Randomized controlled trials are known to be the best research design for estimating the effects of various intervening factors; however, in fact, such trials may not be feasible from the perspective of efforts required and ethics. Propensity score analysis is a method that allows comparison by reducing biases arising from differences in existing background factors, and it is a statistical method expressed as pseudorandomization. In this study, we report the noninferiority of the one-year postoperative success rates of using microhook for Trabectome surgery with propensity scores (Trabectome versus Ab Interno Microhook Trabeculotomy Comparative Study: TramTrac study). Before adjustment, there were differences between the two groups in various factors (e. g., preoperative intraocular pressure), and the effectiveness of the two surgical procedures could not be appropriately estimated by simply comparing postoperative outcomes. The propensity score analysis allowed for an appropriate estimation of effects by equalizing the effects of confounders in the two study groups.
There are various methods for propensity score analysis, such as the matching method, inverse probability of treatment weighting (IPTW), stratification method, and regression method. Notably, the TramTrac study performed the abovementioned four analyses, and the results of all analyses confirmed that the 95% confidence interval for the risk difference was smaller than the set noninferiority margin, thereby statistically proving the noninferiority of microhook for Trabectome.
Based on the abovementioned findings, the propensity score analysis is an excellent method, but reports on this analysis in the field of ophthalmology remain few. In the present study, we not only give an overview of propensity score with the TramTrac study as a representative example but also clearly describe the strengths and weaknesses of each method of analysis. This study can help readers, especially ophthalmologists, to conduct high evidence-level controlled studies using propensity score in the future.
Nippon Ganka Gakkai Zasshi (J Jpn Ophthalmol Soc) 127: 1039-1049, 2023.