Assessing covariate imbalance in meta-analysis studies