It is tempting, but wrong, for health professionals to think that they understand in detail the ways in which their clients' or patients' lives can be affected by their disease. It is essential that this information is sought from patients themselves, the first step in the creation of any system to measure the impact of skin disease on people's lives is to gain this information directly from those affected. This may be by unstructured or structured interview, by seeking information in writing or by using focus groups. Clearly the target audience for the use of the measure must be reflected in the population studied, taking into account age range, sex, language and/or cultural definitions. Some authors of QoL measures also use already published information or seek the views of health care professionals as part of their information gathering exercise. This seems reasonable provided this information is simply to confirm that no aspects have been overlooked (i.e. to validate the primary information gathering from patients); it should not be used as part of the primary information on which to base the proposed measure. Information should continue to be sought from those affected until continuing to interview others is revealing no new information.
Once the body of information has been gathered, very closely related areas of impact on patients' lives need to be merged into single descriptors. By this stage there will still usually be a long list of items that have been identified in this way. The next challenge is to use this information as the basis of a questionnaire tool. There are differing approaches as to how to amalgamate the information into a questionnaire that reflects the reality of the situation from the patients' viewpoint.
One possibility is to identify items which are very closely correlated: one of these items can be discarded in the knowledge that change in the presence or absence of this item will normally be exactly reflected in change in the kept item. This process can be repeated until the number of items reaches a level that is realistic for practical purposes or until no items remain that are correlated closely enough to cut. The problem with this scientifically sound approach is that there is the risk that the acceptability of the final series of items may have reduced comprehensiveness or acceptability for patients. The patient may perceive that the questionnaire reveals a lack of understanding of the subject area as aspects are omitted; their omission being due to close correlation with persisting items.
An alternate approach is less satisfactory from the statistical point of view but may result in greater face validity. The 'long list' of items are classified into various domains so that similar items, for example all those relating to work or to personal relationships, are grouped together. Within each domain the number of items can be reduced by the methods above, thereby at least ensuring a spread of questions viewed by the patient as appropriate. Alternatively the domain groups of questions can be reduced by a subjective approach of coalescing similar items into broader questions that encompass the contributing concepts.
Once a list of questions has been identified a draft questionnaire can be written. The time basis of the questions must be specified: this needs to be short if the instrument is going to be of any value to detect clinical change following intervention.
Was this article helpful?