Property | Definition[1] | Considerations for developmental-behavioral screening tests |
Sensitivity | - Probability of test giving a positive result when the condition is present. (What percentage of children with developmental delay will be flagged by this test?)
| - ≥70%[2] is generally accepted.
|
Specificity | - Probability of the test giving a negative result when the condition is not present. (What percentage of children without developmental delay will "pass" this test?)
| - ≥70%[2] is generally accepted.
|
Positive predictive value (PPV) | - Probability of the condition being present in a positive test. (What percentage of children who are flagged by this test have the developmental delay for which it screens?)
| - If the condition is serious and a follow-up test can distinguish true and false positives, then low PPV (meaning a high number of false positives) may be acceptable.
- PPV is higher if the condition is more common (eg, language delay versus cognitive delay).
- PPV and NPV are less useful in rare or low-prevalence conditions.[3]
|
Negative predictive value (NPV) | - Probability of the condition being absent in a negative test. (What percentage of children who "pass" this test do not have the developmental delay for which it screens?)
|
Likelihood ratio (LR) | - Likelihood that a "flagged" test indicates the presence of the condition compared to its absence. (How much more likely is a "failed" screen to occur in a child with a developmental delay than without a delay?)
| — |
Concurrent validity | - Correlation of the test results to the "gold standard" or reference test results at the same point in time. (If this child has a comprehensive evaluation now, how will it correlate with the screening test?)
| - Childhood development is dynamic. Concurrent validity fails to account for the expected trajectory; some mild delays resolve while others become more marked.
|
Predictive validity | - Correlation of the test results to the "gold standard" or reference test results given at a later point in time. (If this child has a comprehensive evaluation sometime in the future, how will it correlate with the screening test?)
| - Predictive validity can be misleading. The screening test results may affect experiences before the "gold standard" evaluation. Developmental-behavioral evaluations sometimes occur months to years after a screening test, and various medical or psychosocial interventions may have occurred between the two time points.[2]
|