Text this: Risk selection in primary care: a cross-sectional fixed effect analysis of Swedish individual data