Dose-Response Analysis Using Generalized Propensity Score: An Application in Home Health Care

• Activity Number: 413
• Type: Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section for Statistical Programmers and Analysts
• Author(s): Iordan Slavov*+
• Companies: Visiting Nurse Service of New York
• Keywords: Dose-Response ; Generalized Propensity Score ; Home Health Care ; Physical Therapy ; R

Abstract

Dose-response analysis in observational studies can be invalidated by the presence of imbalances in the measured covariates. A popular method for removal of the biases in binary treatment problems is using the propensity score. One recent generalization for continuous treatment is the generalized propensity score (GPS), see e.g. Hirano and Imbens (2004) or Imai and van Dyk(2004).

The important clinical question of whether more intensive physical therapy (PT) improves medical outcomes in a home health care setting, and the limited research in this area, motivated us to study a large sample (n=3650) of patients, who received services at the Visiting Nurse Service of New York in 2009. By applying the GPS methodology, it was shown that increased number of PT visits per week leads to a statistically significant, although small, improvement in the physical functioning of the patients.