Abstract
Purpose :
Clinical trial eligibility criteria are critical for the evaluation of the safety and efficiency of therapeutics. However, there is variability in the criteria used to describe who may participate. We reviewed eligibility criteria for 300 interventional Phase 3/4 studies for diabetic retinopathy and glaucoma to evaluate consistency and specificity.
Methods :
Eligibility criteria for interventional, phase 3/4 trials for diabetic retinopathy (DR) and glaucoma therapeutics completed between Jan. 2011 and Jan. 2019 were retrieved from Clinicaltrials.gov. Eligibility criteria were split by inclusion or exclusion and individual elements in each study labeled according to the following categories: demographics, observations, measures, medications, reproduction, procedures, or not likely to be in health record.
Results :
A total of 4,351 elements were annotated for 137 DR and 163 glaucoma studies. Results are shown in the table. The most common category was observations (29%), followed by procedures (21%). The least encountered criteria were demographics (3%). Overall, there were more exclusion criteria (58%) than inclusion criteria (42%). Inclusion criteria most often related to measures (quantifiable characteristics), where exclusion criteria related to observations (subjective clinical characteristics).
In this analysis we documented variability in the representation of eligibility criteria concepts. We noted inconsistent ways to describe a single concept. For example, there were 77 unique ways to indicate that a participant needed to be age ≥ 18 based on the exact text, capitalization, punctuation. There was a lack of consistency for representation of factors such as visual acuity which included values (i.e. number of letters, logMAR) 161 times but referred to subjectively 145 times (i.e. 'vision impairment'). Reference to eligibility based on the 'opinion of the clinician' was used as a criteria 131 times.
Conclusions :
Our analysis showed high variability in the representation of clinical trial eligibility criteria. Variability hinders the ability to use computational advances to compare results across cohorts and limits the use of standard queries to identify potential eligible trial participants in electronic health records systems. Standardization in clinical trial criteria may reduce the cost of recruitment and accelerate biomedical research by enabling machine actionability.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.