how is the PRESCRIBING pivot modifier_cd constraint supposed to work?

[dckc]

How could this ever work? The modifier_cd in basis and freq are never the same. The result we see, rather than one row with basis and freq columns, is two rows, one with basis, and one with freq:

 from i2b2medfact m inner join pcornet_med mo on m.concept_cd = mo.c_basecode 
...
    left join basis
    on m.encounter_num = basis.encounter_num
    and m.concept_cd = basis.concept_Cd
    and m.start_date = basis.start_date
    and m.provider_id = basis.provider_id
    and m.modifier_cd = basis.modifier_cd

    left join  freq
    on m.encounter_num = freq.encounter_num
    and m.concept_cd = freq.concept_Cd
    and m.start_date = freq.start_date
    and m.provider_id = freq.provider_id
    and m.modifier_cd = freq.modifier_cd

cc @mprittie @rwaitman

[dckc]

It would seem to make more sense to join on instance_num. I see somebody else requested this in #11.

[matthewjoss]

I changed the left joins to use instance_num instead of modifer_cd.

[dckc]

for reference: 58f138dabe46a485b26dacf51d9980a65d0a9cb5