feature: adding observation_period table

phs/data/optum.py

@@ -109,6 +109,20 @@ class Optum:
         }
     }
     
+    _OBSERVATION_PERIOD = """
+        observation_period AS (
+            SELECT DISTINCT
+                t0.Patid AS person_id,
+                t0.Eligeff AS observation_period_start_date,
+                t0.Eligend AS observation_period_end_date,
+                CAST(CASE
+                    WHEN (STRPOS(t0.Zipcode_5, '_') = 0) THEN t0.Zipcode_5
+                        ELSE SUBSTR(t0.Zipcode_5, 1, STRPOS(t0.Zipcode_5, '_') - 1)
+                    END AS INT64) AS zip
+            FROM 
+                {person_table} AS t0
+        )"""
+    
     _PERSON = """
         person AS (
             SELECT *
@@ -117,8 +131,6 @@ class Optum:
                     t0.Patid AS person_id,
                     t0.Gdr_Cd AS gender_source_value,
                     t0.Yrdob AS year_of_birth,
-                    t0.Eligeff AS observation_period_start_date,
-                    t0.Eligend AS observation_period_end_date,
                     CAST(CASE
                         WHEN (STRPOS(t0.Zipcode_5, '_') = 0) THEN t0.Zipcode_5
                             ELSE SUBSTR(t0.Zipcode_5, 1, STRPOS(t0.Zipcode_5, '_') - 1)
@@ -159,9 +171,9 @@ class Optum:
                 p.observation_period_end_date
             FROM (
                 SELECT op.person_id, op.zip, op.observation_period_start_date, op.observation_period_end_date, e.condition_start_date
-                FROM person AS op
+                FROM observation_period AS op
                 JOIN (
-                    SELECT p.person_id, c.condition_start_date
+                    SELECT p.person_id, p.zip, c.condition_start_date
                     FROM person AS p
                     JOIN (
                         SELECT *, row_number() OVER (PARTITION BY person_id ORDER BY condition_start_date) AS ordinal
@@ -171,7 +183,7 @@ class Optum:
                     ON p.person_id = c.person_id
                     {event_occurrence_filters}
                 ) AS e
-                ON op.person_id = e.person_id
+                ON op.person_id = e.person_id AND op.zip = e.zip
                     AND e.condition_start_date >= op.observation_period_start_date
                     AND e.condition_start_date <= op.observation_period_end_date
                 {continuous_observation_filters}
@@ -223,22 +235,23 @@ class Optum:
             SELECT p.person_id, p.zip
             FROM qualifying_event AS p
             JOIN (
-                SELECT op.person_id, op.zip, c.condition_start_date
-                FROM person AS op 
+                SELECT pe.person_id, pe.zip, c.condition_start_date
+                FROM person AS pe 
                 JOIN (
                     SELECT *
                     FROM condition_occurrence
                     WHERE condition_source_value IN ({event_valueset})
                 ) AS c
-                ON op.person_id = c.person_id
+                ON pe.person_id = c.person_id
                 {event_occurrence_filters}
             ) AS e
-            ON p.person_id = e.person_id
+            ON p.person_id = e.person_id AND p.zip = e.zip
                 AND e.condition_start_date >= p.observation_period_start_date
                 AND e.condition_start_date <= p.observation_period_end_date
                 {event_occurrence_before_index_date}
                 {event_occurrence_after_index_date}
         )"""
+
     
     _INCLUSION_EVENT = """
         inclusion_event AS (
@@ -258,7 +271,7 @@ class Optum:
 
     # SQL template for getting a cohort population
     _GET_COHORT_POPULATIONS = """
-        WITH {person}, {condition_occurrence}, {qualifying_event}, {inclusion_criteria}, {inclusion_event},
+        WITH {observation_period}, {person}, {condition_occurrence}, {qualifying_event}, {inclusion_criteria}, {inclusion_event},
         cohort AS (
             SELECT mg.person_id,
                    mg.zip,
@@ -503,6 +516,8 @@ class Optum:
         
         query = textwrap.dedent(
             self._GET_COHORT_POPULATIONS.format(
+                     observation_period=self._OBSERVATION_PERIOD.format(
+                             person_table=self._member_table_id),
                      person=self._PERSON.format(
                              person_table=self._member_table_id,
                              zipcode_valueset=",".join(str(z) for z in zips)),