Purchase this article with an account.
Sophia Y Wang, Kuldev Singh; Informatics for Investigation of Glaucoma Filtration Surgery Outcomes Using Electronic Health Records. Invest. Ophthalmol. Vis. Sci. 2020;61(7):1415.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Controversy exists over the efficacy of various approaches to glaucoma filtration surgery. We used an automated pipeline for extraction of ophthalmic surgical outcomes from electronic health records (EHR) to perform a retrospective observational study investigating long-term survival of trabeculectomies, Ex-PRESS shunts, Baerveldt and Ahmed tube shunts.
Patients undergoing glaucoma filtration surgery at an academic center from 2009-2018 were identified from procedure codes captured in the EHR. Patient features were identified from EHR billing codes, demographic tables, medical orders, clinical notes, and eye exam fields. A previously validated natural language processing pipeline identified from operative notes the laterality of glaucoma surgery and the type of glaucoma shunt and from medication orders the laterality of glaucoma medications. Cox proportional hazards modeling evaluated factors associated with surgery failure, defined as undergoing additional glaucoma surgery or using more glaucoma medications than preoperatively. Standard errors were adjusted for patient-level clustering.
512 patients underwent 711 glaucoma filtration surgeries: 287 trabeculectomies, 47 Ex-PRESS shunts, 274 Baerveldt and 103 Ahmed tube implantations. Median follow-up was 292 days (IQR 192-617). Mean baseline IOP was 24.4 mmHg (SD 10.9); 73.1% were on ≥3 glaucoma medications. Compared to trabeculectomy, tube shunt surgery had higher hazard of failure (Baerveldt: HR 1.44, 95% CI 1.02-2.02; Ahmed: HR 2.01, 95% CI 1.28-3.17). Previous conjunctival-based glaucoma surgery was associated with increased failure (1 previous surgery: HR 1.93, 95% CI 1.32-2.83; ≥2 previous surgeries: HR 2.74, 95% CI 1.62-4.64). Fewer baseline medications was associated with increased failure compared to >3 baseline medications (<3 medications: HR 2.96, 95% CI 2.12-4.13; 3 medications: HR 1.68, 95% CI 1.21-2.34). Men had increased hazard of failure compared to women (HR 1.40, 95% CI 1.03-1.90).
We found that Baerveldt and Ahmed tube shunts were associated with increased failure compared to trabeculectomies. Fewer baseline glaucoma medications, more previous glaucoma surgeries, and male sex were also risk factors for surgical failure. These results demonstrate the utility of applying an informatics pipeline to electronic health records to investigate key clinical questions using real-world evidence.
This is a 2020 ARVO Annual Meeting abstract.
This PDF is available to Subscribers Only