In selecting the best measure the following questions should be considered: (extract from
What is your question?
Appropriate selection of measures follows from the objectives of the research study, knowledge of the patient population and the development of specific research questions. What type of outcome would be a useful measure for the given population, condition or intervention under evaluation?
Is the content relevant?
This can be assessed by examining the instrument and considering each item (and its response options) individually. This is particularly important when considering use of generic measures. For example, both the SF-36 and EQ-5D (EuroQoL 5-Dimension) include items about self-care, which may be irrelevant in a young population with few complications or when comparing two long-acting insulins. Despite the limitations in in some subgroups, the SF-36 and ⁄ or the EQ-5D may still be important to understand people’s health status.
Are any relevant issues missing?
This is particularly important when considering use of generic measures. For example, neither the SF-36 nor the EQ-5D include items regarding dietary freedom, demonstrated to be an important issues for QoL in people with diabetes. Omission of such issues may mean that the full impact of diabetes cannot be assessed and a lost opportunity to demonstrate significant treatment benefits.
Will responses be influenced by other conditions ⁄ factors?
Will responses be influenced by other conditions/factors?
The use of generic measures can be useful but need appropriate interpretation. In an elderly population, co-morbid conditions are likely to affect scores derived from generic measures.
Will the measure be acceptable to respondents?
Some measures may include sensitive issues, or be more complex or lengthy than others, which may impact on their acceptability, particularly if the population is elderly or very young and ⁄ or has low literacy skills.
What is the burden on respondents?
The lack of a ‘gold’ standard instrument and ⁄ or the complexity of the intervention may necessitate administration of several instruments in one questionnaire booklet, e.g. diabetes-specific and generic, measuring various concepts such as QoL, health status, well-being, health beliefs, in order to achieve holistic evaluation. Researchers must be aware of the burden that this may place on respondents, though it is unlikely to be as burdensome as many biomedical procedures that they may be required to undergo.
Has the measure been validated in the given population and/or country/language previously?
Researchers need to be aware of the need to ensure that the measure(s) they select have been validated or be prepared to undertake the psychometric validation themselves.
How will I analyse and interpret the data?
Clinicians and researches need to consider whether or not they feel confident not only in the selection of appropriate measures but also in undertaking their analysis and interpretation. If not, they may be better placed to collaborate (in a multidisciplinary team) with a social scientist experienced in the development, use and interpretation of measures in diabetes.