A Critical Assessment of “Aspirin Use and Risk of HCC and Gastrointestinal Bleeding in Patients With HBV‐Related Cirrhosis: A Landmark Analysis”

We appreciate the study of Kim et al. [1]. The authors provide an in-depth study of the effect of aspirin intake on the associated risk of hepatocellular carcinoma and bleeding of the gastrointestinal tract in patients with HBV-related cirrhosis, which is worth of praise. These results also show statistically significant and consistent findings that the aspirin-treated cohort had a significantly reduced occurrence of hepatocellular carcinoma. Although the study provides worthy input, it has several limitations that were overlooked, and if given special attention in future studies, it will make this research more relevant, meaningful, and impactful. First, the study defined aspirin exposure using NHIS prescription information. Low-dose aspirin is widely sold over-the-counter; therefore, prescription claims do not include it. Exposure misclassification may result. Counting true OTC aspirin users as nonusers contaminates the comparison group. This bias's effect estimates toward the null, making aspirin's apparent protective effect on HCC or bleeding risk less credible [2]. Pharmaco-epidemiology recognizes misclassification of drug exposure due to OTC use and should be discussed. Second, proton-pump inhibitors, nonsteroidal anti-inflammatory medicines, and anticoagulants greatly affect gastrointestinal bleeding risk but were not considered in the analysis. Without adjustment for these drugs, aspirin's increased bleeding risk may be confounded by variations in co-medication use between users and nonusers. The absence of key co-medications in bleeding risk studies is a methodological flaw that impairs association specificity [3]. Third, due to cardiovascular comorbidity, aspirin users may be monitored more often and adopt healthier habits, the healthy-user effect. Both enhance the likelihood of early or frequent outcome detection (e.g., HCC) or baseline risk irrespective of aspirin. Enhanced surveillance may increase outcomes in one group, but healthy aspirin use may reduce HCC risk independently. These biases may misattribute aspirin risk differences. Surveillance bias and healthy-user or adherer bias are well-known in observational research and should be disclosed [4]. Fourthly, the study did not model aspirin exposure status changes, initiation, and withdrawal after the 3-year milestone. Real-world data is dynamic, so time-varying modeling is better. Ignoring exposure changes over time may lead to misclassification (e.g., patients who quit taking aspirin are still considered users) and skewed hazard estimates or duration–effect associations. Thus, pharmaco-epidemiology requires time-varying exposure modeling and flags its absence. Furthermore, the study used ICD-10 codes to classify all gastrointestinal bleeding occurrences without distinguishing upper vs. lower GI bleeds, variceal vs. non-variceal, or severity such major vs. small bleed, transfusion required. In observational safety studies, detailed outcome definitions and clinical relevance are expected, so aggregating bleeding types may obscure clinically meaningful differences, limit interpretability, and weaken the ability to balance aspirin benefit vs. harm [5]. In conclusion, we praise the author's contribution to the investigation of aspirin usage and its correlated risk of hepatocellular carcinoma and gastrointestinal hemorrhage in individuals with HBV-induced cirrhosis. However, potential exposure misclassification of over-the-counter aspirin use, lack of co-medication adjustment, no time-dependent exposure modeling, surveillance and healthy-user bias, and lack of detail on bleeding type and severity are a few concerns, which need to be studied further in future studies to enhance the impact of this study. The authors have nothing to report. The authors declare no conflicts of interest. Data are available on request from the authors.