Generally, percentages are more convenient since most clinicians work with them and are familiar with them. They are also used in the guidelines. The reason is that the percentages are comparable (because they are corrected for age, weight, height and race).
However, it also depends on what you want to study and generally both are presented in most of the studies.
Using either absolute values or %predicted will not make a difference if you are studying a change in the same patient group. However, while comparing two groups I would prefer %predicted over absolute values. I also feel that both the approaches have their own flaws.
The question is really a good one. And I don't know if someone has the answer. When we use % we are trying to reduce differences due to sex, age, etc. Of course makes no sense if you compare the same subjects before x after. Unless you are expecting that the differences are also sex, ... will be also sex and so on dependent. But if you compares two different groups, it becomes complicated. If you compares differences (again same subject after _- before) you can use for example the same way you use the changes in lung forced values (after-before/before). But always remember: the more variable s are used to calculate your final variable, the variability (dispersion) can become worse instead of smaller.
It did not work trying to answer such a question in a very small cel screen. But the best I can answer is: better keep it simple. The more you try to make the numbers look better, the worse they usually look.
If we are comparing the same subjects before and after and you do think the change will be lung volume dependent, as for example before x after bronchodilator, we can use the same principles we use when calculating the change in forced lung volumes after using bronchodilators [Change= ((after-before)/before) x100].
If we compare differences (same subject after - before) that occur the subjects from two different groups (for example two different bronchodilators), and we suppose they will be lung size dependent (that is: for a big lung, 200ml can be almost nothing and for a small one the same 200ml may be quite a change) we can use the same principle again.
On the other side, for each single variable in our "formula", we are adding also a dispersion. And may always happen that the final dispersion of our data will be greater after we make a lot of calculations, than if we used the values we measured.
The different parameters of Pulmonary Function Tests have their own significance. Forced Vital Capacity is static as against Forced Expired Volume at end of first Second which is dynamic, as is time related, which will help in deciding whether the patient is having obstructive type of lung dysfunction or restrictive type. Similarly measurement of Peak Expiratoty Flow Rate helps in assessing strength of muscles of expiration. Absolute values should be considered when one thinks of same patient and when values are measured at intervals to assess the effect of treatment or some training for improving lung function.But when one wants to compare different persons having difference in age, height etc.it is better to consider percentage of predicted values