(P-573) Using Negative Control Outcomes to Assess the Comparability of Treatment Groups Among Postmenopausal Women Switching Osteoporosis Therapy

Background: Prior research suggests that osteoporosis (OP) treatment-naïve women initiating different therapies may not be comparable, likely because therapy choice is influenced by disease severity. Patients switching medications may be more comparable. Bias can be assessed with negative control outcomes (NCOs), which are expected to have no plausible association with OP treatments.

Objectives: Using NCOs, we evaluated the comparability of women switching OP treatments, overall and among women with recent fracture.

Methods: We used Optum Health commercial insurance claims from 2012 through 2018 to identify women aged 55 years or older initiating denosumab, intravenous zoledronic acid (ZA), or an oral bisphosphonate (BP) (risedronate, alendronate, or ibandronate) within 455 days after the use of a different type of OP medication. We required at least 15 months of continuous enrollment before initiation and excluded patients with a history of Paget’s disease or cancer. We compared women switching to oral BPs, denosumab and ZA, with respect to 12 NCOs selected by subject matter experts. We estimated the 12-month cumulative risk difference (RD) and 95% confidence intervals (CI) for all NCOs except early hip fracture, which was measured at 90 days after treatment initiation. We used an augmented inverse-probability treatment and censoring weighted estimating function to control for confounding and dependent censoring by baseline demographics, healthcare utilization, comorbidities, and comedications. Additional analyses stratified the cohort by calendar time (before and after October 2015 to account for ICD coding changes).

Results: We identified 32,336 women, of which 52%, 37%, and 11% switched to an oral BP, denosumab and ZA, respectively. Women switching to denosumab and ZA were older and were more likely to have osteoarthritis and history of fracture compared with women switching to oral BPs. Nearly all NCOs for all exposure comparisons had similar risks in the overall population and among women with recent fracture, indicating that these groups were comparable. There were more differences before October 2015 than after (e.g., RD for influenza vaccine for denosumab versus oral BPs, before 2015: 4.0% [95% CI 1.7-6.3%] versus after 2015: 2.6% [95% CI -0.2-5.5%]).

Conclusions: After adjustment, there was minimal evidence of residual bias for most NCOs among women switching to a new OP treatment, overall and among women with a recent fracture. Women initiating in the later period appeared to be more comparable, suggesting there could be residual bias in the earlier period.