Single global question and psychometrics

It may seem that the only question you need to ask patients is simply 'How is your QL?' and then measure improvements or deterioration. Indeed there are single question 'performance status' scales [44], such as the Karnovsky [45], Eastern Co-operative Oncology Group (ECOG) [46] or World Health Organization (WHO) performance status [47]. However, these cannot be classed as QL scales because, as discussed above, QL is a complex concept and they only address one of the many aspects of QL.

Although superficially attractive, simply asking a single question about QL makes it impossible to disentangle what components of QL are associated with any changes over time or differences between treatments observed. It is perhaps not surprising that when comparisons are made between health-related questions and overall QL there is rarely good correlation, suggesting of course that health is only one aspect of QL.

From the point of view of analysis and interpretation it would appear superficially helpful to reduce QL to a single score so that the QL of two treatments could be readily compared. Consequently, there is a temptation to combine all the answers from a questionnaire into a single score but there are numerous reasons why this is not an appropriate approach, and perhaps could be compared to reporting toxicity but not specifying what type. As stated earlier, the patient's health is only one element of overall QL and it will be influenced by numerous other aspects of daily life. Simply adding together all the scores from a QL questionnaire does not represent overall QL. Not only would one need information on all the factors that make up QL, but also the patients' individual weighting for each factor.

McHorney et al. [48] compared the validity and relative precision of four methods of measuring general health. Although the questionnaire with the most items was considered more reliable, defined more distinct levels of health and better represented the content of health perceptions, this was, of course, at the expense of considerable extra patient and staff time and energy. Shorter questionnaires were not significantly less reliable than the long questionnaires but in turn were much better than single item forms, which failed to detect individual patient changes.

Nevertheless, many questionnaires respond to the demand for a single QL score by supplementing more specific questions with a single question along the lines of 'How would you rate your overall QL?' This is certainly preferable to an aggregated score, but its limitations (as discussed above) must be kept in mind.

Questionnaires have been developed by a number of groups, usually by a long process of surveying patients, compiling lists of questions, and then testing and evolving the questionnaire until they are satisfied with its psychometric (the measurement of human characteristics) properties.

Most of the standard cancer QL questionnaires cover a number of general domains (aspects of QL) and specific symptoms. Usually the scores from a number of questions are combined to give a 'domain' score. The reason for this is that it has repeatedly been shown that in order to get, say, a score of anxiety for a patient, it is much more reliable to ask a series of anxiety-related questions and combine the scores, than simply ask 'what is your level of anxiety?

Usually subscales are made up of items that are positively correlated. However, there maybe problems with the statistical methodology, factor analysis, that often underlie this approach, as described by Fayers and Hand [49] who have attempted to divide symptoms into causal and effect indicators. Thus, increased nausea might cause a decrease in overall QL (i.e. nausea is a causal indicator), whereas poor overall QL might result in depression (i.e. depression is an effect indicator). In addition, some factors, such as insomnia, may be considered as both causal and effect indicators (see Fig. 6.5).

This schema may seem obvious and logical, but Fayers et al. [50] argue that it could have important implications for the development of questionnaires and especially the formation of subscales. They argue that factor analysis may select both causal and effect indicators to be included in subscales, and that it might make more sense to group together items that make sound clinical sense, such as gastro-intestinal symptoms.

Kicking Fear And Anxiety To The Curb

Kicking Fear And Anxiety To The Curb

Kicking Fear And Anxiety To The Curb Can Have Amazing Benefits For Your Life And Success. Learn About Calming Down And Gain Power By Learning Ways To Become Peaceful And Create Amazing Results.

Get My Free Ebook

Post a comment