(293) Use of a Rate-Based Sequence Symmetry Analysis to Assess Recurring Adverse Drug Events

Publication Number: 967

Background: Sequence symmetry analysis (SSA) is well-proven to identify first time onset of an adverse event. Investigating event recurrence may be important clinically and booster statistical power.

Objectives: Using a known association between tumor necrosis factor-α (TNF-α) inhibitors and pneumonia, to compare findings derived from traditional and rate-based SSA, a proposed alternative SSA method.

Methods: This study utilized MarketScan claims database (2005-2019) including TNF-α inhibitors initiators (index date) aged ≥20 with pneumonia diagnosis (360 days of index date) and continuous enrollment (720 days before and 360 days of index date). Pneumonia was identified in any position from inpatient claims only (definition 1), outpatient claims only (definition 2), and inpatient or outpatient claims (definition 3). Crude and adjusted sequence ratio (cSR/aSR) were calculated separately under traditional and rate-based SSA frameworks. Within the traditional framework, cSR was the ratio of the number of patients with first pneumonia occurred after the index date to the number of patients with first pneumonia occurred before the index date; Within the rate-based framework, cSR was the ratio of the number of pneumonia diagnoses after the index date to the number of pneumonia diagnoses before the index date. For both frameworks, aSR was calculated by adjusting for secular trends in pneumonia diagnoses.

Results: Within the traditional framework, 1,601 (Def 1), 4,047 (Def 2), and 4,465 (Def 3) patients were identified. Within the rate-based framework, 1,669 (Def 1), 4,467 (Def 2), and 4,951 (Def 3) patients were identified. Across all three definitions, aSR derived from rate-based framework (Def 1: 1.62 (1.53-1.72)/Def 2: 1.33 (1.28-1.38)/Def 3: 1.32 (1.27-1.36)) were consistently higher than that derived from traditional framework (Def 1: 1.48 (1.40-1.58)/Def 2: 1.21 (1.16-1.26)/Def 3: 1.19 (1.15-1.23)). For infliximab, aSR was Def 1: 1.94 (1.54-2.44)/ Def 2: 1.64 (1.42-1.89)/ Def 3: 1.57 (1.37-1.80)) within the rate-based framework and Def 1: 1.80 (1.39-2.32)/ Def 2: 1.37 (1.15-1.63)/ Def 3: 1.31 (1.11-1.55) within the traditional framework. For etanercept, aSR was Def 1: 1.43 (1.14-1.79)/ Def 2: 1.21 (1.07-1.37)/ Def 3: 1.23 (1.10-1.39) within the rate-based framework and Def 1.31 (1.02-1.68)/ Def 2: 1.10 (0.94-1.27)/ Def 3: 1.11 (0.96-1.28) within the traditional framework. For adalimumab, aSR was Def 1: 1.59 (1.32-1.93)/ Def 2: 1.28 (1.15-1.42)/ Def 3: 1.25 (1.13-1.38) within the rate-based framework and Def 1: 1.45 (1.18-1.79)/ Def 2: 1.23 (1.09-1.38) / Def 3: 1.19 (1.06-1.34) within the traditional framework.

Conclusions: Using TNF-α inhibitors and pneumonia as an example, stronger signals were detected from rate-based than traditional SSA approach.