Extending Difference-in-Difference to Analyze Changes in Bed-Hold Policy for Nursing Homes

• Activity Number: 436
• Type: Invited
• Date/Time: Wednesday, August 6, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Health Policy Statistics Section
• Author(s): Roee Gutman*+ and Orna Intrator and Anthony Lancaster and Chenyang Gu
• Companies: Brown University and University of Rochester and Brown University and Brown University
• Keywords: Difference-in-Difference ; Causal Inference ; Potential Outcomes ; Health Policy ; Multiple Imputation

Abstract

Healthcare researchers are often interested in estimating the impact of policy interventions. To estimate this impact they often compare units affected by the intervention to unaffected units. Difference-In-Differences (DID) methods have been frequently used to account for changes over time unrelated to the intervention. Using DID, the change experienced by the group subjected to the intervention is adjusted by the change experienced by the group not subjected to the intervention. The underlying assumption is that the time trend in the control group is an adequate proxy for the time trend that would have occurred in the treatment group in the absence of the policy intervention. DID may suffer from weaknesses when more than two time points are considered, when the outcomes are not normally distributed or are not scalar, and when the treatment effect is heterogeneous. We propose a new procedure that relies on multiply imputing the potential outcomes using past outcomes to overcome these weaknesses. We apply this method to estimate the impact of states' bed-hold policy for nursing homes, which reserve beds for hospitalized residents, on both mortality and re-hospitalization rate.