An Informatics Tool to Assess Individualized Colon Cancer Risk

Abstract

We outline the methodology used to develop an Internet-based cancer risk assessment tool and describe factors needed to create and post credible risk assessment tools or risk calculators.

Introduction

Healthcare delivery is shifting towards giving patients more responsibility in their choices. Consumer retrieval of health information has never been more prevalent than today, with ready access to large amounts of health information on the Internet. New and experienced users spend 4% – 5% of their online time looking for medical information, a near equivalent to the online time spent viewing news (1). Whereas the physician in the past gave educated filtering of health information to consumers, the Internet, which is fast becoming a health information repository, does not help consumers to critically evaluate this information. Health information on the Internet is not necessarily concordant with evidence-based sources (2). Consumers need valid and reliable sites created using evidence-based medicine, developed with human-centered design guidelines, authored by credible sources, yet providing evidence of review by content experts. We are developing such a site on cancer risk. Colon cancer is our initial model prototype since it is ideally suited to primary and secondary prevention strategies due to the slow progression of the theorized adenoma-carcinoma sequence (3). Thus, it is a disease that potentially can be impacted by screening and lifestyle changes (4).

Tool Development

A colon cancer risk assessment model for the general population is being developed based on a meta-analysis of published epidemiologic studies. A systematic review of the literature identified all published English-language articles that provided estimates of risk. We began by discovering patterns among study results on factors potentially contributing to or decreasing colon cancer incidence. We included in the model only cohort studies that used colon cancer incidence. Experts in clinical gastrointestinal cancer prevention, epidemiology, and biostatistics reviewed the evidence for contributing and protective factors. After establishing the factors, we developed a risk formula for the model based on the relative risks in the included studies, reported population prevalence rates, and SEER cancer rates to estimate future cancer risk within 10-year windows. The outcome is a baseline probability adjusted for relative risks to determine individual colon cancer 10-year risk probability. The next step is to validate the model using a large population cohort.

Website Development

Website development has been an iterative process. In addition to user, task, functional, representational, and comparative analyses, a heuristic evaluation of the site in the design phase ensures a matching of the users’ and designers’ conceptual models of the site.

Discussion

Informatics tools such as risk calculators are becoming prevalent on the Internet. Regardless of their potential to educate consumers on disease risks, these tools must be built using evidence-based medicine, authored by credible sources, and approved by content experts. This requires the expertise of many professionals, such as disease experts, statisticians, epidemiologists, behavioral scientists, computer scientists, and health informaticists. Although these sites are potentially useful for physicians, genetic counselors, and consumers in making informed decisions regarding screening and lifestyle changes, they must be medically reliable and valid and help users filter the vast amount of potentially conflicting information. Their content should be endorsed for accuracy by appropriate, independent professional organizations. To anticipate potential harm by non-credible sources, medical informatics must set the standards for these sites.

This research is supported in part by the National Colorectal Cancer Research Alliance.