In this context, it's worth knowing about the difference between screening and non-screening tests. In general, the latter are meant for people with signs or symptoms of a disease, while the former are applied to individuals without clinical suspicion, although the appropriateness of screening can be affected by age and other factors. For example, screening mammography isn't appropriate in a 20-year-old woman without other risks.
Usually, screening tests maximize sensitivity (low false negatives) over specificity (low false positives). For a disease that causes significant mortality and morbidity if not detected and treated in time, like colon cancer, you want to minimize misses, even if it means sending some people without the condition to colonoscopy, which is very safe. However, if the only way to ascertain whether a person with a positive screening test had the disease was to do surgery, the calculus would be different.
Among other things, the disease being screened for must be sufficiently harmful and prevalent to warrant testing many people. For instance, it would make no sense to screen for thyroid cancer using ultrasound. Thyroid nodules are extremely common (almost 70% of adults), and less than 10% are malignant. Also, most thyroid cancers have an excellent prognosis, and over-diagnosis, defined as detection of disease that would not harm or kill a person during their lifetime if left untreated, is a worldwide problem.
