Wing et al (above) indicate agreement with the main findings from our study. These were that HoNOS may be most suitable for tracking changes over time, but less useful for treatment planning, and should not be used to infer the level of morbidity in a case-load. The Camberwell Assessment of Need Short Appraisal Schedule (CANSAS) (Reference Slade, Thornicroft and LoftusSlade et al, 1999a ), by contrast, indicates when treatment should be commenced or continued, and can be used as a case-load measure, but may be insufficiently sensitive to be used as an outcome measure at the individual level. The concerns expressed in regard to other papers reflect the tension in creating assessments which are both robust and clinically relevant. Developing and implementing outcome measures for use in routine clinical settings will require attention to the construct being assessed, the purpose of the assessment, and the measurement tool used.

Clarity is needed about what constructs are assessed. The National Health Service and Community Care Act 1990 states that services are to be provided on the basis of need, and outcome measured in relation to changes in quality of life. This directive accords with recent research comparing these constructs, which found high need to be associated with low quality of life (Reference Slade, Leese and TaylorSlade et al, 1999b ). The constructs of ‘need’ and ‘quality of life’ should both be assessed, and HoNOS may contribute to measuring the latter.

The type and robustness of data needed will vary, according to the purpose of assessment. One recent model identifies three levels of potential outcome: individual, local and national (Reference Tansella and ThornicroftTansella & Thornicroft, 1998). This ‘matrix model’ provides a framework for identifying potential uses of HoNOS.

Finally, more attention needs to be paid to developing outcome measures suitable for routine use. Recently proposed feasibility criteria suggest that assessment schedules need to be brief, simple, relevant, acceptable, available and valuable (Reference Slade, Thornicroft and GloverSlade et al, 1999c ). HoNOS appears to meet the first five criteria. Whether it proves valuable, and to whom, will depend on a number of factors. The development of an information infrastructure to support data entry, analysis and feedback will be needed, as will the introduction of local and national systems for linking effort in collecting such data with visible benefits for patients and staff. Product champions will need to provide strong local leadership, with ongoing training given a high priority. Perverse incentives will need to be avoided, such as performance-related pay based on health gain.

In the longer term, assessments such as HoNOS and CANSAS which are intended for routine clinical use may contribute to a culture-shift, where reflective, evolving and evaluated mental health services become the norm.