Healthcare & Clinical
CATCH Head-Injury Research: Understanding Sensitivity in a Spreadsheet
Calculate sensitivity from fictional counts and read the original CATCH derivation and later validation studies.
CATCH research and interpretation
The 2010 CATCH study derived a clinical decision rule concerning CT use in children presenting with minor head injury. A later study evaluated the original rule and a refined version in a separate cohort. Performance estimates apply to the study population, eligibility criteria and measured outcome, and are accompanied by uncertainty.
This worksheet does not implement CATCH, assess a child, or determine whether CT or observation is appropriate. Qualified clinicians use current guidance and the full circumstances of an injury. The original research, validation and NICE guidance are linked below.
A fictional example of sensitivity
Sensitivity is the proportion of cases with the target outcome identified by a test or rule: true positives divided by true positives plus false negatives. In an invented data set, 45 identified cases and 5 missed cases give 45/(45+5) = 90.0%. These numbers are not CATCH study results.
A sensitivity percentage alone does not tell a person the chance that a negative result is safe, and a small sample can lead to substantial statistical uncertainty. This arithmetic example does not establish the clinical performance of any rule.
| Fictional quantity | Value |
|---|---|
| True positive cases | 45 |
| False negative cases | 5 |
| Cases with the target outcome | 50 |
| Sensitivity | 90.0% |
Calculate sensitivity from counts
Enter nonnegative whole-number counts in B2 and B3. When both are zero, sensitivity is undefined because there are no outcome-positive cases in the denominator.
=IF(COUNT(B2:B3)<>2,"Enter two numeric counts",IF(OR(MIN(B2:B3)<0,B2<>INT(B2),B3<>INT(B3)),"Enter nonnegative whole counts",IF(SUM(B2:B3)=0,"Sensitivity undefined with zero cases",ROUND(100*B2/SUM(B2:B3),1))))Try the example
Use the original values or change them to test your understanding.