[REDLAN]

Thanks some criticism

always useful helps to tease out the arguments

The dropout rate was for the ENTIRE study group, no differentiation between the ketogenic or the low calorie group.

Yes this is true they do quote the dropout rate for the entire group, which is not very helpful. And you are right I have made the assumption that it was mostly in the ketogenic diet group - The reason for this is that from reading past studies, dropout rates for low carb diets are typically higher than for low fat diets - the israeli study oft quoted displays this characteristic. From reading about ketogenic diets used in children to control epilepsy the authors comment about the high drop out rates. The authors of this study allude to the fact that the high drop out rates is mostly in the ketogenic group with

The diet is not easy for everybody. "This is a therapeutic diet for people who are sick," says Westman.

It is I believe a reasonable assumption - The problem for high drop out rates in a study is that it makes the study results much less reliable. When at the end of 6 months nearly half you cohort regardless of which arm has disappeared this is an issue. It is the highest dropout rate for a diet that I have seen.

Firstly, if the ketogenic diet caused improvement in A1c because it caused weight loss, then to all intents and purposes the ketogenic diet caused the improvement of A1c. Why is that invalidated by being a short chain of events?

because at the 12 month mark weights between the 2 tend to equalise, and so do HBa1C values. The advantage for the ketogenic diet is lost. This is borne out by the studies cited by the Cochrane report as being the best available.

Westman claims...

"Low glycemic diets are good, but our work shows a no-glycemic diet is even better at improving blood sugar control,"

Really???

would the ketogenic diet produce better blood sugar control if there was no weight loss? This study can not answer this crucial question. The reason it matters is because a) ketogenic diets are hard to stick to, and b) they tend to produce the same weight loss at the 12 month mark as a low fat diet (which is easier to follow).

if the improvement is driven by weight loss alone, then you would be better sticking to a diet that is easier to follow, and produces similar results.

Improvement of Hba1c with weight loss in type 2 is not a speculation it is well known that weight reduction causes an improvement in blood glucose control. And it correlates extremely well - greater weight loss correlates with better glucose control.

The only salient part of the "results" section would be "(b) Studies that assessed low-fat diets versus moderate fat or low-carbohydrate diets", yes?

There is no such assertion about the 12+ month mark, here.

from the cochrane report, and the section that looked at dietary advice that did not differ in intent to lose weight.

Hockaday 12 months mean changes in Hba1c were not significant.

Milne at 18 months the difference in HBa1c between the 2 was 0.1% in favour of the low carb group.

Like I said these results unreliable as they are suggest that there is little difference between low carb or low fat. If you are able to draw a different conclusion then please feel free.

The point of quoting Nielsen is that this researcher has managed to run a cohort using a low carb intervention which has shown sustained weight loss, and sustained and significant improvement in Hba1c. He/she doesn't have a proper control group, which is a shame,as it could be very good evidence in favour of a low carb diet. The 44 month length is significant too.

What about short term gains?

The problem I have with this study is that it is all about short term gains. However people do not have diabetes for just 6 months.

In the event the ketogenic diet results in a much faster normalisation of A1c, why is that invalidated?

the dropout rate is the issue. If it is no better than a low fat diet at 12 months, which is easier to follow, why would you want to start your patients on a ketogenic diet when half would dropout before 6 months, and you'd have to start again with some other intervention? Wouldn't it be better to start everyone on the low fat diet to start with? Unless that is you can show that lowering the amount of carb really does improve glucose control all on it's own, which this study doesn't do and nor do the cochrane ones either.

This is what the cochrane report calls for - it's very disappointing, and I think somewhat scandalous that there isn't better research to answer such a basic question as to what kind of diet would be best to treat type 2 diabetes.