Article type
Year
Abstract
Introduction: Conducting systematic reviews of complex interventions is frequently complicated by difficulties in defining the intervention in a way which is accurate, meaningful and applicable. This is a particular problem in rehabilitation research where a clear theoretical basis for describing and classifying interventions is often lacking. While attempting a systematic review of outpatient stroke rehabilitation services we first needed to establish a descriptive taxonomy of interventions.
Objectives: To develop a practical classification of out-patient stroke rehabilitation interventions.
Methods: We began with a review question of whether intervention by a rehabilitation worker would reduce disability and handicap after stroke (compared with no routine intervention) and identified relevant trials in the Cochrane Stroke Group trials register. We then contacted the trialists' to obtain standard descriptive details and applied a Delphi process to develop and apply a simple taxonomy of the trial interventions. The final version included the following intervention characteristics; staffing, type (domain), target patient group, and mode of delivery of the intervention. A similarity index was calculated comparing each trial with all others and a numerical dendrogram used to express the relationship between interventions (Armitage & Berry, Statistical Methods in Medical Research 1987).
Results: Two reviewers identified 29 relevant trials covering a heterogeneous group of 37 interventions. The dendrogram indicated two large clusters of interventions with a high similarity index (>0.8);
1. "therapy rehabilitation" - delivered by a physiotherapist, occupational therapist or team with the aim of altering task-orientated behaviour (e.g. walking, dressing).
2. "stroke liaison worker" - delivered by a nurse, social worker or volunteer to provide information and improve social support.
The remaining interventions differed from the above (similarity index <0.6) and included specific programmes of information provision, psychological treatments (e.g. counselling) and social support.
Discussion: A simple descriptive analysis can be used to help plan systematic reviews. Four reviews are now planned which will use this classification.
Objectives: To develop a practical classification of out-patient stroke rehabilitation interventions.
Methods: We began with a review question of whether intervention by a rehabilitation worker would reduce disability and handicap after stroke (compared with no routine intervention) and identified relevant trials in the Cochrane Stroke Group trials register. We then contacted the trialists' to obtain standard descriptive details and applied a Delphi process to develop and apply a simple taxonomy of the trial interventions. The final version included the following intervention characteristics; staffing, type (domain), target patient group, and mode of delivery of the intervention. A similarity index was calculated comparing each trial with all others and a numerical dendrogram used to express the relationship between interventions (Armitage & Berry, Statistical Methods in Medical Research 1987).
Results: Two reviewers identified 29 relevant trials covering a heterogeneous group of 37 interventions. The dendrogram indicated two large clusters of interventions with a high similarity index (>0.8);
1. "therapy rehabilitation" - delivered by a physiotherapist, occupational therapist or team with the aim of altering task-orientated behaviour (e.g. walking, dressing).
2. "stroke liaison worker" - delivered by a nurse, social worker or volunteer to provide information and improve social support.
The remaining interventions differed from the above (similarity index <0.6) and included specific programmes of information provision, psychological treatments (e.g. counselling) and social support.
Discussion: A simple descriptive analysis can be used to help plan systematic reviews. Four reviews are now planned which will use this classification.