Risk of bias in included studies
The detailed risk of biases and quality of each study are explained
in the Characteristics of included studies table.We identified none
of the included trials as ’low risk of bias’, we assessed three trials
(Graham 1978; Noll 1999; Noll 2000) as ’moderate risk of
bias’, and we assessed three trials (Bjorkqvist 1997; Britton 1985;
Tydeman 1989) as ’high risk of bias’ (Figure 1; Figure 2). None of
the studies were supported by pharmaceutical company funding