Researchers from the Fred Hutchinson Cancer Research Center examine biases stemming from choosing matched controls — people with the same risk factors for disease as the cases but with no disease themselves — in biomarker research studies, and present a statistical method to predict performance of such matched studies. As they report in Clinical Chemistry, the researchers found that the performance of biomarkers alone when applied to the general population was underestimated, but when added to other risk factors, it was overestimated. "To properly gauge prediction performance in the population or the improvement gained by adding a biomarker to known risk factors, matched case control studies must be supplemented with risk factor information from the population and must be analyzed with nonstandard statistical methods," the researchers say.
Also in Clinical Chemistry, the University of Amsterdam's Patrick Bossuyt and his colleagues review different ways in which clinical utility has been defined — from changing clinicians' decisions to affecting health outcomes — and how it should be determined for diagnostic tests. Since randomized trials of diagnostics tests can be complicated, the researchers instead suggest a decision analysis model-based approach to determine the clinical utility of a diagnostic. However, Bossuyt and his team also note that cost-effectiveness studies will be needed to guide decisions about diagnostic tests. "These days, laboratory medicine cannot escape the ongoing paradigm shift, in which evidence that diagnostic testing improves patient outcomes is becoming a requirement before the cost of a test can be reimbursed and the test can be used in practice," they add.