Should all risk factors for a disease be treated as confounders even if they are not related to the exposure? I'm doing a study looking at the association between poor food hygiene  (exposure) and stunting (outcome).
Maternal short stature and low infant birth weight are both risk factors for stunting - however, in my opinion they are not related to the exposure.
Should these factors be controlled for?
Sorry, i'm new to statistics and really appreciate any advice!
Many thanks.
 A: While there is no universal definition, the most generally accepted view (for example see McNamee, 2003) is that for a variable to be a confounder it must be: (i) a cause, or a proxy of a cause, of the outcome; (ii) a cause, or a proxy of a cause, of the main exposure; and (iii) unaffected by the main exposure, thus not a mediator, i.e. not on the causal pathway from exposure to the outcome. Therefore, the answer to your question, is no, you should not treat them as confounders.
However, this does not mean that you should not include them in a regression model. Whether or not to control for them crucially depends on the causal process(es) you assume (or is known) for the data. 
If the other variables are risk-factors that are not on the causal path between your main exposure and the outcome (otherwise known as competing exposures) you should include them, since this will increase the precision of the estimate for your main exposure, unless they are highly correlated with each other, in which case one or the other would suffice. However, if they are on the causal path (ie they are mediators) then you should not include them, as this could introduce bias due to the reversal paradox (of which Simpson's paradox, Lord's paradox and suppression are examples - see Tu, Gunnell and Gilthorpe, 2008 for more details on this).
So, in your case, it seems that you can rule out that maternal short stature or low birth weight cause poor food hygiene, thus, they should not be treated as confounders. If maternal short stature and low birth weight are unaffected by poor food hygiene, then you can include them as competing exposures,  However, if poor food hygiene is a cause of maternal short stature or low birth weight, then you should not include them.
References:
McNamee R. Confounding and confounders. Occup Environ Med 2003; 60(3):227-234.
Tu YK, Gunnell D, Gilthorpe MS. Simpson's Paradox, Lord's Paradox, and Suppression Effects are the same phenomenon - the reversal paradox. Emerg Themes Epidemiol 2008; 5:2.
