Sensitivity and specificity in a normal population
If people present with
- chronic cough or phlegm (at least 3 month per year),
- more dyspnea with everyday activities than people of the same age,
- and/or attacks of dyspnea suggestive of asthma,
it is likely that some of them can be shown to have obstructive lung disease. Airway obstruction is defined as an FEV1%(F)VC below the 5-percentile for persons of the same ages who never had chronic respiratory symptoms and who are lifelong non-smokers. By using the 5-percentile we fix the specificity at 95%, and we accept 5% false positive test results in a reference population. In the Vlaardingen – Vlagtwedde population the sensitivity (the percentage of subjects with respiratory symptoms whose FEV1%(F)VC is below the 5th percentile for the reference population) is as follows:
| Females | Males |
| 13.3% | 27.3% |
In females the reference group is about equally large as the group with chronic respiratory symptoms. If lung function tests are performed without clinical indication, in women for 5% false positive test results there will be only 13.3% true positive results. The signal to noise ratio is far more favorable in men, in part because the reference group is small compared with the group with chronic respiratory symptoms. These results probably compare favorably to e.g. the detection of hypertension from routine measurements of blood pressure.
Can we improve the sensitivity of spirometric tests by using FEV1 or FVC instead of FEV1%FVC? No, we cannot, here are the sensitivities:
| Index | Females | Males |
| FEV1 | 12.9% | 26.6% |
| FVC | 9.1% | 13.8% |
One way to increase the sensitivity is to lower the specificity
of spirometric tests, as shown below.
