Bone densitometry:
All patients had at least two dual X-ray absorptiometry (DXA) exams: one before or at the time of DMAB initiation and one after VF occurrence. Exams were carried out with different machines for different patients, but the same for each patient, allowing us to compare both exams. T-scores were calculated using normal values for the Spanish population. For biochemical determinations, fasting blood was drawn. The biochemical parameters: creatinine, total proteins, calcium and phosphorus were measured using standardized colorimetric methods. Immunochemiluminescence was used to determine the biochemical parameters of bone remodeling: P1NP, beta-crosslaps and osteocalcin.
 VF diagnosis was confirmed by MRI assessed by a radiologist, except in four patients in which it was based on shape changes in X-ray exams as compared to recent previous images.
The study was carried out following the rules of the Declaration of Helsinki (13), the protocol approved by the Insular University Hospital of Gran Canaria Clinical Trials Committee. All patients were informed of the study objectives and gave their informed written consent.
Statistical analysis Univariate analysis. Categorical variables are expressed as frequencies and percentages and continuous as mean and standard deviation (SD). Paired means were compared using the Wilcoxon test for paired data.Poisson models. The effect of each factor (X ) on the number of vertebral fractures after DMAB (nVF ) was analyzed by means of the Poisson model: \(\text{nVFPoisson}\left(\mu\right)\), being:
\begin{equation} \log\left(\mu\right)=\alpha+\beta X\nonumber \\ \end{equation}
Where \(\mu\) is the expected number of vertebral fractures, which may depend on the X factor. When \(X\) is a binary variable indicating presence or absence of a character its values were coded as 1 (presence) and 0 (absence). From this model it follows:
\begin{equation} \frac{\mu\left(X=t+1\right)}{\mu\left(X=t\right)}=exp\left(\beta\right)\nonumber \\ \end{equation}
Where \(\mu\left(X=t\right)\) corresponds to the expected number of vertebral fractures when the factor X is in level t . Therefore, \(\exp\left(\beta\right)\) correspond to the proportion of variation of the expected number of vertebral fractures for each unit that varies X.
Statistical significance was set at p < 0.05. Data were analyzed using the R package, version 3.6.1 (R Development Core Team, 2019).