Which lower limit of normal is best ?
We have compared two test criteria (use the software, click the button COPD criteria and explore the available databases):
- one where one decides whether there is airway obstruction on the basis of scientifically established lower limits of normal for FEV1%VC (LLN criterion) based on gender and age (in other populations standing height has been shown to also affect the LLN);
- one where, out of concern that the above method would be too complicated to handle in daily medical practice, a fixed lower limit for FEV1%VC (GOLD criterion) of 70% was adopted, irrespective of age, gender and standing height.
The comparison shows that the GOLD criterion leads to a considerable percentage of false positive and false negative test results. The most plausible explanation is that this is due to the fact that the GOLD criterion is not appropriate, especially at higher ages. As argued above regarding the ‘surplus’ subjects identified by the GOLD criterion as patients with COPD, and treating them, is unlikely to be of benefit to them even if they had COPD. It may cause them harm, and it burdens the individuals with being tagged as a patient with a chronic disease with a progressive course that can only be favourably influenced by smoking cessation but is unaffected by medical intervention. Medical intervention is therefore not cost-effective and potentially unethical.
Therefore the LLN criterion, apart from being based on good science, is to be preferred over the GOLD criterion.
What about 5-percentiles for the lower limit of normal?
Even so one might wonder whether it is such a good idea to use a LLN which, by the way it was constructed, will single out about 5% of healthy nonsmokers as having an FEV1%VC compatible with airway obstruction. As smoking becomes less accepted in many countries, and therefore the proportion of lifelong nonsmokers in the population increases, so does the absolute number of subjects who might be erroneously tagged as subjects with COPD (false positives), albeit mostly in a mild form. Since the potential benefits of such tagging and associated medical intervention seem to be minimal even in true positives (ref. 1-2), there is good reason to adopt a more conservative LLN. The software allows you to tweak the LLN and see how this affects the proportion of false positive and false negative results. If you agree with published and peer reviewed reports that FEV1%VC is age (and often height) dependent, what you should consider then is adjusting the LLN so that for example only 2.5% of the reference population falls below the limit (prevalence LLN), and then consider how this affects results in a population sample. If data are normally distributed, subtracting 1.96·RSD from the predicted value will give you the 2.5 percentile, and subtracting 2.33·RSD would give you about the 1 percentile.
Optimal lower limit of normal for diagnosing airway obstruction
Ideally one should establish the optimal threshold value using hard evidence about the costs and benefits of treatment, so that an optimal balance between false positives and true positives can be found. However, if you main interest is in COPD, you may decide to apply the 2½ percentile only to individuals aged over 45 yrs and not to younger individuals. This would effectively limit the percentage of false-positives in the older age range.
|1||Highland KB, Strange C, Heffner JE. Long-term effects of inhaled corticosteroids on FEV1 in patients with chronic obstructive pulmonary disease: a meta-analysis. Ann Intern Med 2003; 138: 969-973.|
|2||Burge PS, Lewis SA. So inhaled steroids slow the rate of decline in FEV1 in patients with COPD after all? Thorax 2003; 58: 911-913.|