Development and validation of a stroke-specific quality of life measure

Background and aim: This thesis describes a study to develop and validate a patient-centred, stroke-specific quality of life (QOL) measure. At the outset of the study, no such measure existed that had been developed using patientcentred methods. Such approaches are rapidly becoming recognised as the most valid in constructing patient-based outcome measures. Methods and results: In-depth interviews with 28 people who had had a stroke provided a wealth of qualitative data. These were analysed using the manual analysis tool 'Framework', and discussed with reference to the social model of health and the concept of biographical disruption. The qualitative data subsequently informed the content of the new QOL measure. Ani initial item pool of 418 items was generated. Members of the study team reached a consensus as to which items to retain. Those that were deemed ambiguous or too colloquial were rejected. This process resulted in a pre-test version of 140 items. These items, and their response options, were pre-tested for acceptability, understandability and comprehensiveness with 30 stroke patients using 'cognitive interviewing' techniques. This led to major revisions of the original measure. A 132-item version was interviewer-administered to 100 stroke patients in an item reduction phase. Factor analysis and examination of the worst performing items led to a final measure (the Newcastle Stroke-specific Quality of Life measure (NEWSQOL)) consisting of 56 items across 11 domains (Feelings, Mobility, ADL/Self-care, Communication, Cognition, Emotion, Sleep, Fatigue, Interpersonal Relationships, Vision, and Pain/Sensation). The internal consistency of these domains was good (Cronbach's alpha range 0.72-0.90). Examination of item convergent and item discriminant validity indicated a scaling success rate of 95-100% across domains. In a separate validation phase, NEWSQOL was administered to a further 100 stroke patients, together with the Barthel Index, Nottingham Health Profile, and National Institutes of Health Stroke Scale. Evidence of construct validity was found for all NEWSQOL domains except Cognition. Test-retest reliability was also examined at this stage and the intraclass correlation coefficients were high for all domains (range 0.78-0.92). Internal consistency, and item convergent and discriminant validity, were re-examined at this phase. The value of alpha ranged from 0.71-0.90, and the scaling success rate was acceptable (85-100% across domains). Conclusions: NEWSQOL is an acceptable and reliable stroke-specific QOL measure with evidence of construct validity for all but one of its 11 domains. Moreover, since one of the most meaningful indications of validity is the extent to which relevant groups are involved in generating the content of a measure, one can be confident about NEWSQOL's validity for use in general stroke populations. Responsiveness of the measure was not tested in the present study but must be evaluated in future studies if it is to be used to assess within-person changes over time. As it stands, it can safely be used as a discriminative measure, for example in comparing an intervention group with a control group at one point in time. In addition, NEWSQOL could be used as a descriptive measure in everyday clinical practice to identify areas of QOL affected by stroke, or as a means of improving professional/client interaction.