Background:

Poor handoffs may lead to adverse health outcomes. We sought to determine rates of self–reported handoffs from inpatient to outpatient providers, to examine predictors of successful handoffs, and to determine the association between successful handoffs and 30–day readmissions.

Methods:

Starting in 2010, we mandated that discharging providers electronically document communication (or lack thereof) with outpatient providers in the discharge worksheet of all medicine patients. They had to choose one of four options: (1) successful communication (verbally or electronically); (2) attempted but unsuccessful communication; (3) unsure if communication done; and (4) communication not attempted. When not done, the provider had to indicate the reason using structured choices (e.g., lack of outpatient provider), or free text. We excluded visits for which there was: lack of outpatient provider, outpatient provider was inpatient provider, planned admissions, discharges to other facilities, and in–hospital deaths. Thirty–day readmissions, LOS, and demographics were obtained from administrative databases; we used state–wide readmission data by APR–DRG and associated complexity–of–illness scores to determine expected readmission rates for each patient. We estimated household income using US census data by zip code.

Results:

Eight thousand and two hundred and seventy–seven discharges met inclusion criteria. Successful communication occurred in 36.5% of cases. In 19.6%, providers were unsure, if communication occurred. Unsuccessful attempts were reported in 8.0%. Communication was not done, because providers felt the discharge summary would be adequate (23.7%), outpatient provider was in our health system (8.8%), and patient planned to update the primary provider (2.0%). Only 0.7% felt that it was not needed. Free–text reasons for no communication were listed in <1%. Univariate predictors of successful communication were: patient cared for by hospitalists without housestaff (OR = 2.0; 95% CI 1.8–2.3), high–expected readmission rate (OR = 1.17, 1.10–1.24 per 10%), longer LOS (OR 1.04, 1.03–1.05 per day), and female sex (OR 1.12, 1.03–1.23). Age, race, and income were not associated with successful handoffs. Successful handoff was not associated with readmissions in univariate (OR 1.05, 0.93–1.18) or multivariate analysis adjusting for significant predictors of successful handoff (OR = 1.02, 0.90–1.16).

Conclusions:

Successful communication was reported in only 36% of discharges, with a higher rate on the nonteaching service. These data are consistent with data from outpatient settings suggesting that inpatient providers communicate inconsistently with referring providers about their hospitalized patients. We found no association between self–reported handoff behavior and readmission rates, suggesting that simply increasing the rate of interprovider communication at hospital discharge may not, in isolation, reduce readmission rates.