That study is pretty good... except for the gaping conceptual flaw in its hypothesis, the Evel Kneivel leaps of logic in the conclusion, and the generally piss-poor adherence to a reasonable level of empiricism.
Translation: It's a pile of predeterministic crap that violates the first rule of statistical research. (Correlation ≠ Causation)
In summary, our findings suggest that high consumption of diet soft drink is independently associated with increased risk of PDR in patients with diabetes. Our study adds to the growing body of evidence reporting on the harmful effect of artificially sweetened beverages on a range of health outcomes.
They even say it right in the conclusion... the consumption of diet soft drinks is an independent variable. There is no established dependency between the variable and the result, only a simple association. You could just as legitimately say that the consumption of water is associated with increased risk of PDR. So what do they do? In the very next sentence go right to the "well this adds to the evidence that artificial sweeteners are
de debbil!".
Given the proliferation of low-calorie beverage choices I would have been absolutely stunned if there
wasn't a highly significant statistical link between consumption of low-calorie beverages and people with diabetes-related illnesses. They're testing a single variable which has an inherent association with the behavior in question, so close to 1:1 that the likelihood of finding a diabetic who hasn't consumed an artificially sweetened beverage is essentially zero. That's like testing the link between breathing air and dying.
What they did wasn't research, it was a predetermined result in search of just enough study to ostensibly legitimize their position. They haven't provided, in any way, shape, or form, a single iota of evidence that there is causality involved. A responsible researcher would have ended their summary with something like "We see an association between the consumption of artificially sweetened beverages and PDR. However, this association is independent and should be treated as such pending additional research into confirmation or denial of dependent association."
Brad