgery or endoscopy | 4 (1) | 2 (1) | 2 (1) | 1.00 | ||
| Total | 201 (57) | 112 (64) | 89 (50) | 0.010 | ||
| Other treatments initiated before WHWD | Low-flow oxygen therapy | 354 (100) | 241 (68) | 125 (71) | 116 (65) | 0.24 |
| Peripheral venous catheter | 347 (98) | 169 (96) | 178 (100) | 0.007 | ||
| Venous blood tests | 333 (96) | 155 (88) | 178 (100) | < 0.0001 | ||
| Arterial blood tests | 166 (47) | 75 (43) | 91 (51) | 0.11 | ||
| Subcutaneous infusion | 4 (1) | 4 (2) | 0 (0) | 0.060 | ||
| Antibiotherapy | 229 (65) | 117 (66) | 112 (63) | 0.48 | ||
| Anticoagulation | 104 (29) | 42 (24) | 62 (35) | 0.024 | ||
| Transfusion | 24 (7) | 7 (4) | 17 (10) | 0.088 | ||
| Life-sustaining treatments stopped or not initiated after WHWD | CPR | 354 (100) | 354 (100) | 176 (100) | 178 (100) | 1.00 |
| Orotracheal intubation and mechanical ventilation | 353 (99) | 176 (100) | 177 (99) | 1.00 | ||
| Non-invasive ventilation | 161 (45) | 94 (53) | 67 (38) | 0.003 | ||
| High-flow oxygen therapy | 49 (14) | 33 (19) | 16 (9) | 0.008 | ||
| Catecholamines infusion | 327 (92) | 167 (95) | 160 (90) | 0.076 | ||
| Fluid resuscitation | 45 (13) | 29 (16) | 16 (9) | 0.034 | ||
| Renal replacement therapy | 332 (94) | 167 (95) | 165 (93) | 0.39 | ||
| Emergency surgery or endoscopy | 269 (76) | 138 (78) | 131 (74) | 0.29 | ||
| Other treatments stopped or not initiated after WHWD | Low-flow oxygen therapy | 354 (100) | 13 (4) | 9 (5) | 4 (2) | 0.15 |
| Peripheral venous catheter | 11 (3) | 5 (3) | 6 (3) | 0.77 | ||
| Venous blood tests | 43 (12) | 27 (15) | 16 (9) | 0.067 | ||
| Arterial blood tests | 63 (18) | 35 (20) | 28 (16) | 0.31 | ||
| Subcutaneous infusion | 9 (3) | 6 (3) | 3 (2) | 0.34 | ||
| Antibiotherapy | 18 (5) | 11 (6) | 7 (4) | 0.32 | ||
| Anticoagulation | 31 (9) | 21 (12) | 10 (6) | 0.036 | ||
| Transfusion | 51 (14) | 31 (18) | 20 (11) | 0.088 | ||
| Withdrawal of life-sustaining treatments | 354 (100) | 44 (12) | 26 (15) | 18 (10) | 0.18 | |
| Patient previously followed by the hospital palliative care team** | 354 (100) | 27 (8) | 12 (7) | 15 (8) | 0.57 | |
| Patient assessed by the palliative care | 354 (100) | 25 (7) | 12 (7) | 13 (7) | 0.86 | |
| Assessment of physical or mental pain | 354 (100) | 287 (81) | 142 (81) | 145 (81) | 0.85 | |
| Palliative care initiated | Midazolam | 354 (100) | 74 (21) | 47 (27) | 27 (15) | 0.008 |
| Morphin | 104 (29) | 66 (38) | 38 (21) | 0.0009 | ||
| Scopolamin | 29 (8) | 19 (11) | 10 (6) | 0.078 | ||
CPR: Cardiopulmonary Resuscitation; ED: Emergency department; WHWD: Withholding/Withdrawing
Data are presented as median (interquartile range) for continuous data and as number (percentage) for categorical data. The P-value to the right represent the test for trend determined by either Wilcoxon rank-sum (continuous data) or chisquare [categorical data]
* Specialist refers to a physician with expertise in the patient’s underlying disease (e.g., geriatrician, oncologist, cardiologist, neurologist or other specialty physician). Intensivists and emergency physicians are reported separately** Patient previously followed by the hospital palliative care team refers to patients who had already been evaluated by the hospital palliative care team before the index emergency department visit, either during a previous hospitalization or as outpatients
Regarding outcomes (Table 3), 31% of patients died in the ED, 32% died in downstream units, and 35% were discharged alive. ED mortality was higher in the eWHWD group (39% vs. 22%; p = 0.0005), while downstream unit mortality was higher in the lWHWD group (38% vs. 27%; p = 0.030). Discharge rates were similar between groups. The mortality of the whole population at 7, 28, 90-day and 1-year was 51%, 70%, 82% and 90% respectively. At 7 days, 58% of eWHWD patients had died, compared to 44% in the late group (p = 0.008). Seventy-six per cent of the overall eWHWD patients died within 28 days, compared to 65% in the lWHWD group (p = 0.022). There was no significant difference at 90 days or one year.
Table 3.
Outcomes
| Variable | Available data N (%) | Total | EWHWD (N = 176) |
LWHWD (N = 178) |
P-value | |
|---|---|---|---|---|---|---|
| ED mortality | 353 (99) | 108 (31) | 69 (39) | 39 (22) | < 0.001 | |
| Post-ED disposition | Acute geriatric ward | 245 (69) | 119 (49) | 57 (53) | 62 (45) | 0.19 |
| Internal medicine ward | 33 (13) | 14 (13) | 19 (14) | 0.88 | ||
| Home / Nursing Home | 33 (13) | 18 (17) | 15 (11) | 0.18 | ||
| Palliative care | 14 (6) | 4 (4) | 10 (7) | 0.24 | ||
| Other | 24 (10) | 6 (6) | 18 (13) | 0.052 | ||
| Mortality in admission ward | 349 (98) | 114 (33) | 48 (27) | 66 (38) | 0.030 | |
| Post ward disposition | Home / Nursing Home | 101 (28) | 79 (79) | 20 (47) | 24 (41) | 0.61 |
| Palliative care | 17 (17) | 5 (12) | 12 (21) | 0.21 | ||
| Other | 4 (4) | 0 (0) | 4 (7) | 0.13 | ||
| Discharged Alive | 347 (98) | 124 (32) | 56 (32) | 68 (40) | 0.12 | |
| 7–days mortality | 347 (98) | 177 (51) | 100 (58) | 77 (44) | 0.008 | |
| 28–days mortality | 344 (97) | 242 (70) | 130 (76) | 112 (65) | 0.022 | |
| 90–days mortality | 329 (93) | 270 (82) | 143 (86) | 127 (78) | 0.052 | |
| 1–year mortality | 326 (92) | 295 (90) | 152 (92) | 143 (89) | 0.31 | |
ED: Emergency department; WHWD: Withholding/Withdrawing
Data are presented as median (interquartile range) for continuous data and as number (percentage) for categorical data. The P-value to the right represent the test for trend determined by either Wilcoxon rank-sum (continuous data) or chisquare [categorical data]
The relationship between the timing of a WHWD decision and survival favoured the late group decision regarding the short-term prognostic (Fig. 2). The unadjusted hazard ratio (HR) was lower in lWHWD for 7- and 28-day survival, respectively HR = 0.61; 95% CI (0.45 to 0.82) and 0.66; 95% CI (0.51 to 0.85). The adjusted HR for short-term prognostic stay in favour of late decision group. For long-term prognostic (90 days and 1 year survival), there was no difference between the 2 groups according to the unadjusted HR (Fig. 2). But after adjustment with the age, sex, dementia, active cancer, KNAUS scale, APACHE II score, and home care, the HR at 90-day and one year was significantly in favour of lWHWD group, respectively adjusted HR = 0.76; 95% CI (0.72 to 0.81) and 0.86; 95% CI (0.81 to 0.91).
Fig. 2.
Short & long-term mortality according to the time of Withholding/Withdrawing (WHWD). Unadjusted and adjusted HRs and 95% confidence intervals for the association between mortality and the WHWD timing
Discussion
This study investigated the outcomes of early versus late decisions to limit or withdraw life-sustaining treatments (WHWD) in emergency department (ED) patients. Our findings reveal significant differences in patient characteristics, care pathways, and short-term mortality between early WHWD (eWHWD) and late WHWD (lWHWD) groups, underscoring the complexity of end-of-life (EOL) decision-making in the emergency setting.
Patient profiles and clinical context
Patients in the eWHWD group were more clinically fragile, with significantly higher APACHE II scores, lower Knaus functional status, and more frequent cognitive impairment. These characteristics likely influenced earlier recognition of poor prognosis by emergency physicians. The predominance of respiratory, circulatory, and neurological failures in the eWHWD group also reflects more acute presentations requiring urgent evaluation and decision-making.
These observations are consistent with previous research, which has found that early decisions to forgo life-sustaining treatments often correlate with poor functional baseline and severe acute illness upon ED arrival [6, 13]. Similar profiles have been described in intensive care settings, where early WHWD is often prompted by factors like age, cognitive impairment, and comorbidity burden [14, 15].
Interestingly, active cancer was more frequent in the lWHWD group, suggesting that patients with identifiable chronic disease trajectories may prompt a more deliberate and perhaps collegial process for limiting treatments after stabilisation or further assessment. Conversely, eWHWD patients were often more serious, necessitating quicker decisions in a highly constrained setting.
Timing and mortality outcomes
The median delay to WHWD decision was significantly shorter in the eWHWD group (approximately 2 h) compared to the lWHWD group (7 h). This timing was closely associated with early mortality. Specifically, eWHWD patients had significantly higher mortality at 7 and 28 days. While this may initially suggest a negative impact of earlier decisions, it likely reflects the severity of illness and the acute context in which these decisions were made, rather than premature decision-making. These results align with findings by Hua et al. [16, 17] who reported higher early mortality among patients with rapidly made WHWD decisions in intensive care units.
Notably, after adjusting for confounders, late WHWD remained independently associated with improved survival even at 90 days and 1 year, suggesting that, when possible, a delayed and more comprehensive decision-making process might be beneficial. However, it is critical to note that the latter group may benefit from more complete diagnostic evaluations, discussions with families, and collegial processes, all of which may improve the quality of decision-making rather than merely delay it [18, 19].
Decision-making practices and ethical considerations
Despite legal and professional guidelines emphasising patient involvement, our study revealed a low patient involvement rate (7%) in WHWD decisions, with only 3% having documented advance directives. These findings underscore persistent challenges in anticipatory EOL planning and communication, especially in emergency settings where many patients arrive with impaired consciousness or no clear medical history [20]. The low uptake of advance directives is also echoed in broader French studies [21] and may reflect societal, cultural, or systemic barriers to proactive care planning.
Moreover, while the collegiality principle was applied in 84% of cases consistent with French legal requirements consultation with palliative care teams remained rare (7%). This underutilization has been previously noted [22], and it may stem from the limited availability of palliative care professionals in EDs, especially outside regular working hours. The higher use of comfort measures (e.g., midazolam and morphine) in the eWHWD group reflects a greater emphasis on symptom control in imminently dying patients, but also points to potential variability in how comfort care is applied depending on timing and patient acuity.
Implications for practice
Our findings highlight the need for better EOL care integration into ED protocols. The stark contrast between the rapid pace of emergency care and the reflective process required for WHWD decisions creates systemic tension. To address this, advance care planning must be strengthened in upstream care settings (e.g., general practice, long-term care facilities). This echoes recommendations by international bodies, including the American College of Emergency Physicians and European Resuscitation Council, which emphasise early goals-of-care discussions in chronic disease management [23–25].
Enhanced training for ED staff in palliative care principles, earlier identification of patients with chronic life-limiting illnesses, and broader use of decision aids such as the one introduced in France in 2020 could help align care more closely with patient values [9]. The limited documentation of advance directives and designated trusted persons in our study highlights a gap between national policy and actual clinical practice, underscoring the need for targeted interventions.
Limitations
This study has several limitations. First, its retrospective, single-centre design limits generalizability. Second, we relied on electronic medical records, which may incompletely capture the nuances of clinical judgment and discussions with patients or families. Third, while the timing of WHWD was used to categorise patients, it may also reflect institutional workflow factors not accounted for in our analysis. Finally, unmeasured confounding cannot be ruled out, despite multivariable adjustment.
Conclusion
In this retrospective study, early WHWD decisions were associated with more severe clinical presentations and higher short-term mortality, while late rulings were linked to better adjusted long-term survival. These findings emphasise the need for thoughtful, patient-centred EOL decision-making in EDs and better anticipatory planning across the healthcare continuum. Strengthening palliative care integration and improving communication tools may help reconcile the realities of emergency care with the goals of dignified end-of-life experiences.
Acknowledgements
Not applicable.
Abbreviations
- CI
Confidence Interval
- ED
Emergency Department
- EOL
End-Of-Life
- HR
Hazard Ratio
- IQR.
Interquartile Range
- WHWD
Withholding or withdrawing
Author contributions
JS & Prabakar VA: did the Data curation; JS, Pradeebane VA & Prabakar VA did the Formal analysis and the Statical analysis. Prabakar VA did the validation and supervision of the study. Prabakar VA prepared the Fig. 1. Pradeebane VA prepared the Fig. 2. Prabakar VA wrote the main manuscriptAll authors reviewed the manuscript.
Funding
The authors received no funding from an external source.
Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
The work conformed to the Declaration of Helsinki. The informed consent was waived by the institutional review board called Assistance Publique des Hôpitaux de Paris Clinical Data Warehouse (AP-HP CDW) Scientific and Ethics Committee (Conseil Scientifique et Ethique; IRB number: 00011591).
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Institut national de la statistique et des études économiques. Évolution du nombre de décès entre le 1er janvier et le 31 décembre 2022 – Nombre de décès quotidiens | Insee. 2023. https://www.insee.fr/fr/statistiques/6206305?sommaire=4487854#tableau-figure2. Accessed 3 Sep 2024.
- 2.Inspection Générale des Affaires Sociales. La mort à l’hôpital.|IGAS. 2009. https://www.igas.gouv.fr/La-mort-a-l-hopital.html. Accessed 3 Sep 2024.
- 3.Gomes B, Calanzani N, Gysels M, Hall S, Higginson IJ. Heterogeneity and changes in preferences for dying at home: a systematic review. BMC Palliat Care. 2013;12:7. 10.1186/1472-684X-12-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Commission de réflexion sur la fin de vie en France. Rapport a François hollande, président de la république française; 2012.
- 5.Rothmann C, Evrard D. La mort aux urgences. J européen des urgences. 2005;18:3–9. [Google Scholar]
- 6.Le Conte P, Riochet D, Batard E, Volteau C, Giraudeau B, Arnaudet I, et al. Death in emergency departments: a multicenter cross-sectional survey with analysis of withholding and withdrawing life support. Intensive Care Med. 2010;36:765–72. 10.1007/s00134-010-1800-1. [DOI] [PubMed] [Google Scholar]
- 7.Le Conte P, Batard E, Pinaud V, Evain Y, Potel G. Décisions de limitation ou d’arrêt des thérapeutiques actives dans les services d’urgence. Réanimation. 2008;17:802–6. 10.1016/j.reaurg.2008.09.005. [DOI] [Google Scholar]
- 8.Article R712-65. - Code de la santé publique - Légifrance. https://www.legifrance.gouv.fr/codes/article_lc/LEGIARTI000006802859. Accessed 10 Sep 2024.
- 9.Reignier J, Feral-Pierssens A-L, Boulain T, Carpentier F, Le Borgne P, Del Nista D, et al. Withholding and withdrawing life-support in adults in emergency care: joint position paper from the French Intensive Care Society and French Society of Emergency Medicine. Ann Intensive Care. 2019;9:105. 10.1186/s13613-019-0579-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.SFMU SFAP. Procédure d’aide à la décision de Limitation et Arrêt des thérapeutiques lors d’une prise en charge en urgence. Fiche Mémo. 2020.
- 11.Knaus WA, Zimmerman JE, Wagner DP, Draper EA, Lawrence DE. APACHE-acute physiology and chronic health evaluation: a physiologically based classification system. Crit Care Med. 1981;9:591–7. 10.1097/00003246-198108000-00008. [DOI] [PubMed] [Google Scholar]
- 12.Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE II: a severity of disease classification system. Crit Care Med. 1985;13:818–29. [PubMed] [Google Scholar]
- 13.Azoulay E, Pochard F, Garrouste-Orgeas M, Moreau D, Montesino L, Adrie C, et al. Decisions to forgo life-sustaining therapy in ICU patients independently predict hospital death. Intensive Care Med. 2003;29:1895–901. 10.1007/s00134-003-1989-3. [DOI] [PubMed] [Google Scholar]
- 14.Sprung CL, Cohen SL, Sjokvist P, Baras M, Bulow H-H, Hovilehto S, et al. End-of-life practices in European intensive care units: the Ethicus Study. JAMA. 2003;290:790–7. 10.1001/jama.290.6.790. [DOI] [PubMed] [Google Scholar]
- 15.Chambers N, Curtis JR, Rubenfeld GD. Managing Death in the ICU: The Transition from Cure to Comfort. Crit Care. 2002. 10.1186/CC1479. [DOI] [Google Scholar]
- 16.Hua M, Halpern SD, Gabler NB, Wunsch H. Effect of ICU Strain on Timing of Limitations in Life-Sustaining Therapy and Death. Intensive Care Med. 2016;42:987–94. 10.1007/s00134-016-4240-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Azoulay E, Metnitz B, Sprung CL, Timsit J-F, Lemaire F, Bauer P, et al. End-of-life practices in 282 intensive care units: data from the SAPS 3 database. Intensive Care Med. 2009;35:623–30. 10.1007/s00134-008-1310-6. [DOI] [PubMed] [Google Scholar]
- 18.Bosslet GT, Pope TM, Rubenfeld GD, Lo B, Truog RD, Rushton CH, et al. An Official ATS/AACN/ACCP/ESICM/SCCM Policy Statement: Responding to Requests for Potentially Inappropriate Treatments in Intensive Care Units. Am J Respir Crit Care Med. 2015;191:1318–30. 10.1164/rccm.201505-0924ST. [DOI] [PubMed] [Google Scholar]
- 19.Ferrand E, Robert R, Ingrand P, Lemaire F, French LATAREA, Group. Withholding and withdrawal of life support in intensive-care units in France: a prospective survey. French LATAREA Group. Lancet. 2001;357:9–14. 10.1016/s0140-6736(00)03564-9. [DOI] [PubMed] [Google Scholar]
- 20.Le Guen J, Boumendil A, Guidet B, Corvol A, Saint-Jean O, Somme D. Are elderly patients’ opinions sought before admission to an intensive care unit? Results of the ICE-CUB study. Age Ageing. 2016;45:303–9. 10.1093/ageing/afv191. [DOI] [PubMed] [Google Scholar]
- 21.Auriemma CL, Nguyen CA, Bronheim R, Kent S, Nadiger S, Pardo D, et al. Stability of End-of-Life Preferences: A Systematic Review of the Evidence. JAMA Intern Med. 2014;174:1085–92. 10.1001/jamainternmed.2014.1183. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Baumann A, Audibert G, Claudot F, Puybasset L. Ethics review: End of life legislation – the French model. Crit Care. 2009;13:204. 10.1186/cc7148. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Sudore RL, Fried TR. Redefining the Planning in Advance Care Planning: Preparing for End-of-Life Decision Making. Ann Intern Med. 2010;153:256–61. 10.7326/0003-4819-153-4-201008170-00008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Grudzen CR, Richardson LD, Hopper SS, Ortiz JM, Whang C, Morrison RS. Does palliative care have a future in the emergency department? Discussions with attending emergency physicians. J Pain Symptom Manage. 2012;43:1–9. 10.1016/j.jpainsymman.2011.03.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Grudzen CR, Richardson LD, Morrison M, Cho E, Morrison RS. Palliative care needs of seriously ill, older adults presenting to the emergency department. Acad Emerg Med. 2010;17:1253–7. 10.1111/j.1553-2712.2010.00907.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.