# http://www.fpnotebook.com/

## Test Specificity

*Aka: Test Specificity, Specificity, False Positive Rate*

- See Also
- Screening Test
- Contingency Grid or Cross Tab (includes Statistics Example)
- Bayes Theorem (Bayesian Statistics)
- Fagan Nomogram
- Experimental Error (Experimental Bias)
- Lead-Time Bias
- Length Bias
- Selection Bias (Screening Bias)
- Likelihood Ratio (Positive Likelihood Ratio, Negative Likelihood Ratio)
- Number Needed to Screen (Number Needed to Treat, Absolute Risk Reduction, Relative Risk Reduction)
- Negative Predictive Value
- Positive Predictive Value
- Pre-Test Odds or Post-Test Odds
- Receiver Operating Characteristic
- Test Sensitivity (False Negative Rate)
- U.S. Preventive Services Task Force Recommendations

- Definition
- Screening Test correctly negative in absence of disease
- A test with high Specificity has few false positives
- Independent of disease Prevalence in the community
- Specific Tests allow user to rule in or confirm a condition (mnemonic "SPin")

- Calculation
- Test Specificity
- True negative tests per unaffected patients tested
- Expressed as a percentage
- Test Specificity = P(negative test | no disease)
- Where P (A | B) = Probability of A given B

- False Positive Rate
- Test positive despite absence of condition
- False Positive Rate = 1 - Test Specificity

- Test Specificity
- Example: A new Screening Test for Crohn's Disease
- Patients without Crohn's Disease tested: 255
- Patients without Crohn's Disease who have a negative test: 230
- Specificity = 230/255 or 90%

- Precaution
- Test Specificity can be misleading
- Example
- Condition A is actually present in 150 patients (5%) of the 3000 patients tested
- Therefore 2850 patients do not have condition A
- Test Specificity of 90% would result in a 10% False Positive Rate (of 2850) or 285 patients
- In this case a 90% Test Specificity would result in a false positive result in 285 patients, when only 150 actually had the condition

- Conclusion
- The lower the Prevalence of disease in the cohort tested, the higher the Test Specificity must be to give a reasonable likelihood of correctness
- Positive Predictive Value may be a more valuable measure as it takes the condition Prevalence into account
- Risk stratifying a group prior to testing can concentrate patients more likely to be positive without missing a significant number
- Example: Limit D-Dimer testing to the intermediate likelihood of Pulmonary Embolism group (based on Wells Score)
- This increases the Prevalence in the tested group and reduces the number of patients with false positive results

- References
- Hennekens (1987) Epidemiology Medicine, p.327-47
- Majoewsky (2012) EM:RAP 12(1): 9-11
- Gates (2001) Am Fam Physician 63(3):513-22 [PubMed]
- MacLean (1996) Med Clin North Am 80(1):1-14 [PubMed]
- Nielsen (1999) Med Clin North Am 83(6):1323-37 [PubMed]