Not PEDro's bias: summary quality scores can be used in meta-analysis.