TABLE I.

Subgroup Analyses of Fracture Type and Smoking Status

	Subgroup Effect of Fracture Type*	Subgroup Effect of Smoking Status*
Is the difference suggested by comparisons within rather than between studies?	Yes	Yes
Does the interaction test suggest a low likelihood that chance explains the apparent subgroup effect?	P = 0.011†	P = 0.0013†
Were the subgroup hypothesis and its direction specified a priori?	Yes	No
Is it one of a small number of subgroup hypotheses tested?	1 of 7 subgroup hypotheses	1 of 12 subgroup hypotheses
Is the magnitude of the subgroup effect large?	60% difference in RR reduction	88% difference in RR reduction
Is the observed differential effect consistent across studies?	No	No evidence
Is there indirect evidence supporting the hypothesized differential effects?	Yes	No

^*RR = relative risk.

^†Test for interaction.