Impact of Home Health Nurse Communication with Physicians on Readmissions: An Instrumental Variable Approach

The hospital readmission reduction program (HRRP) is a new financial incentive-based program designed to hold hospitals accountable for high readmission rates. Improving visiting nurse-to-physician communication may reduce hospital readmission rates. We use an instrumental variable model to assess how the success of visiting nurse communication with the patient’s physician impacts a patient's probability of a follow-up readmission. We use Medicare claims data and electronic medical records from the Visiting Nurse Service of New York (VNSNY) in 2008 or 2009. We identify all patients (N=6,741) with congestive heart failure, Medicare fee-for-service, and who had a discharge from a short-term, acute care hospital within 8 days prior to the start of the home health care episode (so-called “post-acute home care”).

In the VNSNY electronic medical records, nurses documented all attempts to communicate with their patients’ physicians and provided additional comments in free-text indicating the result of each attempt.  Natural language processing was used to assign a category of communication “success” for each attempt and indicated that the “success” of communication varied substantially among the 15,518 attempts—from direct nurse-physician contact to failure to communicate or leave a message with the physician or staff. We hypothesize that communication failure increases the patient’s probability of readmission by not allowing the sharing of information necessary for coordination of care, which is an especially important activity after hospital discharge.

We use an outcome variable of if a readmission occurred following each communication attempt, controlling for hospital fixed effects, nurse fixed effects, and patient characteristics to remove potential confounding. Our independent variable of interest is the category of communication “success.” We instrument this category with the timing of the day that the communication occurred. The timing is predictive of communication quality because certain times of the day are busier for physicians and their staffs, or after hours. The timing of the communications was not found to predict readmissions except through the impact it has on the success of the communication. We use this instrument to address potential endogeneity and identify the impact that successful visiting nurse-physician communication has on a patient’s probability of a readmission.