Could salary be used as a surrogate for smoking when sample data are lacking?
In epidemiology cohort studies, information on smoking status is often times lacking for the study sample, which makes it challenging to characterize the effects of other risk factors on endpoints that are causally related to smoking (e.g., lung cancer mortality). Given the strong inverse reported for smoking and income, could salary be used as a surrogate for smoking when direct data are lacking? or would potential relationships between salary other risk factors (e.g., health care access) make such an approach non-viable?
anon-24349
No, I could not imagine under what mental syllogism this association could have been suggested.
B Cox
Of course not. I am surprised the question is even being asked. Relationships with salary are themselves important without investigative desperation to use it to guess smoking status (over the previous 30 years would be needed for most relationships to health).
GAD RENNERT
No. While possibly "true" in some communities, there is absolutely no correlation between income and smoking status at a yes/no level.
Dr Livia Carvalho
No, smoking is a health behaviour and salary a measure of social economic status they are associated but biology is different.
DonO
Absolute nonsense, especially without regard to other social and geographic factors. Would such data pertain to, e.g., Kentucky vs. Nwe York City, or Japan or Denmark vs Canada? It is pseudo-relationships like these, based on confirmation bias and snobbism that blacken the cause of science. As much as, as a toxicologist and pathologist, I believe smoking might be correlated with abject stupidity, income and stupidity are very poorly correlated.
Babak Khoshnood
I am going with the consensus of course and say No. In general, inclusion of any proxy / confounding variables should be guided by the study question and the causal diagram that is supposed to represent the causal links between the outcome, main exposure or predictor of interest and the confounding or intermediate variables
Kerry S Wilson
Salary is one possible factor, particularly in low-income countries with tobacco taxes. Still, the strength of the association and its direction with smoking varies greatly from country to country and may not be linear creating further complications with use as a proxy.
Fernando SL
It's not a good surrogate, there are many confounding issues, and Country, educational and social status, even cultural and religious issues that may influence the results.
SMG
Not without a lot of other covariates, and even then the method will raise eyebrows. There are just too many things that are related to smoking as well as things that are related to income.
Daya
Income cannot be taken as a surrogate for smoking. The statistical association between income and smoking may not be a real association.
ВЕ
Correlation may depend on geographical location and could be stronger in countries where smoking is more prevalent and less stigmatized than in the US. Then it would be a marker for education and socioeconomic status more than for income. But these are just hypotheses. It would be a good study in itself, and there must be survey data available to see the correlation.
J Mal
No, that would be confounding to assume income is correlated with smoking status; there are multiple factors that can be associated with smoking status to different extent but none are reliable to be used as a surrogate factor. I would like to see the study that claims there is a strong correlation between income and smoking status as smoking is seen across all social groups.
Trudy