The question

Although the principles are similar across all types of study question, systematic review of diagnostic tests requires some different approaches, notably in the methods used for combining data from different studies. As with the other types of questions, the starting point for diagnostic studies is an appropriate question, including a description of:

• the disease of interest;

• the test(s) of interest;

• patient features that are likely to alter the test performance characteristics; and

• the performance characteristics of the test compared to the performance characteristics of another test or tests.

If test performance characteristics vary between patient subgroups, this needs to be taken into account when applying the results of a systematic review of diagnostic tests. Common features that affect test performance characteristics include the symptoms, signs, tests and previous triage through the health care system that has got patients to the point at which you wish to evaluate the performance characteristics of a test. This issue is explored further in Section 8.3 on appraising the quality and applicability of studies.

When the performance characteristics of the test are compared to the performance characteristics of another test(s), the situation is analogous to trials in which an intervention is compared to a placebo or to another drug. For example, we may not want to know if the presence of leukocytes in an abdominal fluid aspiration has a high sensitivity and specificity for the diagnosis of appendicitis in people presenting with abdominal pain. Rather, we may want to know its incremental sensitivity and specificity compared to other features that are more easily obtained, for example, rebound tenderness in the right iliac fossa (Caldwell and Watson, 1994).