Determining medications associated with drug-induced pancreatic injury through novel pharmacoepidemiology techniques that assess causation
PI: Ravy Vajravelu, MD, MCSE
Funding Source: NIH/NIDDK
July 1, 2023 - April 30, 2027
Acute pancreatitis causes nearly 300,000 hospitalizations per year in the United States, and its rates are rising. One-third of cases are classified as having unknown cause, leaving patients vulnerable to repeated episodes because they do not know how to alter their lifestyles. Unexpected side effects of prescription medications may be responsible for acute pancreatitis cases with unknown cause. This situation is called drug-induced pancreatic injury (DIPI). Unfortunately, healthcare providers and medical researchers do not know which medications cause DIPI. This is because the majority of research about DIPI comes from descriptions of the experience of individual patients. While these are valuable for providing clues about medications that might cause DIPI, they do not account for other factors that could contribute to acute pancreatitis. Therefore, conclusions from this type of study may falsely label particular medications as dangerous. This may lead to reduced use of medications that are effective for the conditions that they treat, resulting in worse outcomes for patients. There is a critical need to determine which medications do and do not cause DIPI in order to prevent cases of acute pancreatitis and to continue patients on safe essential medications. The recent availability of electronic databases with health information and powerful computer processing has made it possible to study the effects of thousands of medications. Additionally, a new data analysis technique called pharmacopeia-wide association studies (PWAS) has improved the efficiency of these studies. Furthermore, PWAS can be combined with fundamental epidemiology principles to determine whether a study finding demonstrating a medication side effect is true or false. The overall objective of this proposal is to identify medications that cause DIPI by applying PWAS to two large databases of patient health information. Additionally, this proposal will combine PWAS with a research framework called the Bradford Hill criteria to distinguish medications that cause DIPI from false results. The specific aims of this proposal are (1) To identify medications that are strongly associated with DIPI, demonstrate dose response, and exhibit biologic plausibility by applying the PWAS framework to case-control studies; (2) To identify medications that demonstrate consistent temporality and specificity with DIPI through novel applications of the PWAS framework; (3) To identify replicable medication-DIPI associations by repeating Aims 1 and 2 using a second database; and (4) To develop and disseminate an interactive database to integrate the study findings for clinicians and investigators. This research is significant because it will improve patient outcomes by resolving clinical uncertainty about which medications should be stopped after acute pancreatitis and which essential medications are safe to continue. This research is innovative because it combines cutting-edge data analysis techniques with fundamental research principles to comprehensively identify medications that cause DIPI. These techniques will be applied to future studies that aim to identify medications that contribute to other medical conditions.