“…We decided a priori to include age, sex, race, ethnicity, region, urban status, and year in the model, as they have been previously demonstrated to be associated with differences in both ED and psychiatric care. 3,6,7,11,12,16,[19][20][21][22][23][24] For the length of stay (LOS) analysis as our secondary outcome, we tested for trend the median visit length. Median LOS is used in place of the mean LOS as values were not normally distributed.…”