I wanted to present this info: http://carbsanity.blogspot.com/2013/09/low-hba1c-is-as-bad-or-worse-than-high.html
In case the website is not allowed I will post an image.
The data is for NON diabetics.
The mortality likelyhood is modelled by 8 different variants of 3 models for non-diabetics.
The lowest mortality rate among non-diabetics is the group with HbA1C between 5.0 and 5.4, NOT the groups lower than 5.0 as Bernstein believes. The mortality rate of those with an HbA1C between 4.5-4.9 is higher than that of those between 5.0 and 5.4 in all of the model variants. The mortality rate of those lower than 4.5 is as high as for those above 6.0.
the point is that the lowest mortality rate is NOT skewed above 5.0 because of deaths due to diabetic hypoglycemia, because this study does not include diabetics. It COULD be slightly skewed due to very low HbA1C's being related to anemia, but the point remains that there is no evidence that anybody lives longer by taking their HbA1C down below 5.0. Model 2e though I believe accounts for anemia and still shows higher mortality prediction rate at 4.5-4.9 than at 5.0-5.4
Since the 4.5-4.9 group has a little lower increased mortality than the 5.5-5.9 group, it looks to me like the absolute bottom of the graph of mortality to HbA1C would bealmost, but not quite at the very bottom of the 5.0-5.4 group, say 5.1.
It is odd that the groups with HbA1C below 5.0 had higher average mortality but a broader range of mortality that extended from lower than that of those at 5.0-5.4 to higher than that. This might indicate that there is a division between non-diabetics who are below 5.0 into those who are below 5.0 due to pathological reasons.
I will also point out that another study cited shows that ALL CAUSE mortality for non-diabetics is lowest for on the graph for FASTING blood sugar of 99 which is considered by many doctors to be a sign of likely future diabetes. The study did exclude those who were actually diabetic though.
I wanted to add that Dr. Bernstein was clearly ahead of his time, or even the current times with regard to diabetics, but he personally advocates that a normal fasting blood sugar is 83, and an A1C should be below 4.6, and possibly as low as 4.2 (which he claims to have reached). I think he is a genius, and did took the best possible approach for diabetes when insulin was slow and blood sugar tests were slow and not up to the minute. Lower insulin doses for lower carbs reduce the margin of error that can lead to killer hypos, but IF an insulin using diabetic can avoid hypos the evidence still supports the best health in the low end of the 5.0-5.4 HbA1C range, and if someone does not become diabetic, the best fasting blood sugar (lowest mortality) is in the 90s, and clearly NOT in the 80s.