Update: It has been pointed out that the methodology used to arrive at the -0.2% number is incredibly strange. The authors selected seven studies to include in

Update: It has been pointed out that the methodology used to arrive at the -0.2% number is incredibly strange. The authors selected seven studies to include in their calculation, with estimates ranging from 0.0% to -35.3%. The arithmetic average across the seven studies is -7.3% and the median is -2.4%, however the authors instead calculate a "precision-weighted average" where one study is weighted significantly higher than the rest (7390 vs 11-256 for the others, see Table 3 for details) and the number from that one study (0.1%) pretty much overwhelms all the others. I don't think there is a reasonable explanation for doing the calculation this way unless you are intentionally trying to massage the numbers to support a pre-determined outcome. Very suspicious.