Abstract
Purpose :
Spectral domain optical coherence tomography (OCT) is commonly used to study retinal changes as biomarkers for neurodegenerative diseases. To assess current approaches to OCT imaging and statistical analysis in this context, we performed a review of primary literature across 20 journals.
Methods :
In the top ten journals within the categories of clinical neurology and ophthalmology as ranked by SJR, we identified 137 eligible OCT-based studies published between 2013-2023. HS extracted data on the number of subjects and eyes imaged and analyzed, use of confirmatory diagnostic data, imaging method, types of imaging measures, approach to eye inclusion and statistical analysis, and adjustments for multiple measurements and inter-eye correlation.
Results :
Among 80 papers already reviewed studying Alzheimer’s (36%), multiple sclerosis (43%), Parkinson’s (15%), frontotemporal dementia (1%), or other conditions (5%), 33 were cohort studies, 43 cross-sectional studies, 3 case-control studies, and 1 was a randomized controlled trial. Studies used OCT imaging alone (75%), OCT angiography alone (5%), or a combination of OCT and OCT angiography (15%). The median (25th, 75th) number of subjects and eyes analyzed were 129 (66, 268) and 211 (100, 392), respectively. Peripapillary RNFL thickness and macular retinal layer thicknesses were the most common measures (70% and 71% of studies); auto-measurement was used in 91% of studies, of which 23 (30%) included additional manual correction. 71 (89%) studies reported excluding scans that did not meet a quality criteria, and 53 (66%) included a refractive error threshold. Imaging of both of the subjects’ eyes was reported in 73 (91%) studies, with 55 (75%) further analyzing data from both eyes. Bilateral studies most frequently combined data from both eyes (49%) for analysis, but only 29 (53%) of these studies used methods to adjust for inter-eye correlation, with 69% using generalized estimating equations and 31% using mixed effects models. Correction for statistical testing of multiple measurements was performed in 31% of analyses; statistical power was provided in 12.5%.
Conclusions :
OCT has been increasingly used for assessment of retinal changes in neurodegenerative diseases, but study design and statistical analysis can be improved by appropriate inclusion of subject eyes and correction for inter-eye correlation and multiplicity from multiple retinal measures.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.