A structured conversation tool for dialysis decisions did not help, and may have harmed

Design

Multisite cluster randomised trial with nephrologists as the unit of randomisation, 10 US sites, 2021-2023

Population

68 nephrologists and 268 patients aged 60 or over with eGFR 24 or below and estimated survival of 18 months or less

Primary outcome

Receipt of palliative care within 12 months of enrolment

Effect

Palliative care HR 0.94 (95 per cent CI 0.59 to 1.52, P = 0.812); dialysis initiation HR 1.50 (1.00 to 2.25, P = 0.048); quality of life worse (FACIT-Pal -6.51, FACT-G -4.78)

Best Case/Worst Case is a communication tool designed to help clinicians and older patients think through a major decision by describing the best plausible outcome, the worst, and the most likely, for each option. It has been widely promoted in surgery and nephrology. This cluster randomised trial tested it properly.

Sixty-eight nephrologists at 10 US sites were randomised, 36 to training in the tool and 32 to usual care, and 268 patients aged 60 or over with an estimated glomerular filtration rate of 24 or below, estimated survival of 18 months or less, and a live decision about starting dialysis, were followed for up to two years. The primary outcome, receipt of palliative care within 12 months, did not differ (hazard ratio 0.94, 95 per cent CI 0.59 to 1.52, P = 0.812). Patients of trained nephrologists were more likely to start dialysis (1.50, 1.00 to 2.25, P = 0.048) and reported worse quality of life throughout the study on two separate instruments (FACIT-Pal -6.51, -12.20 to -0.81; FACT-G -4.78, -8.35 to -1.22). Intensity of end-of-life treatment, quality of communication and death on study did not differ.

The change to practice is to stop assuming that a structured tool improves a difficult conversation because it makes it feel better organised. Two secondary outcomes moving the wrong way in a properly randomised trial is not proof of harm - they are secondary, the dialysis interval touches 1.00, and a communication intervention cannot be blinded - but it is enough to say the tool has not earned adoption and that scenario planning may push patients towards intervention rather than away from it. What still works is the unglamorous version: ask what the patient understands, ask what matters to them, and say plainly what you expect to happen.

• Do not adopt Best Case/Worst Case as a service standard on the strength of enthusiasm; the trial is negative.
• Note the direction of the secondary outcomes - more dialysis, worse quality of life - before recommending it.
• Continue to document what the patient understands and what matters to them; that was not what was tested.
• Remember palliative care referral was the primary outcome and did not move at all.
• Communication trials cannot be blinded, so treat patient-reported outcomes here with that in mind.

The statistics, in plain English

The primary outcome is a clear null: a hazard ratio of 0.94 with an interval from 0.59 to 1.52 rules out any large effect on palliative care referral. The dialysis initiation result, 1.50 with a lower bound of exactly 1.00 and P = 0.048, is on the boundary and is a secondary outcome, so it should be read as a signal rather than a result. The quality of life differences are larger relative to their intervals and appear on two instruments, which makes them harder to dismiss - though patients and clinicians both knew the allocation.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.