New England Journal of Medicine October 27, 2022

Effect of Colonoscopy Screening on Risks of Colorectal Cancer and Related Death (NordICC Trial)

Michael Bretthauer, Magnus Løberg, Paulina Wieszczy, Mette Kalager, Louise Emilsson, Kjetil Garborg, Maciej Rupinski, Evelien Dekker, Manon Spaander, Michal Bugajski, et al. (NordICC Study Group)

Bottom Line

In a landmark pragmatic trial, an invitation to a single screening colonoscopy modestly reduced the 10-year risk of colorectal cancer but did not significantly reduce colorectal cancer–related mortality, largely due to low participant uptake.

Key Findings

1. In the intention-to-screen analysis, the 10-year risk of colorectal cancer was 0.98% (259 cases) in the invited group versus 1.20% (622 cases) in the usual-care group, representing an 18% relative risk reduction (RR 0.82; 95% CI, 0.70 to 0.93).
2. The risk of colorectal cancer–related death was not significantly reduced in the intention-to-screen analysis: 0.28% (72 deaths) in the invited group versus 0.31% (157 deaths) in the usual-care group (RR 0.90; 95% CI, 0.64 to 1.16).
3. All-cause mortality was similar between the two groups (11.03% vs. 11.04%; RR 0.99; 95% CI, 0.96 to 1.04).
4. Screening uptake was low, with only 42.0% (11,843 of 28,220) of the invited participants actually undergoing the colonoscopy.
5. In the adjusted per-protocol analysis (estimating the effect if all invited participants had completed screening), the risk of colorectal cancer would have been reduced by 31% (RR 0.69; 95% CI, 0.55 to 0.83) and colorectal cancer mortality by 50% (RR 0.50; 95% CI, 0.27 to 0.77).
6. The number needed to invite to screening to prevent one colorectal cancer case over 10 years was 455 (95% CI, 270 to 1429).
7. Colonoscopy was safe; 15 participants experienced major bleeding after polyp removal, with no perforations or screening-related deaths occurring within 30 days.

Study Design

Design
Pragmatic RCT
Open-Label
Sample
84,585
Patients
Duration
10 years
Median
Setting
Poland, Norway, Sweden
Population Presumptively healthy men and women 55 to 64 years of age drawn from population registries, with no prior history of colorectal cancer screening.
Intervention An invitation to undergo a single screening colonoscopy.
Comparator Usual care (no invitation or screening).
Outcome 10-year risk of colorectal cancer and related death.

Study Limitations

Suboptimal participant adherence (only a 42% uptake rate), which severely diluted the intention-to-screen effectiveness estimates compared to true physiological efficacy.
The 10-year follow-up period may be too short to observe the full mortality benefit of colonoscopy, as it takes many years for removed precancerous polyps to translate into averted cancer deaths.
Endoscopist quality varied, with nearly 30% of endoscopists in the trial having adenoma detection rates (ADR) below the globally recommended benchmark of 25%.
The study evaluated only a one-time screening colonoscopy invitation without a continuous, systemic surveillance program, which differs from real-world clinical practice guidelines.

Clinical Significance

The NordICC trial demonstrated that a population-based invitation to screening colonoscopy yielded lower-than-expected reductions in colorectal cancer incidence and mortality. While initially alarming to the medical community, the study highlights a critical public health reality: a screening program only works if patients actually participate. The per-protocol analysis strongly reaffirmed that completing a colonoscopy halves the risk of dying from colorectal cancer. Ultimately, these results emphasize the necessity of improving patient adherence and ensuring high-quality endoscopic performance to maximize screening benefits.

Historical Context

Prior to NordICC, evidence supporting colonoscopy as the premier colorectal cancer screening modality was derived primarily from large observational cohort studies (which are prone to healthy user bias) and randomized trials of flexible sigmoidoscopy. These older studies suggested colonoscopy could lower mortality by up to 68-90%. Because of these estimates, colonoscopy became the predominant screening method in countries like the United States without direct randomized controlled trial (RCT) evidence comparing it to no screening. NordICC was the first large-scale pragmatic RCT to test colonoscopy screening directly, and its modest intention-to-screen results sparked global debate about how to accurately measure, implement, and communicate the benefits of population-level screening.

Guided Discussion

High-yield insights from every perspective

Med Student
Medical Student

Colonoscopy screening is fundamentally based on interrupting a specific pathophysiologic sequence in the colon. What is this sequence, what are the typical genetic mutations involved, and how does the NordICC trial's finding of a reduced cancer incidence reinforce the mechanism of this primary prevention tool?

Key Response

This question connects the clinical trial's population-level findings to the basic science concept of the adenoma-carcinoma sequence. Students must recall that CRC typically develops over 10-15 years through mutations in APC, KRAS, and TP53. The NordICC trial's demonstration of reduced CRC risk directly reflects the physical interruption of this sequence via polypectomy, reinforcing the pathophysiological basis for endoscopic screening.

Resident
Resident

The NordICC trial showed a non-significant mortality benefit in the intention-to-treat analysis, but an estimated 50% reduction in CRC mortality among those who actually completed the colonoscopy. How should you use this distinction between intention-to-treat and per-protocol data when counseling a reluctant 50-year-old patient who cites this study as a reason to avoid screening?

Key Response

Residents must master the difference between a study's 'effectiveness' (the impact of a screening invitation program on a population, driven by the low 42% uptake) and its 'efficacy' (the biological benefit of the procedure itself). This question tests the resident's ability to translate complex clinical trial statistics into clear, persuasive, and evidence-based shared decision-making with a patient.

Fellow
Fellow

Given the NordICC trial's low (42%) uptake for screening colonoscopy and the modest intention-to-treat benefit, how might these findings influence the decision to design programmatic, population-level screening workflows utilizing non-invasive modalities like Fecal Immunochemical Testing (FIT) compared to direct-to-colonoscopy screening?

Key Response

Gastroenterology and oncology fellows must think systematically about screening programs. A test with high efficacy but low adherence (colonoscopy) may perform worse at a population level than a test with moderate efficacy but high adherence (FIT). This question forces fellows to weigh healthcare economics, patient compliance, and the utilization of endoscopic resources.

Attending
Attending

As an attending teaching evidence-based medicine, how do you reconcile the public health interpretation of the NordICC trial—that population-wide colonoscopy invitations yield marginal mortality benefit—with the individual-level imperative to recommend colonoscopy as a gold-standard prevention tool, and how does this reframe our understanding of 'pragmatic' trial designs?

Key Response

Attendings frequently navigate the tension between public health policy and individual patient care. This question challenges them to synthesize trial data conceptually, using NordICC to teach junior doctors that pragmatic trials measure real-world policy implementation (where non-adherence dilutes effect size), which does not negate the profound benefit for the individual patient who chooses to adhere.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD
PhD

To estimate the per-protocol effect of actually undergoing a colonoscopy, the NordICC investigators utilized an instrumental variable analysis to adjust for systematic differences between adherents and non-adherents. What are the core assumptions (e.g., the exclusion restriction) required for this instrumental variable approach, and how might residual 'healthy adherer bias' still threaten the validity of their adjusted mortality estimates?

Key Response

PhD-level researchers must critically evaluate advanced statistical methodologies. This question targets the use of instrumental variables (using the randomized invitation as the instrument) to estimate the complier average causal effect (CACE). It requires deep knowledge of causal inference, unmeasured confounding, and whether the instrument perfectly isolates the effect of the intervention.

Journal Editor
Journal Editor

As a peer reviewer, how would you critically appraise the NordICC trial's 10-year follow-up period and the historical timeline of its interventions (enrolled 2009-2014) as potential threats to construct validity, specifically regarding the natural history of slow-growing colorectal cancers and modern improvements in Adenoma Detection Rates (ADR)?

Key Response

Editors look for methodological limitations that might lead to misinterpretation of results. Given that CRC takes 10-15 years to develop and cause death, a 10-year follow-up is arguably too short to capture the true mortality benefit of removing precancerous polyps. Furthermore, modern endoscopists have significantly higher ADRs than those in 2009, meaning the trial's intervention might underrepresent the efficacy of contemporary colonoscopies.

Guideline Committee
Guideline Committee

Current USPSTF and ACG guidelines strongly recommend CRC screening, often positioning colonoscopy as a premier option. Does the NordICC intention-to-treat data warrant a downgrade in the strength of recommendation for colonoscopy, or does the trial merely highlight a context-specific implementation gap in European populations where primary screening colonoscopy is not the cultural norm, unlike the US?

Key Response

Guideline committees must determine if new trial data invalidates existing recommendations. This question asks whether the lack of ITT mortality benefit in NordICC represents a failure of colonoscopy itself or a failure of the intervention model (a mailed invitation in countries lacking a culture of screening colonoscopy). It requires assessing external validity and deciding if US guidelines, which operate in a different healthcare culture, should remain unchanged.

Clinical Landscape

Noteworthy Related Trials

2010

UK Flexible Sigmoidoscopy Screening Trial

n = 170,432 · Lancet

Tested

Once-only flexible sigmoidoscopy

Population

Adults aged 55-64 years

Comparator

No screening (usual care)

Endpoint

Colorectal cancer incidence and mortality

Key result: A single flexible sigmoidoscopy screening reduced colorectal cancer incidence by 33% and mortality by 43% in the per-protocol analysis.
2012

PLCO Cancer Screening Trial

n = 154,900 · NEJM

Tested

Flexible sigmoidoscopy at baseline and at 3 or 5 years

Population

Adults aged 55-74 years

Comparator

Usual care

Endpoint

Colorectal cancer mortality

Key result: Screening with flexible sigmoidoscopy significantly reduced colorectal cancer incidence by 21% and mortality by 26% compared to usual care.
2012

COLONPREV Trial

n = 53,220 · NEJM

Tested

One-time colonoscopy

Population

Asymptomatic adults aged 50-69 years

Comparator

Biennial fecal immunochemical testing (FIT)

Endpoint

Colorectal cancer mortality (long-term) and detection rates (interim)

Key result: Colonoscopy and FIT had similar CRC detection rates, but colonoscopy found more advanced adenomas while FIT had significantly higher patient participation.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis