Objective Compassionate care continues to be a focus for national and international attention, but the existing evidence base lacks the experimental methodology necessary to guide the selection of effective interventions for practice. This study aimed to evaluate the Creating Learning Environments for Compassionate Care (CLECC) intervention in improving compassionate care.

Setting Ward nursing teams (clusters) in two English National Health Service hospitals randomised to intervention (n=4) or control (n=2). Intervention wards comprised two medicines for older people (MOPs) wards and two medical/surgical wards. Control wards were both MOPs.

Participants Data collected from 627 patients and 178 staff. Exclusion criteria: reverse barrier nursed, critically ill, palliative or non-English speaking. All other patients and all nursing staff and Health Care Assistant HCAs were invited to participant, agency and bank staff were excluded.

Intervention CLECC, a workplace intervention focused on developing sustainable leadership and work-team practices to support the delivery of compassionate care. Control: No educational activity.

Primary and secondary outcome measures Primary—Quality of Interaction Schedule (QuIS) for observed staff–patient interactions. Secondary—patient-reported evaluations of emotional care in hospital (PEECH); nurse-reported empathy (Jefferson Scale of Empathy).

Results Trial proceeded as per protocol, randomisation was acceptable. Some but not all blinding strategies were successful. QuIS observations achieved 93% recruitment rate with 25% of patient sample cognitively impaired. At follow-up there were more total positive (78% vs 74%) and less total negative (8% vs 11%) QuIS ratings for intervention wards versus control wards. Sixty-three per cent of intervention ward patients scored lowest (ie, more negative) scores on PEECH connection subscale, versus 79% of control. This was not a statistically significant difference. No statistically significant differences in nursing empathy were observed.

Conclusions Use of experimental methods is feasible. The use of structured observation of staff–patient interaction quality is a promising outcome measure inclusive of hard to reach groups.

Trial registration number ISRCTN16789770.