Long-term effects of colonoscopy screening on colorectal cancer incidence and mortality: a multicountry, population-based randomised controlled trial.

# Long-term effects of colonoscopy screening on colorectal cancer incidence and mortality: a multicountry, population-based randomised controlled trial *The Lancet 2026* Articles Long-term effects of colonoscopy screening on colorectal cancer incidence and mortality: a multicountry, population- based randomised controlled trial Michal F Kaminski, Mette Kalager, Magnus Løberg, Louise Emilsson, Anna Macios, Faye Samy, Joy Shi, Shona Fielding, Miguel A Hernán, Kjetil Garborg, Maciej Rupinski, Evelien Dekker, Manon Spaander, Øyvind Holme, Ann G Zauber, Nastazja D Pilonis, Joanna Didkowska, Piotr Spychalski, Geir Hoff, Jaroslaw Regula, Hans-Olov Adami, Michael Bretthauer, for the NordICC Study Group* Summary Background We previously reported the 10-year eects of colonoscopy screening on colorectal cancer incidence and Published Online mortality. Here, we report the eects after 13 years of follow-up. May 5, 2026 https://doi.org/10.1016/ S0140-6736(26)00508-8 Methods In this multicountry, population-based randomised controlled trial, 84 583 men and women aged See Online/Comment 55–64 years at enrolment from Norway, Poland, and Sweden were randomly allocated (1:2) to colonoscopy screening https://doi.org/10.1016/ or no screening and analysed. The primary outcomes were colorectal cancer incidence and mortality after 10–15 years S0140-6736(26)00794-4 of follow-up in intention-to-screen analyses, with first analysis after 10 years, and repeated every other year or at *Members listed in the appendix longer intervals. This trial is registered with ClinicalTrials.gov, NCT00883792, and is ongoing. (p 3) Department of Oncological Findings At 13 years of follow-up, colorectal cancer incidence was 375 colorectal cancers (1·46%) of 28 217 individuals Gastroenterology, Maria Skłodowska-Curie National in the screening group and 912 colorectal cancers (1·80%) of 56 366 individuals in the no-screening group. The risk Research Institute of Oncology, ratio (RR) was 0·81 (95% CI 0·71–0·90) in intention-to-screen analyses and 0·55 (0·33–0·81) in per-protocol Warsaw, Poland analyses. The risk for proximal colorectal cancer was 129 (0·51%) in the screening group versus 283 (0·56%) in the (Prof M F Kaminski MD PhD, no-screening group (RR 0·91 [0·71–1·09]), and the risk for distal colorectal cancer was 224 (0·87%) in the screening M Rupinski MD, Prof N D Pilonis MD PhD, group versus 563 (1·11%) in the no-screening group (RR 0·79 [0·65–0·89]; interaction p<0·0001). In men, the Prof J Regula MD PhD); Centre of colorectal cancer risk was 214 (1·69%) of 14 154 in the screening group and 541 (2·19%) of 28 247 in the no-screening Postgraduate Medical group (RR 0·77 [0·64 to –0·88]); in women, the risk was 161 (1·24%) of 14 063 in the screening group Education, Warsaw, Poland versus 371 (1·43%) of 28 119 in the no-screening group (RR 0·87 [0·70 to 1·02]; interaction p<0·0001). Colorectal (Prof M F Kaminski, M Rupinski, Prof N D Pilonis, Prof J Regula); cancer mortality was 106 (0·41%) of 28 217 in the screening group and 236 (0·47%) of 56 366 in the no-screening Clinical Effectiveness Research group (intention-to-screen RR 0·88 [0·68–1·08], per-protocol RR 0·70 [0·26–1·25]). The observed colorectal cancer Group, University of Oslo, Oslo, mortality in the non-screening group (0·47%) was substantially lower than expected at the time of designing the trial Norway (0·82%). (Prof M Kalager MD PhD, Prof M Løberg MD PhD, A Macios PhD, Interpretation One colonoscopy significantly reduced colorectal cancer incidence but not mortality over 13 years. K Garborg MD PhD, Colorectal cancer mortality was lower in both study groups than when the trial was designed. Prof Ø Holme MD PhD, Prof H-O Adami MD PhD, Prof M Bretthauer MD PhD); Funding The Norwegian Research Council, the Nordic Cancer Union, the Norwegian Cancer Society, and the Health Department of Transplantation Fund of South-East Norway. Medicine, Oslo University Hospital, Oslo, Norway Copyright © 2026 Elsevier Ltd. All rights reserved, including those for text and data mining, AI training, and similar (Prof M Kalager, Prof M Løberg, A Macios, K Garborg, technologies. Prof Ø Holme, Prof H-O Adami, Prof M Bretthauer); Department Introduction In 2022, we reported 10-year results from NordICC, a of General Practice, Institute of With almost 2 million new cases every year, colorectal randomised trial on colonoscopy screening.6 In intention- Health and Society, University of Oslo, Oslo, Norway cancer is the third most common cancer worldwide.1 To to-screen analyses, the risk of colorectal cancer at 10 years (L EmilssonM D PhD); decrease the burden of colorectal cancer, many countries was 0·98% in the colonoscopy screening group and Vårdcentralen Värmlands have introduced screening programmes for the general 1·20% in the no-screening group, a risk reduction Nysäter and Centre for Clinical Research, County Council of population.2 A variety of screening tests for colorectal of 18%. Colorectal cancer mortality was, however, not Värmland, Värmland, Sweden cancer are available, with colonoscopy being the most significantly reduced in the colonoscopy screening (L Emilsson); Department of frequently used in the USA and in some countries in group. Adjusted per-protocol analyses suggested a Medical Epidemiology and Europe.3 Colonoscopy screening was, however, introduced reduction in colorectal cancer incidence of 30%, and Biostatistics, Karolinska Institutet, Solna, Sweden without high-quality evidence from randomised trials, of up to 50% for colorectal cancer mortality.6 A (L Emilsson, Prof H-O Adami); when observational studies and disease modelling 2025 randomised trial in Spain reported that colonoscopy Frontier Science Scotland, indicated that colonoscopy might reduce colorectal cancer screening yields similar benefits on colorectal cancer Kingussie, UK (F Samy BSc, incidence and mortality by at least 50% and extend the incidence and mortality as screening with faecal S Fielding PhD); CAUSALab, Department of Epidemiology, length of life.4,5 immunochemical testing, but that trial had no www.thelancet.com Published online May 5, 2026 https://doi.org/10.1016/S0140-6736(26)00508-8 1 Articles Harvard T H Chan School of Research in context Public Health, Boston, MA, USA (J Shi PhD, Evidence before this study follow-up, the analyses did not have enough statistical power to Prof M A Hernán MD PhD); Mongan Institute, Department We searched PubMed from database inception to Dec 31, 2025, analyse important, predefined subgroups, such as women and of Medicine, Massachusetts for randomised controlled trials published in English that men. General Hospital, Boston, MA, assessed the effect of colonoscopy for screening of colorectal USA (J Shi); Department of Added value of this study cancer in individuals with average risk (ie, people aged ≥50 years Gastroenterology and We report results with longer follow-up after colonoscopy Hepatology, Amsterdam without personal history of colorectal cancer or hereditary screening. Due to more colorectal cancer cases and deaths, University Medical Centers, colorectal cancer syndrome) on colorectal cancer incidence or Amsterdam, Netherlands the present report can also provide estimates for men versus mortality. We used the following search terms: (“colorectal (Prof E Dekker MD PhD); women, proximal versus distal colorectal cancers, and people cancer” OR “colon cancer”) AND (“colonoscopy”) AND University of Amsterdam, younger than 60 years versus 60 years and older. Finally, the Amsterdam, Netherlands (“screening”) AND (“mortality” OR “incidence”). Two randomised report uses methodologically improved per-protocol analyses (Prof E Dekker); Department of trials were identified: the SCREESCO trial in Sweden, which Gastroenterology and to estimate screening benefits for individuals who were actually compares colonoscopy, faecal immunochemical testing, and no Hepatology, Erasmus screened, which are important for personal shared decision University Medical Centre, intervention; and our own trial (the NordICC trial). The making. Rotterdam, Netherlands SCREESCO trial has not reported results on colorectal cancer (Prof M Spaander MD PhD); mortality or incidence yet. Our own trial reported results after Implications of all the available evidence Research Unit, Sørlandet 10 years of follow-up in 2022, which showed a risk of colorectal According to these new results, a single colonoscopy reduces Hospital, Kristiansand, Norway (Prof Ø Holme); Department of cancer at 10 years of 0·98% in the screening group and 1·20% in colorectal cancer risk by 0·3% to 0·8% over 13 years. The study Epidemiology and the no-screening group (risk ratio [RR] 0·82 [95% CI 0·70–0·93]) reveals that the risk of colorectal cancer death is significantly Biostatistics, Memorial Sloan in intention-to-screen analyses. Colorectal cancer mortality was lower than was expected when the trial started even without Kettering Cancer Center, 0·28% in the screening group and 0·31% in the no-screening screening, and that screening does not provide additional New York, NY, USA (A G Zauber PhD); Department group (RR 0·90 [0·64–1·16]). The risk of death from any cause reduction of an already low risk for colorectal cancer mortality of Epidemiology & National was 11·03% in the screening group and 11·04% in the no- with longer follow-up than 10 years. These data might help to Cancer Registry, Maria screening group (RR 0·99 [0·96–1·04]). Some screening experts guide policies at the population level, where screening benefits Skłodowska-Curie National have questioned whether 10 years of follow-up was too short to must be weighed against other expenditures that promote Research Institute of Oncology, Warsaw, Poland observe the full benefit of colonoscopy screening. After 10-year public health. (Prof J Didkowska MD); Department of Surgical Oncology, Transplant Surgery comparison group of individuals who were not oered information on the sex of each individual in the and General Surgery, Medical screening.7 population registries. In accordance with data protection University of Gdansk, Gdansk, Poland (P Spychalski MD PhD); It has been claimed that 10 years of follow-up is too laws in the participating countries, we did not receive Department of Research and short to fully ascertain screening benefits.8 In this study, information about race or ethnicity from the national Development, Telemark we report the eects of colonoscopy screening after registries, and we did not request this information from Hospital, Skien, Norway 3 years of further follow-up in intention-to-screen and participants who underwent screening. Public statistics (Prof G Hoff MD PhD); Cancer Registry of Norway, Norwegian per-protocol analyses.9 The current analysis is a priori as from the participating countries do not routinely report Institute of Public Health, Oslo, defined in the protocol (appendix pp 31–54). Due to race or ethnicity but provide country of birth. According Norway (Prof G Hoff) substantially more colorectal cancer cases and deaths to these statistics, more than 90% of individuals in the Correspondence to: during this additional follow-up, we now also provide trial recruitment areas were born in the country where Prof Michael Bretthauer, Clinical estimates for men and women, proximal and distal they live. Effectiveness Research Group, University of Oslo, N-0318 Oslo, colorectal cancers, and people aged 60 years and younger As before, this analysis is based on follow-up of all Norway versus older than 60 years. 84 583 individuals in Norway, Poland, and Sweden michael.bretthauer@medisin. (89·1% of all individuals originally included in the trial; uio.no Methods appendix p 18). Data from the Netherlands were not See Online for appendix Study design and participants included due to continued restrictions in availability of The design of the NordICC trial has been described in trial endpoints. detail elsewhere.6,10 Briefly, NordICC is a population- The study was approved by the ethical committees at all based, randomised trial comparing colonoscopy participating centres (Norway: Regional Ethics screening with no screening. Eligible individuals were Committee of South-East Norway [2010/2187]; Poland: men and women aged 55–64 years living in Norway, Maria Skłodowska Curie Memorial Cancer Center and Poland, Sweden, and the Netherlands. Individuals were Institute of Oncology and Medical Center for drawn directly from the population registries and Postgraduate Education, Warsaw, Poland [30/2007]; randomly allocated (1:2) to either invitation for Sweden: Swedish National Ethics Council [2007/363]; colonoscopy screening (screening group) or no invitation and the Netherlands: Dutch National Health Council (no-screening group). Screening was done between [2009/03WBO]).10 The study was funded by research June 8, 2009, and June 23, 2014.6 Sex of trial participants grants in the participating countries. All authors vouch was determined by the national identifier, which includes for the accuracy of the data and fidelity of the trial to the 2 www.thelancet.com Published online May 5, 2026 https://doi.org/10.1016/S0140-6736(26)00508-8 Articles protocol. This trial is registered with ClinicalTrials.gov, classified as early stage (Dukes A or B), late stage NCT00883792, and is ongoing. (Dukes C or D), or unknown. Tumours with histopathology other than adenocarcinoma were not Procedures counted as events. Proximal colorectal cancer was Individuals were randomly allocated to either invitation defined as cancers proximal for the descending colon, to one screening colonoscopy (screening group) or to no and distal colorectal cancer as cancers in the descending invitation to any screening (no-screening group).10 colon, sigmoid colon, or rectum. Colorectal cancer Individuals randomly allocated to the screening group mortality was defined as colorectal cancer as cause of received a letter of invitation to screening colonoscopy. death registered in the cause-of-death registries in the All individuals receiving the screening provided written participating countries, and a diagnosis of colorectal informed consent before the colonoscopy. Individuals cancer in the national cancer registries. We followed all randomly allocated to the no-screening group did not randomly allocated individuals by linkage to national receive any intervention and were not contacted at study cancer and cause-of-death registries, as previously enrolment. Requirement of consent was waived for described.6 The sample size calculation was based on individuals randomly allocated to the no-screening group intention-to-screen analyses and has been described in by the ethical committees at all participating centres.10 detail elsewhere.10 A colonoscopy quality and training programme was implemented for the trial.10 All lesions detected during Statistical analysis colonoscopy were removed if feasible, and all tumours Follow-up was from randomisation until date of were biopsied. Dedicated histopathologists assessed all emigration, colorectal cancer diagnosis for colorectal polyps and cancers according to WHO classification.11 cancer incidence or colorectal cancer death for colorectal Participants were referred for polyp surveillance after cancer mortality, or death or end of follow-up after screening following national guidelines.6 13 years, whichever came first. Poland had an opportunistic colorectal cancer To estimate the intention-to-screen eect, we calculated screening programme in some geographical areas at trial cumulative incidence as the 13-year risk of colorectal start, but not in the area of the trial. In the other countries, cancer in the screening and no-screening groups using no organised colorectal cancer screening of any kind was Kaplan–Meier estimators, and colorectal cancer mortality available at trial start. Colorectal cancer screening equivalently; we compared risks using risk ratios (RRs) programmes using faecal immunochemical testing were and risk dierences, and annual colorectal cancer gradually introduced in the participating countries incidence and mortality RRs. We did analyses with and region by region during follow-up. Such staggered without censoring of competing events (death from geographical implementation of cancer screening causes other than colorectal cancer). As in our previous programmes is common in Scandinavia.12 To protect the report,6 the results were similar between these analyses integrity of the trial, the geographical areas in which the (appendix p 4), and thus we focused on analyses that NordICC trial was held were not the first to start the new treated non-colorectal cancer deaths as censoring events. screening programmes and, thus, NordICC trial 95% CIs were calculated via bootstrapping. We calculated participants had become too old to be eligible for the the number needed to invite to screening to prevent screening programmes when these programmes were one colorectal cancer as the reciprocal of the 13-year risk introduced in their geographical area. Colonoscopy dierence of colorectal cancer incidence. screening was not available publicly or privately outside To estimate the per-protocol eect, defined as the eect the trial.6,10 Thus, no individuals enrolled in the trial were of screening if all individuals randomly allocated to eligible for any colorectal cancer screening programmes the screening group had been screened, we used an outside the trial during screening or follow-up. instrumental variable approach, which does not require measurement of all confounders to provide valid Outcomes results.9 We present estimates of per-protocol eects The primary outcomes were colorectal cancer incidence based on instrumental variable analyses under additive and mortality after 10–15 years of follow-up in intention- homogeneity—that is, factors that modify the eect of to-screen analyses, with first analysis after 10 years, and screening on the outcome (when estimating risk repeated every other year or at longer intervals (appendix dierences) do not also modify the eect of receiving an pp 31–54). Secondary outcomes were colorectal cancer invitation on undergoing screening. The appendix incidence and mortality in per-protocol analyses, all-cause (pp 7–10) also provides estimates using alternative mortality, subgroup analyses by age and gender, and assumptions: multiplicative homogeneity (analogous to analyses for distal and proximal colorectal cancer. additive homogeneity, but for RRs rather than risk Colorectal cancer diagnosis was defined according to dierences) and monotonicity (ie, that there are no ICD-10 as cancer in the colon or rectum by topography individuals who would undergo screening if not invited codes C18 to C20, combined with ICD-O3 morphology but would choose not to undergo screening if invited). codes for adenocarcinoma. Colorectal cancer stage was Under a homogeneity assumption (additive or www.thelancet.com Published online May 5, 2026 https://doi.org/10.1016/S0140-6736(26)00508-8 3 Articles multi plicative), the per-protocol estimate was interpreted complications of the trial interventions have been as the eect in the entire trial population, whereas under reported previously.13 monotonicity, the estimate was interpreted as the causal 3 years of additional follow-up (from 10 years to eect among those who were screened. 13 years) increased the number of colorectal cancer cases from 259 to 375 (by 45%) in the screening group and Role of the funding source from 622 to 912 (by 47%) in the control group (table 2). The funders of the study had no role in study design, Death from colorectal cancer increased from 72 to 106 data collection, data analysis, data interpretation, writing (by 47%) cases in the screening group, and from 157 to 236 of the report, or the decision to submit for publication. (by 50%) cases in the no-screening group (table 2). The 13-year risk of colorectal cancer was 375 (1·46%) of Results 28 217 in the screening group and 912 (1·80%) of 56 366 in Our analyses were based on 84 583 individuals from the no-screening group in intention-to-screen analyses Norway, Poland and Sweden, of whom 28 217 were (RR 0·81 [95% CI 0·71–0·90]; figure 1A, table 2). randomly allocated to screening and 56 366 to no Figure 1A shows colorectal cancer incidence by group screening. Participant characteristics of age, sex, and and figure 2 provides annual colorectal cancer incidence participation with screening colonoscopy in the rate ratios. The number needed to invite for screening to three countries are provided in table 1. Overall, prevent one colorectal cancer within 13 years was 294 11 841 (42·0%) of 28 217 participants invited to screening (95% CI 185–588). In per-protocol analyses, colorectal attended, varying from 6002 (33·0%) of 18 183 in Poland cancer incidence was 1·00% in the screening group and to 485 (39·8%) of 1219 in Sweden and 5354 (60·7%) of 1·80% in the no-screening group (RR 0·55 [0·33–0·81]; 8815 in Norway.13 The median follow-up for this report table 2, figure 1B). Colorectal cancer incidence in was 13·0 years in both study groups (IQR 12·9–13·0, screening participants and non-participants in the maximum 13·0). Data on adverse events and screening group compared with the no-screening group is shown in the appendix (p 20). The 13-year risk of colorectal cancer mortality was Poland Norway Sweden 106 (0·41%) of 28 217 in individuals randomly allocated to (n=54 527) (n=26 413) (n=3643) screening, and 236 (0·47%) of 56 366 in those randomly Group allocated to no screening (RR 0·88 [95% CI 0·68–1·08]; Attender 6002 (11·0%) 5354 (20·3%) 485 (13·3%) table 2, figure 1C). The appendix (p 19) shows yearly Non-attender 12 181 (22·3%) 3461 (13·1%) 734 (20·1%) colorectal cancer mortality rate ratios by group. In per- Control 36 344 (66·7%) 17 598 (66·6%) 2424 (66·5%) protocol analyses, colorectal cancer mortality was 0·33% in Sex the screening group and 0·47% in the no-screening group Female 27 329 (50·1%) 13 194 (50·0%) 1659 (45·5%) (RR 0·70 [0·26–1·25]; table 2 figure 1D). Colorectal cancer Male 27 198 (49·9%) 13 219 (50·0%) 1984 (54·5%) mortality in screening participants and non-participants in Age at enrolment the screening group compared with the no-screening group is shown in the appendix (p 21). <60 years 28 791 (52·8%) 12 524 (47·4%) 1782 (48·9%) During 13 years of follow-up, 4506 (16·30%) individuals ≥60 years 25 736 (47·2%) 13 889 (52·6%) 1861 (51·1%) died of any cause in the invitation-to-screening group, Data are n (%). Discrepancies in numbers of analysed individuals between this compared with 9017 (16·34%) in the no-screening group analysis and the previously published 10-year follow-up study can be found in the appendix (p 28). (RR 1·00 [95% CI 0·97–1·03]; table 2). In intention-to-screen analyses, the risk for proximal Table 1: Characteristics of trial participants colorectal cancer was 129 (0·51%) of 28 217 in the screening Screening group (n=28 217) No-screening group (n=56 366) Risk difference, % Risk ratio (95% CI) (95% CI) Number 13-year risk, % (95% CI) Number 13-year risk, % (95% CI) of events of events Intention-to-screen analysis Colorectal cancer 375 1·46% (1·32–1·62) 912 1·80% (1·69–1·92) –0·34% (–0·54 to –0·17) 0·81 (0·71–0·90) Death from colorectal cancer 106 0·41% (0·34–0·50) 236 0·47% (0·42–0·54) –0·06% (–0·16 to 0·04) 0·88 (0·68–1·08) Death from any cause 4506 16·30% (15·87–16·74) 9017 16·34% (16·03–16·65) –0·04% (–0·56 to 0·51) 1·00 (0·97–1·03) Per-protocol analysis Colorectal cancer ·· 1·00% (0·63–1·41) ·· 1·80% (1·69–1·91) –0·80% (–1·24 to –0·33) 0·55 (0·33–0·81) Death from colorectal cancer ·· 0·33% (0·13–0·55) ·· 0·47% (0·42–0·53) –0·14% (–0·37 to 0·11) 0·70 (0·26–1·25) Table 2: Primary and secondary endpoints in intention-to-screen and per-protocol analyses 4 www.thelancet.com Published online May 5, 2026 https://doi.org/10.1016/S0140-6736(26)00508-8 Articles A Intention-to-screen analysis 2·00 1·75 1·50 1·25 1·00 0·75 0·50 0·25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Number at risk (censored) Screening group No-screening group Figure 1: Cumulative colorectal cancer incidence and mortality at 13 years in intention-to-screen and per- protocol analyses Colorectal cancer incidence in intention-to-screen analysis (A) and per-protocol analysis (B). Colorectal cancer death in intention-to-screen analysis (C) and per-protocol analysis (D). Risk curves based on per-protocol estimations were smoothed with restricted cubic splines with knots at 0·5, 1, 3, 7, and 11 years of follow-up. RR=risk ratio. www.thelancet.com Published online May 5, 2026 https://doi.org/10.1016/S0140-6736(26)00508-8 5 latceroloc fo ksir evitalumuC )%( recnac B Per-protocol analysis 2·00 1·75 1·50 1·25 1·00 0·75 0·50 0·25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 latceroloc fo ksir evitalumuC )%( recnac C Intention-to-screen analysis 0·50 0·40 0·30 0·20 0·10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Number at risk (censored) Screening group No-screening group latceroloc fo ksir evitalumuC )%( htaed recnac D Per-protocol analysis 0·50 0·40 0·30 0·20 0·10 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Time since randomisation (years) latceroloc fo ksir evitalumuC )%( htaed recnac Screening group No-screening group RR=0·81 (95% CI 0·71–0·90) 28 217 279 47 27 682 27 392 27 104 26 773 26 445 26 113 25 778 25 351 24 910 22 722 21 811 16 043 (0) (210) (233) (261) (270) (309) (308) (312) (311) (395) (417) (2166) (878) (5729) 56 366 558 74 55 367 54 780 54 179 53 516 52 802 52 050 51 342 50 575 496 88 45 260 43 436 31 910 (0) (434) (455) (536) (552) (594) (640) (676) (637) (691) (800) (4338) (1741) (11 450) RR=0·55 (5% CI 0·33–0·81) RR=0·88 (95% CI 0·68–1·08) 28 217 280 04 27 764 27 500 27 217 26 899 26 581 26 256 25 937 25 522 250 92 23 990 21 993 17 641 (0) (210) (234) (261) (270) (309) (309) (318) (312) (403) (421) (1091) (1986) (4346) 56 366 559 25 55 462 54 917 54 351 53 734 53 072 52 370 51 705 50 978 50 143 47 909 43 935 35 148 (0) (437) (457) (538) (555) (601) (645) (682) (643) (700) (809) (2213) (3943) (8759) RR=0·70 (95% CI 0·26–1·25) Articles [0·62–0·85]). The appendix shows cumulative colorectal 3·5 cancer incidence (p 24) and mortality (p 25) according to age group, a forest plot of colorectal cancer incidence 3·0 in subgroups (p 27), and per-protocol estimates for subgroup analyses (pp 7–10). 2·5 Colorectal cancer stages in screening and no-screening groups based on Duke’s stage are provided in the 2·0 appendix (pp 7–10). 61 participants included in this report were diagnosed 1·5 with colorectal cancer at screening. Of these, 47 were distal cancers and 12 had a proximal location (two had 1·0 an unclassified location). In screening participants, 65 colorectal cancers (27 distal and 32 proximal location, 0·5 and six unclassified location) were diagnosed after screening over the course of the 13-year follow-up, for a 0 risk of cancer after colonoscopy in screening participants 0 1 2 3 4 5 6 7 8 9 10 11 12 13 of 65 (0·55%) of 11 841. Time since randomisation (years) Sensitivity analyses gave per-protocol estimates using Figure 2: Yearly rate ratios for colorectal cancer incidence in the screening alternative assumptions (multiplicative homogeneity and group compared with the no-screening group, intention-to-screen analysis monotonicity; appendix p 11) and per-protocol estimates Shading indicates 95% CIs. The dotted line indicates similar incidence rate ratio in the two groups. with deaths from causes other than colorectal cancer not treated as censoring events (appendix p 12). Country- group and 283 (0·56%) of 56 366 in the no-screening specific analyses for Poland and Norway for colorectal group (RR 0·91 [95% CI 0·71–1·09]), and the risk for distal cancer incidence and mortality in intention-to-screening colorectal cancer was 224 (0·87%) in the screening group and per-protocol analyses are given in the appendix and 563 (1·11%) in the no-screening group (RR 0·79 (pp 16–17). [0·65–0·89], interaction p<0·0001; figure 3A, B, appendix pp 5–6). Discussion In intention-to-screen analyses, the risk for death due After 13 years of follow-up, the relative colorectal cancer to proximal colorectal cancer was 33 (0·13%) of 28 217 in incidence reduction was 19% (compared with 18% at the screening group and 67 (0·13%) of 56 366 in the 10 years) and the absolute risk reduction was no-screening group (RR 0·95 [95% CI 0·56–1·35]), and 0·34 percentage points (compared with 0·20 at 10 years) the risk for death due to distal colorectal cancer was for screening versus no screening in intention-to-screen 68 (0·27%) in the screening group and 154 (0·31%) analyses. The corresponding relative colorectal cancer in the no-screening group (RR 0·87 [0·62–1·12]; mortality reduction was 12% and the absolute risk appendix pp 5–6). Cumulative colorectal cancer reduction 0·06 percentage points, but these estimates mortality according to cancer location is shown in the were not statistically significant. appendix (p 22). Per-protocol estimates for proximal Due to the almost 50% greater number of events in versus distal colorectal cancer are provided in the the extended follow-up, we could more precisely appendix (pp 7–10). establish colonoscopy benefits in the distal versus In men, the 13-year colorectal cancer risk was proximal colon, for people younger than 60 years versus 214 (1·69%) of 14 154 in the screening group and 60 years and older, and for men versus women. Our 541 (2·19%) of 28 247 in the no-screening group in data suggest that colonoscopy reduces colorectal cancer intention-to-screen analyses (RR 0·77 [95% CI incidence in the distal colon by 21% in intention-to- 0·64–0·88]), whereas the corresponding risks were 161 screen analyses (appendix pp 5–6), whereas eect (1·24%) of 14 063 versus 371 (1·43%) of 28 119 in women estimates are smaller and imprecise in the proximal (RR 0·87 [0·70–1·02], interaction p<0·0001; figure 3C, D). colon. As expected, colorectal cancer risk was larger in Cumulative colorectal cancer mortality according to sex men than in women in both study groups. Colonoscopy is provided in the appendix (p 23). could be more eective in men than in women, but the Participants aged 55–59 years at enrolment in the trial dierences are not statistically significant (appendix had a 13-year colorectal cancer risk of 167 (1·26%) of pp 5–6). 14 367 in the screening group in intention-to-screen We did not observe a clinically relevant cancer stage analyses, compared with 362 (1·38%) of 28 730 in the shift (detecting cancers earlier in the screening group) no-screening group (RR 0·91 [95% CI 0·74–1·07]) The between the screening and no-screening groups corresponding rates were 208 (1·68%) of 13 850 and (appendix p 11). This finding is not surprising because 550 (2·25%) of 27 636 in participants aged 60–64 years stage shift of cancer due to screening applies only to at enrolment in intention-to-screen analyses (RR 0·75 cancers detected at screening (or surveillance after 6 www.thelancet.com Published online May 5, 2026 https://doi.org/10.1016/S0140-6736(26)00508-8 oitar etar ecnedicnI Articles B Distal colon 2·25 2·00 1·75 1·50 1·25 1·00 0·75 0·50 0·25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Number at risk (censored) Screening group No-screening group Figure 3: Cumulative colorectal cancer incidence at 13 years in intention-to- screen analysis according to cancer location and sex Proximal colorectal cancer incidence* (A), distal colorectal cancer incidence* (B), colorectal cancer incidence in female participants (C), and colorectal cancer incidence in male participants (D). *Proximal colorectal cancer: location proximal for the descending colon. Distal colorectal cancer: location in the descending colon, sigmoid colon, or rectum. www.thelancet.com Published online May 5, 2026 https://doi.org/10.1016/S0140-6736(26)00508-8 7 latceroloc fo ksir evitalumuC )%( recnac RR=0·79 (95% CI 0·65–0·89) C Female 2·25 2·00 1·75 1·50 1·25 1·00 0·75 0·50 0·25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Number at risk (censored) Screening group No-screening group latceroloc fo ksir evitalumuC )%( recnac RR=0·87 (95% CI 0·70–1·02) D Male 2·25 2·00 1·75 1·50 1·25 1·00 0·75 0·50 0·25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Time since randomisation (years) Number at risk (censored) Screening group No-screening group latceroloc fo ksir evitalumuC )%( recnac A Proximal colon 2·25 2·00 1·75 1·50 1·25 1·00 0·75 0·50 0·25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 Number at risk (censored) Screening group No-screening group RR=0·77 (95% CI 0·64–0·88) latceroloc fo ksir evitalumuC )%( recnac Screening group No-screening group RR 0·91 (95% CI 0·71–1·09) 28 217 27 992 27 749 27 473 27 184 26 861 26 542 26 215 25 888 25 469 250 30 22 830 21 931 16 146 (0) (212) (239) (262) (279) (312) (315) (321) (316) (404) (427) (2193) (890) (5772) 56 366 55 911 55 437 54 876 54 303 53 674 52 996 52 277 51 600 50 860 500 03 45 586 43 775 32 157 (0) (438) (460) (542) (559) (610) (655) (693) (657) (718) (826) (4387) (1784) (11 597) 28 217 27 960 27 699 27 425 27 145 26 819 26 494 26 167 25 839 25 417 249 85 22 795 21 889 16 102 (0) (212) (234) (261) (274) (315) (311) (314) (315) (407) (421) (2178) (884) (5765) 56 366 55 895 55 400 54 832 54 242 53 598 52 901 52 172 51 477 50 724 498 63 45 447 43 633 32 047 (0) (436) (460) (537) (560) (601) (648) (682) (648) (706) (809) (4366) (1763) (11 540) 14 063 13 967 13 878 13 775 13 660 13 531 13 416 13 298 13 165 13 014 128 36 11 843 11 426 8492 (0) (74) (79) (91) (109) (117) (106) (107) (121) (137) (169) (984) (401) (2915) 28 119 27 949 27 764 27 566 27 337 27 107 26 835 26 553 26 291 26 007 256 65 23 630 22 805 16 929 (0) (152) (161) (171) (206) (200) (238) (248) (239) (255) (309) (2002) (792) (5846) 14 154 13 980 13 804 13 617 13 444 13 242 13 029 12 815 12 613 12 337 120 74 10 879 10 385 7551 (0) (136) (154) (170) (161) (192) (202) (205) (190) (258) (248) (1182) (477) (2814) 28 247 27 925 27 603 27 214 26 842 26 409 25 967 25 497 25 051 24 568 24 023 21 630 20 631 14 981 (0) (282) (294) (365) (346) (394) (402) (428) (398) (436) (491) (2336) (949) (5604) Articles screening); at 13-year follow-up, screening-detected in colorectal cancer mortality and, given the very low colorectal cancer only amounted to 60 (16%) of risk, it seems questionable whether such an eect would 375 colorectal cancers in the screening group. Cancers be clinically relevant (appendix p 15). detected later were due to clinical symptoms, as there Quality of colonoscopy performance is associated with was no colonoscopy screening outside of the trial and colorectal cancer risk after screening. The risk of screening was only oered once. colorectal cancer for screening participants at this time Our risk estimates were determined by the causal eect of follow-up is 65 (0·55%) cancers in 11 841 participants. of colonoscopy and the participation rates among This risk corresponds well with the risk in a recent individuals invited to screening. Because overall nationwide analysis of the Polish screening programme, participation with screening was only 42%, the results which reported a 10-year risk of cancer after colonoscopy from our per-protocol analyses are important for clinical of 0·38% for all participants, and 0·32% in participants decision making. Whereas previous per-protocol screened by physicians with an adenoma detection rate estimates might have been less reliable due to restricted of 26% or above.16 With 3 years longer follow-up in our access to data on risk factors that predict participation to trial compared with the Polish report, and thus higher the intervention, the per-protocol eect estimates in this risk, the present interval cancer estimates confirm a high report use instrumental variable methodology, which is quality of colonoscopies performed in the NordICC trial. not dependent on the availability of such data.9 Using Strengths of our trial are its randomised design and this approach and three dierent assumptions regarding large size, its population of screening-naive participants, eect homogeneity and monotonicity, colonoscopy the minimal screening contamination of the control screening reduced colorectal cancer incidence by group, and the complete, long-term follow-up of all 0·8 percentage points (from about 1·8 to 1·0, depending randomly allocated individuals. Limitations include the on assumptions), a 45% relative risk reduction (table 2), participation rate of 42% (which was expected when assuming that all people would participate in screening. planning the trial) and the absence of individual data on However, the per-protocol eect estimates of screening colonoscopies after screening. Dierences in benefit on colorectal cancer mortality are still imprecise. ascertainment of colorectal cancer mortality between the Colonoscopy as applied in our trial and in most study groups are unlikely, firstly because the intervention colorectal cancer screening programmes around the (invitation to colonoscopy) happened several years before world is mainly a preventive screening test; its primary death for most participants (appendix p 21), and secondly target is to detect and remove benign polyps to prevent because we did not involve general practitioners (who them from growing into cancer. Thus, the primary often fill in death certificates for their patients) in the potential benefit of colonoscopy screening is reduction in study procedures. Thus, general practitioners were not colorectal cancer incidence. Because only people who get involved in decision making about trial participation or colorectal cancer can die from colorectal cancer, follow-up, which diminishes ascertainment bias for colonoscopy will also reduce colorectal cancer mortality if cause of death. Finally, we cannot rule out that some it reduces colorectal cancer incidence, provided that there individuals in the control group have been screened for are no changes in colorectal cancer treatment that would colorectal cancer on their own initiative outside any alter prognosis of individuals with colorectal cancer. organised screening, although we believe this number to However, prognosis of colorectal cancer has improved be negligible, because we were able to track background dramatically due to advances in chemotherapy, surgery, colonoscopy activity through national registries in and radiotherapy, and the advent of immune therapy. Norway and Poland (the two countries with the highest Since we planned the trial almost 20 years ago, colorectal contributions of participants in the trial) and did not cancer has evolved from a disease with high mortality to observe increased activity.17 a chronic disease that most individuals survive.1 The Continued follow-up of the NordICC trial remains observed colorectal cancer incidence in the non-screening important to document more definitively whether a group (1·80%) is similar to the predicted rate when we mortality reduction is achievable by colonoscopy planned the trial (1·85%), whereas the observed colorectal screening. Longer follow-up of our trial will also cancer mortality in the non-screening group (0·47%) is investigate the duration of a preventive eect of half of what it was when we planned the trial (0·82%; one colonoscopy on colorectal cancer incidence. This appendix pp 15, 26). This finding might also explain the follow-up is important to tailor screening programmes smaller absolute benefit for colorectal cancer mortality in regarding repetition of screening tests, a major driver our trial than in the sigmoidoscopy trials that took of costs and resources. A recent randomised trial in place about 10–15 years before our trial, when colorectal Spain reported similar performance of biennial faecal cancer mortality was higher.14,15 Although a reduction in immunochemical testing and one colonoscopy in colorectal cancer incidence without a subsequent decline intention-to-screen analyses.7 However, participation in colorectal cancer mortality seems biologically with screening was significantly lower in both screening implausible, it is uncertain whether longer follow-up will groups in the Spanish trial compared with our trial. Per- be able to document a statistically significant reduction protocol analyses in the Spanish trial indicated lower 8 www.thelancet.com Published online May 5, 2026 https://doi.org/10.1016/S0140-6736(26)00508-8 Articles colorectal cancer incidence in those screened with Education (501-1-09-12-12/22), the Polish Foundation of colonoscopy than those who underwent faecal testing, Gastroenterology, the Dutch Ministry of Health and Health Care Prevention Program–Implementation (ZonMw 2008), the Netherlands but the analyses were not performed with causal Organisation for Health Research and Development of the Dutch inference methodologies. Ministry of Health (ZonMw 120720012), the Centre for Translational In summary, a single colonoscopy reduces colorectal Molecular Medicine (CTMM DeCoDe-project), the Swedish Cancer cancer risk by 0·3–0·8% during a follow-up of 13 years. Foundation (2010/345 and CAN 2013/553), Regional Forskningsfond i Uppsala-Örebro Regionen, a Karolinska Institutet Distinguished The risk of colorectal cancer death is very low even Professor Award to H-OA (2368/10–221), and AFA (130072). We would without screening, and the hope of finding a screening like to thank research coordinator Siv Isaksen for help with registry data benefit on colorectal cancer mortality with longer access and monitoring. follow-up is so far not substantiated.8 We encourage the References screening community to consider what is most important 1 Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for for a preventive screening test such as colonoscopy. 36 cancers in 185 countries. CA Cancer J Clin 2024; 74: 229–63. Screening might be justified if it prevents people from 2 Helsingen LM, Kalager M. Colorectal cancer screening—approach, getting cancer even without reducing the risk of dying. evidence, and future directions. N Engl J Med Evid 2022; 1: EVIDra2100035. On the population level, screening benefits must be 3 Lauby-Secretan B, Vilahur N, Bianchini F, et al. International weighed against other expenditures that promote public Agency for Research on Cancer Handbook Working Group. health. The IARC perspective on colorectal cancer screening. N Engl J Med 2018; 378: 1734–40. Contributors 4 Nishihara R, Wu K, Lochhead P, et al. Long-term colorectal-cancer MFK, H-OA, GH, JR, AGZ, MK, ML, and MB were involved in incidence and mortality after lower endoscopy. N Engl J Med 2013; conception and trial design. KG, ØH, GH, MFK, JR, MS, ED, and MR 369: 1095–105. were involved in trial interventions. MFK, MB, AM, and GH accessed 5 Knudsen AB, Rutter CM, Peterse EFP, et al. Colorectal cancer and verified the data. AM, FS, SF, JS, and MAH provided statistical screening. An updated modeling study for the US Preventive expertise. MFK, MB, and H-OA were involved in drafting this Services Task Force. JAMA 2021; 325: 1998–2011. manuscript. All authors provided critical revision of the manuscript and 6 Bretthauer M, Løberg M, Wieszczy P, et al. Eect of colonoscopy intellectual content, had full access to all of the data in the study, and screening on risks of colorectal cancer and related death. were responsible for the final decision to submit for publication. N Engl J Med 2022; 387: 1547–56. 7 Castells A, Quintero E, Bujanda L, et al. Eect of invitation to Declaration of interests colonoscopy versus faecal immunochemical test screening on MB reports research funding from the Norwegian Research Council and colorectal cancer mortality (COLONPREV): a pragmatic, the Health Fund of South-East Norway. FS and SF report research randomised, controlled, non-inferiority trial. Lancet 2025; funding from Oslo University Hospital. MFK reports honoraria for 405: 1231–39. speaking and teaching from Fujifilm; and consulting fees from 8 Dominitz JA, Robertson DJ. Understanding the results of a Olympus. MAH reports grants and contracts from the US National randomized trial of screening colonoscopy. N Engl J Med 2022; Institutes of Health, the US Department of Veterans Aairs, and Novo 387: 1609–11. Nordisk; consulting fees from ProPublica; participation on an advisory 9 Shi J, Løberg M, Kalager M, et al. Eect of colonoscopy screening board for Adia; and stock options in Adigens Health. ED reports on risks of colorectal cancer and related death: instrumental consulting fees from Olympus, Fujifilm, Norgine, Exact Sciences, and variable estimation of per-protocol eects. Eur J Epidemiol 2025; ProMedCS; honoraria from Olympus, Norgine, Ipsen (Mayoly), Fujifilm, 40: 419–25. Steris, and Pentax; leadership or fiduciary roles in CRC Screening; 10 Kaminski MF, Bretthauer M, Zauber A, et al. The NordICC Study: committee roles for the World Endoscopy Organization and the national rationale and design of a randomized trial on colonoscopy colorectal cancer screening programme of the Netherlands; guideline screening for colorectal cancer. Endoscopy 2012; 44: 695–702. commissions, position statements, etc for the European Society of 11 Hamilton SR, Aaltonen LA (eds). World Health Organization Gastrointestinal Endoscopy and the European Hereditary Tumour classification of tumors: pathology and genetics of tumors of the Group; and endoscopic equipment from Fujifilm. All other authors digestive tract. Lyon: IARC Press, 2000: 104–19. declare no competing interests. 12 Kalager M, Zelen M, Langmark F, Adami HO. Eect of screening mammography on breast-cancer mortality in Norway. N Engl J Med Data sharing 2010; 363: 1203–10. Individual participant data, including data dictionaries, can be shared 13 Bretthauer M, Kaminski MF, Løberg M, et al. Population-based upon request to the NordICC trial steering committee stating the colonoscopy screening for colorectal cancer: a European purpose and nature of the intended use, provided that the requesting randomized trial. JAMA Intern Med 2016; 176: 894. party has approval for the intended use from the ethics committees that 14 Holme Ø, Schoen RE, Senore C, et al. Eectiveness of flexible approved the trial. The study protocol, including the statistical analysis sigmoidoscopy screening in men and women and dierent age plan, is available from the trial principal investigator (MB) on request. groups: pooled analysis of randomized trials. BMJ 2017; 356: i6673. 15 Senore C, Riggi E, Armaroli P, et al. Long-term follow-up of the Acknowledgments Italian Flexible Sigmoidoscopy Screening Trial. Ann Intern Med This study was funded by the Nordic Cancer Union, the Norwegian 2022; 175: 36–45. Cancer Society, the Norwegian Research Council (197309), and the 16 Pilonis ND, Spychalski P, Kalager M, et al. Adenoma detection rates Health Fund of South-East Norway (5135). Dr Falk Pharma (Freiburg, by physicians and subsequent colorectal cancer risk. JAMA 2025; Germany) provided the bowel preparation free of charge for 333: 400–07. colonoscopies in Norway. This study also benefited from grant support 17 Bretthauer M, Løberg M, Kaminski MF. Colonoscopy screening and from the National Centre for Research and Development of Poland colorectal cancer incidence and mortality. N Engl J Med 2023; (N R13 0024 04), the Polish–Norwegian Research Programme 388: 376–79. (Pol-Nor/204233/30/2013), the Medical Center for Postgraduate www.thelancet.com Published online May 5, 2026 https://doi.org/10.1016/S0140-6736(26)00508-8 9 --- [PDF原文](https://sci-net.xyz/storage/7900171/ca80225c9ffd4a00494db1e10bdff74bffb4f6d2ac0fe6be8bbb5efea823f2b9/Long-term-effects-of-colonoscopy-screening-on-colorectal-cancer-incidence-and-mortality-a.pdf) DOI: 10.1016/S0140-6736(26)00508-8