In contrast the association with poorer glycemic
In contrast, the association with poorer glycemic control for pancreatic cancer was strongly positive, with HR estimates varying from 1.26 to 1.51 across HbA1c and blood Puromycin levels and incident and prevalent diabetes groups (Table 2). There was no evidence of departure from the proportional hazards assumption.
In addition, there was a moderate negative association with poorer glycemic control for prostate cancer. The HR estimates varied from 0.85 to 0.96 according to glycemic control variable and diabetes group (Table 2 and Fig. 1). In these models, there was an indication of a departure from the proportional hazards assumption, with HR estimates more strongly negative close to diabetes diagnosis and increasing towards 1.0 by the end of the follow-up period. See Supplementary Materials Part 4 for more details.
As a graphical check of our modeling results, we computed, for all-sites cancer and prostate cancer, Kaplan-Meier curves according to four categories of HbA1c level (<6.5 6.5–7.25, 7.25–8.0 and ≥8.0%). We observed an apparent lack of association with HbA1c for all-sites cancer, and a negative association for prostate cancer (Fig. 1), in agreement with the Cox model results.
We conducted a series of sensitivity analyses in which we changed the window-of-effect period of glycemic control from the previous 2 years. We investigated windows-of-effect of length 2 years but with a lag of 1 year (1–3 years prior to the current time), or with a lag of 3 years (3–5 years back). HRs for all-sites cancer and prostate cancer were little changed. Those for pancreatic cancer are presented in Table 3. It can be seen that, when averaging HbA1c or blood glucose over the period 3–5 years back, the HRs for pancreatic cancer substantially declined (to 1.08–1.23) from the values (1.26–1.51) obtained when averaging over the 0–2 years period.
We also investigated a window-of-effect period of 5 years (0–5 years back). Colorectal cancer HRs in the incident and prevalent diabetes groups changed by 0.05 or less in no consistent direction (see Table A1 in Supplementary materials Part 2), whereas for pancreatic cancer the HRs all declined, by 0.04 to 0.09, but remained in the range 1.22–1.42 (Table 3).
Conflict of interest statement
Funding This work was supported by a grant of the European Foundation for the Study of Diabetes (EFSD).
Duality of interest
Introduction In 2012, Denmark had the fourth highest age-standardised rate for colorectal cancer in the world (40.5 per 100,000) . Results from randomised controlled studies have shown that guaiac faecal occult blood testing (gFOBT) screening reduces colorectal cancer mortality by 18% . Based on this and the fact that faecal immunochemical test (FIT) screening has proved to have a higher participation rate, detection rate and positive predictive value [, , ], it was decided to set up a national colorectal cancer screening programme in Denmark. The overall aim of colorectal cancer screening is to reduce morbidity and mortality of the disease, but such outcomes are only measurable after several years and rounds of a cancer screening programme. Meanwhile, performance indicators can be set up for continuing evaluation and monitoring of an ongoing screening programme, as e.g. suggested in the European Guidelines for colorectal cancer screening .
Methods The Danish National Colorectal Cancer Screening Programme was implemented 3 March 2014  and is offered free of charge to all residents aged 50–74 years. All residents with a registered address in Denmark, aged 50–74 years in the period 1 January 2014 – 31 December 2017 were invited to have one screening in the first screening round . To gradually build up the colonoscopy capacity, the first screening round was planned to take 46 months (3 March 2014 – 31 December 2017), and subsequent rounds, were all residents aged 50–74 years will be invited once, are planned to take two years. Citizens are invited randomly according to birth month, meaning that first all citizens born in one month were invited, thereafter citizens born in another month etc. The sequence of months was drawn randomly. Non-invited citizens turning 75 during the first invitation round were invited 1 month before turning 75. Citizens turning 50 during the first invitation round were invited 1–3 months before turning 50 . Each of the five regions in Denmark is responsible for invitations and follow-up procedures in their own region.