Introduction and aims To improve end-of-life care we need good evidence. The challenges of carrying out research with patients who are dying are well documented yet this research is vital to determine whether new interventions improve care. We examined the feasibility of recruiting patients into a study examining their care as they died.

Methods We performed a prospective mixed-methods observational study of the care given to patients who were dying. All patients on study wards were screened on admission using the question ‘Is this patient so unwell you feel they could die on this admission?’ If the answer was ‘yes’ the patient (and/or family) was approached to give prior consent to being included in the study ‘should they become more unwell’.

Results Over 2 years, 6703 patients were screened; staff answered ‘yes’ to the screening question for 327 patients (5%). Prior consent (23) or relative assent (94) was obtained for 117 (36%) of these patients. 70 died within the study. Overall, the screening question had a sensitivity of 57% and a specificity of 98%.

Conclusion To our knowledge prior consent for research during the dying process has not been used before within an acute trust. While a large number of patients need to be screened, we have shown that a modified surprise question is helpful in identifying patients who will die during an admission, and it is feasible to gain consent from these patients, or assent from their relatives, for inclusion in a study to examine the dying process.