*4.3. Detection Bias*

Only 7 [7,10,13,22,25,34,37] of the 27 included studies were recorded as having a "low" risk for both signalling questions used to ascertain detection bias. Whilst blinding of the outcome assessor was not necessarily confirmed for these studies, the majority have been considered to be at low risk as they utilised objective, biochemically validated outcome measures. Of the studies that were of high risk due to the lack of blinding of the outcome assessor, it was mainly due to the inclusion of outcomes such as behavioural assessment that required subjective judgment on the part of the outcome assessor. These were considered to have been at a high risk.
