Mitchell 2008.

Methods	RCT; Unit of allocation: Patient
Participants	Patients having a life expectancy of at least 1 month and referred to three specialist palliative care units (note from author: 92.5% patients with cancer). Setting / country: An inner urban, an outer metropolitan and a regional general hospital in Queensland / Australia Type of cancer: Any type Phase of care: Palliative care Sample size at randomisation: 159 (101 GPs)
Interventions	Case conferences between the patient's GPs and a specialist team: the intervention case conference was intended to provide an opportunity for negotiating a treatment plan for the patient, with the GP playing an active part. The case conference was conducted by teleconference, with the GP phoning in to a routine specialist team meeting. At two of the participating services, representatives of domiciliary nursing services routinely attended these meetings. Subsequent communication followed normal practice (i.e. faxed or posted letters, and telephone communication between family physician and specialist, or domiciliary nurses present at specialist team meetings acting as an intermediary). Control: Standard care: case review by the specialist team, with routine communication with the GP thereafter (i.e. faxed or posted letters, and telephone communication between family physician and specialist, or domiciliary nurses present at specialist team meetings acting as an intermediary).
Outcomes	Patient: QoL Informal carer: Carer burden Process: Number of case conference completed
Notes	Length of follow‐up: Until death or the end of the study (min. 1; max. 22) months
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote from ref #1: "The unit of randomisation was the GP?patient dyad: these were randomised by an administrative officer off‐site using a computer‐generated random number list and block randomisation. Subsequent patients of a previously randomised GP were allocated to the same arm as the first patient. We achieved concealment of allocation from both patient and GP by explaining that different means of communication between GPs and palliative care teams were being tested without informing them of the details. Neither the palliative care team nor the research officer at each site could be blinded to the allocation."
Allocation concealment (selection bias)	Low risk	See quote first item.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Quote: "Neither the palliative care team nor the research officer at each site could be blinded to the allocation." Comment: Patients were interviewed for data collection. From the above citation, it seems that interviews were conducted by a research officer, but it cannot be ascertained.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Comment: Proportions of participants who withdrew from the study because of death were similar between intervention (84%) and control groups (89%). Details of the reasons of the other withdrawals are not given for each study group, but for all participants instead. No details on the way the researchers have dealt with withdrawals other than death are provided. Because number of participants in each group used for analysis are not detailed, we could not verify if missing data could have changed the observed effect size.
Selective reporting (reporting bias)	Low risk	All outcomes described in Methods are reported in Results.
Other bias	Low risk	No evidence of other bias.
Baseline outcomes similar?	Low risk	Quotes from ref #1: "The control group had a few QoL subscale outcomes that were better than the intervention group at baseline, and others over the following 3 weeks. (Table 4)" Relative to carer burden: "There were no differences between the groups at baseline." "Two analytical approaches were planned a priori. The first used measures of change from baseline. The second used death as the fixed point and grouped interview data by time from the data collection interviews to death. The purpose of this analysis was to account for differing degrees of severity of illness present at recruitment into the study." Comment: The first analytical approach used takes into account baseline differences in measurements.
Baseline characteristics similar?	Low risk	Quotes from ref #1: "The demographic characteristics of the two groups were similar, except for age on admission (the median intervention group age was 6.6 years older) (Table 2). Colorectal cancer was over‐represented in the control group (24% vs 14%), and prostate cancer over‐represented in the intervention group (14% vs 8%)." "The intervention and control GPs were similar, Table 2." Comment: Few characteristic differed between treatment groups, and these differences in age seem unlikely to bias study results.
Protected against contamination?	Low risk	Comment: The GPs were allocated so contamination was prevented