ABSTRACT Detecting cancer early through screening substantially increases the chance of successful treatment, and thus provides the best opportunity for reducing cancer mortality and alleviating cancer burden in the population. National professional associations such as the US Preventive Services Task Force (USPSTF) and the American Cancer Society have developed screening guidelines for various types of cancers based on extensive reviews of scientific evidence. These guidelines identify cancer screening-eligible populations and provide recommendations on the type and frequency of screening tests. However, even with the evidence- based screening guidelines, a decision to undergo cancer screening is not always straightforward, especially for patients with Alzheimer's disease and related dementias (ADRD). ADRD patients suffer from poor functional status, comorbid illnesses, and shortened life expectancy. Screening ADRD patients for cancer may lead to over-diagnosis and treatment of cancers that otherwise would not have caused symptoms during a patient’s lifetime. On the other hand, the key evidence on the effectiveness of a cancer screening test is the reduction in late-stage cancer incidence and mortality, which is often obtained in randomized clinical trials (RCTs). However, RCTs of cancer screening require a long follow-up. Conducting these RCTs in ADRD patients, a population of compromised physical and cognitive abilities, can be logistically, financially, and ethically challenging. To date, it remains unclear whether cancer screening can actually reduce mortality in this population. In our parent award R01 CA246418, we have developed research-grade computable phenotypes (CP) for identifying individuals who are eligible or ineligible for lung cancer screening (LCS) using both structured and unstructured EHR data in the OneFlorida network. We will soon begin to examine the utilization patterns and predictors of appropriate and inappropriate LCS, and estimate the benefits and cost- effectiveness of LCS. OneFlorida is one of the 9 PCORI-funded research networks contributing to the national PCORnet. It contains ~15 million patient EHRs linked with data from various other sources, including Medicaid and Medicare claims, cancer registries, and vital statistics. In this administrative supplement, we propose to identify a retrospective cohort of ADRD patients in OneFlorida to: (1) examine the utilization patterns and predictors of lung, breast, and colorectal cancer screening among ADRD patients and (2) test whether cancer screening reduces late-stage cancer incidence and mortality among ADRD patients. This study will be the first to examine the effectiveness of cancer screening in reducing late-stage cancer diagnosis and mortality in ADRD patients. Given the infeasibility of RCTs, our study will provide critical evidence that can help clinicians, caregivers, and families understand the effectiveness of cancer screening in quantitative terms and decide ...