That's not how it works. That research shows only correlation does not mean research shows ~causality. Correlation does not imply causation, but it definitely suggests it absent other explanations.

It would be hard to study causality in this instance because of medical ethics. You cannot assign a treatment that you believe to be more likely to cause an infant to die, so RCTs are impossible in this domain.

http://www.medpagetoday.com/Pediatrics/GeneralPediatrics/164... -- co-sleeping babies are 21.7 times more likely to die from SIDS than non-co-sleeping babies, according to this study. That definitely doesn't prove causality. But the correlation is certainly waggling its eyebrows suggestively in causality's direction.

From the study you cite: "Much of the elevated risk appeared to be attributable to drug and alcohol use among the parents".

There are parents that co-sleep because they are disorganized, drunk, and so forth, not even able to put the children in their bed. That's the reason why probably there is no causality.

Wait, so a large portion of the risk comes from drugs and alcohol, therefore you are going to ignore whatever risks come directly from co-sleeping? Am I misunderstanding your argument?

I think what he's saying is that co-sleeping provides benefits to his family. While there are risks, it's difficult to discern what the increased risk is in the absence of drugs and alcohol. He feels that there may not be a causal link between SIDS and co-sleeping for families who take co-sleeping seriously and take steps to minimize the known factors in SIDS.

It's not a great argument necessarily, but people accept additional risk of injury to their kids all the time if the benefits outweigh the risks.