Assessing the Risk of Drug-Induced Influenza-Like Illness: A Disproportionality Analysis of Two Decades of FAERS Data (2004-2025)

Introduction: With the rising incidence of Influenza-like Illness (ILI), comprehensive studies on causative drugs remain lacking. This study aimed to detect drugs associated with ILI through disproportionality analysis of the FDA Adverse Event Reporting System (FAERS) database. materials and methods: Data source and mining We obtained 85 quarters of data spanning Q1 2004 to Q1 2025 from the FAERS database (https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html). The dataset included raw data packets formatted in American Standard Code for Information Interchange (ASCII), which were processed using SAS 9.4 software for data mining and subsequent analysis. Data processing The FAERS database contains seven parts: demographics (DEMO), drug information, AEs, patient outcomes, reporting sources, treatment duration, and indications. As reports were voluntarily submitted, data cleaning was essential. This study strictly complied with FDA guidance for FAERS data cleaning procedures. AE names were described using Preferred Terms (PTs) from the MedDRA (version 27.0). Drugs associated with ILI were identified by searching the PT “influenza like illness” (code:10022004). Drugs demonstrating positive signals were classified using the Anatomical Therapeutic Chemical (ATC) system, with ATC1 (first level) corresponding to anatomical/organ-system categorization and ATC2 (second level) reflecting therapeutic/pharmacological subgroup classification. Statistical analysis We performed a disproportionality analysis using four statistical methods: the ROR, PRR, BCPNN, and EBGM (Table 1) 13,14. Corresponding calculation formulas are summarized in Table 2. Time to onset (TTO) was characterized using median and interquartile range (IQR) as statistical descriptors. Weibull distribution modeling characterized TTO patterns for specific AEs, parameterized by scale (α) and shape (β) 15. Early failure types (βlt;1) indicated decreasing hazard profiles, while random failures (β≈1) suggested constant risk over time. Figure 1 depicts the workflow for detecting drug-induced ILI signals within the FAERS database. Methods: Adverse Event (AE) reports were extracted from the FAERS database for the period spanning 2004 Q1 through 2025 Q1. The Medical Dictionary for Regulatory Activities (MedDRA) was used to identify ILI cases. Four statistical methods were employed to identify drugs with positive signals: the Reporting Odds Ratio (ROR), the Proportional Reporting Ratio (PRR), the Bayesian Confidence Propagation Neural Network (BCPNN), and the Empirical Bayes Geometric Mean (EBGM). results: Demographic Analysis Following data deduplication and screening processes, 77,060 ILI reports were included in this study spanning from Q1 2004 to Q1 2025. The characteristics of drug-induced ILI were shown in table3. Demographic analysis revealed that 50,682 originated from female patients, 22,042 from male patients, and 4,336 were gender-unspecified. The median patient age was 53 years, with the largest proportion (n=24773) in the 45–65-year age group. Additionally, 43,795 reports were submitted by consumers. Among serious cases, the most frequent outcome was other serious outcomes (n=24,786), followed by hospitalization (n=14,520), disability (n=1944), life-threatening (n=1372), and death (n=1139) (Figure 2). The median TTO and IQR associated with ILI was 7 days (0, 144), as illustrated in Figure 3. Disproportionality analysis After analysis with four signal detection methods, 60 drugs showed positive signals, with the top 20 presented in Table 4 and the remainder detailed in Supplementary Table S1. The top 5 drugs with the largest number of reported cases were interferon β-1a n=12,470, ROR 22.12(21.69,22.55), PRR 21.59(21.18,22.00), IC 4.19(4.16), EBGM 18.26(17.91), evolocumab n=2,935, ROR 6.94(6.69,7.20), PRR 6.88(6.63,7.14), IC 2.74(2.68), EBGM 6.66(6.42), ofatumumab n=2,594, ROR 19.13(18.39,19.91), PRR 18.68(17.97,19.42), IC 4.18(4.11), EBGM 18.09(17.39), interferon β-1bn=1,862, ROR 16.69(15.93,17.48), PRR 16.35(15.62,17.11), IC 4.00(3.92), EBGM 15.97(15.25), and zoledronic acidn=1,830, ROR 4.95(4.73,5.19), PRR 4.92(4.70,5.16), IC 2.27(2.20), EBGM 4.83(4.61). According to the ATC1 classification, antineoplastic and immunomodulating agents accounted for the highest proportion of reported cases (n=43,436), followed by nervous system#40;(n=8,941#41;, cardiovascular system#40;(n=6,871#41;, and musculoskeletal system#40;(n=5,577#41;. According to the ATC2 classification, lipid-modifying agents, immunostimulants, and drugs for the treatment of bone diseases have been identified as potential drugs associated with ILI. For the top 20 drugs associated with ILI reports, TTO estimates derived via Weibull distribution modeling and Weibull shape parameter analyses are summarized in Table 5. Our investigation revealed a shape parameter β lt; 1 for most drugs, indicating that drug-induced ILI predominantly manifested early failure types. Results: Our analysis identified 77,060 ILI-associated reports. A total of 60 drugs exhibited positive signals. The 5 most frequently reported drugs were interferon β-1a, evolocumab, ofatumumab, interferon β-1b, and zoledronic acid. Under the ATC1 classification, antineoplastic and immunomodulating agents accounted for the majority of reported cases (n=43,436), followed by nervous system agents (n=8,941), cardiovascular system agents (n=6,871), and musculoskeletal system agents (n=5,577). Weibull shape parameter analyses revealed β<1 for most drugs, indicating that drug-induced ILI predominantly manifested early failure types. discussion: The approval of novel immunotherapeutic agents has sparked increasing interest in drug-induced ILI. In patients co-infected with influenza virus who are taking drugs that may cause ILI, ILI symptoms might be aggravated. Nevertheless, there remains a gap in the systematic investigation of drugs that might induce ILI, and this study aims to address this deficiency in existing research. Consequently, it is crucial to identify drugs that could cause ILI. The result of this study showed that the incidence rate of ILI (0.14%, 77,060/56,173,849) was higher than that reported in previous studies on eosinophilia and pityriasis rosea16,17. Among the 60 drugs that exhibited positive signals, antineoplastic and immunomodulating agents were the most numerous (n=43,436), with interferons constituting the majority of this category. Interferons are established therapeutic agents known for their antiviral, antiproliferative, and immunomodulatory effects 18. Shaygannejad et al. highlighted ILI as the predominant AE in interferon therapy. Specifically, the occurrence of ILI was recorded at 83.33% within the peginterferon β-1a cohort, while it was observed at 79.52% in the interferon β-1a cohort 19. Furthermore, additional research findings have indicated a comparatively elevated occurrence of ILI during the administration of interferon therapy 20-22. In our study, we observed that interferon β-1a and interferon β-1b exhibited the highest frequencies, followed by peginterferon α-2a, and peginterferon β-1a. Notably, all four drugs exhibited strong signals. These findings were not only consistent with the aforementioned studies but also aligned with the adverse reactions reported in the package insert. The precise mechanism by which interferon-induced ILI remains unclear, and might be linked to the extensive regulatory influences of interferons on the immune system and cytokine 23. Interferons promote immune cell activation, leading to the secretion of pro-inflammatory cytokines, including interleukin-1 (IL-1) and tumor necrosis factor-α (TNF-α) 24,25. These cytokines subsequently interact with the thermoregulatory center and the neuromuscular system, which contributes to the symptoms such as fever, headache, and myalgia. In addition to interferons, there has been an increased reporting of ofatumumab and natalizumab. Ofatumumab is a completely human anti-CD20 monoclonal antibody (mAb). Initial clinical trials reported favorable tolerability of ofatumumab with low ILI incidence 26. However, our pharmacovigilance data from 2,594 cases revealed significant ILI correlations, necessitating large-scale real-world studies to validate causality. Natalizumab is a humanized IgG4 monoclonal antibody. The most commonly reported AEs are headache, nausea, and infusion-related responses; however, ILI has not been documented in the literature 27,28. Given that Natalizumab has been available on the market for a short period, comprehensive safety data remain limited. Thus, systematic pharmacovigilance investigations are required to evaluate its potential ILI induction risk in clinical applications. In our research, the PCSK-9 inhibitors demonstrated positive signals in the cardiovascular system category. As a novel lipid-lowering class, they bind PCSK9 to block low-density lipoprotein(LDL) receptor interactions, enhancing cholesterol clearance 29. A retrospective analysis conducted on the Danish population indicated that after one year of treatment wit