Recruitment bias and clinical characterstics of participants with severe cerebral palsy in a cross-sectional survey

Aim.  This article is a report of recruitment bias in a sample of 5–25-year-old patients with severe cerebral palsy.

Background.  The way in which study participants are recruited into research can be a source of bias.

Method.  A cross-sectional survey of 5–25-year-old patients with severe cerebral palsy using standardized questionnaires with parents/carers was undertaken in 2007/2008. A case register was used as the sampling frame, and 260 families were approached: 178/260 (68%) responded and 82/260 families never replied (non-respondents). Among responders: 127/178 (71%) opted in to the study, but only 123/127 were assessed, and 82/178 were opted out (or refused). Multivariable logistic regression giving odds ratios was used to study the association between participant characteristics and study outcomes (responders vs. non-responders; opting in vs. opting out; assessed vs. eligible, but not assessed).

Results.  Responders (compared with non-responders) were significantly more likely to have a family member with cerebral palsy who was male and resident in more affluent areas. Families who opted in (compared with those opting out and refusing) were more likely to have a family member with cerebral palsy and intellectual impairment and to reside in certain geographical areas. Families who were actually assessed (compared with all eligible, but not assessed) were more likely to have a family member with cerebral palsy and intellectual impairment.

Conclusion.  Several sources of bias were identified during recruitment for this study. This has implications for the interpretation and conclusions of surveys of people with disabilities and complex needs.