br Methods As we were not privy to the
Methods As we were not privy to the median year of enrolment, for all selected studies, the mean enrolment year ([start of enrolment+end of enrolment]/2) was calculated. If these dates were unavailable, the corresponding authors or sponsoring companies were contacted. When all methods to gather this information were exhausted, for studies that did not report their enrolment period, the average of the mean enrolment year for the same intervention drug reported in literature was used. Here, an assumption was made that intervention drugs tend to be tested over approximately the same years. To normalize the distribution of SMR, natural log-transformation was applied. Graphs were created depicting SMR (log-scale) as a function of the year of enrolment for placebo arms and treatment arms. Due to the different enrolment numbers in each study, the SMR log-values were weighed accordingly (also known as weighted least squares). Weighted linear regression modified the standard linear regression model (minimizing the square of the error between predicted value and the actual value Y) to , where W is the weighted value for each study (known as number of patients per study).These weighted linear regression models over time were constructed and p-values less than 0.05 were considered as statistically significance. Negative coefficients of time (slope) indicate that the average SMR (log-scale) is decreasing over enrolment year. The interaction term for the slopes of placebo versus intervention patients was calculated as well. We also conducted the above models with accounting for histology and/or drug as confounding factors due to heterogeneous studies. Histology was treated as a categorical variable with different primary cancer sites including breast, bladder, lung (other solid tumors), prostate, and renal cell carcinoma. For intervention treatment, different drug mechanism were accounted for, these included denosumab, ibandronate, pamidronate, and zoledronic acid. R2 was calculated for each model, with higher values of the R2 demonstrating better model fitting.
Discussion This analysis finds a National Survey in the occurrence of SMRs over the time frame from 1990 to 2007. SMRs of both placebo and intervention arms decreased over time, but remained relatively constant in patients receiving the third generation bisphosphonate zoledronic acid. In studies with the most recent bone targeted agent denosumab, an additional relative reduction in SMR of 22% was seen in comparison to zoledronic acid . This suggests not only that newer bisphosphonates and bone-targeted agents are improving outcomes for patients, but management strategies and awareness of such disease is improving at the same time. As is evident by analysis of SMR, improvements are being made in the treatment of bone metastases. This is likely by reducing osteoclast-mediated bone resorption and subsequently normalizing calcium levels. Bisphosphonates have been developed through three generations, with current research focusing on the monoclonal antibody denosumab. Although few direct comparisons were found, second generation bisphosphonates were shown to decrease the risk of SRE similarly to first generation bisphosphonates but with longer lasting pain relief . This can be exemplified by how significantly better bone-pain scores were reported with 90mg intravenous pamidronate compared with a 1600mg/day oral clodronate regimen, in patients with bone metastases from a variety of primary sites . In a number of later trials, zoledronic acid has been shown to decrease the number of SREs in comparison to earlier bisphosphonates. As Rosen et al. found, long-term treatment in lung cancer patients and other patients with solid tumor using zoledronic acid (4mg) has shown an additional 20% reduction in the risk of skeletal complications in comparison to treatment by pamidronate . Recently, as indicated by Stopeck et al. , denosumab significantly delayed the time to first trial SRE and reduced the risk of subsequent SREs when compared to zoledronic acid in breast cancer patients. In a separate trial, Fizazi et al.  accrued similar findings with prostate cancer patients, with the median time to first SRE being 20.7 months compared to 17.1 monthly, for denosumab and zoledronic acid, respectively.