Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

SMBG in type 2 diabetes

Non-fatal endpoints
All cause mortality

Bandolier has revisited this topic on a number of occasions, whenever there appears to be something useful to add (Bandolier 84, 93, 134). The problem has been that in type 2 diabetes where insulin is not used, overwhelming evidence of benefit with self-monitoring of blood glucose (SMBG) is absent. We have a rash of small, short, largely pointless RCTs that tell us little. We have a number of observational studies, some strong, that indicate that SMBG is associated with lowering HbA 1c, and that this is a good thing that might save money. We can now add to these observational studies another long-term study [1] pointing to better clinical outcomes with SMBG.


Information on 3,000 patients was obtained from 192 randomly selected practices in Germany. Patients included had to be over 45 years of age at time of diagnosis of type 2 diabetes (1995-1999), and have full information on therapy at the time of diagnosis and for at least one subsequent year. Patients were evaluated from diagnosis to an outcome or to the end of 2003.

Predefined endpoints were nonfatal endpoints of myocardial infarction, blindness (one or both eyes), or renal failure requiring dialysis. All cause mortality was the other endpoint.


Patients had an average age at baseline of 62 years, an average BMI of 30 kg/m 2, a fasting blood glucose of 9.3 mmol/L, and HbA 1c of 7.7%. The average follow up was 6.5 years. After diagnosis of type 2 diabetes, over half of patients had no specific therapy, though this changed over the course of the observation period, with increases in insulin and oral hypoglycaemic therapy (Table 1).

Table 1: Treatments for type 2 diabetes after diagnosis and at the end of observation period (mean 6.5 years)

Percent receiving treatment
No pharmacological treatment
Insulin monotherapy
Insulin + hypoglcaemic
Metformin (alone or + others)

Patients were allocated to the SMBG group if SMBG was documented for at least one year during the observation period and before a non-fatal endpoint. Overall, 45% of the total of 3,268 patients were classified as using SMBG, though the cumulative initiation of SMBG meant that most were using it by the end of the observation period (Figure 1). Patients classified as using SMBG were significantly younger (by four years), and had higher fasting blood glucose (10 vs 8.7 mmol/L) and HbA 1c (8.1 vs 7.2%).

Figure 1: Cumulative initiation of SMBG during full eight years of observation

For all patients, both fasting blood glucose and HbA 1c levels fell over the first year. In those not using SMBG they rose slightly over subsequent years, while in those using SMBG the levels fell slightly. Fasting blood glucose and HbA 1c levels were always significantly higher in SMBG users (HbA 1c 7.2% vs 6.9% at end of observation, for instance).

Non-fatal endpoints

These were experienced by 293 patients (9% overall), with an incidence of 7.2% (107/1479) in those using SMBG and 10.4% (186/1789) in those not using SMBG, and the difference was present over the whole of the observation period.

In a subgroup of 2,500 who never received insulin, the incidence of non-fatal events was 6.7% (58/808) in those using SMBG and 10.4% (177/1707) in those not using SMBG. This reduction was due to a large decrease in macrovascular endpoints (heart attack, stroke, 6% vs 10%), while microvascular endpoints like amputation, dialysis, loss of vision occurred almost twice as often in the SMBG group (2.5% vs 1.5%).

All cause mortality

One hundred and twenty patients died (3.7% overall), with an incidence of 2.7% (41/1479) in those using SMBG and 4.6% (79/1789) in those not using SMBG, and the difference was present over the whole of the observation period.

In the subgroup never receiving insulin, death occurred in 2.5% (22/866) in those using SMBG and 4.3% (177/1649) in those not using SMBG.


Patients using SMBG were not the same as those who were not. They had worse diabetes, with higher fasting blood glucose and HbA 1c, and this difference remained over the course of observation, despite the use of SMBG. In consequence, the higher rate of microvascular complications was not unexpected.

The reduction in heart attacks and strokes, and in all cause mortality, for those using SMBG over those who were not, despite have less well controlled blood sugar and HbA 1c is interesting. After adjusting for possible confounding the study showed a 32% reduction in non-fatal endpoints and a 50% reduction in mortality. These are hard, important, clinical outcomes in a large cohort observed for a long period.

Of course, the patients were not randomised, but we will wait in vain for a randomised trial of this size and duration. Patients were assigned to SMBG by their doctors, and the evidence is that the doctors made a pretty good fist of it. They chose patients at greater risk, with worse diabetes, and their results were pretty good. They appeared to be maximising benefit while minimising costs. There may be lessons in this tale.


  1. 1 S Martin et al. Self-monitoring of blood glucose in type 2 diabetes and long-term outcome: an epidemiological cohort study. Diabetalogia 2006 49: 271-278.

previous or next story