Measurement protocol development for At Home/Chez Soi: A randomized trial of Housing First for homeless persons with mental illness

Background Selection of appropriate measures for health intervention research is usually straightforward. In a large multi-site multi-methods trial of a complex Housing First intervention for homeless persons with mental illnesses we had some unique challenges and opportunities. First, autonomous research teams and a range of stakeholders, including consumers had differing values regarding measures. Second, most of the proposed measures had not been used in this specialized population, including newer outcome measures such as recovery.

Methods A systematic multi-site, multi-stakeholder process for the development of a measurement framework, and for identification and selection of instruments for key variables was used. In a next stage, selected instruments were tested for relevance, comprehension/recall and sensitivity in 30 homeless adults with mental illnesses recruited from service agencies in Vancouver, Winnipeg and Toronto, Canada using cognitive interviewing.

Results The consultation process resulted in a consensus set of core measures and site-specific measures. For the chosen instruments, pretesting demonstrated that many items on the questionnaires were not applicable to homeless people. Comprehension varied considerably, particularly for participants with acute symptoms of mental illness or cognitive impairment. Sensitivity of items ranged widely across the sample, but findings were consistent with previous literature which indicates that ‘how you ask' is as important as ‘what you ask'.

Conclusions Consideration of multiple perspectives, including the perspectives of consumers resulted in consensus on a measurement protocol. Pre-testing using cognitive interviewing methods was very important in ensuring that instruments were appropriate and valid for our unique study population.