Table 4.

Credibility of subgroup effects for risk of bias for outcome days to radiographic healing in studies of low intensity pulsed ultrasound for bone healing

Criteria22	Rating (yes means higher credibility)
Is subgroup variable a characteristic measured at baseline or after randomization?	Not applicable for risk of bias
Is effect suggested by comparisons within rather than between studies?	No, between studies
Was subgroup effect specified a priori?	Yes, specified in our protocol
Was direction of subgroup effect specified a priori?	Yes, we expected larger effects for studies at high risk of bias
Is there indirect evidence that supports hypothesized interaction (biological rationale)?	Not applicable for risk of bias
Was subgroup effect one of a small number of hypothesized effects tested?	Yes, one of three
Does interaction test suggest low likelihood that chance explains the apparent subgroup effect?	Yes, significant in univariable subgroup analysis (P<0.001)
Is significant subgroup effect independent?	Yes, significant in multivariable meta-regression (P<0.01)
Is size of subgroup effect large?	Yes, 31.8% acceleration in high risk of bias trials versus 1.7% acceleration in low risk of bias trials
Is interaction consistent across closely related outcomes within study?	Yes, risk of bias explained heterogeneity in outcomes weight bearing and pain
Is interaction consistent across studies?	Yes, high risk of bias studies consistently showed large effects, low risk of bias studies small effects