Assessing non-response bias in pediatric palliative care research.

ABSTRACT

National experts have recognized a need for increased research in pediatric palliative care. However, when conducting research it is important to use rigorous methods, report significant and non-significant findings, and include information on responders and non-responders. Most studies do not present information on non-responders, yet this is critical as the results many not be generalizable if there are inherent differences between the two groups. Using survey data from parents whose children with life-limiting illnesses were enrolled in Florida's publicly funded pediatric palliative care program called Partners in Care Together for Kids; this study investigates whether non-response bias exists, and if so, what characteristics are associated with non-response. Bivariate and multivariate analyses were conducted to determine whether individual characteristics differed between responders and non-responders. Throughout our analyses, we conducted the analyses using different ways in which 'non-response' can be defined. Our results suggest that regardless of how non-response is defined, Black, non-Hispanic parents were less likely to participate than White non-Hispanic parents. However, we also found that of the Black, non-Hispanic parents who did not participate, their primary reason for doing so was that they had non-working or disconnected phone numbers. Only 3% of the Black, non-Hispanic parents who did not participate flatly refused. Information from this study can be used to design interventions aimed at increasing minority participation in pediatric palliative care research.