Epstein 2017.
| Study characteristics | ||
| Methods |
Aim: to determine the effects of a combined intervention (involving oncologists, patients and carers) on patient‐centred communication and other outcomes including shared understanding, patient‐physician relationships, QoL and aggressive treatments in the last 30 days of life Study design: cluster‐RCT, multisite (VOICE study); 2 arms (intervention, usual care) Unit of randomisation: clinician Consumer involvement: QPL was based on previous study developing this for cancer patients in palliative care; refined on the basis of a focus group and based on semistructured interviews with "demographically diverse patients with advanced cancer" (Rodenbach 2017) Funding source: grants from National Cancer Institute, National Institutes of Health supported the research. Funders had no involvement in design, conduct, data analysis or interpretation, preparation of manuscript or approval or decisions about submission. Authors report no conflicts of interest |
|
| Participants |
Participants: patients with advanced cancer, and their carers. Oncologists Setting and geographic location: community‐based cancer clinics, academic medical centres, community hospitals. Western New York, Sacramento, California; USA Methods of recruitment: Physicians: medical oncologists caring for non‐haemotologic cancer patients, recruited at practice meetings at participating clinics Patients: research assistants reviewed, with clinic staff, clinic rosters for enrolled clinicians to identify potentially eligible patients Carers: identified by patients (family member, partner, friend or other involved in health care, preferably a person who attended physician appointments with the patient) Patients recruited, consented and enrolled based on allocation of their physicians to intervention or control group All participants provided informed consent (written) From supplementary file 1 (protocol): page 23 "Method of Subject Identification and Recruitment: All Phase 1 and Phase 2 patients will be identified by research assistants working closely with participating physicians and their clinic staff by reviewing the clinic roster in detail to ascertain that all potentially eligible patients are identified. Potentially eligible patients will receive a brochure that describes the study (see attached study brochure). Office staff will explain to the patient that a research assistant will be calling him/her in the next two weeks to find out if he/she might be interested in participating in the study. Patients who do NOT want to be contacted about the study will be asked to return an enclosed opt‐out card to the study office within 4 days of receiving the study brochure. A research assistant will only call patients who have not returned the opt‐out card within the stated time period" Selection criteria for participation in study: Inclusion:
Exclusion:
Diagnosis of person approaching EoL: either stage IV non‐haematologic or stage III cancer "and whose physician 'would not be surprised’ if the patient were to die within 12 months" (page 93) Target of intervention: carers (family member, partner, friend or other involved in health care) 73% of enrolled patients nominated a carer Age: Physicians: 44 years Patients: 64.4 years Carers: not reported Gender: Physicians: 29% female Patients: 55% female Carers: not reported Ethnicity/culture/language: Physicians: 45% White, 42% Asian, 13% other race Patients: 11.5% non‐White Carers: not reported Other PROGRESS aspects: Those unable to understand spoken English or to provide written informed consent were excluded; participants were 89% White: those from minority groups, non‐English speaking, lower health and general literacy groups may not be well represented Numbers of participants: see Additional Table 1 |
|
| Interventions |
Intervention: combined patient‐centred communication (training) intervention Aim of intervention: to improve patient‐centred communication between physicians and patients/carers, and related outcomes Interventions developed based on previous studies on training of physicians, development and use of QPLs for patients Comparison: control (oncologist meets with research assistant but receives no training) Delivered by: physicians and patient coaches Trainers (for physicians) and coaches (for patients/carers) received 3 days of on‐site training Physician training included Standardized Patient Instructors (SPIs) adopting role of patient with advanced cancer with life expectancy of 12 months or less Setting: intervention setting not described Training for clinicians occurred in their clinical office. Training for patients occurred at the centre where their oncologist practiced Materials, procedures, content: 2 components to the intervention:
Physician and patient interventions focused on the same 4 elements of patient‐centred communication: engaging patients in consultations, responding to emotions, informing patients about choices related to treatment and prognosis, and framing information in a balanced (unbiased) way When and how much: Physician training: 2 educational outreach sessions. 1st session 1 hour; 2nd booster session 45 minutes 1 month later Patients/carers: coaching session (approx. 35 to 40 minutes in duration) prior to oncology consultation; follow‐up phone calls (up to 3 at monthly intervals) Tailoring: coaching was tailored to patient/carer priorities and concerns i.e. coaches helped patients to identify their most pressing questions in order to help these to be raised and addressed in physician consultation Modified during study: not applicable Co‐intervention(s): none Fidelity assessed: assessed and reported as 94% or higher (assessed by review of audio recordings of intervention sessions). All intervention physicians completed both training sessions; all intervention patients received in‐person coaching Theoretical base: based on previous work to develop interventions targeting patients |
|
| Outcomes |
Primary outcomes Patient‐centred (patient‐doctor) communication (composite measure) [Evaluation of the communication] Method: composite of 4 communication measures (engaging, responding, informing, framing of decisions) Audiotaped physician consultation, coded by trained university students (audited continuously, blinded to study aims and assignments) Timing: first physician visit following coaching session (intervention group) or study entry (control) Scale and scoring: Active Patient Participation Coding (engagement); Verona VR‐CoDES (response to emotions); Prognostic and Treatment Choices (PTCC) Informing subscale; PTCC Balanced Framing subscale. Component scores for each scale transformed to z scores; 4 scores averaged to give overall composite measure (authors report better sensitivity and precision than component individual scales) Patient‐physician relationship [Evaluation of the communication] Method: patient‐physician relationship Timing: shortly after audio recorded consultation (2 to 4 days, then quarterly) supplement 2 page 12 Scale and scoring: Human Connection Scale (THC); Health Care Communication Questionnaire (HCCQ); Perceived Efficacy in Patient‐Physician Interactions (PEPPI) scale Decision regret (caregiver) [Evaluation of the communication] Method: Modified Decision Regret Scale Timing: 2 months post‐death Scale and scoring: 8 items Shared understanding of prognosis (discordance between ratings) [Knowledge and understanding] Method: research‐administered questionnaire/interview Timing: shortly after audio recorded consultation Scale and scoring: 7‐point scale; discordance defined as difference of 2 or more categories of difference (i.e. between category ratings) Primary outcomes ‐ adverse events None reported Secondary outcomes QoL composite score [Quality of life] Method: research‐administered questionnaire/interview Timing: 3‐montly from study entry to 3 years Scale and scoring: composite QoL score as average of 5 z‐scored subscales: McGillQoL scale single item, McGill Psychological Well‐Being subscale, McGill Existential Well‐Being subscale, FACT‐G Physical Functioning subscale, FACT‐G Social Functioning subscale Treatments and hospice use in last month of life [Hospital admissions etc.] Method: trained nurse and physician‐abstracted data from medical records Timing: last 30 days of life Scale and scoring: composite score of 3 indicators of aggressive treatment in last 30 days of life: "chemotherapy, potentially burdensome interventions, emergency department [ED]/hospital admission) and hospice utilization" (page 95) Caregiver evaluation of quality of EoL care [Quality of EoL care] Method: caregiver Evaluation of Quality of EoL Care Scale Timing: 2 months post‐death Scale and scoring: 6 items |
|
| Notes | Clinical trial registration number NCT01485627 Mean survival of studied population was 16 months (19 months intervention, 14 months control group) See supplement 3 page 5 for detailed outline of all intervention components and delivery Cluster‐RCT; unit of randomisation: clinician, unit of analysis: clinician‐patient dyad (communication), patient ICC for all outcomes (except aggressive care at EoL) was < 0.1 "This is a cluster‐randomized trial, where our primary communication outcomes (Aim 1a) are measured at the level of the physician‐patient dyad and our secondary outcomes (Aims 1b, 2 & 3) are measured at the level of the patient. Analyses are based on published guidelines for group (cluster) randomized controlled trials" (supplement 1, page 16) "The physician‐patient dyad will be the unit of analysis, as measured in a single audio‐recorded clinical encounter. Because patients are clustered within physicians, in the data analysis, we may add random effects for physicians to account for the within‐physician correlation of each dyad. If analysis of the Phase 1 data identifies plausible confounding by physician (communication style) or patient factors (demographic, clinical status), these factors will be eligible for inclusion in final analyses as described above" (supplement 1 page 17) "As described in the BMC Protocol, we will primarily rely on regression models for clustered data to account for the stratified 325 cluster randomised longitudinal study design" (supplement 2 page 21) Seems likely that analyses were appropriately adjusted Treatments and hospice use in last month of life, assessed as a composite score of indicators of aggressive treatment in last 30 days of life was judged as primarily clinical in focus and data were not extracted for analysis |
|
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | Random number sequence used Randomised by physician (physicians as the unit of randomisation) and stratified by 2 sites and by oncologist subspeciality Within strata physicians were randomly assigned 1:1 to intervention or control Patients enrolled based on allocation of their physicians to intervention or control group |
| Allocation concealment (selection bias) | Low risk | All but study statistician blinded to random number sequence and group assignment "To preserve blinding, assignment to the treatment or control conditions is maintained by the study statisticians and project manager, and not explicitly revealed to research assistants, transcriptionists, or coders of the audio‐recorded office visits" (supplement 2, page 12) |
| Blinding of participants and personnel (performance bias) All outcomes | Unclear risk | Not possible to blind intervention group physicians; unclear what effect this might have Potentially patients and carers were aware of their treatment assignment; again not clear what effect this might have on outcomes sought (all but health services use (medical records) may have been influenced by knowledge of group assignment) |
| Blinding of outcome assessment (detection bias) All outcomes | Low risk | Only study statistician was aware of random number sequence and assignment: blinding preserved amongst transcriptionists, coders, abstractors |
| Incomplete outcome data (attrition bias) All outcomes | Low risk | "Fewer than 3% of follow‐up questionnaires were missing" (page 95) Data seems otherwise complete for outcomes reported in main paper and in supplement 3 |
| Selective reporting (reporting bias) | Low risk | Changes in subscale measured for primary outcome acknowledged in trial report (supplement 2 analysis plan), also other changes to outcome measures acknowledged in this report Several outcomes described in the protocol were reported in related papers |
| Other bias | Low risk | Selective recruitment of cluster participants: clusters were patients of an oncologist (unit of randomisation), these were identified before randomisation, therefore this risk of bias seems low Baseline imbalances: there were no differences at baseline between physician or patient groups at baseline. Low risk |