Talk:Smoking status in Beneris
|Fact discussion: .|
|Opening statement: Socioeconomic and smoking status should be continuous variables.
Closing statement: Accepted.
(A closing statement, when resolved, should be updated to the main page.)
←--1P: . During the workshop in Berlin it has been decided that socioeconomic status and smoking status should be discrete variables having respectively three (low/medium/high) and two (smoking/not-smoking) states. Moreover, as indicated in the discussion on this website, distributions of these variables are age-group and country specific. However, if socioeconomic status and smoking status are discrete variables then the modeling of dose-response functions become more complex and imposes higher requirements on data which in some cases may not be available. For example, suppose that we want to estimate the (remaining) lifetime cancer risk of people in the third age group (18-55yr) in a selected population. In order to determine the dose-response function in this case one needs to specify, among others, the background (or average) level of the (remaining) lifetime risk of cancer in the considered age group. If smoking and socioeconomic statuses impact cancer risk via background risk and are assumed to be discrete variables (with states as above) then the background risk is a function of these statuses and takes six different values. These values have to be extracted from data. If there is not enough data on background risk levels in different socioeconomic and smoking groups within a given age group we propose to replace discrete variables by continuous ones (they could be defined as daily intake of nicotine smoke and number of years of schooling received). This replacement implies that smoking and socioeconomic statuses won’t interact through background risk but they will be included in the dose-response function as additional variables, whose slopes in relation to the health effect studied have to be determined from data. This replacement will also allow capturing and quantifying effects of much larger group of interactions including interactions between intakes of fish constituents and smoking (and/or socioeconomic) status and interaction among smoking and socioeconomic statuses. --Patrycja Gradowska 17:13, 26 October 2007 (EEST) (type: truth; paradigms: science: defence)