Owen Yang
This may not have been the first time I speak about this, but I will not stop speaking until the tide changes.
When I say descriptive reports I mean just describing how things go: how many people have cancer every year, and which type, over time, or across different sub-population such as children, female, elderly, etc. This is contrast to the studies that generate ‘inferences.’ This means studies investigating whether factor A causes Factor B, or whether factor A is associated with factor B. Inferencing studies are historically concluded with a p value that tells you yes or no. Nowadays we do not like p values very much because it has been misused and abused, but we are still waiting for the old generation of scientists to fade away. This is a slow process, because not all new generations of medical scientists have access to the new education, and therefore they are still influenced by the old generations. Maybe you are one of them.
Anyway, a descriptive report does not care about whether factor A causes factor B. It simply tells you how many people out there, and where and when, who has factor A. Say hepatitis virus causes liver cancer. A descriptive study would intend to tell you who carries hepatitis virus, and/or how many individuals develop liver cancer in a population every year.
A norm as a good example
It does not need to be a yes/no factor. For example, last time there was a brilliant obstetrician who told he he wants to do the norm (or normal values) of number of blood cells of premature babies. You would agree it is a good idea: without a normal value, how do we know it is abnormal if we do not know what is normal?
But his colleagues did not care about what he wanted to do, because they could not see the scientific value of it. I think this basically just because they think there is no ‘hypothesis’ in there. It is interesting because they seem to quote a some sort of scientific principle, but selectively forget the stage of ‘observation’ in their quotation.
General undervalue of descriptive reports
In the science world (or perhaps just medical science) there is a general undervalue of descriptive reports. The science world would disagree because you see them on Lancet journals all the time. However, you almost only see them on Lancet journals, which only publishes descriptive reports when you are a well-funded global research. Nowadays the common name is Global Burden of Diseases. They are great in a sense that they provides a picture of what is going on in the world. The trouble is anything below this scale seems unpalatable.
Perhaps it is not the duty of scientific journals to publish descriptive results. The distorted reality here is scientists need publication to survive, and therefore anything that does not attract publication is not worthwhile doing. Perhaps it is the duty of the research funder to appreciate these descriptive reports. Perhaps it is the research institutions that should fix this. I do not have a solution. It is such a complex.
Is there a bias?
Journals need readership and needs citation, and to some degree should be allowed some preferences. However, I do think journals need to be explicitly aware of their biases (or preferences). It is very common that a descriptive report from the US or from the Europe (and nowadays China) could be self-sufficient. A descriptive report from Taiwan, or Ghana, would need to justify itself ‘why it can be generalised to the rest of the world’ or ‘why it is lack of hypothesis.’ They need to be aware of the fact that they select certain reports not because of the scientific value, but because of their market value.
A paper that describes the association between statin and cancer from Taiwan can be accepted more easily than a paper the describes the pattern of statin prescription from Taiwan. There is no right or wrong, but there is an issue in there.
We need more descriptive reports from under-reported countries so that they could be understood.
