Benefit of Screening Colonoscopy Underestimated in NordICC

TOPLINE:

Delayed reporting of cancer diagnosis by registries probably led to the weaker-than-expected effects of screening colonoscopy in preventing colorectal cancer (CRC) observed in the NordICC trial, a new analysis found.

METHODOLOGY:

• The NordICC trial randomly assigned 85,179 adults aged 55-64 years in a 1:2 ratio to receive or not receive an invitation for a single screening colonoscopy and determined the risk for CRC diagnosis and death over 10-15 years of follow-up using cancer registries. After randomization, the trial excluded 221 adults who had CRC at baseline but who did not yet appear in a cancer registry at the time of randomization.
• The trial found that CRC risk and associated mortality were lower in adults who had colonoscopy, though only modestly so, which generated considerable controversy.
• Because registration delays are a known concern with population-based cancer registries but the trial did not account for them, researchers on the current study postulated that delays might have led to an underestimation of the impact of colonoscopy on CRC risk.
• They estimated the magnitude of delayed reporting of CRC diagnosis to cancer registries by comparing the 221 exclusions with expected CRC diagnoses per year. They explored the impact that delays may have had on the results of the trial’s intention-to-screen analysis and adjusted per-protocol analysis.

TAKEAWAY:

• The trial’s post hoc exclusion of 221 adults who had CRC at baseline but who did not yet appear in a cancer registry at the time of randomization suggests delays of 2-3 years in registration.
• With no assumed delay in cancer registration, the 10-year reported CRC risk difference was 0.22% in intention-to-screen and 0.38% in adjusted per-protocol analyses. With a mean delay in cancer registration of 2 years, the risk difference rose to 0.44% in the intention-to-screen analysis and 0.76% in the adjusted per-protocol analysis.
• Assuming no delay in cancer registration, the number needed to invite for screening colonoscopy and number needed to undergo the procedure to prevent one CRC diagnosis/death were 455 and 263, respectively. These numbers decreased to 227 and 132, respectively, with a 2-year reporting delay.
• Registration delays of 1, 2 or 3 years led to an underestimated risk for CRC by 25%, 50%, and 75%, respectively.

IN PRACTICE:

“Updated analyses ensuring complete 10- and 15-year follow-up will be crucial to derive the true reductions of CRC risk and mortality in the trial’s predefined interim and primary analysis. In the meantime, available estimates are to be interpreted with caution, as they likely severely underestimate true screening colonoscopy effects,” the authors concluded.

The lag in reporting found by the study raises questions about the time interval needed beyond the end of a study to assure its completeness, which varies across registries, wrote Chyke A. Doubeni, MD, MPH, The Ohio State University Wexner Medical Center, Columbus, and colleagues in an invited commentary. “Publication guidelines should be strengthened to ensure affirmation of completeness and quality of cancer registries used for outcomes ascertainment to minimize uncertainties,” they added.

SOURCE:

The study, with first author Hermann Brenner, MD, MPH, German Cancer Research Center, Heidelberg, was published online in JAMA Network Open, as was the invited commentary.

LIMITATIONS:

Exact quantification of and correction for registration delays were not possible.

DISCLOSURES:

The study was partially funded by the German Federal Ministry of Education and Research and German Cancer Aid. The authors reported no conflicts of interest.