If you want to make a recommendation on a new diagnostic test, then it is essential that you consider the potential of overdiagnosing.
A substantial challenge with diagnostic tests are that there is a risk that you may diagnose a patient with a disease, that may never have developed into a serious condition. It is not possible to know if a patient with a positive test result is overdiagnosed, and therefore these patients are offered treatment. By definition, this treatment is redundant. Adverse events are found for a majority of treatments, and thus overdiagnosis may potentially hurt patients. In addition to the physical consequences, that may be present both at short and longterm follow-up, there may also be serious psychological consequences of being given a diagnosis. Overdiagnosis is especially present with tests, where you screen for risk factors for a disease (e.g increased colesterol, blodpressure), since the fraction of people that eventually develop the disease is often low.
A new diagnostic test and the potential of overdiagnosing, what to look for:
- You need to assess if the relevant disease is clearly defined. This is the case if the defintion has not changed over time and if the definition does not mainly rely on subjective assessments.
- There should be no considerable inter-observer variations when applying the new test.
- The reference standard needs to be reliable. This can be evaluated using QUADAS-2