a solution to de-seasonalizing data? input appreciated

hi to all,
i'm trying to fit a model for the admission rate to our ED by using time series in STATA, which is fairly new to me and i would like to take your input about deseasonalizing.
data consists of 2 years of univariate data with weekly seasonality, where saturdays and sundays show more admission as seen in attachment
to overcome this 7 day seasonality, first i differenced by 7 and built the model with d=7 but the residuals failed to show white noise with portmanteau test
sarima model with 7 days differencing resulted same
but i modelel with regression analysis with seasonal dummy variables (d1-d7 for 7 days of week (with the command regress totalhasta d1 d2 d3 d4 d5 d6 d7 L(1/7).VAR_NAME, noconstant) (thanks to Prof Hansen and his lecture @ https://www.ssc.wisc.edu/~bhansen/390/390Lecture14.pdf), residuals show white nose and model shows better fit with lower AIC and BIC tests compared to other 2 models.
so it it good practice to use dummy variables iinstead of differencing with the seasonality length, sarima models to overcome seasonality?
thanks in advance...