Despite great progress in cancer therapeutics, cancer continues to place a tremendous burden on patients’ physical, emotional, and financial health. Although better supportive care has made treatments more tolerable, the lives of patients and their families are still disrupted. In addition, many effective treatments never reach those in need. These aspects of cancer care lie outside the realm of most therapeutic clinical trials and thus have not been studied nearly as well as traditional clinical end points of efficacy. Outcomes research (OR) is a field of study that encompasses important aspects of cancer care that extend beyond traditional survival end points. Goals of OR are to fully characterize how cancer therapy impacts patients, assess the quality of care being delivered, determine why it is delivered in that manner, and include analyses of how health care is organized and delivered. Many OR investigations approach these questions from the patient’s perspective, although OR also investigates cancer care from the perspective of other decision makers, including physicians, health care payers, and policy makers.

Although some may argue about what technically falls under its rubric, OR typically uses patient-oriented end points, such as quality of life (QOL), long-term impact, patient preferences, or patient satisfaction as ways to evaluate delivered care. By comparison, cancer clinical trials classically use the more clear-cut measures of survival, diseasefree survival, and toxicity (Table 11.1) (1,2). Many studies can be defined as outcomes research, and therefore OR can be categorized along a number of dimensions. The four most common classifications are theme, end point, arena of application, and method (1,3). For example, clinical decision making and quality of care are two broad themes in OR. However, a single theme, such as quality of care, might be measured by the distinct end points of patient satisfaction, receipt of recommended care, or survival. Another classification scheme uses arenas of application to subdivide OR studies into the following: macrolevel analyses that examine population trends in end points such as QOL; mesolevel analyses that examine a diverse range of topics such as effectiveness, cancer impact, and care utilization; and microlevel analyses that focus on improving patient–physician decision making (3).

Finally, OR investigations may be categorized according to the method they employ. In some circumstances, OR is embedded alongside a traditional clinical trial; in this case, the methods are analogous to those used for trials designed with primary clinical end points. For example, a recent randomized clinical trial evaluating the efficacy of laparoscopic colectomy had patient survival as its primary outcome but included both QOL and cost effectiveness as secondary outcomes (4). Because OR is very concerned with understanding real world practice and how therapies delivered in the clinical trial setting translate into real world practice, outcomes researchers often rely on observational study designs. These studies are of most interest when they describe the experience of large segments of the population and when the experience of the studied segment is perceived to be relevant and generalizable to the population as a whole. The most notable example of this type of OR is secondary database analysis. These studies rely on resources such as the Surveillance, Epidemiology, and End Results (SEER) registries and administrative data from health insurance providers, including Medicare. These resources can be used to evaluate disparities of care and to characterize experience in real world settings. The drawback of these studies is that the lack of random assignment of therapies makes inferences about the effectiveness based on comparisons of treated and untreated persons extremely challenging. Another methodology, meta-analyses of clinical trials, which are predominantly used to answer therapeutic trial questions that are unanswerable by a smaller trial, are sometimes classified as OR. Health services research is a particular type of OR that typically focuses on health care delivery and organization outside the clinical trial setting.

The concept of using patient-oriented outcomes to assess care dates back to the early 20th century when practitioners suggested that care should be measured not by the numbers of patients being treated, but by the impact of treatment on the health of patients (1). It was not until the 1990s, however, that OR gained traction and recognition as an important way to investigate cancer care. During this decade, the number of outcomes studies focused on cancer and cancer care has markedly increased, and systematic reviews show that this effort has
been largely led by investigators in the field of breast cancer (1). In gastrointestinal (GI) malignancy, the initial forays into OR methodology were undertaken by those studying colorectal cancer (CRC) screening, generally not practicing oncologists (5). These trends have begun to change, however, and outcomes methods—such as secondary analyses of large databases and the incorporation of QOL measures as secondary end points of clinical trials—are commonly used today. Mainstream oncology groups have begun to recognize the importance of this emerging field to the study of cancer. The American Cancer Society, the National Cancer Institute, and the American Society of Clinical Oncology, among others, have initiated OR teams or have integrated health services researchers into their organizations. Increasingly, there are dedicated research funds to support this type of research. Despite this gathering momentum, there are still large gaps in the application of outcomes me thods to treatment of cancer, and specifically to GI malignancy.

This chapter provides an overview of OR by describing some representative typologies in CRC. It is not intended to be a comprehensive review, but rather this chapter is meant to convey why OR is valuable for a comprehensive understanding of cancer care. Because OR has been applied less frequently to noncolorectal GI cancers, CRC studies are highlighted for this purpose. However, this is unfortunate because gastroesophageal and pancreatic cancers are diseases for which treatments are both rigorous and inadequate. In such a setting, OR topics such as communication of patient preference, information exchange, and impact of treatment on QOL should all be integral components of the treatment process. Hopefully, subsequent editions of this text will comprise separate chapters describing the newer OR pertaining to each GI cancer.

Patterns of care and access to care studies both aim to assess what treatments cancer patients are actually receiving outside the realm of the clinical trial. In addition, they hope to uncover what factors predict the receipt of care. The primary methodology for this research question is secondary analysis of existing databases. These investigations allow for the identification of treatments or interventions that are not being used as recommended; in addition, because they are clinically annotated, they allow for the identification of groups—such as the elderly, the poor, and racial minorities—that are less likely to receive recommended care. However, although a study using a database may demonstrate an association between a clinical characteristic (e.g., black race) and an outcome (e.g., survival), such a study is unable to identify the cause of the outcome disparity. Thus, the associations that emerge from analyses of patient outcomes from large databases are generally used to generate hypotheses. These investigations identify target areas for further study and future interventions that might improve the quality of care for cancer patients.