To compare costs and health consequences and to assess the cost-effectiveness of using low-dose oral long-acting morphine in people with chronic breathlessness.

Within-trial planned cost-consequences and cost-effectiveness analysis of data from a multisite, parallel-group, double-blind, randomised, placebo-controlled trial of low-dose, long-acting morphine.

11 hospital outpatients across the UK.

Consenting adults with chronic breathlessness due to long-term cardiorespiratory conditions.

5–10 mg two times a day oral long-acting morphine with a blinded laxative for 56 days.

Mean and SD of healthcare resource use (HRU) by trial arm; mean differences and 95% CI of costs between trial arms.

Mean differences in 28- and 56-day quality-adjusted life years (QALYs based on EuroQol five-dimension five-level score), Short Form-six dimensional scores and ICEpop CAPability-Supportive Care Measure scores; cost-utility of long-acting morphine for chronic breathlessness.

143 participants (75 morphine and 67 placebo) were randomised; 140 (90% power, males 66%, mean age 70.5 (SD 9.4)) formed the modified intention-to-treat population (participants receiving at least one dose of study medication). There were more inpatient and fewer outpatient services used by the morphine group versus the placebo. In the base-case analysis at 56 days, long-acting morphine was associated with similar mean per-patient costs and QALYs. There was an increase of £24 (95% CI –£395 to £552) and 0.002 (95% CI –0.004 to 0.008) QALYs. Hospitalisations were the main driver of cost differences. The corresponding incremental cost-effectiveness ratio was £12 000/QALY, with a probability of cost-effectiveness of 54% at a £20 000 willingness-to-pay threshold. In the scenario analysis that excluded costs of adverse events considered unrelated to long-acting morphine by site investigators and researchers, the probability of cost-effectiveness increased to 73%.

Oral morphine for chronic breathlessness is likely to be a cost-effective intervention provided adverse events are minimised, but the effect on outcome is small and cautious interpretation is warranted.

ISRCTN87329095.