Add testing to check for overlap between medical claim / pharmacy claim and eligibility (#943)

* first pass at overlap checks

* update checks

* linting

* gitkeep

* refactor testing, update severity level

* lint

* database compatibility

* swap coalesce for case when

* fix join

Author: krishyfishy

tests/.gitkeep

@@ -0,0 +1,70 @@
+{{ config(
+     enabled = var('claims_preprocessing_enabled',var('claims_enabled',var('tuva_marts_enabled',False)))
+ | as_bool,
+     tags = ['dqi', 'tuva_dqi_sev_2', 'dqi_service_categories', 'dqi_ccsr', 'dqi_cms_chronic_conditions',
+            'dqi_tuva_chronic_conditions', 'dqi_cms_hccs', 'dqi_ed_classification',
+            'dqi_financial_pmpm', 'dqi_quality_measures', 'dqi_readmission'],
+     severity = 'warn'
+   )
+}}
+
+with eligibility as (
+    select distinct
+        person_id
+        , data_source
+    from {{ ref('input_layer__eligibility') }}
+)
+
+, medical_claims as (
+    select distinct
+        person_id
+        , data_source
+    from {{ ref('input_layer__medical_claim') }}
+)
+
+, mc_records_check as (
+    select
+        data_source
+        , count(*) as n_rows
+    from medical_claims
+    group by data_source
+)
+
+, elig_records_check as (
+    select
+        data_source
+        , count(*) as n_rows
+    from eligibility
+    group by data_source
+)
+
+, overlap_check as (
+    select
+        m.data_source
+        , count(e.person_id) as n_rows
+    from medical_claims as m
+    left outer join eligibility as e
+    on m.person_id = e.person_id
+    and m.data_source = e.data_source
+    group by m.data_source
+)
+
+, final as (
+    select
+        oc.data_source
+        , oc.n_rows as n_overlapping_records
+        , case when (mc.n_rows = 0 or mc.n_rows is null) then 1 else 0 end as is_mc_empty
+        , case when (ec.n_rows = 0 or ec.n_rows is null) then 1 else 0 end as is_elig_empty
+    from overlap_check as oc
+    left outer join mc_records_check as mc
+    on oc.data_source = mc.data_source
+    left outer join elig_records_check as ec
+    on oc.data_source = ec.data_source
+)
+
+select
+    data_source
+    , n_overlapping_records
+from final
+where not (is_mc_empty = 1 or is_elig_empty = 1)
+and n_overlapping_records = 0

tests/medical_claim_eligibility_overlap_check.sql

@@ -0,0 +1,70 @@
+{{ config(
+     enabled = var('claims_preprocessing_enabled',var('claims_enabled',var('tuva_marts_enabled',False)))
+ | as_bool,
+     tags = ['dqi', 'tuva_dqi_sev_2', 'dqi_service_categories', 'dqi_ccsr', 'dqi_cms_chronic_conditions',
+            'dqi_tuva_chronic_conditions', 'dqi_cms_hccs', 'dqi_ed_classification',
+            'dqi_financial_pmpm', 'dqi_quality_measures', 'dqi_readmission'],
+     severity = 'warn'
+   )
+}}
+
+with eligibility as (
+    select distinct
+        person_id
+        , data_source
+    from {{ ref('input_layer__eligibility') }}
+)
+
+, pharmacy_claims as (
+    select distinct
+        person_id
+        , data_source
+    from {{ ref('input_layer__pharmacy_claim') }}
+)
+
+, pc_records_check as (
+    select
+        data_source
+        , count(*) as n_rows
+    from pharmacy_claims
+    group by data_source
+)
+
+, elig_records_check as (
+    select
+        data_source
+        , count(*) as n_rows
+    from eligibility
+    group by data_source
+)
+
+, overlap_check as (
+    select
+        p.data_source
+        , count(e.person_id) as n_rows
+    from pharmacy_claims as p
+    left outer join eligibility as e
+    on p.person_id = e.person_id
+    and p.data_source = e.data_source
+    group by p.data_source
+)
+
+, final as (
+    select
+        oc.data_source
+        , oc.n_rows as n_overlapping_records
+        , case when (pc.n_rows = 0 or pc.n_rows is null) then 1 else 0 end as is_pc_empty
+        , case when (ec.n_rows = 0 or ec.n_rows is null) then 1 else 0 end as is_elig_empty
+    from overlap_check as oc
+    left outer join pc_records_check as pc
+    on oc.data_source = pc.data_source
+    left outer join elig_records_check as ec
+    on oc.data_source = ec.data_source
+)
+
+select
+    data_source
+    , n_overlapping_records
+from final
+where not (is_pc_empty = 1 or is_elig_empty = 1)
+and n_overlapping_records = 0