Update CCSR mart to v2025.1 mappings (#967)

dbt_project.yml

@@ -69,8 +69,8 @@ vars:
   # Set to "ip" for inpatient data or "op" for outpatient data (more info in the README).
   record_type: "ip"
   # CCSR version numbers from HCUP.
-  dxccsr_version: "'2023.1'"
-  prccsr_version: "'2023.1'"
+  dxccsr_version: "'2025.1'"
+  prccsr_version: "'2025.1'"
 
   ## CMS HCC data mart vars
   # The payment_year variable has a default of the current year.
@@ -258,12 +258,12 @@ seeds:
 
     value_sets:
       ccsr:
-        ccsr__dxccsr_v2023_1_body_systems:
-          +post-hook: "{{ load_versioned_seed('value_sets','ccsr__dxccsr_v2023_1_body_systems.csv') }}"
-        ccsr__dxccsr_v2023_1_cleaned_map:
-          +post-hook: "{{ load_versioned_seed('value_sets','ccsr__dxccsr_v2023_1_cleaned_map.csv') }}"
-        ccsr__prccsr_v2023_1_cleaned_map:
-          +post-hook: "{{ load_versioned_seed('value_sets','ccsr__prccsr_v2023_1_cleaned_map.csv') }}"
+        ccsr__dxccsr_v2025_1_body_systems:
+          +post-hook: "{{ load_versioned_seed('value_sets','ccsr__dxccsr_v2025_1_body_systems.csv') }}"
+        ccsr__dxccsr_v2025_1_cleaned_map:
+          +post-hook: "{{ load_versioned_seed('value_sets','ccsr__dxccsr_v2025_1_cleaned_map.csv') }}"
+        ccsr__prccsr_v2025_1_cleaned_map:
+          +post-hook: "{{ load_versioned_seed('value_sets','ccsr__prccsr_v2025_1_cleaned_map.csv') }}"
       ahrq_measures:
         pqi__measures:
           +post-hook: "{{ load_versioned_seed('value_sets','ahrq_measures__pqi_measures.csv') }}"

models/claims_preprocessing/encounters/staging/encounters__stg_medical_claim.sql

@@ -69,7 +69,7 @@ with ccs_release_year as (
     and g.duplicate_row_number = 1
   inner join {{ ref('encounters__patient_data_source_id') }} as d on m.person_id = d.person_id
     and m.data_source = d.data_source
-  left outer join {{ ref('ccsr__dxccsr_v2023_1_cleaned_map') }} as dx on m.diagnosis_code_1 = dx.icd_10_cm_code
+  left outer join {{ ref('ccsr__dxccsr_v2025_1_cleaned_map') }} as dx on m.diagnosis_code_1 = dx.icd_10_cm_code
   left outer join {{ ref('provider_data__provider') }} as p on m.facility_npi = p.npi
   left outer join {{ ref('terminology__nitos') }} as n on m.hcpcs_code = n.hcpcs_code
   left outer join {{ ref('terminology__ms_drg') }} as msdrg on m.drg_code_type = 'ms-drg' and m.drg_code = msdrg.ms_drg_code

models/claims_preprocessing/service_category/staging/service_category__stg_medical_claim.sql

@@ -43,7 +43,7 @@ with ccs_release_year as (
     , n.modality
     , m.data_source
   from {{ ref('normalized_input__medical_claim') }} as m
-  left outer join {{ ref('ccsr__dxccsr_v2023_1_cleaned_map') }} as dx on m.diagnosis_code_1 = dx.icd_10_cm_code
+  left outer join {{ ref('ccsr__dxccsr_v2025_1_cleaned_map') }} as dx on m.diagnosis_code_1 = dx.icd_10_cm_code
   left outer join {{ ref('provider_data__provider') }} as p on m.facility_npi = p.npi
   left outer join {{ ref('terminology__nitos') }} as n on m.hcpcs_code = n.hcpcs_code
   left outer join {{ ref('terminology__revenue_center') }} as r on m.revenue_center_code = r.revenue_center_code

models/data_marts/benchmarks/pre_predict/final/benchmarks__inpatient_input.sql

@@ -346,7 +346,7 @@ inner join first_last as fl on e.person_id = fl.person_id
   er.data_source = fl.data_source
   and
   fl.year_nbr = c.year
-left outer join {{ ref('ccsr__dxccsr_v2023_1_cleaned_map') }} as ccsr on e.primary_diagnosis_code = ccsr.icd_10_cm_code
+left outer join {{ ref('ccsr__dxccsr_v2025_1_cleaned_map') }} as ccsr on e.primary_diagnosis_code = ccsr.icd_10_cm_code
 left outer join {{ ref('reference_data__ansi_fips_state') }} as st_ab on p.state = st_ab.ansi_fips_state_abbreviation
 left outer join {{ ref('reference_data__ansi_fips_state') }} as st_full on p.state = st_full.ansi_fips_state_name
 left outer join {{ ref('terminology__race') }} as r on p.race = r.description

models/data_marts/ccsr/final/ccsr__long_condition_category.sql

@@ -17,7 +17,7 @@ with ccsr__dx_vertical_pivot as (
 
 , dxccsr_body_systems as (
 
-    select * from {{ ref('ccsr__dxccsr_v2023_1_body_systems') }}
+    select * from {{ ref('ccsr__dxccsr_v2025_1_body_systems') }}
 
 )
 

models/data_marts/ccsr/intermediate/ccsr__dx_vertical_pivot.sql

@@ -15,7 +15,7 @@ with codes as (
         {%- endfor %}
         , default_ccsr_category_ip
         , default_ccsr_category_op
-    from {{ ref('ccsr__dxccsr_v2023_1_cleaned_map') }}
+    from {{ ref('ccsr__dxccsr_v2025_1_cleaned_map') }}
 
 )
 

models/data_marts/ccsr/intermediate/ccsr__procedure_category_map.sql

@@ -16,6 +16,6 @@ select
     , ont.device
     , ont.qualifier
     , cast('{{ var('tuva_last_run') }}' as {{ dbt.type_timestamp() }}) as tuva_last_run
-from {{ ref('ccsr__prccsr_v2023_1_cleaned_map') }} as ccsr_map
+from {{ ref('ccsr__prccsr_v2025_1_cleaned_map') }} as ccsr_map
 left outer join {{ ref('terminology__icd10_pcs_cms_ontology') }} as ont
     on ccsr_map.icd_10_pcs = ont.icd10pcs_code

models/data_quality/dqi/mart_review/final/mart_review__ed_classification.sql

@@ -58,5 +58,5 @@ left outer join cte on e.facility_npi = cte.location_id
 left outer join {{ ref('ccsr__dx_vertical_pivot') }} as p
   on e.primary_diagnosis_code = p.code
   and p.ccsr_category_rank = 1
-left outer join {{ ref('ccsr__dxccsr_v2023_1_body_systems') }} as b on p.ccsr_parent_category = b.ccsr_parent_category
+left outer join {{ ref('ccsr__dxccsr_v2025_1_body_systems') }} as b on p.ccsr_parent_category = b.ccsr_parent_category
 where e.encounter_type = 'emergency department'

models/data_quality/dqi/mart_review/final/mart_review__inpatient.sql

@@ -42,7 +42,7 @@ with cte as (
     left outer join {{ ref('ccsr__dx_vertical_pivot') }} as p
       on e.primary_diagnosis_code = p.code
       and p.ccsr_category_rank = 1
-    left outer join {{ ref('ccsr__dxccsr_v2023_1_body_systems') }} as b
+    left outer join {{ ref('ccsr__dxccsr_v2025_1_body_systems') }} as b
       on p.ccsr_parent_category = b.ccsr_parent_category
     left outer join {{ ref('terminology__ms_drg_weights_los') }} as weights
       on e.drg_code = weights.ms_drg and {{ date_part('year', 'e.encounter_start_date') }} = cast (weights.fiscal_year as {{ dbt.type_numeric() }})

seeds/value_sets/ccsr/README_v2025_1_publish.md

@@ -0,0 +1,41 @@
+# CCSR v2025.1 seed data — publish to S3 before release
+
+The CCSR seeds in this folder are header-only. Their data is loaded at
+`dbt seed` time from the versioned `value-sets` bundle in public object storage
+via the `load_versioned_seed` post-hooks in `dbt_project.yml`.
+
+**This update bumps the CCSR mart from v2023.1 to v2025.1.** The v2025.1 data
+files are NOT yet published to object storage. Until they are published, any
+`dbt seed`/`dbt build` (including CI) will fail to load these three seeds.
+
+## Required before merge / release
+
+Publish these gzipped files to the `value-sets` bundle in all mirrored buckets
+(S3, GCS, Azure), alongside the existing `value-sets/<bundle-version>/` files:
+
+- `ccsr__dxccsr_v2025_1_body_systems.csv.gz`
+- `ccsr__dxccsr_v2025_1_cleaned_map.csv.gz`
+- `ccsr__prccsr_v2025_1_cleaned_map.csv.gz`
+
+Object keys must match the post-hook filenames, e.g.
+`s3://tuva-public-resources/value-sets/<bundle-version>/ccsr__dxccsr_v2025_1_cleaned_map.csv.gz`.
+Bump `tuva_seed_versions.value_sets` if publishing under a new bundle version.
+
+## Source and transformation
+
+Generated from HCUP CCSR v2025-1 (ICD-10 codes valid through 2025-09-30):
+
+- DXCCSR: `DXCCSR_v2025-1.zip` -> `DXCCSR_v2025-1.csv`
+- PRCCSR: `PRCCSR_v2025-1.zip` -> `PRCCSR_v2025-1.csv`
+- https://hcup-us.ahrq.gov/toolssoftware/ccsr/dxccsr.jsp
+- https://hcup-us.ahrq.gov/toolssoftware/ccsr/prccsr.jsp
+
+Cleaning applied to match the seed schema: strip the single quotes wrapping
+codes, rename headers to snake_case, drop the DXCCSR "Rationale for Default
+Assignment" column, and convert empty CCSR category cells to null. Row counts:
+dxccsr cleaned map 75,238; prccsr cleaned map 82,121; body_systems 22.
+
+`ccsr__dxccsr_v2025_1_body_systems` was carried over from v2023.1 (the CCSR
+parent-category set is stable). Re-verify the parent-category counts in each
+`parent_category_description` against `DXCCSR-Reference-File-v2025-1.xlsx`
+before release.

seeds/value_sets/ccsr/ccsr__dxccsr_v2025_1_body_systems.csv

@@ -0,0 +1 @@
+body_system,ccsr_parent_category,parent_category_description

seeds/value_sets/ccsr/ccsr__dxccsr_v2023_1_cleaned_map.csv → seeds/value_sets/ccsr/ccsr__dxccsr_v2025_1_cleaned_map.csv

@@ -1 +1 @@
-icd_10_cm_code,icd_10_cm_code_description,default_ccsr_category_ip,default_ccsr_category_description_ip,default_ccsr_category_op,default_ccsr_category_description_op,ccsr_category_1,ccsr_category_1_description,ccsr_category_2,ccsr_category_2_description,ccsr_category_3,ccsr_category_3_description,ccsr_category_4,ccsr_category_4_description,ccsr_category_5,ccsr_category_5_description,ccsr_category_6,ccsr_category_6_description
+icd_10_cm_code,icd_10_cm_code_description,default_ccsr_category_ip,default_ccsr_category_description_ip,default_ccsr_category_op,default_ccsr_category_description_op,ccsr_category_1,ccsr_category_1_description,ccsr_category_2,ccsr_category_2_description,ccsr_category_3,ccsr_category_3_description,ccsr_category_4,ccsr_category_4_description,ccsr_category_5,ccsr_category_5_description,ccsr_category_6,ccsr_category_6_description

seeds/value_sets/ccsr/ccsr_seeds.yml

@@ -1,14 +1,14 @@
 version: 2
 
 seeds:
-  - name: ccsr__dxccsr_v2023_1_body_systems
+  - name: ccsr__dxccsr_v2025_1_body_systems
     description: |
       A table containing descriptions of the top level categories used in the CCSR condition mapper.
       This was manually extracted from the User Guide PDF included with the DXCCSR SAS program.
     config:
       schema: |
         {%- if var('tuva_schema_prefix',None) != None -%}{{var('tuva_schema_prefix')}}_value_sets{% else %}value_sets{%- endif -%}
-      alias: _value_set_dxccsr_v2023_1_body_systems
+      alias: _value_set_dxccsr_v2025_1_body_systems
       tags:
         - ccsr
       enabled: "{{ var('claims_enabled', var('clinical_enabled', False)) }}"
@@ -27,15 +27,15 @@ seeds:
       - name: parent_category_description
         description: Description of the parent category in the CCSR condition mapper.
 
-  - name: ccsr__dxccsr_v2023_1_cleaned_map
+  - name: ccsr__dxccsr_v2025_1_cleaned_map
     description: |
       A table containing the CSV mapping file included with the DXCCSR SAS program.
       The original CSV was not natively seedable, and the file was transformed with the coded included in the README.
       For each ICD-10-CM code, the table contains up to six CCSR category codes and a default CCSR category code.
     config:
       schema: |
         {%- if var('tuva_schema_prefix',None) != None -%}{{var('tuva_schema_prefix')}}_value_sets{% else %}value_sets{%- endif -%}
-      alias: _value_set_dxccsr_v2023_1_cleaned_map
+      alias: _value_set_dxccsr_v2025_1_cleaned_map
       tags:
         - ccsr
       enabled: "{{ var('claims_enabled', var('clinical_enabled', False)) }}"
@@ -117,14 +117,14 @@ seeds:
       - name: ccsr_category_6_description
         description: Description of the sixth CCSR category code
 
-  - name: ccsr__prccsr_v2023_1_cleaned_map
+  - name: ccsr__prccsr_v2025_1_cleaned_map
     description: |
       A table containing the CSV mapping file included with the PRCCSR SAS program.
       The original CSV was not natively seedable, and the file was transformed with the coded included in the README.
     config:
       schema: |
         {%- if var('tuva_schema_prefix',None) != None -%}{{var('tuva_schema_prefix')}}_value_sets{% else %}value_sets{%- endif -%}
-      alias: _value_set_prccsr_v2023_1_cleaned_map
+      alias: _value_set_prccsr_v2025_1_cleaned_map
       tags:
         - ccsr
       enabled: "{{ var('claims_enabled', var('clinical_enabled', False)) }}"