Copyright 2016 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
In Reply Dr Kalakoti and colleagues raise 2 important questions about our analysis of outcomes among patients hospitalized during an ED return visit. The first is the effect of missing data in the HCUP data sets and the second is the potential for unadjusted confounding to bias the results of the study.
First, the amount of missing data for the variables used in our regression models was low in HCUP data sets, amounting to less than 0.01% for age, sex, and primary payer. In the Florida State Inpatient Database, race was missing in approximately 1% of cases, whereas in New York, race was missing in less than 0.01% of cases. These low amounts of missing data are unlikely to have systematically affected the assessment of outcomes in the study and the use of multiple imputation, as suggested, would have added little value.
Sabbatini AK, Basu A, Hsia RY. Hospitalization After a Return Visit to the Emergency Department—Reply. JAMA. 2016;316(2):223-224. doi:10.1001/jama.2016.5253