SQL OOH Point In Time Race

-- this is the table all point in time measures come from
select cd_race_census ,count(distinct id_prsn_child) from prtl.ooh_point_in_time_child 
where  point_in_time_date='2011-01-01' 
and date_type=2 and qry_type=2 
group by cd_race_census
order by cd_race_census

-- this is CA table
select rp.cd_race_census,count(distinct plc.child) from base.rptPlacement_Events  plc
join base.rptPlacement rp on rp.id_removal_episode_fact=plc.id_removal_episode_fact
join ref_last_dw_transfer dw on dw.cutoff_date=dw.cutoff_date
where  rp.removal_dt < '2011-01-01'  
		and  iif( rp.[18bday] < rp.discharge_dt 
					and rp.discharge_dt<=cutoff_date,rp.[18bday],rp.discharge_dt)>='2011-01-01' 
			and plc.begin_date <='2011-01-01'   and plc.end_date>='2011-01-01' 
	and not exists (select * from vw_nondcfs_combine_adjacent_segments nd 
								where nd.id_prsn=rp.child
									and rp.removal_dt between nd.cust_begin  and nd.cust_end 
									and rp.discharge_dt between nd.cust_begin and nd.cust_end )
and rp.age_at_removal_mos <  12*18 and rp.age_at_removal_mos >=0 
group by rp.cd_race_census

These do not match exactly as in the top table, the episodes are "spliced" removing the "nondcfs" segments.