Presentation (CTI)

OBJECTIVES: Economic modelling in ophthalmology has several features that distinguish ophthalmology from other disease areas. As such the economic model structures have unique features that have pragmatically evolved to reflect long term data, new data sources and HTA feedback. However, since the first ophthalmic NICE submissions over 10 years ago there has been significant development in methods used in other disease areas as well as understanding of what’s important to patients.
METHODS: Publicly available trial data was analysed to assess if a regression based approach could be used to estimate health state occupation for DMO using publicly available data from the DRCR. The regression analyses focused on assessing correlation between treated and untreated eyes in BCVA, fitting distributions to the BVCA in the treated eye and estimating the likelihood of blindness given the BCVA in the treated eye. Separate analyses were conducted for patients who were bilaterally impacted by DMO. A descriptive assessment of the proportion of patients who were treated in the worst-seeing eye (WSE), over time and by treatment arm was completed. Goodness of fit was assessed based on the mean absolute error and mean square error comparing the regression approach to established MSM approaches.
RESULTS: The proportion of trial participants who were treated in the WSE was comparable at baseline, however by week 40 there was a 15% difference between the treatment arms. To investigate further, the correlation between change from baseline in BCVA was assessed also by treatment arm; the correlation was not statistically different between treatment arms. Fitted distributions showed potential to perform similarly to MSM methods.
CONCLUSIONS: The analysis showed the importance of modelling both eyes and the interaction between eyes. Economic model structures using mean change BCVA instead of time dependent transition probabilities has potential simplify the cost effectiveness model structures.