Faculty Bibliography

PURPOSE/OBJECTIVE:Anti-vascular endothelial growth factor medications are a cornerstone in the treatment of many macular diseases in modern ophthalmology. These medications accrue significant economic burdens to individuals and health systems, and recent data on health-system level practice patterns involving agent selection is lacking. We sought to examine utilization of intravitreal anti-VEGF agents from 2017 to 2023 across various payors and diagnoses in the United States in order to analyze trends in usage. METHODS:A retrospective cross-sectional study was performed. Data were obtained from the Epic Cosmos database, comprised of de-identified data from the electronic health record, Epic, of 238 million patients from all 50 states from January 1, 2017 to December 31, 2023. Encounters where a patient received an intravitreal injection in an ophthalmology clinic were categorized by intravitreal medication ordered, visit diagnosis, and primary insurance payor. The overall usage rates of each medication across different diagnoses and primary payor were compared. Total number of injections per year of anti-VEGF medication, the rate per 100000 ophthalmology encounters (OE), and the rate in different retinal conditions and with various insurance providers were the main outcome measures. RESULTS:From 2017-2023, there were 571,545 anti-VEGF injections administered, with 53.6% aflibercept 2 mg (306,247/571,545; 625.8/000 OE), 34.8% bevacizumab 1.25 mg (198,696/571,545; 414.7/100000 OE), 7.83% ranibizumab 0.3 mg and 0.5 mg (44,803/571,545; 99.2/100000 OE), and 3.43% faricimab 6 mg (19,624/571,545; 32.8/100000 OE). Brolucizumab 6 mg, ranibizumab-eqrn 0.5 mg, ranibizumab-nuna 0.5 mg, and aflibercept 8 mg accounted for <1% of all injections. From 2017-2023, the rate of aflibercept 2 mg increased from 361.3 injections/100000 OE to 913.1, bevacizumab increased from 274.1 to 538.1, and ranibizumab decreased from 92.4 to 64.0. The rate of faricimab increased from 37.9/100000 OE to 189.9 from 2022-2023. Patients with Medicare received aflibercept 2 mg at higher rates than bevacizumab (364.3/100000 OE vs 187.1, respectively) [t(6) = 5.67, p = 0.001] and faricimab (19.4/100000 OE), a pattern not seen in commercially insured [t(6)=2.07, p > 0.05] patients. Medicaid patients had a marginally significant difference in use of bevacizumab (M = 19.9/100000 OE, SD = 7.32) and aflibercept 2 mg (16.1/100000 OE, SD = 7.22) [t(6)]=2.95, p = 0.026]. CONCLUSIONS:Aflibercept 2 mg was the most common drug used overall from 2017-2023. Faricimab had the highest absolute and relative increase in utilization from 2022-2023. Patients with Medicare B or C were significantly more likely to receive aflibercept over bevacizumab.

BACKGROUND:Delayed diagnosis of venous thromboembolism (VTE) is prevalent among hospitalized patients, yet case identification is challenging and feedback limited. OBJECTIVE:To develop a large language model (LLM)-based electronic-trigger to identify VTE diagnostic delays. METHODS:All admissions to internal medicine (IM) residents at NYU Langone Health between January 2022 and December 2023 (n = 20,843) were included. Using an open-source LLM, prompts were validated to detect (1) residents considering VTE in admission notes and (2) VTE confirmation in five types of imaging reports (n = 100 for each prompt validation set). The validated prompts were applied to determine discordance between admission note differential omitting VTE and imaging report confirming VTE. Two hospitalists reviewed discordant cases using a validated tool to identify diagnostic delays. Hospitalizations were labeled as diagnostic delays, in-hospital complication, or false-positive. Based on in-hospital complication and false-positive patterns, exclusion criteria were implemented. Positive predictive value (PPV) and negative predictive value (NPV) were calculated. RESULTS:The LLM prompts correctly classified admission notes and VTE imaging studies with high accuracy (range 98%-100%, n = 699 VTE cases identified). Of the 137 diagnostic delays the LLM-based electronic-trigger identified, 31 were true-positives, 60 in-hospital complications, and 46 false-positives. 4.4% of all VTE hospitalizations had a diagnostic delay. With the exclusion criteria, the PPV was 48% (95% confidence interval [CI], 35%-62%) and NPV was 95% (95% CI, 87%-98%). CONCLUSIONS:We developed the first LLM-based electronic-trigger to identify VTE diagnostic delays, with higher performance than existing non-LLM electronic-triggers. LLM-based approaches can facilitate diagnostic performance feedback and are scalable to other conditions and institutions.

Idiopathic intracranial hypertension (IIH) is a condition in which elevated pressure in the cerebrospinal fluid can lead to optic nerve head (ONH) dysfunction and subsequent visual impairment. Physicians are currently limited in their ability to monitor and manage this condition, as clinical symptoms and exam findings are often delayed in response to changes in intracranial pressure. In order to find other biomarkers of disease, researchers are using imaging modalities such as optical coherence tomography (OCT) to observe microscopic changes in the eye in this condition. OCT can create 2-dimensional and 3-dimensional high definition images of the retina of the ONH and has been used to study various conditions such as glaucoma and multiple sclerosis. Numerous studies have used OCT in IIH as well, and they have shown that certain retinal layers and the ONH change in thickness and shape in both the short and long term with intracranial pressure changes. OCT is a promising modality for clinical and scientific evaluation of IIH as it is a noninvasive and practical tool to obtain in depth images. This review will discuss how OCT can be used to assess a patient with IIH, both before and after treatment, along with its limitations and future applications.