Expansion
The condition, the test, the treatment and the programme
Expansion
The condition
- An important health problem, with a recognisable latent or early symptomatic stage
- Its natural history adequately understood
The test
- Suitable, simple, safe and acceptable to the population
- Valid, with adequate sensitivity and specificity, and an agreed cut-off
The treatment
- An effective treatment exists, and earlier treatment gives a better outcome than treatment at the usual time of diagnosis
- An agreed policy on whom to treat
- Facilities for diagnosis and treatment available
The programme
- Cost effective, with the cost balanced against overall expenditure
- Continuous, not a one-off
- The benefit must outweigh the harm, including anxiety, false positives, and the consequences of investigation
- Evidence, ideally from randomised trials, that it reduces mortality or morbidity
The three biases that make screening look better than it is
- Lead time bias: diagnosis is earlier, so survival from diagnosis appears longer even if the date of death is unchanged
- Length time bias: screening preferentially detects slow growing disease, which has a better prognosis anyway, since aggressive disease presents between rounds
- Overdiagnosis: detecting disease that would never have caused harm, so every case detected is counted as a success and treated, with all the associated harm and none of the benefit
Selection bias compounds it, because those who attend screening are healthier than those who do not. These are the reasons that mortality in a randomised population, not five year survival among those diagnosed, is the only credible measure of a screening programme.