Description:
<jats:p>To the Editor.—</jats:p>
<jats:p>Rosenberg et al1 have made a commendable attempt to define early markers of maltreated children. Their effort demonstrates how difficult it is to develop an instrument to identify families with complicated, low-prevalence psychosocial disease. The sensitivity and positive predictive value of such assessment tools are usually very low. This example is no exception. The sensitivity of a test is defined as the "test positivies" in patients with the disease (eg, abused children in the sample who are unkempt and abnormally parented in the emergency room) divided by all the patients with the disease (all abused children in the sample).2</jats:p>