Real-world test of colonoscopy screening program shows cancer reduction but no mortality benefit
Colonoscopy is often called the “gold standard” of colon cancer screening. Now, the gold standard of screening has been put to the test in the gold standard of clinical trials — a randomized controlled trial. The NordICC trial (Nordic-European Initiative on Colorectal Cancer) was recently published in the New England Journal of Medicine, and it’s already creating a big debate. Let’s break down the results.
What did this trial measure?
While there have been many observational studies of colonoscopy (comparing outcomes among people who received a colonoscopy and who did not), there has never been a randomized controlled trial. That’s important because there may be differences between patients that get colonoscopies and those who do not that cannot be adjusted for in observational studies (such as access to healthcare, income, etc). By randomizing people to either get an invitation to screen or not, we can better evaluate the effectiveness of the program in a real-world setting.
Before looking at the results, it’s important to establish what this trial was designed to measure. In the NordICC trial, about 85,000 people in Poland, Norway, Sweden, and the Netherlands were randomized to either get an invitation to have a colonoscopy or not get an invitation. Cancer incidence, mortality from colon cancer, and all-cause mortality were compared after ten years.
The NordICC trial is a test of a public health intervention– if we invite people to get screened, what will be the impact on colon cancer incidence and mortality? This is why the study authors call it a “pragmatic” randomized controlled trial; they designed it to find out what will happen in the real world if they implement this screening program. This means the study takes into account aspects of screening programs that impact the outcome of the program, like how many people are able and willing to get screened. For example, a screening test that is more effective but few people want to do could have the same effect on population health as a test that is less effective but easier for more people to do.
What were the results?
The results of what are called the “intention to screen” analysis (comparing the control and intervention groups) were disappointing. While the group invited to have a colonoscopy screening had lower rates of cancer incidence (relative risk reduction of 18%, absolute rate reduction from 1.2% to 0.98%), their rates of mortality from colon cancer and all-cause mortality were not significantly different from the control group after ten years. To prevent one case of colon cancer, 455 people would need to be invited to undergo screening.
In previous observational studies of colonoscopy, the test reduced rates of colon cancer and cancer-specific mortality by 60%, so these findings were sobering for many doctors. “I think we were all expecting colonoscopy to do better,” said Samir Gupta, a gastroenterologist at the University of California, San Diego and the VA, in an interview with StatNews. The study authors speculate that there wasn’t as large a decrease as expected because colon cancer rates have declined in the trial countries and treatments for colon cancer have improved.
However, the study also showed some positive aspects of colonoscopy. The invitation to screen with colonoscopy did reduce rates of colon cancer. The rate of major bleeding from the procedure was very low (0.13%) and there were no bowel perforations (a serious adverse event). Previous studies have showed a bowel perforation rate of 3 per 10,000 procedures and an incident cardiovascular event rate of 120 per 10,000 procedures, so the lack of adverse events in this study was remarkable. It’s possible that the doctors in this trial took more care than usual, or that these procedures have become safer over time.
Low uptake problem
The biggest criticism of the study was the relatively low number of people invited to screen who went through with the procedure. Fewer than half (42%) of those invited to screen actually received a colonoscopy. In the US, about 72% of adults 50-75 are up to date with colon cancer screening. However, looking only at the study results from Norway, which had a take-up rate of 61%, the results for colon cancer incidence were similar to the full sample (relative risk reduction of 24%, absolute rate reduction from 1.57% to 1.2% over ten years).
Given the low take-up rate, the authors also conducted a “per-protocol analysis” to estimate the rate of cancer incidence and mortality if everyone who was invited to get a colonoscopy actually got one. They found that rates of colon cancer would have been lower (relative risk reduction 31%, absolute reduction 1.22% to 0.84%) as well as rates of colon cancer mortality (relative risk reduction 50%, absolute reduction 0.30% to 0.15%).
A 50% decrease in death from colon cancer sounds great, but it’s important to point out the context. The absolute risk of colon cancer death only decreased by 0.15%, compared to an 11% risk of dying from any cause. We also don’t know for sure whether this reduction is due to the impact of the procedure itself or to underlying characteristics among people who get colonoscopies.
The fact that uptake was lower than expected is itself an important finding, because it demonstrates a key limitation of colonoscopy screening programs: it doesn’t matter how much colonoscopies reduces deaths if nobody wants to get one! Using the “intention to treat” analysis, we see how well the colonoscopy screening program works in the real world, taking into account the barriers that people face to getting the screening. That allows us to better compare colonoscopies to other types of screening that may be easier for patients to undergo.
The NordICC trial shows that colonoscopy screening programs may not be as effective as expected, especially in comparison to other types of colon cancer screening. For example, the study authors note that the results put colonoscopy on par with flexible sigmoidoscopy, a cheaper colon cancer screening test that does not require sedation.
The results raise several questions for public health leaders to consider:
- Given the high cost of colonoscopy ($2000 on average for privately insured patients and as much as $10,000 in some cases), how much should we be investing in colonoscopy compared to other colon cancer screening methods or interventions?
- Are there ways we can better target those at higher risk of colon cancer with types of screening that have higher uptake rates than colonoscopy?
- How much are we spending on colonoscopy screening for younger low-risk people that we could use to increase awareness of early colon cancer symptoms or toward addressing the root causes of colon cancer?
The NordICC trial prompts us to reconsider colonoscopy as the “gold standard,” and to rethink how our healthcare system invests in prevention and screening around colon cancer.