Consecutive Successes in a Time Frame

dkernen What about this:

Recovery for Count = 
VAR __referralDT = [Referral_DT Measure]
VAR __group = MAX('dimODisp'[ORecovered_N])
VAR __tmpTable1 = FILTER(ALLSELECTED('dimODisp'),[Referral_DT]<=__referralDT)
VAR __tmpTable1a = ADDCOLUMNS(__tmpTable1,"__Index",COUNTROWS(FILTER(__tmpTable1,[Referral_DT]<=EARLIER([Referral_DT]))))
VAR __tmpTable1b = FILTER(__tmpTable1a,[ORecovered_N] = __group)
VAR __tmpTable2 = 
    ADDCOLUMNS(
        __tmpTable1b,
        "__diff",[__Index] - MAXX(FILTER(__tmpTable1b,[__Index]<EARLIER([__Index])),[__Index]))
VAR __max = MAXX(__tmpTable2,[__Index])
VAR __maxStart = MAXX(FILTER(__tmpTable2,[__diff]>1),[__Index])
VAR __tmpTable3 = FILTER(__tmpTable2,[__Index]>=__maxStart)
RETURN
    SWITCH(TRUE(),
        __group = 0,0,
        ISBLANK(__max),1,
        __max=__maxStart,1,
        COUNTROWS(__tmpTable3)
    )

The final solution for this was posted as a Quick Measure Gallery Entry, Bride of Cthulhu!