When we do studies concerning association between infectious disease and climate change, do you think it is better to use onset dates rather than notified dates and why? I might prefer to use onset dates, what do you think?
There can be multiple time variables in epidemiological data sets e.g. date of onset, date diagnosed, receipt of laboratory specimen, date of notification etc. Date of notification is commonly used for notifiable infectious diseases - other variables can have limited data completeness. But it would be worth checking the level of completeness with respect to the time variables in your data.
It would make more sense to me to use the date of onset if you have reliable data for that. If a climatic variable has a contemporaneous effect on the incidence of a disease, and the date of notification is delayed from onset, your estimate of the strength of the relationship may be biased downward.
But as Ronan says the problem is data availability. Do you actually have full reliable data for date of onset? In some cases the date may be missing, a rough approximation (i.e., a guess by the recorder), or "partial dates" may be present (e.g., month or year known, day unknown). If the onset-date data is very unreliable, it may make more sense to use the notification date, especially if the delay between the two is usually small.
Thanks Ronan and Matt! In the dataset, we have multiple time variables as Ronan said. Although the dataset have 'reliable' date of onset (at least the data provider told like this), I still consider the notifiable date is more reliable. Our team next meeting may talk about it. Thank you so much!
A very genuine question, and one of the most critical issue in these kind of studies. As the date of "onset" entirely depend on the patient how he or she remembered and recorded in the hospital/Nursing. Hence, I second you to consider the notifiable date.
However, in some of our studies with dengue, HFMD and climate, it was interesting to see the results considering a lag between the climatic data and the disease data. Hence, sometimes without knowing the onset date, it is critical to set the period of lag (as onset to notifiable date is not always same) and hence difficult to argue the analysis results. In such case, you might consider the onset date (although with little uncertainity).
Whatever it is, it entirely depends on your study objectives and assumptions which you will want to make at the beginning itself.
I think onset. Global climate change is projected to increase the frequency and intensity of extreme climatic events such as floods, droughts and cyclones, and these climatic conditions have been associated with increased disease risk.
The accuracy and consistency with which the data are reported is probably going to be the best way to guide such a decision and then adjust for factors such as lag time.
You should strive for onset date of infection, if you can obtain it. But agree with Wittert: use accurate, consistent information. Ideally you would adjust for lag time if you are not using onset date. How you adjust needs to be thoughtful.
Yes, onset dates is important to study the association between infectious disease and climate change, when there is climate change it lead to favorable conditions , that increase the density of infectious agent at critical level, then we can predict the disease in the form of epidemic /outbreak onset, hence it is important to use association of climate change with the onset of disease. So that there should be preparedness to deal, for this we can take the public health measures to control infectious disease or its outbreak/ epidemic.Here, we can also use the geographical information system.
It is ideal to use the onset date as compared to notified date, but it may not be possible to get the onset date accurately. The notified date does not reflect the actual climatic conditions at which the transmission occurred.
It also depends on the incubation period of the disease. If the difference between onset date and reporting date is minimised, the accuracy would be more.