Tight Glycemic Control for Type 2 Diabetes (Over Five Years)

Benefits in NNT

None were helped (prevented death)
None were helped (prevented stroke)
None were helped (prevented heart attack)
None were helped (prevented kidney failure)
1 in 250 were helped (prevented limb amputation)
0% were helped by preventing death
0% were helped by preventing stroke
0% were helped by preventing heart attack
0% were helped by preventing kidney failure
0.4% were helped by preventing limb amputation

Harms in NNT

1 in 6 (This is a linear extrapolation to 5 years from 12 months, 3.5% per 12 months; at 10 years the rate would be 35%, NNH 3.) were harmed (severe hypoglycemia requiring hospitalization)
17.5% (This is a linear extrapolation to 5 years from 12 months, 3.5% per 12 months; at 10 years the rate would be 35%, NNH 3.) were harmed by severe hypoglycemia requiring hospitalization
View As:

Efficacy Endpoints

Mortality, Heart attack, Stroke, Kidney Failure, Limb Amputation

Harm Endpoints



Type 2 diabetes is increasingly common in the industrialized world. The most common allopathic treatment strategy is glycemic control using hemoglobin A1C levels as a target. Tight glycemic control attempts to keep glucose levels rigidly controlled (typically HbA1C of 6.5-7.0). Standard control is less rigid and allows higher levels (usually 7.5-8.0). The American Diabetes Association and others maintain that tight glycemic control is life-saving. There are, however, potential dangers including hypoglycemia, and the patient effort, resources, and time consumed by this approach are not insignificant, making the implications of its true impact far-reaching.

This summary utilizes the Cochrane Collaboration comprehensive review, updated in 2013, to address the impact of tight glycemic control compared to standard glycemic control on the outcome of particular interest to patients with diabetes. The review includes a total of 28 trials enrolling nearly 35000 subjects.

Overall there was no improvement in deaths, heart attacks, strokes, or kidney failure. There was an improvement in limb amputation (NNT 250). Unfortunately, the most identifiable harm of tight glycemic control, hypoglycemia requiring medical care, was common (NNH 6).

The time period over which these outcomes were assessed varies, but is typically in the 5-year range. In some cases the median time period is shorter (as little as 2 years) despite some trials for the outcome being substantially longer, though most were tracked for a median of 5 years or more.


These are data estimates from randomized trials, which tend to represent a best case scenario for a drug’s beneficial impact. The included trials were, according to the Cochrane authors, at some risk of bias, which may magnify this effect and raises the possibility that harms may have been more significant than reported here.

There are no data to support the statement that tight glycemic control is lifesaving, and indeed these considerable data suggest that it is not. The median time for mortality outcomes was short, roughly two years, and perhaps over many years of using this approach there may be an identifiable mortality benefit, though if so it is likely to be very small based on the point estimates and sequential analyses performed by the Cochrane authors.

Hypoglycemia is a major problem for diabetics, and can in extreme cases be fatal or neurologically devastating. This problem did not, however seem to increase mortality which is reassuring for those at higher risk of limb amputation, or any others for whom this approach may be used or considered.

These data should not be interpreted to mean that any attempts to control glucose levels have been proven not to work. While it is true that glucose controls are not the cause of type 2 diabetes, but rather a symptom of an underlying metabolic disorder, treating this measurable symptom may have benefits. Unfortunately at this point even this remains unproven, despite being intuitively likely. Trials examining diet or lifestyle approaches versus directed glucose control are badly needed to determine the degree to which treating glucose levels is beneficial in comparison to other approaches.

Finally, we did not address microvascular complications here because we find these not to be patient-oriented. Nephropathy (protein in the urine) and retinopathy (retinal changes on exam) may both be harbingers of later problems, and both are reduced by tight glycemic control, however existing data argue strongly that clinically important outcomes like kidney failure and vision loss occur far less than cardiovascular outcomes and to-date these endpoints have not been shown to be impacted by tight glucose control.

The original manuscript was published in Medicine by the Numbers, American Family Physician as part of the partnership between TheNNT.com and AFP.


David Newman, MD; Peer-Reviewed by James McCormack, MD and Jerome Hoffman, MD


July 20, 2014