- Data could be saved in pg using jsonb and preserving model format (aka fhirbase for FHIR).
- Most of cql rules could be expressed as views and/or CTE and/or queries in postgresql.
- Terminology could be part of database
- Some dynamic code could be done by plv8/js
So the result could be CQL database, which could efficiently eval CQL rules.
Here is example:
using QUICK
valueset "Asthma": '2.16.840.1.113883.3.117.1.7.1.271'
valueset "Asthma Management Plan": '2.16.840.1.113883.3.117.1.7.1.131'
valueset "Home Discharge Disposition": 'TBD'
valueset "Doctor": 'TBD'
parameter MeasurementPeriod default Interval[DateTime(2013, 1, 1, 0, 0, 0, 0), DateTime(2014, 1, 1, 0, 0, 0, 0))
context Patient
define "In Demographic":
AgeInYearsAt(start of MeasurementPeriod) >= 2 and AgeInYearsAt(start of MeasurementPeriod) <= 17
define "Asthma Encounters":
["Encounter": "Asthma"] E
where E."class" = 'inpatient'
and E."length" <= 120 days
and E."hospitalization"."dischargeDisposition" in "Home Discharge Disposition"
and E."period" ends during MeasurementPeriod
define "Asthma Encounters with Plan":
"Asthma Encounters" E
with ["CommunicationRequest"] C
such that (C."encounter" as "Encounter")."id" = E."id"
and exists ((C."reason") R where R in "Asthma Management Plan")
and exists (((C."sender" as "Practitioner")."practitionerRole") R where R."role" in "Doctor")
and (First(C."recipient") as "Patient")."id" = (C."subject" as "Patient")."id"
define "In Initial Patient Population":
"In Demographic" and exists ("Asthma Encounters")
context Population
define "Denominator": "Asthma Encounters"
define "Numerator": "Asthma Encounters with Plan"which could be compiled into CTE
WITH doctor_vs AS (
select code from codes where system = '<oid>' and code in ('<code1>')
), asthma_vs AS (
select code from codes where system = '<oid>' and code in ('<code1>')
), population AS (
SELECT patient.*
FROM Patient
WHERE age_in_years_at(?, (resource->>'birthDate')::timestamp) >= 2
and age_in_years_at(?, (resource->>'birthDate')::timestamp) <= 17
), astma_encounters AS (
SELECT e.*
FROM Encounter e
JOIN population p ON reference(e, '{patient}') = p.id
WHERE resource->>'class' == 'inpatient'
AND resource->>'length' <= 120 -- days
AND extract(e, '{hospitalization,discargeDisposition}', 'token') ilike '%HomeDischarge%'
AND extract(e, '{period,end}', 'date') <@ '[2010-01-01 14:30, 2010-01-01 15:30)'::tsrange
), AS encounters_with_plan (
SELECT e.*
FROM astma_encounters e
JOIN population p p ON reference(e, '{patient}', p)
JOIN CommunicationRequest cr ON reference(cr, '{encounter}', e)
WHERE extract(cr, 'reason', 'code') in (select code from astma_vs)
AND extract(cr, 'role', 'code') in (select code from doctor_vs)
AND extract(cr, 'recipient', 'reference') = p.id
)
SELECT (select count(*) from astma_encounters) as Denominator,
(select count(*) from encounters_with_plan) as Numerator,
(select count(*) from population) as Population;
Libraries could be implemented as schemas & views:
CREATE VIEW <module_schema>.astma_encounters AS (
SELECT e.*
FROM Encounter e
WHERE resource->>'class' == 'inpatient'
AND resource->>'length' <= 120 -- days
AND extract(e, '{hospitalization,discargeDisposition}', 'token') ilike '%HomeDischarge%'
AND extract(e, '{period,end}', 'date') <@ '[2010-01-01 14:30, 2010-01-01 15:30)'::tsrange
);
Copyright © 2016 niquola
Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.