Intended for healthcare professionals


Self monitoring of blood glucose in type 2 diabetes

BMJ 2007; 335 doi: (Published 19 July 2007) Cite this as: BMJ 2007;335:105
  1. Simon R Heller, professor of clinical diabetes
  1. School of Medicine and Biosciences, University of Sheffield, Sheffield S10 2RX
  1. s.heller{at}

    Clinicians should stop patients doing this if it has no benefit

    Self monitoring of blood glucose costs the NHS more than £100m (€150m; $200m) each year and the cost is rising.1 For many people with insulin treated diabetes and their families, blood glucose self monitoring is an essential tool, enabling them to confirm hypoglycaemia or high glucose concentrations and to take corrective action. Yet large numbers of patients diligently record the results and then do nothing with them.

    In this week's BMJ Farmer and colleagues report the results of a primary care trial in patients with well controlled type 2 diabetes who were not taking insulin. They found no evidence of an effect of blood glucose self monitoring on glycaemic control, with and without structured education, compared with usual care.2 This study confirms that the contribution of self monitoring is not clear in type 2 diabetes, particularly for those treated with diet alone or oral agents other than sulphonylureas. Furthermore, there is wide geographical variation in the use of self testing by such patients.3

    One view is that providing such technology to diabetic patients treated with tablets or diet is a waste of time and money, because there is little an individual can do with the results.4 Others believe that the information provided by blood glucose testing is a powerful motivating factor,5 encouraging self management of diabetes by allowing patients to measure directly the impact of their behaviour, such as the effect of eating on postprandial glucose or the glucose lowering effect of exercise. Some,6 7 but not all,8 observational studies have shown that, even in patients treated by diet alone, those who measure their blood glucose more often have better outcomes, including HbA1c concentration and mortality. Such positive associations may simply show, however, that those who are highly motivated (reflected in the frequency of blood testing) are likely to do well in the long term.

    A limited number of prospective studies have randomised patients to blood glucose self monitoring or to no monitoring. A recent meta-analysis reported a modest mean improvement in HbA1c concentration of around 0.3%, but the confidence intervals were so wide that this difference was not significant.9 Importantly, the meta-analysis comparing blood and urine testing found no difference in HbA1c concentrations. This suggests that blood glucose self monitoring has little effect on glycaemic control in patients treated with diet or metformin. Structured education on using the information obtained from self monitoring to adjust insulin dosing, however, leads to sustained improvements in glycaemic control in type 1 diabetes,10 and this might also apply to those with type 2 diabetes.

    In the diabetes glycaemic education and monitoring trial (DIGEM), Farmer and colleagues directly test the contribution of blood glucose self monitoring on glycaemic control, with and without structured education, in 450 people in primary care with diabetes treated by tablets or diet, with relatively tight glycaemic control.2 Patients were randomised to receive usual care (and were asked not to test their blood), basic information on self management and limited blood self testing, or training in self management and encouragement to undertake more intensive blood monitoring. At one year, HbA1c concentration was unchanged in the usual care group, and marginally and equally improved in the other two groups, with no significant difference among the three.

    The trial was well designed and conducted but had some limitations. Patients who were already testing their blood more than twice a week were excluded (possibly removing those who found glucose monitoring valuable and leaving individuals who had already used and rejected it). Furthermore, only around 15% of those eligible entered the study, thus limiting the generalisability of the findings. In one arm of the trial the authors embedded blood glucose self monitoring within an educational intervention designed to enhance self management, yet glycaemic control did not improve. This may be because patients with relatively tight control were included, in contrast to previous studies, or because the intensive intervention was ineffective. Indeed, fewer patients randomised to the intensive arm ended up using a glucose meter than in the less intensive arm, an unexpected outcome among patients who were trained to monitor more frequently. Finally, patients seem to prefer blood glucose monitoring to urine tests,11 and different conclusions might have been reached if patients' views had been taken into account.

    The DIGEM trial has shown that in patients with established diabetes relatively well controlled by oral drugs who monitor blood glucose infrequently, little is gained in promoting blood glucose testing even in conjunction with an education programme.2 Whether self monitoring is useful in patients at diagnosis and whether it offers advantages over urine testing (which is much cheaper) remains uncertain. None the less, the results of this study should encourage clinicians to discuss the value of glucose testing with their patients and give them the confidence to discontinue it if it is providing no benefit.


    • Competing interests: SRH is principal investigator in an ongoing randomised controlled trial comparing blood glucose to urine testing in newly diagnosed individuals with type 2 diabetes.

    • Provenance and peer review: Commissioned; not externally peer reviewed.


    View Abstract