(179) Developing a Framework for Rapid Analytics: A Self-Controlled Case Series Study Replication

Publication Number: 1224

Background: The ability to replicate prior studies in other data sources and time periods in a rapid and reproducible manner for assessment of confounding and bias is needed for well-known and novel associations. Use of rapid analytic tools can accelerate these processes.

Objectives: To replicate and update a landmark self-controlled case series study on association of antidepressants and risk of hip fracture, conducted using The UK Clinical Practice Research Datalink data over 1987-1999 (Hubbard 2003), in a large US claims database using Instant Health Data (IHD) rapid analytics platform.

Methods: Using Optum’s de-identified Clinformatics® Data Mart Database, all patients with diagnosis of hip fracture or fractured neck of femur between 2010-01-01 and 2019-12-31 were identified. Event date was defined as date of first recorded diagnosis for hip fracture. Person-time was segmented into days 0-14, 15-42, and >43 following first antidepressant prescription, with end date of exposure defined as last prescription day plus days supplied, allowing for treatment gaps up to 30 days. Two 91-day washout periods were included following end of treatment. Remaining person time was classified as baseline. Following Hubbard, incidence rate ratios (IRRs) were calculated for exposure to tricyclic antidepressants (TCAs) and selective serotonin reuptake inhibitors (SSRIs) separately.

Results: We identified 177,961 patients with diagnosis of hip fracture. The mean(SD) age of cases was 75.95(14.83) and 68.9% were female. Rates of exposure to TCAs pre-index were 4.24% and 2.76% for cases and controls, while corresponding rates for SSRI exposure pre-index were 22.6% and 14.17%. Rates for TCAs were higher in Hubbard et al (cases: 17.8%; controls: 11.8%) and lower for SSRIs (5.5% and 3.0%).

IRRs for fracture in days 0-14, 15-42, and >43 of tricyclic treatment were IRR(CI): [1.51 (1.30,1.76), 1.23(1.07,1.41), 1.23(1.14,1.32)] and for SSRIs: [1.77(1.67,1.87), 1.52(1.44,1.59), 1.37(1.33,1.41)]. Results reported by Hubbard et al were directionally similar but higher in magnitude for TCAs: [2.30 (1.82,2.90), 1.88(1.55,2.28), 1.18 (1.06,1.31)] and for SSRIs: [1.96(1.35,2.68), 2.03 (1.53,2.68), 1.54(1.33,1.78)]. Increased risk of fracture was observed in the first washout period for both TCAs and SSRIs with IRR(CI) of 1.15(1.06,1.26) and 1.33(1.28,1.37), but no increased risk was observed in the second washout period. The same finding was observed by Hubbard et al. Risk of hip fracture increased with age and was significantly elevated in patients 85 years or older.

Conclusions: In this study, we were able to confirm the association between use of antidepressants and the incidence of hip fracture using contemporary data and modern analytic tools.