Why GPs should strive further on stroke
Blood self-testing strips are a huge cost for the NHS, and some question their benefit. Bandolier editors Dr Andrew Moore and Professor Henry McQuay outline two new RCT studies that make the case for their continued use
In some parts of the world there is controversy about whether blood or urine testing for glucose in patients on oral treatment with type 2 diabetes is worthwhile. There are some large and very positive observational studies demonstrating an association between increased blood glucose monitoring and lower levels of HbA1c and between lower levels of HbA1c and reduced complications and cost. But purists demand randomised trials showing benefit from glucose monitoring.
A systematic review in 20001 concluded there was no evidence of clinical effectiveness of self-monitoring. It was correct in that there were only 230 patients in four trials providing evidence on glycosylated haemoglobin and blood glucose monitoring, insufficient to make a decision.
But three of the four trials had lower HbA1c levels, with an overall reduction of about 0.23 per cent more than control. While this was not significantly different from control, the analysis was confounded by inclusion of results from urine monitoring in one of the studies. Despite the small sample, the trend was for lower HbA1c levels with self-monitoring of blood glucose.
For some, lack of evidence from RCTs trumps good evidence from good observational studies.
New RCTs have been published in the intervening years, which makes it interesting to revisit our thinking on blood glucose monitoring. Two new trials provide almost four times more information than the meta-analysis did.
Blood glucose monitoring
Type 2 diabetes outpatients with BMI over 25, HbA1c values between 7.7 and 10 per cent, and treated with diet or diet plus oral hypoglycaemic agents, were enrolled into a randomised trial of self-monitoring of blood glucose or not monitoring for six months, plus a further six months of follow-up. There was a range of sensible exclusion criteria.
The self-monitoring group was requested to measure blood glucose six times (before and one hour after meals) on two days a week (12 measurements), and to keep results in a diary. They were seen every four weeks, and received a defined algorithm focusing on self-perception, reflection and regulation. The control group received non-standardised counselling with a focus on diet and lifestyle.
Patients in both groups (233 total) were aged about 60 years on average, had an average BMI of about 31, initial HbA1c levels of about 8.4 per cent, and about half were women. HbA1c levels fell in both groups, by an average of 1 per cent with self-monitoring, and 0.5 per cent with control, a statistically greater improvement in HbA1c levels with self-monitoring.
It was suggested that there were three subgroups in the self-monitoring group: those who were continuously successful (58 per cent), with an average end of study HbA1c level below 7 per cent; those with delayed success (18 per cent), who after an initial rise had an average final HbA1c level of about 7.8 per cent; and failures (24 per cent), with an average final HbA1c level of 8.6 per cent.
Self-monitoring led to significant improvements in measures like depression and wellbeing. Cholesterol, triglycerides and microalbumin improved more with self-monitoring, but not significantly so. Self-monitoring patients actually tested themselves 24 times a week, and kept accurate diaries. Over the six-month follow-up, 90 per cent of patients randomised to self-monitoring of blood glucose continued to do so.
Blood glucose monitoring
This randomised study in general practice included patients aged between 40 and 75 years with type 2 diabetes taking oral hypoglycaemic medicines, and with HbA1c levels between 7.5 and 11 per cent. Controls were seen as usual, including laboratory tests every 12 weeks.
The experimental group additionally were required to perform at least six blood glucose measurements a week. The duration was six months, with monthly visits. For analysis there had to be at least two evaluations of HbA1c levels.
Patients in both groups (689 total) were aged about 61 years, had an average BMI of 30, initial HbA1c levels of 9.0 per cent, and about half were women. HbA1c levels fell in both groups, by an average of 0.9 per cent with self-monitoring, and 0.6 per cent with control, a statistically greater improvement in HbA1c levels with self-monitoring. At the end of six months 57 per cent of patients who self-monitored had improved their HbA1c level by more than 0.5 per cent, compared with 47 per cent with control. This implies an NNT for blood glucose self-monitoring over six months of 10 (95 per cent CI 5.7 to 39).
There were no differences between the groups in weight or other blood measurements. There were no serious episodes of hypoglycaemia in the study, but symptomatic or asymptomatic hypoglycaemia was reported in 10 per cent of patients in the self-monitoring group, compared with 5 per cent in the control group.
Pulling it all together
It is possible to combine the results of these two new large studies with four small studies in the previous
In all, self-monitoring of blood glucose has been tested in 1,142 patients in six studies, 80 per cent of whom were in the two new ones published since the meta-analysis. Consistently lower end-of-study HbA1c levels were found for blood glucose self-monitoring in five trials with results, across a range of HbA1c levels (see figure 1). With blood glucose self-monitoring, the weighted average reduction in HbA1c level was 0.82 per cent, 0.30 per cent more than with control.
Other supporting information
Intensive blood glucose monitoring in type 2 diabetes can lead to reduced HbA1c levels. For instance, a randomised trial in Korea of internet monitoring versus normal monitoring led to a doubling in the number of blood glucose measurements made by patients, and an average improvement in HbA1c levels of 0.7 per cent.
A cohort study4 in Turkey examined the impact of a combination of educational programme plus blood glucose self-monitoring on all type 2 diabetic patients with and without retinopathy in one clinic over one year. The average age was 58 years, 60 per cent were women and the average duration of diabetes was nine years. Implementation of the programme led to large falls in HbA1c levels in all groups (see figure 2).
Do averages apply here?
Randomised trial evidence may not be the whole story when looking for evidence. In 2000 there was no convincing RCT evidence that blood glucose self-monitoring was beneficial in type 2 diabetes, but neither did the existing RCTs show evidence that there was no benefit. Absence of evidence of benefit was not evidence of absence of benefit.
Observational studies supported benefit, and even the few small studies we had could be interpreted as telling us that there was likely to be a small benefit. New RCTs confirm that there is indeed a significant benefit of blood glucose self-monitoring in type 2 diabetics. We can be more confident of this now than then, but new RCTs have not changed the overall conclusion, just supported it. Perhaps for some the size of the benefit might be an issue.
The German RCT tells us that while the average benefit may be small, that is because some people have little fall in HbA1c levels perhaps because they do not follow diet or medications. Some patients benefit much more, with larger reductions in HbA1c level and consequent reduction in future complications. This is another example where average results are of less interest than knowing that an intervention benefits some people very much. For those individuals, the cost of blood glucose monitoring is very likely to be offset by lower costs of care over a lifetime.
1 S Coster et al. Self-monitoring in type 2 diabetes mellitus: a meta-analysis. Diabetic Medicine 2000 17:755-61
2 U Schwedes et al. Meal-related structured self-monitoring of blood glucose. Effect on diabetes control in non-insulin-treated type 2 diabetic patients. Diabetes Care 2002 25:1928-32
3 B Guerci et al. Self-monitoring of blood glucose significantly improves metabolic control in patients with type 2 diabetes mellitus: the auto-surveillance intervention active (ASIA) study. Diabetes Metabolism 2003 29:687-94
4 Ozmen, S Boyvada. The relationship between self-monitoring of blood glucose control and glycosylated haemoglobin in patients with type 2 diabetes with and without diabetic retinopathy. Journal of Diabetes and its Complications 2003 17: 128-34
Andrew Moore is honorary professor of health sciences at University College Swansea, and editor-in-chief of Bandolier
Henry McQuay is professor of pain relief at the Oxford pain relief unit and co-editor of Bandolier
Bandolier(www.ebandolier.com) is an independent monthly journal on evidence-based health care. Subscription costs £36 for 12 issues and subscribers receive the print journal three months before articles are available on the website. A subscription form is available on the website or from: firstname.lastname@example.org