OBJECTIVES: Indirect Treatment Comparisons (ITC) are used to contrast the effectiveness of two or more treatments, and are usually undertaken in the absence of head-to-head information. However, these indirect comparisons are less effective in situations where baseline patient characteristics (e.g. age, disease duration) differ between studies. Any clinically meaningful variation in these characteristics between the studies should be adjusted for in the statistical analyses in order to arrive at less biased estimates of the treatment differences. At present, many ITCs use a comparison of a sponsor’s Individual Patient Data (IPD) with study-level summary information (typically means and SDs) from their competitors’ studies. Various methods currently exist which allow for the matching between studies of the baseline characteristics means, but crucially not their variances.  METHODS: We outline a novel approach which allows for the matching of both means and variances across multiple baseline patient characteristics. Our approach involves fitting higher-order polynomials separately to each of the baseline parameters with the aim of estimating a single weight for each individual patient. The weighted means and variances of the IPD are then compared with the (target) summary-level data. Simulation is used in order to arrive at the combination of polynomial functions which give the ‘best fit’.  RESULTS: The method is highlighted with a case study of anti-VEGF therapies in the treatment of visual impairment due to diabetic macular edema. Our proposed method successfully matches both the means and variances across three important predictors of post-baseline changes in visual acuity. CONCLUSIONS: The ability to match IPD variability with study-level summary variability is critical in order to accurately estimate the statistical significance of treatment differences. To our knowledge, current comparative effectiveness methods fail to do so - our novel approach provides a possible solution to this problem.