Does selective migration bias the health impact assessment of urban regeneration programmes in cross-sectional studies? Findings from a Dutch case study.

Abstract

We examined if the assessment of the health impact of a national Dutch regeneration programme depends on using either a repeated cross-sectional or longitudinal study design. This is important as only the latter design can incorporate migration patterns. For both designs, we compared trends in medication use between target and control districts. We found differences in medication use trends to be modest under the longitudinal design, and not demonstrable under the repeated cross-sectional design. The observed differences were hardly influenced by migration patterns. We conclude that in the Netherlands migration patterns had little effect on the health impact assessment of this national urban regeneration programme, so either the cross-sectional or longitudinal evaluation study design will do.