Keywords: Difference-in-differences, causal inference, medical home
Difference-in-differences (DID) is a popular approach in longitudinal observational and quasi-experiment studies to estimate effects discrete treatment statuses. In many studies the treatment arm can have a range of dosage or exposure levels which can be approximated by a continuous distribution. For example, medical homeness is a continuous score for the extent to which a patient-centered medical home model is achieved. We discuss a causal difference-in-differences (DID) approach to estimating the semi-continuous treatment dosage effect. The proposed semiparametric approach extends the existing causal DID approach for discrete treatment and the generalized propensity score (GPS) in cross-sectional studies. The proposed approach allows for mixed-type designs as well as different propensity models. We applied the proposed approach to evaluate the dosage effect of medical homeness on health care utilization and quality rating outcomes.