Improvement in Chronic Kidney Disease (CKD) diagnosis and management after automatic loading of CKD in the electronic health record problem list
En palabras de los autores
Background Chronic kidney disease (CKD) is often underrecognized and undiagnosed. We evaluate the impact of an algorithm-based automatic loading process for CKD in the electronic health record (EHR) problem list at a large health system. Methods Study populations included patients receiving primary or nephrology care at Geisinger (intervention group) and 4 large health systems in the Optum Labs Data Warehouse (OLDW), a longitudinal, real-world data asset with de-identified administrative claims and electronic health record (EHR) data,(comparator group) from 2008-2023. Since 2013, an automated process has been regularly implemented at Geisinger, adding CKD to problem lists of patients meeting laboratory criteria for CKD (outpatient eGFR <60 ml/min/1.73m2 on 2 lab values for [≥] 3 months without intervening normal values). We conducted difference-in-differences (DiD) and difference-in-difference-in-differences (DiDiD) analyses comparing CKD diagnosis rates, CKD-related lab tests, use of renin-angiotensin-aldosterone system inhibitors (RAASi) and statins for those >50 years within 1 year of incident CKD (index date) at Geisinger vs. OLDW health systems before (2008-2012) and after (2013-2023) CKD auto-load implementation. Analyses relied on summary-level data and did not permit adjustment for individual-level characteristics. Results Among patients with incident, undiagnosed CKD (n=21,995 in Geisinger; n=208,402 in OLDW), CKD diagnosis within 1 year of index date increased from 28.4% in 2008-2012 to 65.7% in 2013-2023 at Geisinger whereas CKD diagnosis decreased from 22.6% to 20.7% at OLDW (DiD estimate 39.1%; 95% CI: 38.1%, 40.1%; p<0.001). Among CKD-related labs, DID estimates favored Geisinger for follow-up eGFR, parathyroid hormone, and phosphorus testing (all p values <0.001). CKD auto-load was associated with increased use of RAASi and statins (p values <0.05). Conclusions An algorithm-based problem list auto-load process was associated with higher CKD diagnosis, some CKD-related labs, and use of recommended medications. This scalable strategy should be validated in other health systems.
Apareció: lunes, 16 de noviembre. medRxiv. Preprint, todavía sin revisión por pares.