The median age at FS was 60 years, 53% of participants were males and 11% had ≥ 1 first degree relative with CRC (Table 1). Bowel preparation quality was excellent for 43%. Median insertion and withdrawal times were 2.4 (IQR 1.7–3.4) and 1.9 (IQR 1.2–3.4) minutes, respectively. Most examinations reached the descending colon or further (78%) and 29% of participants reported feeling no pain during the examination (Table 1).

Variables were examined by centre, synonymous with endoscopist, among the 70% of participants with negative examinations (Supplementary Table 3). ‘Excellent’ bowel preparation quality varied between 9.6% (centre 9) and 68.2% (centre 4). Median insertion time varied from 1.45 (IQR 1.03–2.07; centre 4) to 3.88 (IQR, 2.92–5.50; centre 10) minutes and median withdrawal time varied from 0.88 (IQR 0.65–1.27; centre 4) to 2.38 (IQR 1.90–3.06; centre 5) minutes. Examinations reaching the descending colon varied between 27.7% (centre 13) and 84.1% (centre 8) and participants reporting severe pain varied between 0.2% (centre 4) and 4.1% (centre 8). Despite these differences between centres, there were no clear associations between these factors and endoscopist ADR when examining by ascending order of ADR (Supplementary Table 3).

Among negative examinations, the proportion of participants reporting quite a lot/severe pain tended to decrease with further segment reached (p-trends < 0.001). Females were more likely to report quite a lot/severe pain than males (15.5% vs. 8.0%, respectively, among exams reaching a maximum of the SF) and to have a longer time to maximum insertion for each section of the bowel reached (SF: median 2.37 min (IQR 1.75–3.45) vs. 2.14 min (IQR 1.58–2.90)) (Supplementary Table 4).

In complete-case analyses, 3,349 (14.4%) negative examinations reached at least the SF. Females were less likely to have an examination reaching the SF (10.9%) than males (18.2%) (multivariable: OR 0.57, 95%CI 0.53–0.62). Among those with negative exams, the odds of reaching the SF were 75% lower with ‘poor’ bowel preparation compared to ‘excellent’ (multivariable: OR 0.25 95%CI 0.13–0.50) and 47% lower with the reporting of severe pain compared to no pain (multivariable: OR 0.53 95%CI 0.38–0.73) (Supplementary Table 5).

There were 4,104 (12.0%) participants with ≥ 1 distal adenoma detected (Table 1). In all models, there were increased odds of distal adenoma detection with increasing age (multivariable: OR 1.03, 95%CI 1.01–1.04) (Table 1), with a family history of CRC, compared to without (multivariable: OR 1.40, 95%CI 1.21–1.62), and decreased odds in females compared to males (multivariable: OR 0.62, 95%CI 0.56–0.69).

Although there was no association in the full dataset, in complete-case models there were increased odds of distal adenoma detection for those with ‘poor’ bowel preparation compared to ‘excellent’ (multivariable: OR 2.88, 95%CI 1.25–6.60; Table 1), and lower odds for those with ‘good’ (multivariable: OR 0.84, 95%CI 0.74–0.95) or ‘adequate’ bowel preparation (multivariable: OR 0.56, 95%CI 0.49–0.65).

In all models, increasing insertion and withdrawal times were associated with distal adenoma detection (multivariable: OR ≥ 4.00 vs. < 2.00 min: 1.96, 95%CI 1.68–2.29; 32.79, 95%CI 28.22–38.11, respectively). In comparison to reaching the sigmoid/descending junction, reaching more proximally was associated with higher odds of distal adenoma detection in univariable models (full dataset, descending colon: OR 1.43, 95%CI 1.31–1.56; SF: OR 1.66, 95%CI 1.47–1.88); however, this attenuated in multivariable models.

In complete-case univariable models, there were lower odds of distal adenoma detection with increasing pain (severe compared to none: OR 0.69, 95%CI 0.51–0.95; Table 1) but this was not evident in the other models. In the full dataset, individuals whose FS screening occurred after their endoscopist’s first 500 examinations had increased odds of distal adenoma detection compared to those whose took place earlier (OR 1.32, 95%CI 1.20–1.45); multivariable models were not possible due to missing data.

There were 919 (4.8%) participants with multiple and/or advanced distal adenomas in the complete-case dataset (Supplementary Table 6). Age, sex, family history, bowel preparation quality, insertion and withdrawal time, segment reached, patient pain and the order of FS occurrence were similarly associated with the detection of advanced and/or multiple adenomas as of any distal adenoma.

During a median follow-up of 17 years, 168 (0.5%) distal CRCs were diagnosed (Table 2). In the full dataset, females had a lower risk of distal CRC than males (HR 0.62, 95%CI 0.45–0.85) and those with a family history of CRC had a higher risk than those without (HR 1.65, 95%CI 1.09–2.50) (Table 2, Supplementary Fig. 1A-B); these effects attenuated in complete-case models.

Age, bowel preparation quality, segment of bowel reached, patient-reported pain, and order of examination occurrence were not associated with distal CRC incidence (Table 2, Supplementary Fig. 1C–F). Although overall the associations for insertion and withdrawal times were not statistically significant, those in the top category of ≥ 4.00 min (versus < 2.00 min) had an increased risk of distal CRC (multivariable: HR 1.81, 95%CI 1.00–3.27; HR 1.93, 95%CI 1.14–3.24, respectively) (Table 2, Supplementary Fig. 1G-H).

Compared to those examined by high-detectors, individuals examined by low-detectors had an increased risk of distal CRC (multivariable: HR 4.71, 95%CI 2.65–8.38), as did those examined by intermediate-detectors in complete-case models only (multivariable: HR 2.16, 95%CI 1.22–3.81) (Table 2, Supplementary Fig. 1I).

Excluding participants with multiple FS examinations (n = 1,810) did not materially alter the results for distal adenoma detection or long-term colorectal cancer incidence in any of the models.