Incorporating "first differences" into itsa after identifying unit root

[Karl Johnson]

Hi -- I am performing an interrupted time series using the itsa command in STATA. After testing for autocorrelation (using actest, as suggested by Linden) and incorporating the proper lag value into my itsa model, I ran dfuller (augmented dickey-fuller test) to test for stationary and discovered that I indeed have a unit root. Multiple online forums suggest taking the first difference to eliminate the unit root (which I have done through the "D." prefix before my primary outcome variable, see here pg 4), but I don't know how to then properly incorporate the first differenced values into the itsa command. Any help would be much appreciated.