Text this: Meta-analysis under imbalance in measurement of confounders in cohort studies using only summary-level data.