Although people with diabetes mellitus may prefer blood glucose self-monitoring, evidence does not suggest improved outcomes relative to urine testing

OBJECTIVE

To assess the clinical- and cost-effectiveness of different methods for monitoring blood glucose control in diabetes mellitus.

METHOD

Systematic review of any intervention study, but with particular focus on randomised controlled trials (RCTs).

LITERATURE REVIEW

Comprehensive search, including MEDLINE and EMBASE, generally completed up to 1999. Hand searches of the journals Diabetes Care and Diabetic Medicine (both 1990-99) and contact with manufacturers and British Diabetic Association were also undertaken.

INCLUSION AND EXCLUSION CRITERIA

Studies considering self-monitoring by participants and near-patient or laboratory testing in healthcare settings were included. Participants could have any type of diabetes mellitus and outcomes (satisfaction and quality of life) were particularly targeted.

DATA EXTRACTION AND SYNTHESIS

Retrieved papers were evaluated for quality by two independent reviewers. Data were abstracted and synthesised using meta-analysis where possible.

MAIN RESULTS

Evaluation of blood glucose monitoring devices

There is no standard protocol for evaluating blood glucose monitoring devices. Published evaluations have often only evaluated a limited number of aspects of meter performance and have not always used appropriate methods to analyse the reliability of measurements.

Self-monitoring in type-2 diabetes mellitus

Eight RCTs were initially identified, but two were formally excluded from the meta-analyses. A random effects meta-analysis of data from three studies showed that the difference in glycosylated haemoglobin for those performing self-monitoring of blood glucose compared with those performing urine testing was −0.03% (95% CI, −0.5% to +0.5%). Published information on patient outcomes and the avoidance of hypoglycaemia was extremely limited. Blood testing was noted to be more costly than urine testing.

Self-monitoring in type-1 diabetes mellitus

Eight controlled trials were identified. Only one clearly suggested a benefit of blood testing on glycosylated haemoglobin. A meta-analysis of data from four studies that compared blood monitoring with urine monitoring in children or adults suggested a mean difference in glycosylated haemoglobin of approximately −0.6% (95% CI, −1.1 to −0.1). Blood testing was noted to be more costly than urine testing but was preferred by participants, possibly because it provided better information.

Self-monitoring in diabetes mellitus in pregnancy

Eleven papers, including five RCTs, were identified. The studies generally included small numbers of subjects. Maternal and foetal outcomes appeared to be as good with home self-monitoring as with hospital inpatient admission in late pregnancy, but the studies did not have sufficient power to give conclusive results.

Laboratory and near-patient testing

Results from the Diabetes Control and Complications Trial (DCCT) in type-1 diabetes mellitus and the UK Prospective Diabetes Study in type-2 diabetes mellitus have demonstrated the clinical effectiveness of using glycosylated haemoglobin estimations to monitor blood glucose control. Data from the DCCT suggest that the overall package of intervention employed would have acceptable cost-effectiveness. No unconfounded studies have addressed the optimal testing frequency for glycosylated haemoglobin. Near-patient testing for glycosylated haemoglobin is being developed, but it is too early to judge its value.

AUTHORS' CONCLUSIONS

A standard protocol should be drawn up for conducting and reporting evaluations of blood glucose monitoring devices. Blood glucose self-monitoring is well established in clinical practice but the optimal use of the technique has not been established. Present evidence suggests that it may not be essential for all people with diabetes mellitus.