Background In Denmark, the Cancer Patient Pathway for Non-Specific Signs and Symptoms (NSSC-CPP) has been implemented with variations: in some areas, general practitioners (GPs) do the initial diagnostic work-up (GP paradigm); in other areas, patients are referred directly to the hospital (hospital paradigm). There is no evidence to suggest the most beneficial organisation. Therefore, this study aims to compare the occurrence of colon cancer and the risk of non-localised cancer stage between the GP and hospital paradigms.Material and Methods In this registry-based case-control study, we applied multivariable binary logistic regression models to estimate the odds ratios (OR) of colon cancer and non-localised stage associated with the GP paradigm and hospital paradigm. All cases and controls were assigned to a paradigm based on their diagnostic activity (CT scan or CPP) six months before the index date. As not all CT scans in the control group were part of the cancer work-up as a sensitivity analysis, we investigated the impact of varying the fraction of these, which were randomly removed using a bootstrap approach for inference.Results The GP paradigm was more likely to result in a cancer diagnosis than the hospital paradigm; ORs ranged from 1.91-3.15 considering different fractions of CT scans as part of cancer work-up. No difference was found in the cancer stage between the two paradigms; ORs ranged from 1.08-1.10 and were not statistically significant.Conclusion Patients in the GP paradigm were diagnosed with colon cancer more often, but we cannot conclude that the distribution of respectively localised or non-localised extent of disease is different from that of patients in the hospital paradigm.