This is an important story that touches on all sorts of issues involved in public health, including how to best to screen for a disease that takes a terrible toll, how to determine who is at highest risk and how to get the best use out of a test (colonoscopy) that is both expensive and invasive.
The story does a good job summarizing the study authors’ efforts to devise a scoring algorithm that might answer some of those questions, but makes only perfunctory references to cost and doesn’t acknowledge the long history of similar calculators that have been developed and never widely used. That’s pertinent context for this new effort. Despite leaving some room for improvement, the story makes a strong effort at interpreting a relatively simple algorithm with complex ramifications, and draws on two outside experts for perspective — something we wish more stories would take the time to do.
The concept of “risk stratification” — which is what this story and the related research paper address — is difficult for physicians and patients alike. Risk prediction tools are often created and then not used by physicians because they are cumbersome or their use historically has not been incorporated into guidelines because they have important limitations, including that even low-risk people still sometimes develop the cancer or condition. Since this new tool appears to suffer from these same limitations, it’s not clear if it will fare any differently than previous efforts, but the issue is worth exploring.
The question of cost is cited in the final line of the story, which suggests that the scoring system can “address escalating health care cost issues.” But we didn’t think that was specific enough to warrant a Satisfactory score. The story doesn’t clue readers in to the fact that the financial burden can be measured in billions of dollars — a single screening test recommended for millions of Americans can cost thousands of dollars apiece. The other cost issue specific to this test is that it takes valuable time for providers to retrieve a reputable risk calculator when sitting with a patient and enter in the specific data points, including in this case waist circumference, which clinicians often do not measure as a ‘vital sign’ per se.
The story cites a study that credits colonoscopy with reducing the risk of death from colon cancer by about 50 percent, by removing polyps. It would have been helpful for the story to explain what that relative reduction in risk means in absolute terms (e.g. did the rate of cancer death go from 50% to 25% or from 2% to around 1%?)
The story also notes that the study found that patients with a low to intermediate risk score “still have a risk between of between 1.9 percent to 9.9 percent of harboring a polyp that can develop into cancer.”
We’ll give credit here for the story’s provision of these key statistics, which do give a sense as how well the test accurately identifies truly low-risk individuals. However, we’d note that the benefit here should ideally refer to the ability of the algorithm to classify people at all risk levels (including high-risk), and therefore target the highest-risk patients for colonoscopy. The story doesn’t really get into discussion of the predictive power of the test, referred to as a likelihood ratio, which is in the study manuscript. However, this is a difficult concept for most health care providers and patients, and we think it’s understandable that the story did not include a full discussion of this topic.
The harm of a screening test such as this really is “getting it wrong” in terms of predicting someone to be at low risk and finding them to have a higher risk polyp or malignancy. As noted above, the story cautions that “low- to intermediate-risk patients still have a risk between of between 1.9 percent to 9.9 percent of harboring a polyp that can develop into cancer.”
This story accurately explains that the research was about predictive modeling, and explains the components of the risk score as well as the study design. There are some cautionary notes in the text, including a statement from outside expert that “the data from this study is not strong enough to spur any changes in current screening recommendations,” an opinion shared by an editorialist on the study. We think this is sufficient for a Satisfactory rating.
The story does not hype the incidence or seriousness of the disease. The prevalence statistics cited are consistent with these National Cancer Institute figures.
The story contains two independent voices. The perspective they provide is a particular strength of the article. We wish that more stories would reach out to experts not affiliated with the studies being reported on for context.
The story notes that a variety of screening tests exist, but the relevant alternative here is actually other risk calculators, which the authors of the study describe and cite but are not routinely used. The story doesn’t mention these other calculators and why they are seldom used — context that would have deepened reader understanding of the issue.
Although availability isn’t explicitly addressed, this risk calculator is now available (implicitly) by virtue of publication of the paper. It is a simple scoring system that can be implemented based on the information in the paper, and the story makes this clear. We’ll rate this Satisfactory.
The story makes it sound as if this is a first-of-its-kind tool. However, risk calculation tools for colon cancer have been in circulation for many years.
We could not find a news release related to this study. But in any case, the comments from two independent experts assure us that the story wasn’t based on a news release.