OBJECTIVES: The ISPOR Task force (TF) on Good Research Practices for RCT-CEA aims to foster improvements in the conduct of trial-based economic analysis. The TF recognizes the sample size of randomized clinical trials (RCT) as one of the challenges for trial-based economic analysis, as it is typically based on the primary clinical outcomes only. In the case of vitreomacular traction (VMT), using RCT individual patient-level data (IPD) to establish model starting distributions within visual acuity (VA) health-states magnifies this challenge due to the small patient numbers within each relevant VA health-state. Our objective was to develop an innovative approach to robustly estimate patient within-category VA health-state starting distributions.  METHODS: A baseline VA-adjusted ordered logit model used RCT IPD to predict a patient’s VA starting distribution as a function of treatment allocation, macular hole, vitreomacular adhesion and previous vitrectomy status. The observed ordinal variable consisted of 6 response categories i.e. VA state as a function of an unmeasured, continuous, latent variable Y whose values determine the patient’s VA-state dependent specific VA thresholds. RESULTS: Treatment allocation was not a significant predictor for within-category VA health-state starting distributions (at the 5% significance level), while MH, VMA and previous vitrectomy status were significant and retained in the final model. The proportional odds assumption was tested using a likelihood ratio test and confirmed that the relationship between each pair of VA health-states was the same (chi2 = 0.0906). CONCLUSIONS: In eye-disorders like VMT, estimating within-category VA health-state starting distributions requires a different approach due to the small number of IPD in each VA health-state. Using an ordered logit model allows a more accurate and robust estimation of within-category VA health-state starting distributions. Macular hole, VMA and previous vitrectomy status were significant predictors of a patient’s within-category VA health-state starting distribution, while treatment allocation was not.