Statistical analyses
We assessed the homogeneity of the included RCTs by
their characteristics of PICOS – participants, interventions,
comparators, outcomes and study methods,17
RCTs with similar characteristics were included for
meta-analyses and indirect comparisons.18 We used the
statistic I² to measure the heterogeneities between the
studies; a rough guide to interpretation of I² takes 50%
and more as substantial heterogeneity