Presented: ICPE 2025 in Washington DC, USA
Authors: Tzu-Hsuan Yang, Rebecca Persson, Zarena Jafry, Katrina Wilcox Hagberg, Catherine Vasilakis-Scaramozza, Susan S Jick
Abstract
Background: Information on cardiovascular safety of osteoporosis treatments is important to help patients and clinicians choose between osteoporosis treatment options. Studies have reported increased risks of fatal stroke among users of selective estrogen receptor modulators (SERMs), an osteoporosis treatment option, but information is limited.
Objective: To describe the risks of stroke, including fatal stroke, among female users of various osteoporosis treatments.
Methods: We conducted a cohort study of female users of osteoporosis treatments, age 40–89, in the Clinical Practice Research Datalink Aurum (2000–2020). Exposed patients were those with a first prescription for a SERM (raloxifene), a bisphosphonate (alendronate, ibandronate, risedronate, zoledronic acid, etidronate), denosumab, teriparatide or calcitonin in 2000 or later. The common reference group comprised patients with an osteoporosis-related code and vitamin D (Vit D) prescriptions in 2000 or later. Patients with a history of cardiovascular disease (CVD) were excluded. Person-time was accumulated by study drug from first prescription through censor date (event, other CVD diagnosis, major comorbidity, age 90, death, end of record, or 31 Dec 2020). A patient was considered currently exposed from prescription date through prescription duration + 30 days. We calculated incidence rates (IRs) per 10,000 patient-years (PY) of stroke and fatal stroke for each osteoporosis treatment and adjusted incidence rate ratios (adjIRRs) with 95% confidence intervals (CI) for each exposure, controlling for age, calendar year, osteoporotic disease type and smoking, compared with Vit D-only group.
Results: Among 306,457 patients (median age 68 years [interquartile range, IQR, 59–77], median follow-up 6 years [IQR 3–11]), there were 7312 stroke and 739 fatal stroke cases. Most exposure was to alendronate (77%) or risedronate (12%). Zoledronic acid, teriparatide, and calcitonin use was too low to estimate stable IRs. Across all exposures, IRs per 10,000 PY for stroke and fatal stroke were 43.2 and 4.4, respectively. The incidence of stroke was highest for etidronate (140 cases; IR = 59.9 per 10,000 PY [95% CI 50.4–70.6]) and lowest for current users of raloxifene (71 cases; IR = 36.4 per 10,000 PY [28.4–45.9]). No difference in stroke risk among users of any study drug was observed after adjusting for covariates. Among study drugs with more than 5 exposed cases, there was no increased risk of fatal stroke for any study drug.
Conclusion: There was no significant difference in stroke or fatal stroke risk between osteoporosis treatments. However, the number of fatal stroke cases was too small to make robust comparisons between treatments for most study drugs, including raloxifene.
Online Methods Appendix
Data Resource: UK Clinical Practice Research Datalink (CPRD) Aurum, general practice database, and linked Hospital Episode Statistics, in-patient database
Study Population: Female patients aged 40-89 in CPRD Aurum 2000-2020 with no history of cardiovascular disease in two cohorts:
- Treated cohort: new users of osteoporosis treatments: alendronate, ibandronate, risedronate, zoledronic acid, etidronate, denosumab, raloxifene, teriparatide, calcitonin
- Untreated cohort: prescribed vitamin D (VitD) users with osteoporosis-related clinical codes (osteoporosis, osteopenia, or osteoporosis monitoring)
Cohort Entry Definition:
- Treated cohort:date of the first ever osteoporosis treatment
- Untreated cohort: first VitD prescription on or after age 40 with an osteoporosis-related code (osteoporosis, osteopenia, dual-energy X-ray absorptiometry tests, monitoring, supporting codes) recorded at any time before or up to 90 days after.
Exposure Definition: Current use of each osteoporosis treatment was calculated based on recorded durations, quantities, and instructions (oral treatments) or standard dosing schedules (infusions and injections). We assigned exposure for each day from cohort entry through censor to mutually exclusive categories of study drug use. Exposure categories were osteoporosis treatment monotherapy (separately), 2+ osteoporosis treatment (concomitant use or overlap of exposed time during treatment switching [results for this small exposure category not presented in poster]), past use of any osteoporosis treatment (no current use), and VitD only.
Follow-up: Person-time was accumulated by exposure category from cohort entry until the earliest of stroke diagnosis or censor date (any cardiovascular diagnosis, major comorbidity, age 90, death, end of record, or 31 December 2020).
Case Definitions: Stroke cases includedall patients with stroke diagnosis code recorded in CPRD Aurum or HES primary diagnosis field at least 7 days after cohort entry. Stroke diagnoses were also included as cases if recorded after censoring due to death or if recorded within 7 days after censoring due to transient ischemic attack or unspecified cerebrovascular disease. Stroke diagnoses were not included as cases if recorded within 30 days of traumatic stroke, or within 7 days after a major accident or fall. Fatal strokes were stroke cases recorded within 14 days of death.
Case exposure was assigned as the study drug exposed on stroke date minus 7 days.
Statistical Analysis: For each exposure, we estimated incidence rates with 95% confidence intervals (CIs) and incidence rate ratios (adjusted for age, calendar year, osteoporotic disease type, and smoking status) with 95% CI compared to VitD only (common reference group).
