I think that's more of a category/behavior thing than the result of a product.
Overweight people eat a lot of unhealthy things, then switch to diet drinks because they think it will help them cut back. It's like saying "Race car drivers die from auto accidents, but they use seat belts. Therefore, seatbelts may be dangerous."
I understand an argument from causation; Though your inability, or unwillingness, to accept the potential complexity of the situation shows that you may not fully appreciate the depth of the conversation.
To use your race-car-driver metaphor.
It's like arguing that armor-all causes health problems; evidence is in the increase in health problems among race-car drivers.
You counter Armor-all does NOT cause cancer: FACT!
I say: well no, but it is linked with increases in OCPD and there is a good theoretical reason to think that it would cause OCPD.
You counter: yea, but race-car drivers inhale tire-smoke, car-exhaust and lots of other shit; so OCPD is caused by something other than armor-all.
If you think that pointing to potential third-causes counters a strong theoretical argument you've lost the ability to engage in reasoned discourse. Given the strength of the correlation and given strong theoretical reasoning surrounding why the correlation may be causal; you must now not only offer an equal or better theoretical third-factor, but you must show that they are mutually exclusive.
That is: just because the race-car driver breaths tire-smoke doesn't mean armor-all isn't doing something to him; you need evidence that there is no effect of armor-all outside of the effect that you would expect from tire-smoke alone.
Such is where the conversation is at: not that we don't know if diet soda is correlated with diabetes, but that while other factors are part of why diet-soda drinkers get diabetes nothing has been presented to show that those factors exclude diet soda from having an impact. This is problematic for the 'diet is fine' argument because of the theoretical link between tricking your brain into thinking it had something sweet (thus releasing insulin without sugar for it to process) and diabetes.
Strong theory is the basis upon which we assume causation until we have experimental evidence to the contrary or better theory.