Fig. 3

Observed fracture incidence vs Kaplan-Meier estimated fracture incidence in sensitivity analyses. We compared the observed cumulative fracture incidence in crossover and long-term denosumab-treated subjects up to Extension year 5 (year 8 overall) vs the Kaplan-Meier estimated fracture incidence using a multiple imputation approach with the assumption that subjects who withdrew had double the fracture risk after study discontinuation compared with those who remained on the study. N is the number of subjects who remained at risk at the beginning of each year