patient

One row per person per source system. This is the demographic spine of the warehouse: every clinical and financial fact ultimately hangs off a person_id found here. Because the grain includes data_source, a single real-world individual appears once per contributing system until identity resolution collapses them; use person_id_crosswalk to trace how source identifiers were mapped onto a person_id.

At a glance

Schema	dax_core
Table	patient
Layer	Core Data Model
Object type	table
Columns	26
Grain	person_id, data_source
Tags	core

Columns

Column	Type	Key	Terminology	Constraints	Description
person_id	varchar	PK	—	not null	Dax identifier for the individual. Stable across refreshes and the join key used by every other table in the core layer.
sex	varchar		Sex	—	Normalized administrative sex.
race	varchar		Race	—	Normalized race value.
birth_date	date		—	—	Date of birth, used to derive age and age_group.
death_date	date		—	—	Date of death where reported by a source system.
death_flag	integer		—	—	Set when any source reported the person as deceased. Check this rather than testing death_date for null, since some sources report the fact of death without a date.
name_suffix	varchar		—	—	—
first_name	varchar		—	—	Given name. Direct identifier.
middle_name	varchar		—	—	—
last_name	varchar		—	—	Family name. Direct identifier.
social_security_number	varchar		—	—	Direct identifier. Restrict access to this column; it is present to support identity resolution, not analytics.
address	varchar		—	—	—
city	varchar		—	—	—
state	varchar		—	—	State or province of residence.
zip_code	varchar		—	—	Postal code of residence, commonly used for geographic rollups.
county	varchar		—	—	County of residence.
latitude	float		—	—	—
longitude	float		—	—	—
phone	varchar		—	—	—
email	varchar		—	—	—
ethnicity	varchar		Ethnicity	—	Normalized ethnicity value.
age	integer		—	—	Age in years as of the most recent warehouse build.
age_group	varchar		—	—	Banded age, for cohorting and reporting without exposing exact age.
ingest_datetime	timestamp		—	—	When the source record was loaded into the input layer.
dax_last_run	timestamp		—	—	Timestamp of the warehouse build that produced this row. Present on every core table, so you can confirm you are reading current data.
data_source	varchar	PK	—	not null	Label for the contributing system. Part of the grain, so always include it when joining or de-duplicating across sources.

Querying

select *
from dax_core.patient
limit 100;