Lautrette 2007.

Study characteristics
Methods	Aim: to evaluate a proactive EoL conference and brochure to determine effects of bereavement Study design: 2 arms (intervention, usual care) Consumer involvement: none described Funding source: supported by grants from Assistance Publique‐Hopitaux de Paris and the French Society for Critical Care Medicine; supported by grant from the National institute of Nursing Research The last author disclosed funding support from Pfizer; no other potential conflicts of interest were reported
Participants	Participants: surrogate decision‐makers for patients in ICU and expected to die within days Setting and geographic location: 22 ICUs in France; multisite ICUs: 68% teaching hospitals. Both medical and surgical ICUs Methods of recruitment: in each ICU unit, local investigator agreed to include surrogate decision‐makers of 6 consecutive patients expected to die within a few days Surrogates were either those designated by the patient or the person ranked highest in decision‐making hierarchy according to French law (spouse > parents/children > others) Oral informed consent obtained from surrogates Selection criteria for participation in study: Inclusion: physician belief that patient would die within days. Aged 18 years or older Exclusion: patients younger than 18 years. Those with insufficient French for phone interview Diagnosis of person approaching EoL: variable: included acute respiratory failure, coma, shock, acute renal failure, cardiac arrest Target of intervention: surrogates (primarily family members), approximately 40% spouses, approximately 48% children of patients Age: Patients: intervention median age 68 years, control 74 years Surrogates: intervention median 54 years, control 54 years Gender: Patients: intervention 41% female, control 48% female Surrogates: intervention 77% female, control 70% female Ethnicity/culture/language:around 90% of patients and surrogates in both study groups were of French descent Other PROGRESS aspects:those with inadequate French to enable telephone interviews were excluded; this may have restricted participation of other ethnic/minority groups (with almost 90% of participating surrogates were of French descent) Numbers of participants: see Additional Table 1
Interventions	Intervention: proactive communication family conference Aim of intervention: to decrease the effects of bereavement in family members of patients dying in the ICU (by improving communication between family members and ICU staff and to support families with decisions) Comparison: routine family conference Delivered by: investigators at each ICU attended meeting about the intervention; given a copy of the VALUE guidelines and of supporting research articles 1 member of study team visited each site to discuss the guidelines and ensure the differences between the intervention and routine family conference were understood Setting: in quiet separate room with seating for all; beepers and mobile phones off Usual care was delivered in ICU room or another available room (not specially designated for purpose; may or may not be in a separate room) Materials, procedures, content: Proactive communication strategy conference: conference conducted according to specific guidelines, also provision of bereavement information leaflet Participants planned several hours in advance; participants included senior and junior physicians, nurses, a psychologist, other health professionals, unrestricted number of family and friends; social worker and spiritual representative invited if requested by the family Guidelines for the conference based on previous research: based on detailed conference procedure, provided information on diagnosis, prognosis, treatment and discussed appropriateness of treatment limitation with family members. Intensivist leading the conference sought to achieve the 5r values outlined by the VALUE mnemonic: Value and appreciate things family says, Acknowledge emotions, Listen, ask questions that allow Understanding of who the patient is as a person, and Elicit questions from family At end of family conference family member was given a bereavement information leaflet, with content explained orally. 15 pages, explained EoL care, possible reactions after the death of a family member, how to communicate with other family members, where to find help Used previously but modified for this study to focus on adult ICU patients and optimising EoL care Usual care: routine EoL conference: held to inform family that death is imminent and to describe treatment‐limitation decisions and consequences of these. Family members may share in decisions if wished, but these decisions are under the authority of physicians and are made collegially by ICU team Led by senior physician in charge of patient; nurses may or may not attend. Conference may or may not be held in separate room Previous studies show mean duration is 10 minutes Occurred when at least 1 family member in the ICU When and how much: once, following 3 information meetings provided to all families Tailoring: not stated explicitly but family members had the opportunity to ask questions, discuss treatment options with physician and others (in both intervention and UC groups) Modified during study: not stated Co‐intervention(s): participating ICUs were members of the FAMIREA study group; 3 formal early information meetings held for all families (prior to randomisation). First at 24 hours (general information on diagnosis, prognosis, treatments) plus information leaflet; second at 48 hours (answering questions, additional information check family understanding of situation); third at day 3 to 5 (treatments etc. explained and prognosis explained, questions by family answered) If patient expected to die (following these 3 meetings) or shift to palliative care is indicated an EoL conference is held (i.e. intervention or routine conference) Co‐interventions delivered to all participants (fairly extensive information provision). Authors note that this may have lessened differences between intervention and routine care groups for some outcomes assessed by the study Fidelity assessed: quality of intervention: investigator attended all 3 EoL intervention conferences to ensure consistency of the conference format (prior to delivery of the intervention) Authors note that differences on various outcomes in the conduct of the conferences indicates that the guidelines for the conferences were followed No assessment was made of how many read the bereavement brochure, or how well understood the material (content) was Theoretical base: prior studies developing the communication guideline are cited
Outcomes	Primary outcomes Effectiveness of information provided [Evaluation of the communication] Method: surrogate interview via telephone (ratings of time allotted to provide information, clarity of information, and whether additional information was requested by family members) Timing: 90 days after death  Scale and scoring: unclear Primary outcomes ‐ adverse events None reported Secondary outcomes Duration of family conference [Health system impacts] Method: unclear Timing: at time of family conference with ICU staff Scale and scoring: unclear
Notes	Trial number NCT00331877 Co‐interventions delivered to all participants (fairly extensive information provision), and routine conference meeting with UC group. Authors note that this may have lessened differences between intervention and routine care groups for some outcomes assessed by the study Authors also note that the patient‐doctor relationship in France is typically more paternalistic than elsewhere, and since the standard consultation relied on this interaction that the effects of the intervention in France may have been greater than might be found in other countries where models of shared decision‐making are more commonly practiced. However, authors also note that interactions in control group were similar to those reported in North America and Europe One further issue may be that the control conferences were at least as good as routine care reported in other studies (i.e. longer duration, information needs of families largely met) Trial also reported the following outcomes for surrogates at 90 days after death of the patient: PTSD symptoms (Impact of Event Score), depression and anxiety (Hospital Anxiety and Depression Scales). Trial reported several outcomes rated by physician observation of family members during family conferences: expression of emotions by family members, family belief that patient's symptoms were controlled, family‐reported conflict with ICU staff, or ICU staff reported conflict with family members. ICU and patient medical data were also reported. None of these outcomes were judged as relevant for this review, based on the review's focus on communication at the end of life and were not included in analyses
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"The study biostatistics department generated a randomization list stratified on the ICUs, using permutation blocks of six" (page 3 supplementary appendix)
Allocation concealment (selection bias)	Low risk	"Sealed consecutively numbered envelopes containing the name of the assigned group were sent to each ICU, with bereavement information leaflets. The leaflets were not in the sealed envelopes, so that the blind design was not broken. In each ICU, surrogate decision‐makers who consented to the study were assigned a study number, and the investigator opened the envelope bearing that number to determine group assignment" (page 3 supplementary appendix)
Blinding of participants and personnel (performance bias) All outcomes	High risk	Not possible to blind investigators/providers of the intervention, or participants (surrogates) Possible that investigators with strong positive feelings about the intervention may have influenced family member interactions
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Interviewer assessing surrogate outcomes was blinded to group assignment Unclear whether researchers were blinded to assignment when recording ICU and patient characteristics although the effects of this on clinical and treatment outcomes seem likely to be negligible
Incomplete outcome data (attrition bias) All outcomes	Low risk	Loss to follow‐up and withdrawals were acceptably low and comparable across study groups: 52/63 (83%) completed interviews at 90 days intervention group, 56/63 (89%) in control group Reasons for withdrawal/loss were similar across groups (not answering telephone, refused interview), although higher rates of severe emotional distress in intervention group (n = 5) than control group (n = 1)
Selective reporting (reporting bias)	Unclear risk	Outcomes stated in methods are reported completely, however no protocol available for the trial
Other bias	Low risk	Baseline imbalances: groups were comparable on key features at baseline