Oral Presentation ANZBMS-MEPSA-ANZORS 2022

Bayesian Network Meta-Analysis of All-Cause Mortality in Trials of Osteoporosis Therapies (#21)

Alexander H Seeto 1 , Kevin Leow 2 , Joshua Long 1 , Mina Tadrous 3 , Abadi Kahsu Gebre 4 , Joshua R Lewis 4 , Howard A Fink 5 , Peter R Ebeling 6 , Alexander J Rodriguez 6
  1. Griffith University, Southport, QLD, Australia
  2. The Canberra Hospital, Canberra
  3. Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto
  4. Institute for Nutrition Research, School of Medical and Health Sciences, Edith Cowan University, Joondalup
  5. Geriatric Research Education and Clinical Center, VA Health Care System, Minneapolis
  6. Bone and Muscle Health Research Group, Monash University, Clayton

Fractures appear to have an adverse impact on mortality with controversy over whether this is beyond pre-existing comorbidities. There are also data from observational studies and randomised controlled trials suggesting anti-osteoporosis therapies may reduce mortality. Conventional meta-analyses have examined this question in bisphosphonates, with remaining uncertainties for other anti-osteoporosis therapies. Therefore, we conducted a network meta-analysis (NMA) examining all-cause mortality data from randomised trials of anti-osteoporosis therapies in postmenopausal women.

Trials were identified from recent NMAs of osteoporosis therapies for fracture reduction, updated to December 2020. Included trials enrolled >100 women per study, randomised to placebo or an active comparator (bisphosphonate, denosumab, parathyroid [PTH] analogues, romosozumab, or menopausal hormone therapy [MHT] such as selective estrogen receptor modulators or direct estrogen and/or progesterone supplementation), and reported primary skeletal outcomes. We investigated all-cause mortality and all adverse events (for internal validity). Data were synthesized in a random-effects NMA using Bayesian modelling. Relative to placebo, point estimates for the odds ratio with 95% credible intervals were generated. Probabilistic ranking of treatment safety was performed.

We identified 71 trials, enrolling 143,388 women. Over a mean follow-up of 24 months, 1,990 deaths occurred in all studies (1.39%). There were 835 deaths in 56,460 women randomised to placebo (1.48%) and 1161 deaths in 90,722 women randomised to active therapies (1.28%). Compared with placebo, no osteoporosis therapy was significantly associated with mortality or overall adverse events. Probabilistic ranking found bisphosphonates were less likely to have increased risk of mortality than placebo, while denosumab, romosozumab, MHT, and PTH analogues ranked more likely to have increased risk of mortality than placebo.

Despite known fracture reduction benefits, we found no pooled direct/indirect evidence that anti-osteoporosis therapies significantly affected mortality risk relative to placebo. More data are now needed to understand mortality risks in the age of sequential therapies.

6292e439cbf23-Mortality_Forest_Figure.jpg