Thats a great meta-analysis referenced there (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4951141/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4951141/), but

Thats a great meta-analysis referenced there (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4951141/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4951141/), but the table of studies only contains baseline measures (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4951141/bin/pmed.1002087.s003.pdf https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4951141/bin/pme...) - I can't find the original non-aggregate data on the after-effect (benchmark) measurements of each of the analysed studies. One thing it seems to assume is that effects of carbohydrate reduction are linear, which might not be the case. The study analysed the effect of "replacing 5% energy from carbohydrate with SFA" and found it had no significant effect on fasting glucose - this might be true even if ketosis works perfectly, since on average the trials listed there are medium-to-high carb. It would be very useful if they also had listed a table with the individual study outcome data. Another thing that seems weird is that they deem an increase in insulin secretion capacity as good. Increasing insulin secretion capacity and insulin sensitivity will lower blood glucose, but at the expense of increased obesity - the blood sugar is going somewhere and its either the muscles or fat. So it seems to me that in T2D the increase of secretion capacity only postpones the problem. Ideally what you want is the (fasting) insulin levels to drop while maintaining the same or somewhat lower blood glucose.