Incidence of Cancer, Depression, and Economic Burden of Prescription NSAIDs among Older Adults with Osteoarthritis: Statistical and Machine Learning Approaches

Abstract

Understanding the role of chronic inflammation among older adults is critical because of its implications on chronic inflammatory conditions as well as its interaction with anti-inflammatory medications. Osteoarthritis (OA) is a heterogeneous multi-faceted joint disease with multi-tissue involvement of varying severity, and increasing evidence demonstrates a key role of chronic inflammation in its pathogenesis. Approximately 30 million adults in the United States have OA, and it is reported that a large proportion (43%) of them are older than 65 years. OA is typically treated using non-steroidal anti-inflammatory drugs (NSAIDs) for minimizing pain and reducing inflammation. NSAIDs have consistently shown clinically relevant efficacy in anti-inflammation and pain reduction. However, due to their anti-inflammatory properties, NSAIDs may also have spillover effects on chronic inflammation, resulting in an effect on chronic inflammatory conditions such as cancer and depression. Published studies have shown mixed findings about NSAIDs role in disease prevention as well as exacerbation. In addition, NSAIDs are also well-established for their potential adverse effects on cardiovascular, gastrointestinal, and renal systems, especially among older adults. The literature is limited on the effect of NSAIDs on chronic inflammatory conditions such as cancer and depression, and its overall economic burden taking into account its anti-inflammatory effect and adverse effects. Thus, it is imperative to assess the incidence of chronic inflammatory conditions in association with NSAIDs use and its economic burden among older adults with OA in a real-world care setting. Therefore, this dissertation pursued three related aims: 1) to identify whether prescription NSAIDs is one of the leading predictors of incident cancer and assess its direction of association with incident cancer among older adults with OA; 2) to assess the association of prescription NSAIDs with incident depression among older adults with OA and cancer; and 3) to estimate high-cost use, examine factors associated with high-cost users among older adults with OA and identify the leading predictors of healthcare costs among older NSAID users with OA. This study leveraged data from multiple diverse sources to have a complete understanding of the outcomes. For Aims 1 and 2, a retrospective cohort study was designed using Surveillance, Epidemiology, and End Results (SEER) - Medicare claims data, 5% non-cancer random sample of Medicare beneficiaries from SEER regions, zip-code and census tract-level data for the beneficiaries living in the SEER-region, and Area Healthcare Resource Use Files (AHRF) from the years 2006 through 2016. Aim 3 utilized three panels (2016 -