Is A Colonoscopy Still Effective? My thoughts on the recent NEJM article

This morning, a patient asked me about the recent NEJM Group study because she was hesitant to undergo colorectal cancer screening. Her husband sent her the article causing her to wonder whether the study concluded that a colonoscopy was ineffective.

I told her that the study had the GI community in an uproar, but not because it proved colonoscopies ineffective, but because of its misguided framings concerning its research. The recent New England Journal of Medicine RCT, randomized 85,000 individuals to receive an invitation for a colonoscopy or no screening. Many news outlets have manipulated the results for clickbait, purporting that the study showed no reduction in cancer death and only an 18% reduction in colorectal cancer for patients who were randomized to get screened (colonoscopy).

One of the most valuable skills I have developed as a researcher (including serving as a PI on RCTs) is interpreting misguided framings of research studies.

Most headlines reporting on this study failed to mention critical aspects of the study, which are required to interpret the results correctly.

1. Only 42% of patients in the intervention (“invitation”) arm actually underwent colonoscopy. Yet, most headlines report the impact of colonoscopy on the entire intervention arm, including those who never received a colonoscopy (ITT analysis).
2. Consenting patients to enroll in the study occurred after randomization.
3. 30% of endoscopists in this trial had technical skill ratings for detecting polyps below the recommended minimum threshold.

Many of my colleagues have debated the pros/cons of using an intention-to-treat (ITT) vs. per-protocol (PP) analysis to assess a cancer screening intervention. An ITT analysis is essential as it considers an individual’s willingness/ability to undergo the procedure. Colonoscopy is probably the most burdensome and invasive screening tool due to the cost, bowel prep, missed work, the necessity for a 2nd person to drive you home, etc. When evaluating the impact of a screening tool, we must consider whether people will actually go through with the test. This study used a weak invitation process (3 mailed letters) which is inadequate for informed patient decision-making. A better approach would have been to discuss the benefits and risks of the colonoscopy with the patient at a PCP visit. For example, in this study, patients who actually underwent a colonoscopy, there was a 50% reduction in cancer death and a 31% reduction in colorectal cancer (per-protocol analysis).

The endoscopists in this study had lower adenoma detection rates (ADR) – almost a third of the endoscopists had an ADR of less than 25% (the minimum requirement). The average ADR amongst US endoscopists is above 40%. My colleague, Harry Thomas, noted that every 1% increase in the ADR is associated with a 5% decrease in colorectal cancer death (his excellent post on this topic here). While we know the endoscopists in this study had lower ADRs, what about other widely accepted best practices such as split-dose prep / prep-quality, cecal retroflexion, or water infusion?

In addition to these critical aspects, there are many other nuances of this study design and population to consider. One must understand that colorectal cancers grow slowly and may not be impacted in a 10-year analysis. Impact on cancer incidence and mortality may take decades to appear – it will be interesting to see the NordICC 15-year data. Enrolled patients with symptoms or at higher risk for colorectal cancer are likely over-represented in the colonoscopy group because they would be more likely to follow through with the invitation to undergo a colonoscopy.

Studies have shown that guaiac fecal occult blood test (FOBT), sigmoidoscopy, and colonoscopy  all have reduced colorectal cancer mortality. The NEJM isn’t the first study to question the benefit of screening colonoscopy, however the study has been misinterpreted by patients like the lady who questioned me earlier. By undermining the significant progress that has been made in advocating for CRC screening.

More research is always necessary. Colonoscopies can have risks, costs, and significant patient barriers. There are many stakeholders involved with financial incentives and conflicts of interest. These are incredibly complex topics that require a trusted clinician-patient relationship. My patient from this morning appreciated going through the trial results and will see me later this year for her screening colonoscopy.