Self monitoring of blood glucose in type 2 diabetesBMJ 2008; 336 doi: http://dx.doi.org/10.1136/bmj.39538.469421.80 (Published 22 May 2008) Cite this as: BMJ 2008;336:1139
- Martin Gulliford, professor of public health
In the accompanying papers, O’Kane and colleagues report a randomised controlled trial of blood glucose self monitoring in people with newly diagnosed diabetes (ESMON trial) and Simon and colleagues report a cost effectiveness analysis that ran alongside the previously published blood glucose self monitoring in type 2 diabetes (DiGEM) trial.1 2 3
The question of whether people with type 2 diabetes who do not use insulin should monitor their own blood glucose has been the subject of some lively exchanges in the BMJ for more than 10 years. In 1997, Marilyn Gallichan wrote, “The inappropriate use of self monitoring of glucose is wasteful of NHS resources and can cause psychological harm. There is no convincing evidence that self monitoring improves glycaemic control, nor that blood testing is necessarily better than urine testing.”4 These challenging observations were disputed by correspondents who suggested that self monitoring might help people with diabetes improve blood glucose control, avoid hypoglycaemia, improve quality of life, and enhance long term prognosis.5
The NHS Health Technology Assessment programme responded by commissioning a systematic review and a clinical trial.3 6 7 In the systematic review, a meta-analysis of four randomised trials showed that in subjects with type 2 diabetes who were not treated with insulin, self monitoring of blood or urine glucose gave an estimated mean difference in glycated haemoglobin (HbA1c) of −0.25% (95% confidence interval −0.61% to 0.10%) when compared with people who did not self monitor.6 A meta-analysis of three randomised controlled trials showed that, when compared with urine testing, blood glucose self monitoring did not improve glycated haemoglobin.6 In a second meta-analysis published in 2005, the estimated difference in HbA1c between blood glucose self monitoring groups and no monitoring groups was −0.39% (−0.56% to −0.21%),8 but two of the studies included in this meta-analysis had recognised methodological problems.9 Despite apparently minimal clinical benefit, the use of blood glucose self monitoring has increased.10
In 2007, the BMJ published the results of the DiGEM trial of blood glucose self monitoring in type 2 diabetes.3 The investigators responded to criticisms of earlier trials by including clearly structured advice to participants on how to use the results of self monitoring to adjust their diet, level of physical activity, and drug adherence. The difference in HbA1c associated with more intensive self monitoring compared with no monitoring was −0.17% (−0.37% to 0.03%) over 12 months. This result is consistent with the initial meta-analysis,6 and it suggests that blood glucose self monitoring has little or no effect on medium term blood glucose control in type 2 diabetes not treated by insulin. Mild hypoglycaemic symptoms were uncommon and were more often recorded in the self-monitoring groups. Only one patient in the control group had a serious hypoglycaemic episode.
The BMJ published 33 rapid responses to the DiGEM trial, and nearly all criticised the trial’s conclusions. Some respondents argued that self monitoring would be especially beneficial in people with newly diagnosed diabetes. Another concern was that increased detection of hypoglycaemia in the self monitoring groups might represent a true clinical benefit. People with diabetes expressed concerns that restriction of self monitoring would limit their freedom to manage their own illness and deprive them of perceived benefits.
O’Kane and colleagues’ ESMON trial found no significant difference in HbA1c between people with newly diagnosed diabetes allocated to blood glucose self monitoring and controls managed according the same well defined algorithm without self monitoring at 12 months’ follow-up.1 The two groups showed no significant difference in hypoglycaemia. Patients who were allocated to self monitoring reported greater self rated depression than controls.
Simon and colleagues’ cost effectiveness analysis of the DiGEM trial confirms that subjects in the self monitoring group had reduced self rated quality of life, perhaps as a result of increased anxiety and depression associated with blood glucose self monitoring.2 It also shows that the additional healthcare costs associated with blood glucose self monitoring amount to about £90 (€114; $180) for each patient each year. This is mostly attributable to the costs of monitoring materials.
These results put the debate concerning self monitoring into a new ethical perspective. The DiGEM economic evaluation and the ESMON study draw attention to potential harms from self monitoring. Self monitoring is associated with reduced quality of life and increased depression for people with type 2 diabetes. The healthcare costs of self monitoring have been estimated comprehensively and, with diabetes now affecting some 3-4% of the population of the United Kingdom, the total healthcare cost of self monitoring may now exceed £100m each year in the UK.11 This represents a substantial opportunity cost in terms of alternative interventions that might have improved the health of people with diabetes. For patients, self monitoring carries an opportunity cost in terms of the attention that they might have given to more effective disease control measures aimed not just at blood glucose but also at blood pressure, cholesterol, smoking, body weight, and physical activity.
The statistician Austin Bradford Hill held as an ethical principle that health interventions must be evaluated before they are introduced into practice (WW Holland, personal communication, 2008). Self-monitoring of blood glucose in type 2 diabetes provides an example of the difficulties that arise if this principle is not followed.
It is 25 years since the cardiologist, John Hampton, pronounced the end of clinical freedom, observing, “if we do not have resources to do all that is technically possible, then medical care must be limited to what is of proved value and the medical profession will have to set opinion aside.”12
Competing interests: None declared.
Provenance and peer review: Commissioned; not peer reviewed.