2.3. Statistical methods
Dichotomous data were expressed as relative risks, and continuous
data were expressed as weighted mean differences, or standardised
mean differences wherever appropriate, with corresponding
95% confidence intervals. Statistical pooling was limited to clinically
homogeneous studies for which the study designs, populations, outcomes
and interventions were considered to be similar by the
reviewers. The chi-square and I
2 tests were used to measure statistical
heterogeneity, with a chi-square p-value of >0.1 and an I
2 value
of