I left longitudinal primary care practice before metformin. Back then tight control of Type II diabetes was just about impossible. If we pushed insulin patients just got heavier. In the rare event that we got reasonable control we feared the that occasional hyopglycemia could be deadly.
Times changed. Metformin and subsequent medications transformed Type II DM care. Now it's possible, with a dedicated and disciplined patient, to achieve tight control. Studies of intermediate measures (heart disease, renal failure, eye disease) in patients with both Type I and Type II diabetes showed the value of tight control. Physicians were financially penalized for patients who didn't get good control, and roundly chastised for a lack of energy in pursuing this goal.
There was only one problem. We didn't really know that reducing the rates of nerve, kidney, heart, vessel and eye disease would actually reduce mortality. It certainly seemed that it should...
For decades, researchers believed that if people with diabetes lowered their blood sugar to normal levels, they would no longer be at high risk of dying from heart disease. But a major federal study of more than 10,000 middle-aged and older people with Type 2 diabetes has found that lowering blood sugar actually increased their risk of death, researchers reported Wednesday...
Even the control group, who weren't under "tight" control, had very low glucose levels by the standards of the bad old days. So we're not talking about a return to the dark ages. The question instead is how hard to push, I think this study alone will cause payors to back off on financial penalties for "good" rather than "great" glucose levels.
Incidentally, a similar finding has come up many times over the past 20 years in studies of cholesterol reduction and all cause mortality. We know that reducing cholesterol lowers the risk of heart disease, but it doesn't reduce the risk of death in patients who do not have known heart disease or diabetes (1990:
... Mortality from coronary heart disease tended to be lower in men receiving interventions to reduce cholesterol concentrations compared with mortality in control subjects (p = 0.06), although total mortality was not affected by treatment. No consistent relation was found between reduction of cholesterol concentrations and mortality from cancer, but there was a significant increase in deaths not related to illness (deaths from accidents, suicide, or violence) in groups receiving treatment to lower cholesterol concentrations relative to controls (p = 0.004).
Later studies suggest that, on balance, persons with diabetes or known vascular disease benefit from simvastatin. Maybe a lot. There's still the suspicion that the harm may outweigh the benefit for non-diabetic patients with no known vascular disease (primary prevention) though.
These are tough questions, and in this domain my much loved animal model studies aren't that helpful. All cause mortality can only be studied in humans.
2/15/2008: It occurred to me that results like these could suggest the possibility of unsuspected quality issues with the medications we consume.