How to assemble cohort using data from 2 separate tables PostgreSQL - sql

Question

I am trying to assemble a cohort of patients who meet a set of certain criteria (using data from 2 different tables). I am trying to create a list of patients who
Have been seen for a drug overdose
Encounter occur after 07-15-1999
Age is between 18 and 35 at the time of the encounter
Every patient in this table must meet all of these conditions. I have created a new table (dcohort) to insert the information for all of these patients. I have already figured out how to determine which patients meet the first two conditions, but am struggling to figure which meet the age condition because age is not a listed element in either of the 2 provided tables. Age must be calculated using the birthdate from one table (patients) and the encounter date from another other table (encounters). I want to know how to go about altering my code below to filter for patients who meet the age requirement. The code I have written thus far is:
CREATE TABLE dcohort (
PATIENT_ID VARCHAR(50) NULL
,ENCOUNTER_ID VARCHAR(50) NULL
,HOSPITAL_ENCOUNTER_DATE DATE NULL
,AGE_AT_VISIT NUMERIC(2,0) NULL
,DEATH_AT_VISIT_IND BIT NULL
,COUNT_CURRENT_MEDS NUMERIC(2,0) NULL
,CURRENT_OPIOID_IND BIT NULL
,READMISSION_90_DAY_IND BIT NULL
,READMISSION_30_DAY_IND BIT NULL
,FIRST_READMISSION_DATE DATE NULL
);
----------
INSERT INTO dcohort (patient_id, encounter_id, hospital_encounter_date, age_at_visit)
SELECT encounters.patient, encounters.encounterid, encounters.start, [placeholder]
FROM encounters
JOIN patients
ON encounters.patient = patients.id
WHERE reasondescription = 'Drug overdose'
AND start > '1999-7-15'

Answer 1

I would be starting with something like this:
SELECT DISTINCT ON (e.patient) e.patient, e.encounterid, e.start,
age(p.birthdate, e.start)
FROM encounters e JOIN
patients p
ON e.patient = p.id
WHERE e.start > '1999-07-15' AND
e.reasondescription = 'Drug overdose' AND
e.start >= p.birthdate + interval '18 year' AND
e.start < p.birthdate + interval '36 year'
ORDER BY e.patient, e.start;
This uses DISTINCT ON to get only one record per patient -- regardless of the number of eligible encounters that the patent has.
Note that this does not use age() to calculate the age. Instead, it uses direct comparisons between the dates. That is usually more accurate.