Variables for assessing shoulder pain following stroke: A Modified Delphi survey

A survey (using the Modified Delphi Technique) of a group of physical therapists with expertise in stroke management was conducted to generate variables to be considered for inclusion in a clinical instrument for the assessment of shoulder pain following stroke. Three groups of experts (clinicians, academics and authors) were selected (N = 21) to participate in the survey. A Modified Delphi Technique was used. In Questionnaire #1 experts were asked to generate variables, which were important in assessing individuals with painful shoulders post-stroke. In Questionnaire #2 they ranked the variables within each category in order of importance and then in Questionnaire #3 re-ranked a reduced number of variables using the same ranking system. Results: Questionnaire #1. A total of 19 questionnaires were completed. The number of variables generated were as follows: 25 demographic variables, 28 diagnostic groups, 34 evaluative variables and 33 predictive variables. Questionnaire #2 consisted of a list of all variables identified in the first questionnaire. These were rank ordered by each respondent. The number of items were reduced based on the mean ranks and the standard deviations. The items retained were compiled for Questionnaire #3. The mode was used to rerank the variables. There were 13 diagnostic variables, 16 evaluative variables, 11 predictive variables and 11 diagnostic groups. The modified Delphi technique has enabled us to generate variables of clinical importance to provide a framework for designing a comprehensive clinical instrument.