A model to predict the risk of aspirin/non-steroidal anti-inflammatory drugs-related upper gastrointestinal bleeding for the individual patient

Upper gastrointestinal bleeding is a feared complication of using non-steroidal anti-inflammatory drugs (NSAIDs) or aspirin. Studies predicting the incidence rate for individuals with a given set of characteristics are lacking. The aim of this study was to develop a risk model to predict the incidence rate of upper gastrointestinal bleeding (UGIB) in users of aspirin/NSAID based on presence of well-defined risk factors for the individual patient. Methods: The model was developed from data from a case-control study, sampled from a well-defined source population, residents of the Funen County 1995-2006. All cases and controls were characterized in terms of factors known to affect the risk of UGIB. By using census data, we rescaled the control group, so their composition accurately reflected age and sex distribution of the source population. Only persons using NSAIDs or/and aspirin and no PPI were included in the analysis. As reference group, we chose 80- to 89-year-old women with no ulcer history, using NSAID, but neither aspirin, other platelet inhibitors, vitamin K antagonists, selective serotonin reuptake inhibitors nor corticosteroids. Results: We identified 1388 cases among non-users of PPIs. We found a modelled baseline incidence rate of 10.7 per 1000 person-years for the reference group. The strongest associations were found for ADP inhibitors (OR 5.80), followed by anticoagulants treatment (OR 2.62) and prior ulcer (OR 2.68). The model performed well in terms of calibration and discriminatory power. Conclusion: This study is the first to describe a model, which estimates the incidence rate of UGIB for patients using aspirin/NSAID, based on the specific combination of risk factors. Risk of upper gastrointestinal bleeding for a given patient can be accurately estimated using this model.