Table 2. Answers to the a priori key questions for the feasibility pilot and implications for future development.

Question	Result	Implication for future iterations
Feasibility of the intervention
1. How acceptable is the intervention?	Patients were generally positive about the assessment, openly discussing concerns and coping strategies in all domains. Professionals were interested in the concept of HELP-COPD and considered that the intervention was feasible but, noted some practical issues (e.g., overlap with discharge planning and communication with GPs)	The core concept was acceptable, but communication/integration with other services needs further consideration
2. In what proportion of patients was an action identified?	Twenty-three patients received the intervention. Overall, 37 action points were noted: an average of 1.6 per assessment. Of these: Five were referrals (one each to pulmonary rehabilitation, the home oxygen service, carer support, social care direct and a request to the GP). Fifteen were contacts for the patient to follow-up (e.g., social support agencies) Seventeen people were given information about COPD, and local services.	The reasons for fewer actions than anticipated needs to be explored and addressed. Key issues were: overlap with discharge planning, the priority attached to retaining independence and adaptation to long-term disability.
3. What actions were identified/triggered by the HELP-COPD assessment?	Physical symptoms were discussed in 21 assessments, but only one required action (referral to GP about morphine to relieve breathless) Psychological issues were discussed in 19 assessments, but no actions were generated. Most felt well supported by their family. A range of social issues were discussed in most assessments, but only one referral (for help with social care). Maintaining independence was the priority. Fifteen spoke of pleasure in family life and/or the importance of their faith. Five had put their affairs in order recognising ‘We all have to die sometime’.	Although few actions were generated, the physical, psychological, social and spiritual aspects of the assessment resonated with the participants and should be retained as the underpinning structure of the assessment.
4. Did follow-up help ensure that referrals occurred in a timely manner?	The follow-up telephone calls generated few actions and were perceived as achieving very little. In contrast, clinicians from community teams highlighted on-going care as important for identifying problems.	Integration with existing services (e.g., primary care, community services) could enable on-going care.
5. What are the barriers to implementation of actions?	Offers of help were declined for a number of reasons: Over-riding preference for independence (often with family support) Previous bad experience of social services Services perceived as inappropriate, inaccessible and/or unavailable	The burden on the carers needs to be explicitly considered/addressed in future iterations of HELP-COPD
6. How may the intervention be improved?	The intervention did not seem to work quite as intended: suggestions for future included: An on-going relationship may facilitate detection of need Avoid duplication of services Detecting problems and offering help at an earlier stage in the disease process may be better	Integrating HELP-COPD with existing services providing long-term care for people with COPD (e.g., primary care, community services) would enable incremental detection of needs within the context of an on-going relationship
Question	Result	Implication for future research
Feasibility of the trial procedures
7. What proportion of eligible patients will agree to participate?	Detailed records were kept of ward admissions for 5 months during which there were 189 respiratory admissions. Fifty-nine occurred during a period when the ward was isolated for infection control reasons (we were not allowed to recruit) or when the respiratory nurse was unavailable. Of the remaining 130, 57 patients had other lung conditions and 22 were unable to provide valid consent (e.g., because of cognitive impairment) or complete questionnaires in English. Forty-seven potentially eligible patients were approached: 19 (40%) were recruited.	We recruited 40% of eligible COPD admissions (15% of all respiratory admissions)
8. What factors encouraged (or discouraged) participation?	A key motivator was ‘to help with research’ because ‘not enough is done about COPD’. Reasons for declining included ‘I don’t need/want anything’, ‘too many other agencies involved’ and ‘other clinical/domestic problems more important’.	Avoidance of the word ‘help’ might improve uptake. E.g., HELP-COPD might be better called ‘COPD-plus’. Simplifying services might avoid duplication or confusion due to multiple agencies.
9. What is the likely attrition rate?	Forty-four patients provided consent during their admission: Twelve were unable to participate in the baseline assessment 2 weeks post discharge because of death (n=3), readmission (n=3), too ill (n=3), uncontactable (n=3). We randomised 32 patients. Twenty-four were allocated to the HELP-COPD intervention: one was too ill to participate in the intervention at 4 weeks. On-going attrition throughout the 6 months: 7 of 23 participants in the HELP-COPD group and five of eight in the usual care group mostly for reasons of ill-health.	This very substantial attrition, mainly related to severity of disease and the fact that we were recruiting at a time when the disease had just exacerbated, is an important consideration for researchers recruiting people with COPD at the time of an admission.
10. What is the most appropriate duration for a trial?	Documentary evidence from the HELP-COPD records suggested that referrals to or uptake of pulmonary rehabilitation (one of the actions we identified) or other referral agencies (oxygen, social services) take time to implement.	Our experience suggests that an RCT should be at least 6 months in duration.
11. How feasible/practical are the proposed outcome measures?	Questionnaire completion proved burdensome for some patients, with 13 unable/declining to complete follow-up questionnaires (3 had died, 3 were too ill, 2 had family illness/bereavement, 4 gave no specific reason). For those well enough to complete the questionnaire, FACT-L (the outcome we were considering as a future primary outcome) appeared to be acceptable.	In this frail group of patients use of questionnaires needs to be rationalised. FACT-L covers all domains and obviates the need for multiple questionnaires.
12. What is the expected change in the primary outcome measure?	The FACT-L showed a trend to a difference at 6 months (See Supplementary Information). The change from baseline (85 to 97) in the intervention group is substantial and at least twice the minimum clinically important difference of 5.5	The substantial attrition means that numbers were very small, and difficult to interpret with confidence
13. Which outcomes matter to patients/carers?	The Lay Advisory Group considered that important outcomes for them were greater independence, ability to improve their social lives and to live fuller lives.	FACT-L reflects this and endorses our decision to use it.
14. What is  ‘usual care’?	Usual care in the context of the HELP-COPD intervention is (i) discharge planning and (ii) community services designed to help prevent admissions.	Future iterations of HELP-COPD need to integrate, not overlap, with these services.

Abbreviations: FACT-L, Functional Assessment of Cancer Therapy-Lung; GP, general practitioner; HELP-COPD, HELPing older people with very severe chronic obstructive pulmonary disease; RCT, randomised controlled trial.