Skip to main content

Table 2 Summary of findings of intervention studies (n=18)

From: Improving hospital-based end of life care processes and outcomes: a systematic review of research output, quality and effectiveness

Reference, Country Design

Sample & Setting

Inclusion & Exclusion

Intervention

Outcome measures and time points

Findings

End of life processes

Ahronheim et al. 2000 [36]

USA

RCT

Sample: 99

Response rate:

Setting: Hospital

Diagnostic group: Advanced dementia

Mean age: 84.8 (63–100)

Inclusion: FAST ≥6d, stable at least one month.

Type: Palliative care consultation, discussion with primary care team, family meeting and care recommendations

Control: primary care team only

Primary: Mortality, site discharge, length of stay, readmissions DNR orders, systemic antibiotics

Secondary: Decision to forgo treatments, decision to adopt palliative care plan.

Follow-up: Discharge or in-hospital death

No differences in mortality, readmissions, length of stay.

Intervention patients discharged more often with palliative care plan; and more decisions to forgo treatments

Bailey et al. 2014 [19]

USA

Stepped wedge

Sample: 6066

Response Rate:

Setting: 6 VAMSs

Diagnostic group: Veterans who died in centres – any illness

Mean age: 71.2 (SD=12)

Staff: 1621 physicians, nurses, residents, allied health, pharmacy, mental health, admin and other

Patients: all who died in 6 year period; and subset all who died in 12 month pre and 12 month post-intervention.

Type: Site visits, staff education, decision support tool (CCOS), follow-up consultations during 4 month training period

Control: Retrospective chart audit

Primary: % with: Opioid order, DNR order, Location death; Nasogentric tube,

IV line infusing, Restraints

Secondary: Antipsychotic (order, given)

Opioid given, Death rattle medication

Benzodiazepine (order, given),

PC consult, Pastoral visit, Advance directive, Sublingual admin

Follow-up: 12 month-post intervention

Improved orders for opioid, antipsychotic medication, benzodiazepines, death rattle medication, and advance directives.

Intervention effects were modest but statistically significant.

Cugliari et al. 1995 [20]

USA

CCT

Sample: 419

Setting: 2 hospitals

Diagnostic group: mixed

Mean age: 53 (SD=16.6)

Inclusion criteria: >18 years, planned admission

Exclusion criteria: emergency, obstetrics, psychiatric admission, insufficient English

Type: Written information + 18 min videotape of interviews with adults about experience with advance care planning; and instructions on completing ADs.

Control: Written information (usual care) on law on advance directives and health care proxy.

Recall of information

Attitudes about ADs and decision to complete proxy

No difference between the groups in recall, understanding of proxy form, completion of form, or perceived importance of ADs.

No difference in intention to complete form later.

El-Jawhari et al. 2010 [21]

USA

RCT

Sample: 150

Setting: Inpatient internal medicine in two hospitals

Diagnostic group: Mixed

Mean age: 76 (SD=13)

Inclusion criteria: >60 years, ability to provide consent; communicate in English; advanced cancer, heart failure, COPD, other advanced illness or prognosis <12 months.

Type: 3 minute video on CPR and intubation

Preferences for CPR and intubation communicated to 1+ physician

Control: Asked for CPR and intubation preferences only

Primary: CPR and intubation preferences

Secondary: CPR/intubation orders, documented discussions with providers, patient knowledge CPR/intubation

Follow-up: Post-video

Intervention patients more likely not to want CPR and intubation; have documented orders for CPR and intubation; documented discussions of preferences; and higher mean knowledge scores

Grimaldo et al. 2001 [22]

USA

RCT

Sample: 185

Setting: Pre-operative evaluation clinic

Diagnostic group: scheduled for elective surgery

Median age: 73.3

Inclusion criteria: English speaking; ≥65 years old; scheduled for elective surgery or overnight stay

Type: Usual care + 5–10 minute anaesthetist led information session focusing on the importance of patient-proxy communication about EoL care. Asked if had an AD and offered DPOA paperwork.

Control: Standard pre-operative screening and counselling.

Primary: Increased dialogue between pt and proxies in clinic

Follow-up: Pre- and post-operatively

Intervention group:

Greater proportion who had DPOA at post-operative.

More changed response no to yes about DPOA discussions.

More likely to have discussions about medical care with proxy

Jacobsen et al. 2011 [23]

USA

CCT

Sample: 899

Setting: Two general medical wards

Diagnostic group: general medicine, seriously ill

Mean age: 63.5

Inclusion criteria: stable and unstable seriously

ill patients admitted to general medicine ward.

Type:

Stable pts: patient, family and provider meet to assess knowledge, preferences and experiences.

Unstable pts: provider an family meet to recommend about treatment and prognostication.

Control: Usual care not specified

Primary:

% pts admitted full code without discussion documented

% pts admitted full code with discussion documented

% pts ACP discussion and order of WLST documented

Follow-up: At discharge (record review)

Intervention ward significant better across all outcomes.

Meier et al. 1996 [24]

USA

RCT

Sample: 190

Setting: Geriatric inpatient unit

Diagnostic group: Elderly

Mean age:

Inclusion criteria: >65 years, met Medicare Prospective Payment guidelines, complex care problems

Type: Counselling about advance directives and provided opportunity to complete health care proxy, charting of advance directives and proxy forms.

Control: Usual care

Primary: Documentation: (a) copy of proxy form; (b) patient proxy recorded; (c) advance directive notation

Secondary: Self-reported quality of care

Follow-up: One month

Intervention more likely to complete new proxy or have previously completed proxy identified.

Nicolasora 2006 [25]

RCT

Sample: 297

Setting: Medical ward of teaching hospital

Diagnostic group: Mixed

Median age: 65

Exclusion criteria: cardiac catheterization or admission to ICU; documented dementia or delirium (control); of judged by physician to have impaired cognitive function (intervention)

Type: Script about CPR, mechanical ventilation delivered by physician; asked about CPR status; changes communicated to physician; wishes to prepare ADs and assisted with completing ADs

Control: Medical records surveyed for in-hospital outcomes and AD status

Primary: Willingness to listen to script; acceptability of information; frequency of changing or choosing CPR status; rate of completion of ADs

Secondary:

Follow-up: Post-script

98% willing to discuss CPR and 82% useful

36% intervention had documented code status at admission (vs 34% control)

92% intervention clarified preferences

13/102 without ADs created one after intervention (vs 1/128 control)

Reilly et al. 1995 [26]

USA

ITS

Sample: 1780

Setting: Hospital

Diagnostic group: mixed

Inclusion criteria: Not reported

Patients and staff

Type: Education phase (Reminders, education and feedback to providers); Intervention phase (Standardised AD documentation form placed in medical charts)

Control: Not specified

Primary: Frequency and content of ADs documented in charts;

Secondary: Pt attitudes about ADs; Provider attitudes about ADs

Follow-up: 10 time points (4 control, 3 education and 3 intervention)

Proportion ADs highest during intervention phase (63% vs 23%E vs 25% C)

Frequency of ICU ADs and CPRs greater during intervention phase

87% ADs concordant with pt preferences

Sampson et al. 2011 [27]

UK

RCT (pilot)

Sample: 33

Response Rate:

Setting: Two acute medical wards

Diagnostic group: Advanced primary dementia

Mean age: 59 (SD=13)

Inclusion criteria: Unplanned admission for treatable acute illness; presence of surrogate that was able to provide informed consent. (FAST stage 6d or worse)

Type: Component 1: 30 minute pt assessment and formulation of management plan.

Component 2: Consultations with carers to discuss pts current situation and to provide education about dementia, ACP and PC (these occurred in hospital and in the community post d/c).

Control:

Primary: No. carers with ACP

Secondary: Carer measures: Distress (KD-10); health status QoL (EQ-5D); Decision making (DCS); decisional satisfaction (DSI); Anger (SAS); Life satisfaction (LSQ); EoL satisfaction (SWC-EOLCD).

Patient measures: Pain, distress

Follow-up: Patient- 6 weeks; 6 months. Carer-3 months after death.

Seven ACP’s were made in the I group

Attrition precluded statistical comparison of groups

Teno et al. 1997 [25]

USA

Sample: 4804

Response Rate:

Setting: Five teaching hospitals.

Diagnostic group: acute respiratory failure; multiple organ failure with sepsis or malignancy; coma; COPD; CHF; cirrhosis; metastatic colon cancer; NSCLC

Mean age: 62 (SD=16)

Inclusion criteria: Presence of diagnosis and ≥18 years of age.

Exclusion criteria: d/c or died <48 hours; admitted scheduled d/c <72 hours; non-English speaking; psychiatric ward; AIDS diagnosis; pregnancy; acute burns, trauma.:

Type: Connors et al. + PSDA mandated pt education and documentation of AD’s

Control: Usual care not specified

Primary:

Awareness, completion and documentation of AD’s; effectiveness of AD’s on decision-making about resuscitation; Physicians role in using AD’s; Surrogates perspectives of AD’s

Secondary:

Follow-up: 6 months.

Increase in AD documentation in the Post/I group, otherwise the intervention did not affect the pt familiarity with or the use of AD’s.

End of life outcomes

The SUPPORT Principal Investigators 1995 [30]

(SUPPORT)

USA

Sample: 4804

Setting: Five teaching hospitals.

Diagnostic group: acute respiratory failure; multiple organ failure with sepsis or malignancy; coma; COPD; CHF; cirrhosis; metastatic colon cancer; NSCLC

Inclusion criteria: Presence of diagnosis and ≥18 years of age.

Exclusion criteria: d/c or died <48 hours; admitted scheduled d/c <72 hours; non-English speaking; psychiatric ward; AIDS diagnosis; pregnancy; acute burns, trauma.:

Type: Nurse led intervention: provision of prognostic information to improve communication and decision making. Pt and family EoL preferences elicited and documented.

Control: Usual care not specified

Primary:

Timing of DNR orders; Pt/physician concordance CPR preferences; Days in ICU before death; Pain; Hospital resource use

Secondary:

Follow-up: Audit on Days 1, 3, 7, 14 and 35.

Small improvement in Pt/physician concordance in the group.

Small increase in reported pain.

Costantini et al. 2014 [31]

Italy

CRCT

Sample: 308

Setting: 16 general medicine wards

Diagnostic group: Oncology patients and families

Mean age:75.6 (SD=10.8)

Inclusion criteria (wards): 25+ cancer deaths per year, hospital consent, specialist palliative care team.

Patients who died from cancer and their family member

Type: Liverpool care pathway; training of ward staff and palliative care unit staff; leaflets for family members on emotional and practical issues; audits and feedback; documentation.

Control: Usual care

Primary: Overall mean score on toolkit after bereavement interview

Secondary: Decision making, ACP, respect, emotional support, coordination care, self-efficacy, quality of care, control of symptoms, processes of care

Follow-up: 2–4 months post-bereavement (family); processes 6 months post-implementation.

No difference in overall rating of care.

Improvements in 2/9 secondary outcomes only (respect, and control of breathlessness)

No difference in survival or medication prescribed

Detering et al. 2010 [36]

AUS

RCT

Sample: 309

Setting: Single teaching hospital

Diagnostic group: Cardiac, Respiratory, Falls, General admission

Inclusion criteria: ≥80 years old; admitted under internal medicine, cardiology or respiratory medicine; English.

Exclusion criteria: Not competent; <80 years of age; expected to die or be d/c within 24 hours; had previous ACP; no family.

Type: Received ACP from trained facilitator: multi-disciplinary collaborative approach to ACP; involvement of a surrogate; documentation of EoL care preferences including CPR

Control: Usual medical care, no ACP advice unless specifically requested.

Primary: % pts EOL wishes known and respected

Secondary: Patient satisfaction; Impact of death on relatives

Follow-up: Baseline, death or discharge of patient; 3 and 6 months post death or discharge.

More decedents in intervention groups had EoL wishes known and respected compared to control (86% vs 30%)

A greater proportion of family in I group were satisfied with the quality of death of the patient compared to the C group (83% vs 48%)

Gade et al. 2008 [32]

RCT

Sample: 517

Setting: Two hospitals

Diagnostic group: Mixed

Mean age: 73.6 (SD=12.6)

Inclusion criteria: 18+ years, hospitalised with 1+ life-limiting illness, attending physician judgement of prognosis <12 months

Excluded: cognitive status impaired, no surrogate or currently enrolled in hospice or PC studies.

Type: Palliative care consultation (IPCS) assessing symptoms, assisting goals of care discussions, discharge planning issues,

Control: Not specified

Primary: Symptom control, emotional and spiritual support, satisfaction and health care costs

Secondary:

Follow-up: 2 weeks discharge, 6 months

No difference in hospital LOS

IPCS had longer median hospice stays, fewer ICU admissions

Higher % IPCS completed ADs at discharge

No difference in symptoms, emotional, spiritual support

No difference in survival - more IPCS died during index admission

Satisfaction higher for intervention patients

Total costs lower by $6766 per patient.

Hanks et al. 2002 [33]

UK

RCT

Sample: 261

Setting: Hospital

Diagnostic group: Cancer and non-malignant inpatients

Mean age: 68.4 (26–93)

Inclusion: All inpatient referrals to palliative care team.

Exclusion: unable to give consent or complete baseline, unaware of diagnosis, likely to die or be d/c in 24 hours, or needed to be seen very urgently

Type: Full-PCT – assessment by specialist doctor/nurse, provision of advice to team verbally and documented, telephone and in-person follow-up. At least weekly reviews, and liaison with community teams post-discharge.

Control: Telephone consultation with senior member of PCT and referring doctor and senior nurse and ward nursing staff

Primary: Symptom control, HrQoL, LOS hospital and rate of re-admission

Secondary: Satisfaction patient/family & provider, use of health services.

Follow-up: 1 week post-recruitment

Improvement over time in scores for all items in FPCT; and smaller improvements in control

No difference between the groups.

FPCT discharged home spent fewer days at home, but received more GP visits

Sidebottom et al. 2015 [34]

USA

RCT

Sample: 232

Response Rate:

Setting: Tertiary hospital

Diagnostic group: Heart failure

Mean age: 73.4 (SD=13.0)

Inclusion criteria: Acute hearty failure, 18+ years.

Exclusion: ICU, undergoing evaluation for transplant or LVAD, post-LVAd or transplant, actively dying, cognitive impairments, insufficient English, existing PC order.

Type: PC consult within 24 hour. Differed to usual consult (1) baseline assessment results available to providers; (2) subsequent consults billed to patients.

Referral to ACP process at discharge, post-discharge telephone call if ACP not completed.

Control: Usual care

Primary: Symptom burden, depressive symptoms, quality of life

Secondary: ACP, inpatient 30 day readmission, hospice use, death

Follow-up: 1 and 3 months

Intervention: greater reduction in ESAS distress; improvements in SOB, anxiety and tiredness (1 and 3 mths); pain (3 mths only); lower depression score; higher QoL score.

Intervention 2.87 times more likely to complete ACP

No difference in readmissions, hospice use, death

Song et al. 2005 [35]

USA

RCT

Sample: 32 dyads

Response Rate:

Setting: Cardiothoracic surgery clinic

Diagnostic group: Cardiac condition requiring surgery

Mean age: 69.8 (SD= 8.6)

Inclusion criteria: scheduled for semi-elective surgery in 12 or more hours; had decision-making capacity; >50 years of age; had a surrogate >18 years of age willing to participate.

Type: Patient-Centered Advance Care Planning (PC-ACP) interview (20–45 minutes) by trained nurse: i) representational assessment; ii) exploring concerns planning for future medical decision-making; iii) creating conditions for conceptual change; iv) disease-specific statement of treatment preferences; v) summary.

Control: Information card on ADs. Offered more information about AD’s if requested. Pts referred to pastoral counselling at their own request.

Primary: Congruence (measured over 3 scenarios)

Secondary: Anxiety (SAI); Pt decision making difficulty (16 item decisional conflict scale); Knowledge of ACP.

Follow-up: Baseline and post-interview

Intervention group had significantly higher congruence and lower decisional conflict

No other differences.