First, just because it's hard to disentangle correlation and causation, doesn't mean we should throw up our hands and just assume causation into existence.
Second, your objection would equally wreck a study that just looks at correlation only.
Third, presumably quantity matters, even if everybody is getting at least some secondhand smoke, that doesn't mean everyone is getting the same amount. The study found a way to measure this: they are looking at parental smoking, which presumably is connected with higher total amounts of secondhand smoke on average.
The problem with teasing out causation of A leading to B is that you need some mechanism that cuts of the possibilities of B causing A, and some third factor C causing both A and B. It's easiest to do that, if you know that A was caused by the coin flip of the researcher that put subjects into either treatment or control group. But there are other clever approaches, like 'natural designs'.
