Implementation period (step-wise)

From an epidemiological perspective implementation entails more than simply carrying out the screening process and onward referral for assessment whenever required. The particular epidemiological concerns at the early, implementation phase focus on the complete and accurate recording of all indi-vidual data pertaining to every participant, the screening test, its result, the decisions made as a consequence and their eventual outcome in terms of diagnosis and treatment and monitoring the causes of death.

Pilot demonstration projects have been carried out in some European countries to assess the feasibility of national programmes and their impact on routine services and to test whether the short-term outcomes of RCTs could be achieved in a context of routine care by a programme covering the whole target population (UK Colorectal Cancer Screening Pilot Group 2004; Goulard et al. 2008).

A new screening programme should be implemented in such a way that effectiveness can be evaluated. This can be achieved using individual-level randomisation into screening and control groups at the phase when the programme is new and resources and practical limitations prohibit the full coverage of the target population. This step-wise implementation, in which the target population is gradually taken into the programme as available resources expand, is both feasible and accepted when the available resources are used to their full extent. 

A randomised screening design is helpful in the start-up phase when all the healthcare services and the infrastructure have not been evaluated within the screening programme, and since there cannot be certainty that the desired outcome and quality will be reached in that particular programme. In the first years of screening, an invitation scheme that gradually expands to cover more regions and age groups over the years can be used. Individuals in the control group will be offered screening later after the first years. This provides an unbiased comparison group. 

A model from Finland is based on individual-level randomisation over the first six years (Malila, Anttila & Hakama 2005). For a six-year implementation phase it was expected that the number of colorectal cancer deaths will accumulate during 10 years from launching the programme in a population of around 3 million and a colorectal cancer mortality rate of approximately 15/100 000. Meanwhile, feasibility can be studied and the programme monitored with various process indicators such as attendance rates, proportion of test positives, detection rates, and positive predictive values. 

A randomised screening design can also be used to assess the impact of alternative policies, such as different methods of invitation, or different target age groups. The randomised approach may also represent an acceptable and feasible alternative to assess the impact of a new screening test or to compare cost-effectiveness of different screening strategies, when a clinical randomised trial to evaluate the reduction in cancer occurrence or mortality is deemed impractical. 

Recommendation 

 Ideally, any new screening programme should be implemented using individual-level randomisation into screening and control groups in the phase when resources and practical limitations prohibit the full coverage of the target population (VI - A).