Merge branch 'master' into v5.1

Impala/OMOP_Parquet.sql

@@ -0,0 +1,461 @@
+-- Use the search/replace regex in an editor to fix DATE columns:
+-- ([^ ]+) VARCHAR\(8\), \-\- DATE
+-- TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST($1 AS STRING), 1, 4), SUBSTR(CAST($1 AS STRING), 5, 2), SUBSTR(CAST($1 AS STRING), 7, 2)), 'UTC') AS $1,
+
+CREATE TABLE omop_cdm_parquet.concept
+STORED AS PARQUET
+AS
+SELECT
+ concept_id,
+ concept_name,
+ domain_id,
+ vocabulary_id,
+ concept_class_id,
+ standard_concept,
+ concept_code,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_start_date AS STRING), 1, 4), SUBSTR(CAST(valid_start_date AS STRING), 5, 2), SUBSTR(CAST(valid_start_date AS STRING), 7, 2)), 'UTC') AS valid_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_end_date AS STRING), 1, 4), SUBSTR(CAST(valid_end_date AS STRING), 5, 2), SUBSTR(CAST(valid_end_date AS STRING), 7, 2)), 'UTC') AS valid_end_date,
+ invalid_reason
+FROM omop_cdm.concept;
+
+CREATE TABLE omop_cdm_parquet.vocabulary
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.vocabulary;
+
+CREATE TABLE omop_cdm_parquet.domain
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.domain;
+
+CREATE TABLE omop_cdm_parquet.concept_class
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.concept_class;
+
+CREATE TABLE omop_cdm_parquet.concept_relationship
+STORED AS PARQUET
+AS
+SELECT
+ concept_id_1,
+ concept_id_2,
+ relationship_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_start_date AS STRING), 1, 4), SUBSTR(CAST(valid_start_date AS STRING), 5, 2), SUBSTR(CAST(valid_start_date AS STRING), 7, 2)), 'UTC') AS valid_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_end_date AS STRING), 1, 4), SUBSTR(CAST(valid_end_date AS STRING), 5, 2), SUBSTR(CAST(valid_end_date AS STRING), 7, 2)), 'UTC') AS valid_end_date,
+ invalid_reason
+FROM omop_cdm.concept_relationship;
+
+CREATE TABLE omop_cdm_parquet.relationship
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.relationship;
+
+CREATE TABLE omop_cdm_parquet.concept_synonym
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.concept_synonym;
+
+CREATE TABLE omop_cdm_parquet.concept_ancestor
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.concept_ancestor;
+
+CREATE TABLE omop_cdm_parquet.source_to_concept_map
+STORED AS PARQUET
+AS
+SELECT
+ source_code,
+ source_concept_id,
+ source_vocabulary_id,
+ source_code_description,
+ target_concept_id,
+ target_vocabulary_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_start_date AS STRING), 1, 4), SUBSTR(CAST(valid_start_date AS STRING), 5, 2), SUBSTR(CAST(valid_start_date AS STRING), 7, 2)), 'UTC') AS valid_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_end_date AS STRING), 1, 4), SUBSTR(CAST(valid_end_date AS STRING), 5, 2), SUBSTR(CAST(valid_end_date AS STRING), 7, 2)), 'UTC') AS valid_end_date,
+ invalid_reason
+FROM omop_cdm.source_to_concept_map;
+
+CREATE TABLE omop_cdm_parquet.drug_strength
+STORED AS PARQUET
+AS
+SELECT
+ drug_concept_id,
+ ingredient_concept_id,
+ amount_value, -- NUMERIC
+ amount_unit_concept_id,
+ numerator_value, -- NUMERIC
+ numerator_unit_concept_id,
+ denominator_value, -- NUMERIC
+ denominator_unit_concept_id,
+ box_size,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_start_date AS STRING), 1, 4), SUBSTR(CAST(valid_start_date AS STRING), 5, 2), SUBSTR(CAST(valid_start_date AS STRING), 7, 2)), 'UTC') AS valid_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_end_date AS STRING), 1, 4), SUBSTR(CAST(valid_end_date AS STRING), 5, 2), SUBSTR(CAST(valid_end_date AS STRING), 7, 2)), 'UTC') AS valid_end_date,
+ invalid_reason
+FROM omop_cdm.drug_strength;
+
+CREATE TABLE omop_cdm_parquet.cohort_definition
+STORED AS PARQUET
+AS
+SELECT
+ cohort_definition_id,
+ cohort_definition_name,
+ cohort_definition_description, -- TEXT
+ definition_type_concept_id,
+ cohort_definition_syntax, -- TEXT
+ subject_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cohort_initiation_date AS STRING), 1, 4), SUBSTR(CAST(cohort_initiation_date AS STRING), 5, 2), SUBSTR(CAST(cohort_initiation_date AS STRING), 7, 2)), 'UTC') AS cohort_initiation_date
+FROM omop_cdm.cohort_definition;
+
+CREATE TABLE omop_cdm_parquet.attribute_definition
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.attribute_definition;
+
+CREATE TABLE omop_cdm_parquet.cdm_source
+STORED AS PARQUET
+AS
+SELECT
+ cdm_source_name,
+ cdm_source_abbreviation,
+ cdm_holder,
+ source_description, -- TEXT
+ source_documentation_reference,
+ cdm_etl_reference,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(source_release_date AS STRING), 1, 4), SUBSTR(CAST(source_release_date AS STRING), 5, 2), SUBSTR(CAST(source_release_date AS STRING), 7, 2)), 'UTC') AS source_release_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cdm_release_date AS STRING), 1, 4), SUBSTR(CAST(cdm_release_date AS STRING), 5, 2), SUBSTR(CAST(cdm_release_date AS STRING), 7, 2)), 'UTC') AS cdm_release_date,
+ cdm_version,
+ vocabulary_version
+FROM omop_cdm.cdm_source;
+
+CREATE TABLE omop_cdm_parquet.person
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.person;
+
+CREATE TABLE omop_cdm_parquet.observation_period
+STORED AS PARQUET
+AS
+SELECT
+ observation_period_id,
+ person_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(observation_period_start_date AS STRING), 1, 4), SUBSTR(CAST(observation_period_start_date AS STRING), 5, 2), SUBSTR(CAST(observation_period_start_date AS STRING), 7, 2)), 'UTC') AS observation_period_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(observation_period_end_date AS STRING), 1, 4), SUBSTR(CAST(observation_period_end_date AS STRING), 5, 2), SUBSTR(CAST(observation_period_end_date AS STRING), 7, 2)), 'UTC') AS observation_period_end_date,
+ period_type_concept_id
+FROM omop_cdm.observation_period;
+
+CREATE TABLE omop_cdm_parquet.specimen
+STORED AS PARQUET
+AS
+SELECT
+ specimen_id,
+ person_id,
+ specimen_concept_id,
+ specimen_type_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(specimen_date AS STRING), 1, 4), SUBSTR(CAST(specimen_date AS STRING), 5, 2), SUBSTR(CAST(specimen_date AS STRING), 7, 2)), 'UTC') AS specimen_date,
+ specimen_time,
+ quantity, -- NUMERIC
+ unit_concept_id,
+ anatomic_site_concept_id,
+ disease_status_concept_id,
+ specimen_source_id,
+ specimen_source_value,
+ unit_source_value,
+ anatomic_site_source_value,
+ disease_status_source_value
+FROM omop_cdm.specimen;
+
+CREATE TABLE omop_cdm_parquet.death
+STORED AS PARQUET
+AS
+SELECT
+ person_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(death_date AS STRING), 1, 4), SUBSTR(CAST(death_date AS STRING), 5, 2), SUBSTR(CAST(death_date AS STRING), 7, 2)), 'UTC') AS death_date,
+ death_type_concept_id,
+ cause_concept_id,
+ cause_source_value,
+ cause_source_concept_id
+FROM omop_cdm.death;
+
+CREATE TABLE omop_cdm_parquet.visit_occurrence
+STORED AS PARQUET
+AS
+SELECT
+ visit_occurrence_id,
+ person_id,
+ visit_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(visit_start_date AS STRING), 1, 4), SUBSTR(CAST(visit_start_date AS STRING), 5, 2), SUBSTR(CAST(visit_start_date AS STRING), 7, 2)), 'UTC') AS visit_start_date,
+ visit_start_time,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(visit_end_date AS STRING), 1, 4), SUBSTR(CAST(visit_end_date AS STRING), 5, 2), SUBSTR(CAST(visit_end_date AS STRING), 7, 2)), 'UTC') AS visit_end_date,
+ visit_end_time,
+ visit_type_concept_id,
+ provider_id,
+ care_site_id,
+ visit_source_value,
+ visit_source_concept_id
+FROM omop_cdm.visit_occurrence;
+
+CREATE TABLE omop_cdm_parquet.procedure_occurrence
+STORED AS PARQUET
+AS
+SELECT
+ procedure_occurrence_id,
+ person_id,
+ procedure_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(procedure_date AS STRING), 1, 4), SUBSTR(CAST(procedure_date AS STRING), 5, 2), SUBSTR(CAST(procedure_date AS STRING), 7, 2)), 'UTC') AS procedure_date,
+ procedure_type_concept_id,
+ modifier_concept_id,
+ quantity,
+ provider_id,
+ visit_occurrence_id,
+ procedure_source_value,
+ procedure_source_concept_id,
+ qualifier_source_value
+FROM omop_cdm.procedure_occurrence;
+
+CREATE TABLE omop_cdm_parquet.drug_exposure
+STORED AS PARQUET
+AS
+SELECT
+ drug_exposure_id,
+ person_id,
+ drug_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(drug_exposure_start_date AS STRING), 1, 4), SUBSTR(CAST(drug_exposure_start_date AS STRING), 5, 2), SUBSTR(CAST(drug_exposure_start_date AS STRING), 7, 2)), 'UTC') AS drug_exposure_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(drug_exposure_end_date AS STRING), 1, 4), SUBSTR(CAST(drug_exposure_end_date AS STRING), 5, 2), SUBSTR(CAST(drug_exposure_end_date AS STRING), 7, 2)), 'UTC') AS drug_exposure_end_date,
+ drug_type_concept_id,
+ stop_reason,
+ refills,
+ quantity, -- NUMERIC
+ days_supply,
+ sig, -- TEXT
+ route_concept_id,
+ effective_drug_dose, -- NUMERIC
+ dose_unit_concept_id,
+ lot_number,
+ provider_id,
+ visit_occurrence_id,
+ drug_source_value,
+ drug_source_concept_id,
+ route_source_value,
+ dose_unit_source_value
+FROM omop_cdm.drug_exposure;
+
+CREATE TABLE omop_cdm_parquet.device_exposure
+STORED AS PARQUET
+AS
+SELECT
+ device_exposure_id,
+ person_id,
+ device_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(device_exposure_start_date AS STRING), 1, 4), SUBSTR(CAST(device_exposure_start_date AS STRING), 5, 2), SUBSTR(CAST(device_exposure_start_date AS STRING), 7, 2)), 'UTC') AS device_exposure_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(device_exposure_end_date AS STRING), 1, 4), SUBSTR(CAST(device_exposure_end_date AS STRING), 5, 2), SUBSTR(CAST(device_exposure_end_date AS STRING), 7, 2)), 'UTC') AS device_exposure_end_date,
+ device_type_concept_id,
+ unique_device_id,
+ quantity,
+ provider_id,
+ visit_occurrence_id,
+ device_source_value,
+ device_source_concept_id
+FROM omop_cdm.device_exposure;
+
+CREATE TABLE omop_cdm_parquet.condition_occurrence
+STORED AS PARQUET
+AS
+SELECT
+ condition_occurrence_id,
+ person_id,
+ condition_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(condition_start_date AS STRING), 1, 4), SUBSTR(CAST(condition_start_date AS STRING), 5, 2), SUBSTR(CAST(condition_start_date AS STRING), 7, 2)), 'UTC') AS condition_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(condition_end_date AS STRING), 1, 4), SUBSTR(CAST(condition_end_date AS STRING), 5, 2), SUBSTR(CAST(condition_end_date AS STRING), 7, 2)), 'UTC') AS condition_end_date,
+ condition_type_concept_id,
+ stop_reason,
+ provider_id,
+ visit_occurrence_id,
+ condition_source_value,
+ condition_source_concept_id
+FROM omop_cdm.condition_occurrence;
+
+CREATE TABLE omop_cdm_parquet.measurement
+STORED AS PARQUET
+AS
+SELECT
+ measurement_id,
+ person_id,
+ measurement_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(measurement_date AS STRING), 1, 4), SUBSTR(CAST(measurement_date AS STRING), 5, 2), SUBSTR(CAST(measurement_date AS STRING), 7, 2)), 'UTC') AS measurement_date,
+ measurement_time,
+ measurement_type_concept_id,
+ operator_concept_id,
+ value_as_number, -- NUMERIC
+ value_as_concept_id,
+ unit_concept_id,
+ range_low, -- NUMERIC
+ range_high, -- NUMERIC
+ provider_id,
+ visit_occurrence_id,
+ measurement_source_value,
+ measurement_source_concept_id,
+ unit_source_value,
+ value_source_value
+FROM omop_cdm.measurement;
+
+CREATE TABLE omop_cdm_parquet.note
+STORED AS PARQUET
+AS
+SELECT
+ note_id,
+ person_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(note_date AS STRING), 1, 4), SUBSTR(CAST(note_date AS STRING), 5, 2), SUBSTR(CAST(note_date AS STRING), 7, 2)), 'UTC') AS note_date,
+ note_time,
+ note_type_concept_id,
+ note_text, -- TEXT
+ provider_id,
+ visit_occurrence_id,
+ note_source_value
+FROM omop_cdm.note;
+
+CREATE TABLE omop_cdm_parquet.observation
+STORED AS PARQUET
+AS
+SELECT
+ observation_id,
+ person_id,
+ observation_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(observation_date AS STRING), 1, 4), SUBSTR(CAST(observation_date AS STRING), 5, 2), SUBSTR(CAST(observation_date AS STRING), 7, 2)), 'UTC') AS observation_date,
+ observation_time,
+ observation_type_concept_id,
+ value_as_number, -- NUMERIC
+ value_as_string,
+ value_as_concept_id,
+ qualifier_concept_id,
+ unit_concept_id,
+ provider_id,
+ visit_occurrence_id,
+ observation_source_value,
+ observation_source_concept_id ,
+ unit_source_value,
+ qualifier_source_value
+FROM omop_cdm.observation;
+
+CREATE TABLE omop_cdm_parquet.fact_relationship
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.fact_relationship;
+
+CREATE TABLE omop_cdm_parquet.`location`
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.`location`;
+
+CREATE TABLE omop_cdm_parquet.care_site
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.care_site;
+
+CREATE TABLE omop_cdm_parquet.provider
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.provider;
+
+CREATE TABLE omop_cdm_parquet.payer_plan_period
+STORED AS PARQUET
+AS
+SELECT
+ payer_plan_period_id,
+ person_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(payer_plan_period_start_date AS STRING), 1, 4), SUBSTR(CAST(payer_plan_period_start_date AS STRING), 5, 2), SUBSTR(CAST(payer_plan_period_start_date AS STRING), 7, 2)), 'UTC') AS payer_plan_period_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(payer_plan_period_end_date AS STRING), 1, 4), SUBSTR(CAST(payer_plan_period_end_date AS STRING), 5, 2), SUBSTR(CAST(payer_plan_period_end_date AS STRING), 7, 2)), 'UTC') AS payer_plan_period_end_date,
+ payer_source_value,
+ plan_source_value,
+ family_source_value
+FROM omop_cdm.payer_plan_period;
+
+
+/* The individual cost tables are being phased out and will disappear soon
+
+CREATE TABLE omop_cdm_parquet.visit_cost
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.visit_cost;
+
+CREATE TABLE omop_cdm_parquet.procedure_cost
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.procedure_cost;
+
+CREATE TABLE omop_cdm_parquet.drug_cost
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.drug_cost;
+
+CREATE TABLE omop_cdm_parquet.device_cost
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.device_cost;
+*/
+
+CREATE TABLE omop_cdm_parquet.cost
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.cost;
+
+CREATE TABLE omop_cdm_parquet.cohort
+STORED AS PARQUET
+AS
+SELECT
+ cohort_definition_id,
+ subject_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cohort_start_date AS STRING), 1, 4), SUBSTR(CAST(cohort_start_date AS STRING), 5, 2), SUBSTR(CAST(cohort_start_date AS STRING), 7, 2)), 'UTC') AS cohort_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cohort_end_date AS STRING), 1, 4), SUBSTR(CAST(cohort_end_date AS STRING), 5, 2), SUBSTR(CAST(cohort_end_date AS STRING), 7, 2)), 'UTC') AS cohort_end_date
+FROM omop_cdm.cohort;
+
+CREATE TABLE omop_cdm_parquet.cohort_attribute
+STORED AS PARQUET
+AS
+SELECT
+ cohort_definition_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cohort_start_date AS STRING), 1, 4), SUBSTR(CAST(cohort_start_date AS STRING), 5, 2), SUBSTR(CAST(cohort_start_date AS STRING), 7, 2)), 'UTC') AS cohort_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cohort_end_date AS STRING), 1, 4), SUBSTR(CAST(cohort_end_date AS STRING), 5, 2), SUBSTR(CAST(cohort_end_date AS STRING), 7, 2)), 'UTC') AS cohort_end_date,
+ subject_id,
+ attribute_definition_id,
+ value_as_number, -- NUMERIC
+ value_as_concept_id
+FROM omop_cdm.cohort_attribute;
+
+CREATE TABLE omop_cdm_parquet.drug_era
+STORED AS PARQUET
+AS
+SELECT
+ drug_era_id,
+ person_id,
+ drug_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(drug_era_start_date AS STRING), 1, 4), SUBSTR(CAST(drug_era_start_date AS STRING), 5, 2), SUBSTR(CAST(drug_era_start_date AS STRING), 7, 2)), 'UTC') AS drug_era_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(drug_era_end_date AS STRING), 1, 4), SUBSTR(CAST(drug_era_end_date AS STRING), 5, 2), SUBSTR(CAST(drug_era_end_date AS STRING), 7, 2)), 'UTC') AS drug_era_end_date,
+ drug_exposure_count,
+ gap_days
+FROM omop_cdm.drug_era;
+
+CREATE TABLE omop_cdm_parquet.dose_era
+STORED AS PARQUET
+AS
+SELECT
+ dose_era_id,
+ person_id,
+ drug_concept_id,
+ unit_concept_id,
+ dose_value, -- NUMERIC
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(dose_era_start_date AS STRING), 1, 4), SUBSTR(CAST(dose_era_start_date AS STRING), 5, 2), SUBSTR(CAST(dose_era_start_date AS STRING), 7, 2)), 'UTC') AS dose_era_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(dose_era_end_date AS STRING), 1, 4), SUBSTR(CAST(dose_era_end_date AS STRING), 5, 2), SUBSTR(CAST(dose_era_end_date AS STRING), 7, 2)), 'UTC') AS dose_era_end_date
+FROM omop_cdm.dose_era;
+
+CREATE TABLE omop_cdm_parquet.condition_era
+STORED AS PARQUET
+AS
+SELECT
+ condition_era_id,
+ person_id,
+ condition_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(condition_era_start_date AS STRING), 1, 4), SUBSTR(CAST(condition_era_start_date AS STRING), 5, 2), SUBSTR(CAST(condition_era_start_date AS STRING), 7, 2)), 'UTC') AS condition_era_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(condition_era_end_date AS STRING), 1, 4), SUBSTR(CAST(condition_era_end_date AS STRING), 5, 2), SUBSTR(CAST(condition_era_end_date AS STRING), 7, 2)), 'UTC') AS condition_era_end_date,
+ condition_occurrence_count
+FROM omop_cdm.condition_era;
+

Impala/OMOP_Parquet_v5.1.sql

@@ -0,0 +1,490 @@
+-- Use the search/replace regex in an editor to fix DATE columns:
+-- ([^ ]+) VARCHAR\(8\), \-\- DATE
+-- TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST($1 AS STRING), 1, 4), SUBSTR(CAST($1 AS STRING), 5, 2), SUBSTR(CAST($1 AS STRING), 7, 2)), 'UTC') AS $1,
+
+CREATE TABLE omop_cdm_parquet.concept
+STORED AS PARQUET
+AS
+SELECT
+ concept_id,
+ concept_name,
+ domain_id,
+ vocabulary_id,
+ concept_class_id,
+ standard_concept,
+ concept_code,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_start_date AS STRING), 1, 4), SUBSTR(CAST(valid_start_date AS STRING), 5, 2), SUBSTR(CAST(valid_start_date AS STRING), 7, 2)), 'UTC') AS valid_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_end_date AS STRING), 1, 4), SUBSTR(CAST(valid_end_date AS STRING), 5, 2), SUBSTR(CAST(valid_end_date AS STRING), 7, 2)), 'UTC') AS valid_end_date,
+ invalid_reason
+FROM omop_cdm.concept;
+
+CREATE TABLE omop_cdm_parquet.vocabulary
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.vocabulary;
+
+CREATE TABLE omop_cdm_parquet.domain
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.domain;
+
+CREATE TABLE omop_cdm_parquet.concept_class
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.concept_class;
+
+CREATE TABLE omop_cdm_parquet.concept_relationship
+STORED AS PARQUET
+AS
+SELECT
+ concept_id_1,
+ concept_id_2,
+ relationship_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_start_date AS STRING), 1, 4), SUBSTR(CAST(valid_start_date AS STRING), 5, 2), SUBSTR(CAST(valid_start_date AS STRING), 7, 2)), 'UTC') AS valid_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_end_date AS STRING), 1, 4), SUBSTR(CAST(valid_end_date AS STRING), 5, 2), SUBSTR(CAST(valid_end_date AS STRING), 7, 2)), 'UTC') AS valid_end_date,
+ invalid_reason
+FROM omop_cdm.concept_relationship;
+
+CREATE TABLE omop_cdm_parquet.relationship
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.relationship;
+
+CREATE TABLE omop_cdm_parquet.concept_synonym
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.concept_synonym;
+
+CREATE TABLE omop_cdm_parquet.concept_ancestor
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.concept_ancestor;
+
+CREATE TABLE omop_cdm_parquet.source_to_concept_map
+STORED AS PARQUET
+AS
+SELECT
+ source_code,
+ source_concept_id,
+ source_vocabulary_id,
+ source_code_description,
+ target_concept_id,
+ target_vocabulary_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_start_date AS STRING), 1, 4), SUBSTR(CAST(valid_start_date AS STRING), 5, 2), SUBSTR(CAST(valid_start_date AS STRING), 7, 2)), 'UTC') AS valid_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_end_date AS STRING), 1, 4), SUBSTR(CAST(valid_end_date AS STRING), 5, 2), SUBSTR(CAST(valid_end_date AS STRING), 7, 2)), 'UTC') AS valid_end_date,
+ invalid_reason
+FROM omop_cdm.source_to_concept_map;
+
+CREATE TABLE omop_cdm_parquet.drug_strength
+STORED AS PARQUET
+AS
+SELECT
+ drug_concept_id,
+ ingredient_concept_id,
+ amount_value, -- NUMERIC
+ amount_unit_concept_id,
+ numerator_value, -- NUMERIC
+ numerator_unit_concept_id,
+ denominator_value, -- NUMERIC
+ denominator_unit_concept_id,
+ box_size,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_start_date AS STRING), 1, 4), SUBSTR(CAST(valid_start_date AS STRING), 5, 2), SUBSTR(CAST(valid_start_date AS STRING), 7, 2)), 'UTC') AS valid_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(valid_end_date AS STRING), 1, 4), SUBSTR(CAST(valid_end_date AS STRING), 5, 2), SUBSTR(CAST(valid_end_date AS STRING), 7, 2)), 'UTC') AS valid_end_date,
+ invalid_reason
+FROM omop_cdm.drug_strength;
+
+CREATE TABLE omop_cdm_parquet.cohort_definition
+STORED AS PARQUET
+AS
+SELECT
+ cohort_definition_id,
+ cohort_definition_name,
+ cohort_definition_description, -- TEXT
+ definition_type_concept_id,
+ cohort_definition_syntax, -- TEXT
+ subject_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cohort_initiation_date AS STRING), 1, 4), SUBSTR(CAST(cohort_initiation_date AS STRING), 5, 2), SUBSTR(CAST(cohort_initiation_date AS STRING), 7, 2)), 'UTC') AS cohort_initiation_date
+FROM omop_cdm.cohort_definition;
+
+CREATE TABLE omop_cdm_parquet.attribute_definition
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.attribute_definition;
+
+CREATE TABLE omop_cdm_parquet.cdm_source
+STORED AS PARQUET
+AS
+SELECT
+ cdm_source_name,
+ cdm_source_abbreviation,
+ cdm_holder,
+ source_description, -- TEXT
+ source_documentation_reference,
+ cdm_etl_reference,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(source_release_date AS STRING), 1, 4), SUBSTR(CAST(source_release_date AS STRING), 5, 2), SUBSTR(CAST(source_release_date AS STRING), 7, 2)), 'UTC') AS source_release_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cdm_release_date AS STRING), 1, 4), SUBSTR(CAST(cdm_release_date AS STRING), 5, 2), SUBSTR(CAST(cdm_release_date AS STRING), 7, 2)), 'UTC') AS cdm_release_date,
+ cdm_version,
+ vocabulary_version
+FROM omop_cdm.cdm_source;
+
+CREATE TABLE omop_cdm_parquet.person
+STORED AS PARQUET
+AS
+SELECT
+ person_id,
+ gender_concept_id,
+ year_of_birth,
+ month_of_birth,
+ day_of_birth,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', CAST(year_of_birth AS STRING), SUBSTR(CONCAT('0', CAST(month_of_birth AS STRING)), -2), SUBSTR(CONCAT('0', CAST(day_of_birth AS STRING)), -2)), 'UTC') AS birth_datetime,
+ race_concept_id,
+ ethnicity_concept_id,
+ location_id,
+ provider_id,
+ care_site_id,
+ person_source_value,
+ gender_source_value,
+ gender_source_concept_id,
+ race_source_value,
+ race_source_concept_id,
+ ethnicity_source_value,
+ ethnicity_source_concept_id
+FROM omop_cdm.person;
+
+CREATE TABLE omop_cdm_parquet.observation_period
+STORED AS PARQUET
+AS
+SELECT
+ observation_period_id,
+ person_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(observation_period_start_date AS STRING), 1, 4), SUBSTR(CAST(observation_period_start_date AS STRING), 5, 2), SUBSTR(CAST(observation_period_start_date AS STRING), 7, 2)), 'UTC') AS observation_period_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(observation_period_start_date AS STRING), 1, 4), SUBSTR(CAST(observation_period_start_date AS STRING), 5, 2), SUBSTR(CAST(observation_period_start_date AS STRING), 7, 2)), 'UTC') AS observation_period_start_datetime,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(observation_period_end_date AS STRING), 1, 4), SUBSTR(CAST(observation_period_end_date AS STRING), 5, 2), SUBSTR(CAST(observation_period_end_date AS STRING), 7, 2)), 'UTC') AS observation_period_end_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(observation_period_end_date AS STRING), 1, 4), SUBSTR(CAST(observation_period_end_date AS STRING), 5, 2), SUBSTR(CAST(observation_period_end_date AS STRING), 7, 2)), 'UTC') AS observation_period_end_datetime,
+ period_type_concept_id
+FROM omop_cdm.observation_period;
+
+CREATE TABLE omop_cdm_parquet.specimen
+STORED AS PARQUET
+AS
+SELECT
+ specimen_id,
+ person_id,
+ specimen_concept_id,
+ specimen_type_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(specimen_date AS STRING), 1, 4), SUBSTR(CAST(specimen_date AS STRING), 5, 2), SUBSTR(CAST(specimen_date AS STRING), 7, 2)), 'UTC') AS specimen_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(specimen_date AS STRING), 1, 4), SUBSTR(CAST(specimen_date AS STRING), 5, 2), SUBSTR(CAST(specimen_date AS STRING), 7, 2)), 'UTC') AS specimen_datetime,
+ quantity, -- NUMERIC
+ unit_concept_id,
+ anatomic_site_concept_id,
+ disease_status_concept_id,
+ specimen_source_id,
+ specimen_source_value,
+ unit_source_value,
+ anatomic_site_source_value,
+ disease_status_source_value
+FROM omop_cdm.specimen;
+
+CREATE TABLE omop_cdm_parquet.death
+STORED AS PARQUET
+AS
+SELECT
+ person_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(death_date AS STRING), 1, 4), SUBSTR(CAST(death_date AS STRING), 5, 2), SUBSTR(CAST(death_date AS STRING), 7, 2)), 'UTC') AS death_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(death_date AS STRING), 1, 4), SUBSTR(CAST(death_date AS STRING), 5, 2), SUBSTR(CAST(death_date AS STRING), 7, 2)), 'UTC') AS death_datetime,
+ death_type_concept_id,
+ cause_concept_id,
+ cause_source_value,
+ cause_source_concept_id
+FROM omop_cdm.death;
+
+CREATE TABLE omop_cdm_parquet.visit_occurrence
+STORED AS PARQUET
+AS
+SELECT
+ visit_occurrence_id,
+ person_id,
+ visit_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(visit_start_date AS STRING), 1, 4), SUBSTR(CAST(visit_start_date AS STRING), 5, 2), SUBSTR(CAST(visit_start_date AS STRING), 7, 2)), 'UTC') AS visit_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(visit_start_date AS STRING), 1, 4), SUBSTR(CAST(visit_start_date AS STRING), 5, 2), SUBSTR(CAST(visit_start_date AS STRING), 7, 2)), 'UTC') AS visit_start_datetime,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(visit_end_date AS STRING), 1, 4), SUBSTR(CAST(visit_end_date AS STRING), 5, 2), SUBSTR(CAST(visit_end_date AS STRING), 7, 2)), 'UTC') AS visit_end_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(visit_end_date AS STRING), 1, 4), SUBSTR(CAST(visit_end_date AS STRING), 5, 2), SUBSTR(CAST(visit_end_date AS STRING), 7, 2)), 'UTC') AS visit_end_datetime,
+ visit_type_concept_id,
+ provider_id,
+ care_site_id,
+ visit_source_value,
+ visit_source_concept_id
+FROM omop_cdm.visit_occurrence;
+
+CREATE TABLE omop_cdm_parquet.procedure_occurrence
+STORED AS PARQUET
+AS
+SELECT
+ procedure_occurrence_id,
+ person_id,
+ procedure_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(procedure_date AS STRING), 1, 4), SUBSTR(CAST(procedure_date AS STRING), 5, 2), SUBSTR(CAST(procedure_date AS STRING), 7, 2)), 'UTC') AS procedure_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(procedure_date AS STRING), 1, 4), SUBSTR(CAST(procedure_date AS STRING), 5, 2), SUBSTR(CAST(procedure_date AS STRING), 7, 2)), 'UTC') AS procedure_datetime,
+ procedure_type_concept_id,
+ modifier_concept_id,
+ quantity,
+ provider_id,
+ visit_occurrence_id,
+ procedure_source_value,
+ procedure_source_concept_id,
+ qualifier_source_value
+FROM omop_cdm.procedure_occurrence;
+
+CREATE TABLE omop_cdm_parquet.drug_exposure
+STORED AS PARQUET
+AS
+SELECT
+ drug_exposure_id,
+ person_id,
+ drug_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(drug_exposure_start_date AS STRING), 1, 4), SUBSTR(CAST(drug_exposure_start_date AS STRING), 5, 2), SUBSTR(CAST(drug_exposure_start_date AS STRING), 7, 2)), 'UTC') AS drug_exposure_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(drug_exposure_start_date AS STRING), 1, 4), SUBSTR(CAST(drug_exposure_start_date AS STRING), 5, 2), SUBSTR(CAST(drug_exposure_start_date AS STRING), 7, 2)), 'UTC') AS drug_exposure_start_datetime,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(drug_exposure_end_date AS STRING), 1, 4), SUBSTR(CAST(drug_exposure_end_date AS STRING), 5, 2), SUBSTR(CAST(drug_exposure_end_date AS STRING), 7, 2)), 'UTC') AS drug_exposure_end_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(drug_exposure_end_date AS STRING), 1, 4), SUBSTR(CAST(drug_exposure_end_date AS STRING), 5, 2), SUBSTR(CAST(drug_exposure_end_date AS STRING), 7, 2)), 'UTC') AS drug_exposure_end_datetime,
+ drug_type_concept_id,
+ stop_reason,
+ refills,
+ quantity, -- NUMERIC
+ days_supply,
+ sig, -- TEXT
+ route_concept_id,
+ effective_drug_dose, -- NUMERIC
+ dose_unit_concept_id,
+ lot_number,
+ provider_id,
+ visit_occurrence_id,
+ drug_source_value,
+ drug_source_concept_id,
+ route_source_value,
+ dose_unit_source_value
+FROM omop_cdm.drug_exposure;
+
+CREATE TABLE omop_cdm_parquet.device_exposure
+STORED AS PARQUET
+AS
+SELECT
+ device_exposure_id,
+ person_id,
+ device_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(device_exposure_start_date AS STRING), 1, 4), SUBSTR(CAST(device_exposure_start_date AS STRING), 5, 2), SUBSTR(CAST(device_exposure_start_date AS STRING), 7, 2)), 'UTC') AS device_exposure_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(device_exposure_start_date AS STRING), 1, 4), SUBSTR(CAST(device_exposure_start_date AS STRING), 5, 2), SUBSTR(CAST(device_exposure_start_date AS STRING), 7, 2)), 'UTC') AS device_exposure_start_datetime,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(device_exposure_end_date AS STRING), 1, 4), SUBSTR(CAST(device_exposure_end_date AS STRING), 5, 2), SUBSTR(CAST(device_exposure_end_date AS STRING), 7, 2)), 'UTC') AS device_exposure_end_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(device_exposure_end_date AS STRING), 1, 4), SUBSTR(CAST(device_exposure_end_date AS STRING), 5, 2), SUBSTR(CAST(device_exposure_end_date AS STRING), 7, 2)), 'UTC') AS device_exposure_end_datetime,
+ device_type_concept_id,
+ unique_device_id,
+ quantity,
+ provider_id,
+ visit_occurrence_id,
+ device_source_value,
+ device_source_concept_id
+FROM omop_cdm.device_exposure;
+
+CREATE TABLE omop_cdm_parquet.condition_occurrence
+STORED AS PARQUET
+AS
+SELECT
+ condition_occurrence_id,
+ person_id,
+ condition_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(condition_start_date AS STRING), 1, 4), SUBSTR(CAST(condition_start_date AS STRING), 5, 2), SUBSTR(CAST(condition_start_date AS STRING), 7, 2)), 'UTC') AS condition_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(condition_start_date AS STRING), 1, 4), SUBSTR(CAST(condition_start_date AS STRING), 5, 2), SUBSTR(CAST(condition_start_date AS STRING), 7, 2)), 'UTC') AS condition_start_datetime,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(condition_end_date AS STRING), 1, 4), SUBSTR(CAST(condition_end_date AS STRING), 5, 2), SUBSTR(CAST(condition_end_date AS STRING), 7, 2)), 'UTC') AS condition_end_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(condition_end_date AS STRING), 1, 4), SUBSTR(CAST(condition_end_date AS STRING), 5, 2), SUBSTR(CAST(condition_end_date AS STRING), 7, 2)), 'UTC') AS condition_end_datetime,
+ condition_type_concept_id,
+ stop_reason,
+ provider_id,
+ visit_occurrence_id,
+ condition_source_value,
+ condition_source_concept_id
+FROM omop_cdm.condition_occurrence;
+
+CREATE TABLE omop_cdm_parquet.measurement
+STORED AS PARQUET
+AS
+SELECT
+ measurement_id,
+ person_id,
+ measurement_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(measurement_date AS STRING), 1, 4), SUBSTR(CAST(measurement_date AS STRING), 5, 2), SUBSTR(CAST(measurement_date AS STRING), 7, 2)), 'UTC') AS measurement_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(measurement_date AS STRING), 1, 4), SUBSTR(CAST(measurement_date AS STRING), 5, 2), SUBSTR(CAST(measurement_date AS STRING), 7, 2)), 'UTC') AS measurement_datetime,
+ measurement_type_concept_id,
+ operator_concept_id,
+ value_as_number, -- NUMERIC
+ value_as_concept_id,
+ unit_concept_id,
+ range_low, -- NUMERIC
+ range_high, -- NUMERIC
+ provider_id,
+ visit_occurrence_id,
+ measurement_source_value,
+ measurement_source_concept_id,
+ unit_source_value,
+ value_source_value
+FROM omop_cdm.measurement;
+
+CREATE TABLE omop_cdm_parquet.note
+STORED AS PARQUET
+AS
+SELECT
+ note_id,
+ person_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(note_date AS STRING), 1, 4), SUBSTR(CAST(note_date AS STRING), 5, 2), SUBSTR(CAST(note_date AS STRING), 7, 2)), 'UTC') AS note_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(note_date AS STRING), 1, 4), SUBSTR(CAST(note_date AS STRING), 5, 2), SUBSTR(CAST(note_date AS STRING), 7, 2)), 'UTC') AS note_datetime,
+ note_type_concept_id,
+ note_text, -- TEXT
+ provider_id,
+ visit_occurrence_id,
+ note_source_value
+FROM omop_cdm.note;
+
+CREATE TABLE omop_cdm_parquet.observation
+STORED AS PARQUET
+AS
+SELECT
+ observation_id,
+ person_id,
+ observation_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(observation_date AS STRING), 1, 4), SUBSTR(CAST(observation_date AS STRING), 5, 2), SUBSTR(CAST(observation_date AS STRING), 7, 2)), 'UTC') AS observation_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(observation_date AS STRING), 1, 4), SUBSTR(CAST(observation_date AS STRING), 5, 2), SUBSTR(CAST(observation_date AS STRING), 7, 2)), 'UTC') AS observation_datetime,
+ observation_type_concept_id,
+ value_as_number, -- NUMERIC
+ value_as_string,
+ value_as_concept_id,
+ qualifier_concept_id,
+ unit_concept_id,
+ provider_id,
+ visit_occurrence_id,
+ observation_source_value,
+ observation_source_concept_id ,
+ unit_source_value,
+ qualifier_source_value
+FROM omop_cdm.observation;
+
+CREATE TABLE omop_cdm_parquet.fact_relationship
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.fact_relationship;
+
+CREATE TABLE omop_cdm_parquet.`location`
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.`location`;
+
+CREATE TABLE omop_cdm_parquet.care_site
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.care_site;
+
+CREATE TABLE omop_cdm_parquet.provider
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.provider;
+
+CREATE TABLE omop_cdm_parquet.payer_plan_period
+STORED AS PARQUET
+AS
+SELECT
+ payer_plan_period_id,
+ person_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(payer_plan_period_start_date AS STRING), 1, 4), SUBSTR(CAST(payer_plan_period_start_date AS STRING), 5, 2), SUBSTR(CAST(payer_plan_period_start_date AS STRING), 7, 2)), 'UTC') AS payer_plan_period_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(payer_plan_period_end_date AS STRING), 1, 4), SUBSTR(CAST(payer_plan_period_end_date AS STRING), 5, 2), SUBSTR(CAST(payer_plan_period_end_date AS STRING), 7, 2)), 'UTC') AS payer_plan_period_end_date,
+ payer_source_value,
+ plan_source_value,
+ family_source_value
+FROM omop_cdm.payer_plan_period;
+
+
+/* The individual cost tables are being phased out and will disappear soon
+
+CREATE TABLE omop_cdm_parquet.visit_cost
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.visit_cost;
+
+CREATE TABLE omop_cdm_parquet.procedure_cost
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.procedure_cost;
+
+CREATE TABLE omop_cdm_parquet.drug_cost
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.drug_cost;
+
+CREATE TABLE omop_cdm_parquet.device_cost
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.device_cost;
+*/
+
+CREATE TABLE omop_cdm_parquet.cost
+STORED AS PARQUET
+AS
+SELECT * from omop_cdm.cost;
+
+CREATE TABLE omop_cdm_parquet.cohort
+STORED AS PARQUET
+AS
+SELECT
+ cohort_definition_id,
+ subject_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cohort_start_date AS STRING), 1, 4), SUBSTR(CAST(cohort_start_date AS STRING), 5, 2), SUBSTR(CAST(cohort_start_date AS STRING), 7, 2)), 'UTC') AS cohort_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cohort_end_date AS STRING), 1, 4), SUBSTR(CAST(cohort_end_date AS STRING), 5, 2), SUBSTR(CAST(cohort_end_date AS STRING), 7, 2)), 'UTC') AS cohort_end_date
+FROM omop_cdm.cohort;
+
+CREATE TABLE omop_cdm_parquet.cohort_attribute
+STORED AS PARQUET
+AS
+SELECT
+ cohort_definition_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cohort_start_date AS STRING), 1, 4), SUBSTR(CAST(cohort_start_date AS STRING), 5, 2), SUBSTR(CAST(cohort_start_date AS STRING), 7, 2)), 'UTC') AS cohort_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(cohort_end_date AS STRING), 1, 4), SUBSTR(CAST(cohort_end_date AS STRING), 5, 2), SUBSTR(CAST(cohort_end_date AS STRING), 7, 2)), 'UTC') AS cohort_end_date,
+ subject_id,
+ attribute_definition_id,
+ value_as_number, -- NUMERIC
+ value_as_concept_id
+FROM omop_cdm.cohort_attribute;
+
+CREATE TABLE omop_cdm_parquet.drug_era
+STORED AS PARQUET
+AS
+SELECT
+ drug_era_id,
+ person_id,
+ drug_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(drug_era_start_date AS STRING), 1, 4), SUBSTR(CAST(drug_era_start_date AS STRING), 5, 2), SUBSTR(CAST(drug_era_start_date AS STRING), 7, 2)), 'UTC') AS drug_era_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(drug_era_end_date AS STRING), 1, 4), SUBSTR(CAST(drug_era_end_date AS STRING), 5, 2), SUBSTR(CAST(drug_era_end_date AS STRING), 7, 2)), 'UTC') AS drug_era_end_date,
+ drug_exposure_count,
+ gap_days
+FROM omop_cdm.drug_era;
+
+CREATE TABLE omop_cdm_parquet.dose_era
+STORED AS PARQUET
+AS
+SELECT
+ dose_era_id,
+ person_id,
+ drug_concept_id,
+ unit_concept_id,
+ dose_value, -- NUMERIC
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(dose_era_start_date AS STRING), 1, 4), SUBSTR(CAST(dose_era_start_date AS STRING), 5, 2), SUBSTR(CAST(dose_era_start_date AS STRING), 7, 2)), 'UTC') AS dose_era_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(dose_era_end_date AS STRING), 1, 4), SUBSTR(CAST(dose_era_end_date AS STRING), 5, 2), SUBSTR(CAST(dose_era_end_date AS STRING), 7, 2)), 'UTC') AS dose_era_end_date
+FROM omop_cdm.dose_era;
+
+CREATE TABLE omop_cdm_parquet.condition_era
+STORED AS PARQUET
+AS
+SELECT
+ condition_era_id,
+ person_id,
+ condition_concept_id,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(condition_era_start_date AS STRING), 1, 4), SUBSTR(CAST(condition_era_start_date AS STRING), 5, 2), SUBSTR(CAST(condition_era_start_date AS STRING), 7, 2)), 'UTC') AS condition_era_start_date,
+ TO_UTC_TIMESTAMP(CONCAT_WS('-', SUBSTR(CAST(condition_era_end_date AS STRING), 1, 4), SUBSTR(CAST(condition_era_end_date AS STRING), 5, 2), SUBSTR(CAST(condition_era_end_date AS STRING), 7, 2)), 'UTC') AS condition_era_end_date,
+ condition_occurrence_count
+FROM omop_cdm.condition_era;
+

Impala/README.md

@@ -43,9 +43,16 @@ hadoop fs -chmod +w synpuf
 impala-shell -d omop_cdm -f DataImport/OMOP_CDM_synpuf_load_Impala.sql --var=OMOP_SYNPUF_PATH=/user/$USER/synpuf
 ```
 
-4. Run simple queries to sanity check.
+4. Convert to Parquet format.
 
 ```bash
-impala-shell -d omop_cdm -q 'SELECT COUNT(1) FROM concept'
+impala-shell -q 'CREATE DATABASE omop_cdm_parquet'
-impala-shell -d omop_cdm -q 'SELECT COUNT(1) FROM person'
+impala-shell -f OMOP_Parquet.sql
+```
+
+5. Run simple queries to sanity check.
+
+```bash
+impala-shell -d omop_cdm_parquet -q 'SELECT COUNT(1) FROM concept'
+impala-shell -d omop_cdm_parquet -q 'SELECT COUNT(1) FROM person'
 ```

Oracle/OMOP CDM constraints - Oracle.sql

@@ -154,13 +154,7 @@ Standardized health economics
 
 ALTER TABLE payer_plan_period ADD CONSTRAINT xpk_payer_plan_period PRIMARY KEY ( payer_plan_period_id ) ;
 
-ALTER TABLE visit_cost ADD CONSTRAINT xpk_visit_cost PRIMARY KEY ( visit_cost_id ) ;
+ALTER TABLE cost ADD CONSTRAINT xpk_visit_cost PRIMARY KEY ( cost_id ) ;
-
-ALTER TABLE procedure_cost ADD CONSTRAINT xpk_procedure_cost PRIMARY KEY ( procedure_cost_id ) ;
-
-ALTER TABLE drug_cost ADD CONSTRAINT xpk_drug_cost PRIMARY KEY ( drug_cost_id ) ;
-
-ALTER TABLE device_cost ADD CONSTRAINT xpk_device_cost PRIMARY KEY ( device_cost_id ) ;
 
 
 
@@ -474,39 +468,9 @@ Standardized health economics
 
 ALTER TABLE payer_plan_period ADD CONSTRAINT fpk_payer_plan_period FOREIGN KEY (person_id)  REFERENCES person (person_id);
 
+ALTER TABLE cost ADD CONSTRAINT fpk_visit_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
 
-ALTER TABLE visit_cost ADD CONSTRAINT fpk_visit_cost_id FOREIGN KEY (visit_occurrence_id)  REFERENCES visit_occurrence (visit_occurrence_id);
+ALTER TABLE cost ADD CONSTRAINT fpk_visit_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-ALTER TABLE visit_cost ADD CONSTRAINT fpk_visit_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE visit_cost ADD CONSTRAINT fpk_visit_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_id FOREIGN KEY (procedure_occurrence_id)  REFERENCES procedure_occurrence (procedure_occurrence_id);
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_revenue FOREIGN KEY (revenue_code_concept_id)  REFERENCES concept (concept_id);
-
-
-ALTER TABLE drug_cost ADD CONSTRAINT fpk_drug_cost_id FOREIGN KEY (drug_exposure_id)  REFERENCES drug_exposure (drug_exposure_id);
-
-ALTER TABLE drug_cost ADD CONSTRAINT fpk_drug_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE drug_cost ADD CONSTRAINT fpk_drug_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-
-ALTER TABLE device_cost ADD CONSTRAINT fpk_device_cost_id FOREIGN KEY (device_exposure_id)  REFERENCES device_exposure (device_exposure_id);
-
-ALTER TABLE device_cost ADD CONSTRAINT fpk_device_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE device_cost ADD CONSTRAINT fpk_device_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-
-
-
 
 
 /************************

PostgreSQL/OMOP CDM constraints - PostgreSQL.sql

@@ -155,14 +155,7 @@ Standardized health economics
 
 ALTER TABLE payer_plan_period ADD CONSTRAINT xpk_payer_plan_period PRIMARY KEY ( payer_plan_period_id ) ;
 
-ALTER TABLE visit_cost ADD CONSTRAINT xpk_visit_cost PRIMARY KEY ( visit_cost_id ) ;
+ALTER TABLE cost ADD CONSTRAINT xpk_visit_cost PRIMARY KEY ( cost_id ) ;
-
-ALTER TABLE procedure_cost ADD CONSTRAINT xpk_procedure_cost PRIMARY KEY ( procedure_cost_id ) ;
-
-ALTER TABLE drug_cost ADD CONSTRAINT xpk_drug_cost PRIMARY KEY ( drug_cost_id ) ;
-
-ALTER TABLE device_cost ADD CONSTRAINT xpk_device_cost PRIMARY KEY ( device_cost_id ) ;
-
 
 
 /************************
@@ -475,39 +468,9 @@ Standardized health economics
 
 ALTER TABLE payer_plan_period ADD CONSTRAINT fpk_payer_plan_period FOREIGN KEY (person_id)  REFERENCES person (person_id);
 
+ALTER TABLE cost ADD CONSTRAINT fpk_visit_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
 
-ALTER TABLE visit_cost ADD CONSTRAINT fpk_visit_cost_id FOREIGN KEY (visit_occurrence_id)  REFERENCES visit_occurrence (visit_occurrence_id);
+ALTER TABLE cost ADD CONSTRAINT fpk_visit_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-ALTER TABLE visit_cost ADD CONSTRAINT fpk_visit_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE visit_cost ADD CONSTRAINT fpk_visit_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_id FOREIGN KEY (procedure_occurrence_id)  REFERENCES procedure_occurrence (procedure_occurrence_id);
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_revenue FOREIGN KEY (revenue_code_concept_id)  REFERENCES concept (concept_id);
-
-
-ALTER TABLE drug_cost ADD CONSTRAINT fpk_drug_cost_id FOREIGN KEY (drug_exposure_id)  REFERENCES drug_exposure (drug_exposure_id);
-
-ALTER TABLE drug_cost ADD CONSTRAINT fpk_drug_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE drug_cost ADD CONSTRAINT fpk_drug_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-
-ALTER TABLE device_cost ADD CONSTRAINT fpk_device_cost_id FOREIGN KEY (device_exposure_id)  REFERENCES device_exposure (device_exposure_id);
-
-ALTER TABLE device_cost ADD CONSTRAINT fpk_device_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE device_cost ADD CONSTRAINT fpk_device_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-
-
-
 
 
 /************************

README.md

@@ -1,6 +1,7 @@
 Common-Data-Model
 =================
 
+v5.1
 See full CDM specification file on our [Wiki](http://www.ohdsi.org/web/wiki/doku.php?id=documentation:cdm:single-page) or in the [CDM V5 PDF](https://github.com/OHDSI/CommonDataModel/blob/master/OMOP%20CDM%20v5.pdf)
 
 Release Notes
@@ -44,3 +45,10 @@ This version is bases on this CDM working group  [proposal](http://www.ohdsi.org
 **NOTE**  
 * note_datetime, not required
 
+=======
+This repo contains the definition of the OMOP Common Data Model. It supports the 4 SQL technologies: Impala, Oracle, Postgres and SQL Server. For each, the DDL, constraints and indexes (if appropirate) are defined. 
+
+Versions are defined using tagging and versioning. Full versions (V6, 7 etc.) are released irregularly after a major strategy change or use case coverage. It will be issued during an OHDSI Symposium. Major version (V5.1, 5.2 etc.) are released half yearly (1-Jul and 1-Jan). Those versions are not guaranteed to be backward compatible. Minor versions (V5.1.1, V5.1.2 etc.) are released irregularly and often, and contain small hot fixes or backward compatible changes to the last Major Version.
+
+
+See full CDM specification file on our [Wiki](http://www.ohdsi.org/web/wiki/doku.php?id=documentation:cdm:single-page).

Sql Server/OMOP CDM constraints - SQL Server.sql

@@ -155,14 +155,7 @@ Standardized health economics
 
 ALTER TABLE payer_plan_period ADD CONSTRAINT xpk_payer_plan_period PRIMARY KEY NONCLUSTERED ( payer_plan_period_id ) ;
 
-ALTER TABLE visit_cost ADD CONSTRAINT xpk_visit_cost PRIMARY KEY NONCLUSTERED ( visit_cost_id ) ;
+ALTER TABLE cost ADD CONSTRAINT xpk_visit_cost PRIMARY KEY NONCLUSTERED ( cost_id ) ;
-
-ALTER TABLE procedure_cost ADD CONSTRAINT xpk_procedure_cost PRIMARY KEY NONCLUSTERED ( procedure_cost_id ) ;
-
-ALTER TABLE drug_cost ADD CONSTRAINT xpk_drug_cost PRIMARY KEY NONCLUSTERED ( drug_cost_id ) ;
-
-ALTER TABLE device_cost ADD CONSTRAINT xpk_device_cost PRIMARY KEY NONCLUSTERED ( device_cost_id ) ;
-
 
 
 /************************
@@ -475,39 +468,9 @@ Standardized health economics
 
 ALTER TABLE payer_plan_period ADD CONSTRAINT fpk_payer_plan_period FOREIGN KEY (person_id)  REFERENCES person (person_id);
 
+ALTER TABLE cost ADD CONSTRAINT fpk_visit_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
 
-ALTER TABLE visit_cost ADD CONSTRAINT fpk_visit_cost_id FOREIGN KEY (visit_occurrence_id)  REFERENCES visit_occurrence (visit_occurrence_id);
+ALTER TABLE cost ADD CONSTRAINT fpk_visit_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-ALTER TABLE visit_cost ADD CONSTRAINT fpk_visit_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE visit_cost ADD CONSTRAINT fpk_visit_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_id FOREIGN KEY (procedure_occurrence_id)  REFERENCES procedure_occurrence (procedure_occurrence_id);
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-ALTER TABLE procedure_cost ADD CONSTRAINT fpk_procedure_cost_revenue FOREIGN KEY (revenue_code_concept_id)  REFERENCES concept (concept_id);
-
-
-ALTER TABLE drug_cost ADD CONSTRAINT fpk_drug_cost_id FOREIGN KEY (drug_exposure_id)  REFERENCES drug_exposure (drug_exposure_id);
-
-ALTER TABLE drug_cost ADD CONSTRAINT fpk_drug_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE drug_cost ADD CONSTRAINT fpk_drug_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-
-ALTER TABLE device_cost ADD CONSTRAINT fpk_device_cost_id FOREIGN KEY (device_exposure_id)  REFERENCES device_exposure (device_exposure_id);
-
-ALTER TABLE device_cost ADD CONSTRAINT fpk_device_cost_currency FOREIGN KEY (currency_concept_id)  REFERENCES concept (concept_id);
-
-ALTER TABLE device_cost ADD CONSTRAINT fpk_device_cost_period FOREIGN KEY (payer_plan_period_id)  REFERENCES payer_plan_period (payer_plan_period_id);
-
-
-
-
 
 
 /************************

Sql Server/OMOP CDM indexes required - SQL Server.sql

@@ -185,3 +185,178 @@ CREATE INDEX idx_dose_era_concept_id ON dose_era (drug_concept_id ASC);
 CREATE CLUSTERED INDEX idx_condition_era_person_id ON condition_era (person_id ASC);
 CREATE INDEX idx_condition_era_concept_id ON condition_era (condition_concept_id ASC);
 
+/****************************************************************************************************************************************
+*****************************************************	Non Cluster Index  **************************************************************
+ ****************************************************************************************************************************************
+ **********************************/
+
+ /* PERSON */
+
+USE [Camel_DB]
+GO
+	CREATE NONCLUSTERED INDEX [<PERSON_1, sysname,>]
+		ON [dbo].[PERSON] ([person_id])
+			INCLUDE ([year_of_birth],[gender_concept_id] );
+				GO; 			
+
+	CREATE NONCLUSTERED INDEX [<PERSON_2, sysname,>]
+		ON [dbo].[PERSON] ([location_id])
+			INCLUDE ([person_id]);
+				GO; 
+
+	CREATE NONCLUSTERED INDEX [<PERSON_3, sysname,>]
+		ON [dbo].[PERSON] ([provider_id]);
+				GO;
+
+	CREATE NONCLUSTERED INDEX [<PERSON_4, sysname,>]
+		ON [dbo].[PERSON] ([care_site_id]);
+			GO;
+
+/* OBSERVATION */ 
+
+USE [Camel_DB];
+GO;
+
+	CREATE NONCLUSTERED INDEX [<OBSERVATION_1, sysname,>]
+		ON [dbo].[OBSERVATION] ([provider_id]);
+		GO;
+
+	CREATE NONCLUSTERED INDEX [<OBSERVATION_2, sysname,>]
+		ON [dbo].[OBSERVATION] ([visit_occurrence_id]);
+		GO;
+
+	CREATE NONCLUSTERED INDEX [<OBSERVATION_3, sysname,>]
+		ON [dbo].[OBSERVATION] ([value_as_number],[unit_concept_id])
+			INCLUDE ([observation_concept_id]);
+			GO; 
+
+/* OBSERVATION_PERIOD */
+
+	CREATE NONCLUSTERED INDEX [<OBSERVATION_PERIOD_4, sysname,>]
+		ON [dbo].[OBSERVATION_PERIOD] ([PERSON_ID])
+			INCLUDE ([OBSERVATION_PERIOD_START_DATE]);
+			GO; 
+
+	CREATE NONCLUSTERED INDEX [<OBSERVATION_PERIOD_5, sysname,>]
+		ON [dbo].[OBSERVATION_PERIOD] ([OBSERVATION_PERIOD_START_DATE],[OBSERVATION_PERIOD_END_DATE])
+			INCLUDE ([PERSON_ID]);
+			GO; 
+
+/* VISIT */
+
+	CREATE NONCLUSTERED INDEX [<VISIT_1, sysname,>]
+		ON [dbo].[VISIT_OCCURRENCE] ([care_site_id]);
+		GO; 
+
+/* CONDITION */
+
+	CREATE NONCLUSTERED INDEX [<CONDITION_1, sysname,>]
+		ON [dbo].[CONDITION_OCCURRENCE] ([provider_id]);
+		GO;
+
+	CREATE NONCLUSTERED INDEX [<CONDITION_2, sysname,>]
+		ON [dbo].[CONDITION_OCCURRENCE] ([visit_occurrence_id]);
+		GO;
+
+/* CONDITION_ERA */
+
+	CREATE NONCLUSTERED INDEX [<CONDITION_ERA_1, sysname,>]
+		ON [dbo].[CONDITION_ERA] ([person_id])
+			INCLUDE ([condition_concept_id],[condition_era_start_date]);
+			GO;
+
+/* PROCEDURE */
+
+	CREATE NONCLUSTERED INDEX [<PROCEDURE_1, sysname,>]
+		ON [dbo].[PROCEDURE_OCCURRENCE] ([provider_id], [visit_occurrence_id]);
+		GO;
+		
+
+/* DRUG */
+
+	CREATE NONCLUSTERED INDEX [<DRUG_1, sysname,>]
+		ON [dbo].[DRUG_EXPOSURE] ([provider_id]);
+		GO;
+ 
+	CREATE NONCLUSTERED INDEX [<DRUG_2, sysname,>]
+		ON [dbo].[DRUG_EXPOSURE] ([visit_occurrence_id])
+		GO;
+
+	CREATE NONCLUSTERED INDEX [<DRUG_3, sysname,>]
+		ON [dbo].[DRUG_EXPOSURE] ([days_supply])
+			INCLUDE ([drug_concept_id])
+			GO;
+
+	CREATE NONCLUSTERED INDEX [<DRUG_4, sysname,>]
+		ON [dbo].[DRUG_EXPOSURE] ([refills])
+			INCLUDE ([drug_concept_id])
+			GO;
+
+	CREATE NONCLUSTERED INDEX [<DRUG_5, sysname,>]
+		ON [dbo].[DRUG_EXPOSURE] ([quantity])
+			INCLUDE ([drug_concept_id])
+			GO;
+
+	CREATE NONCLUSTERED INDEX [<DRUG_6, sysname,>]
+		ON [dbo].[DRUG_EXPOSURE] ([drug_concept_id])
+			INCLUDE ([drug_source_value])
+			GO;
+
+/* DRUG_ERA */
+
+	CREATE NONCLUSTERED INDEX [<DRUG_ERA_1, sysname,>]
+		ON [dbo].[DRUG_ERA] ([person_id])
+			INCLUDE ([drug_concept_id],[drug_era_start_date])
+		GO;
+
+/* MEASUREMENT */
+
+	CREATE NONCLUSTERED INDEX [<MEASUREMENT_1, sysname,>]
+		ON [dbo].[MEASUREMENT] ([person_id])
+			INCLUDE ([measurement_concept_id],[measurement_date]);
+			GO;
+
+	CREATE NONCLUSTERED INDEX [<MEASUREMENT_2, sysname,>]
+		ON [dbo].[MEASUREMENT] ([provider_id]);
+		GO;
+
+	CREATE NONCLUSTERED INDEX [<MEASUREMENT_3, sysname,>]
+		ON [dbo].[MEASUREMENT] ([visit_occurrence_id]);
+		GO;
+
+	CREATE NONCLUSTERED INDEX [<MEASUREMENT_4, sysname,>]
+		ON [dbo].[MEASUREMENT] ([value_as_number],[value_as_concept_id]);
+		GO;
+
+	CREATE NONCLUSTERED INDEX [<MEASUREMENT_5, sysname,>]
+		ON [dbo].[MEASUREMENT] ([value_as_number],[unit_concept_id])
+			INCLUDE ([measurement_concept_id]);
+			GO;
+
+	CREATE NONCLUSTERED INDEX [<MEASUREMENT_6, sysname,>]
+		ON [dbo].[MEASUREMENT] ([value_as_number],[unit_concept_id],[range_low],[range_high])
+			INCLUDE ([measurement_concept_id]);
+			GO;
+		
+	CREATE NONCLUSTERED INDEX [<MEASUREMENT_7, sysname,>]
+		ON [dbo].[MEASUREMENT] ([value_as_number]);
+			GO;
+
+/* PROVIDER */
+
+	CREATE NONCLUSTERED INDEX [<PROVIDER_1, sysname,>]
+		ON [dbo].[PROVIDER] ([care_site_id]);
+			GO;
+
+/* PAYER_PLAN_PERIOD */
+
+	CREATE NONCLUSTERED INDEX [<PAYER_PLAN_PERIOD_1, sysname,>]
+		ON [dbo].[PAYER_PLAN_PERIOD] ([person_id])
+			INCLUDE ([payer_plan_period_start_date],[payer_plan_period_end_date]);
+			GO;
+
+/* ACHILLES_results */
+
+	CREATE NONCLUSTERED INDEX [<ACHILLES_RESULTS, sysname,>]
+		ON [dbo].[ACHILLES_results] ([count_value]);
+		GO;