Long-term effects of diabetes pharmacotherapy and multiple medication use on clinical and economic outcomes from claims datasets

Background: Pharmacoepidemiology is the study of the use, beneficial and adverse effect of medications in defined population. Using claims data in pharmacoepidemiologic studies allows us to observe the association of medication use and subsequent health and economic outcomes in more closely resemble treated populations in real-world settings. Objectives: This dissertation aimed to examine (1) the association between the time from first-line therapy to insulin initiation and cardiovascular disease (CVD) risk, (2) the association between insulin initiation and within-person change in healthcare utilization, and (3) the association between chronic multiple medication use (thereafter “persistent polypharmacy”) and treated fall injury. Methods: Our study populations included (1) middle-aged Pennsylvania Medicaid enrollees with type 2 diabetes (T2DM) and (2) community-dwelling older adults in the Health, Aging and Body Composition Study with linked Medicare Fee-For-Service claims. Conditional Cox models proposed by Prentice, Williams, and Peterson with inverse probability treatment weighting, generalized estimating equations models with propensity score matching, and multivariable Cox proportional hazard models with time-varying polypharmacy were applied to examine three objectives, respectively. Results: First, insulin initiation, even in the early course of T2DM pharmacotherapy, did not prevent CVD risk or reduce healthcare burden possibly due to CVD events that developed before insulin for middle-aged Medicaid enrollees with T2DM. Secondly, community-dwelling older adults aged 70-79 years with sustained multiple medications had an increased risk of treated fall injury; but the relationship was attenuated when accounting for performance-based quadriceps strength and diabetes. Conclusions: These findings suggest that further studies are needed to focus on prevention efforts such as controlling CVD risk factors in middle-aged Medicaid enrollees with T2DM to improve their health outcomes and reduce healthcare burden. In addition, clinicians should consider musculoskeletal strength and diabetes in older adults with persistent polypharmacy for increasing fall injury risk. Although several challenges are present when using claims data, the public health relevance of these findings is providing new insights in the effect of medication use in populations who are not typically included in clinical trials because of comorbidity, socioeconomic status, or logistics, and informing disease prevention efforts and resource planning in public health