Brief report
Individual patient data meta-analysis of trials of self-monitoring of blood glucose in non-insulin treated type 2 diabetes: Protocol for a systematic review

https://doi.org/10.1016/j.pcd.2009.03.002Get rights and content

Abstract

Background and aim of the study

Recent trials have provided differing estimates of the benefits of self-monitoring of blood glucose (SMBG) for non-insulin treated patients with type 2 diabetes. Designing and conducting randomised trials to demonstrate the efficacy of complex interventions is challenging. Variations in the components of the intervention delivered, target population selected, and reporting methods used have limited the conclusions drawn in recent systematic reviews. We will systematically examine these factors within the trials of SMBG to enable further investigation of the data through an individual patient data analysis (IPD).

Methods/design

The IPD analysis will include data from randomised trials comparing blood glucose self-monitoring and decisions about self-management versus a control group with standardised or routine care. The data requested for each trial will include outcomes (HbA1c, blood glucose and quality of life), potential moderators of effect (e.g. demographic variables, clinical data and psychosocial factors) and intervention descriptors. The primary outcome of interest will be HbA1c. Secondary outcomes include alternative measures of glycaemia, cardiovascular risk factors, persistence with monitoring, and measures of health status, quality of life and psychosocial factors. Analysis of patient sub-groups defined by age at randomisation, gender, prior use of monitoring and health status will be carried out. An intention to treat analysis will be performed and assessed for clinical and statistical heterogeneity.

Discussion

With current uncertainty about the extent of benefit from SMBG for non-insulin treated patients with type 2 diabetes, this study will provide the best estimates to date of overall effectiveness, effectiveness within potential target populations, and optimal components of the intervention.

Section snippets

Background

Over the next 30 years, the number of people worldwide with diabetes is projected to double to 300 million [1]. Diabetes care is a major contributor to health service costs, and research into better management of this long-term condition is an important priority.

Glucose control together with blood pressure and lipid control is one of the main goals of diabetes management. For patients taking insulin, self-monitoring of blood glucose (SMBG) has been generally accepted as a means of improving

Methods/design

We propose to carry out an individual patient data analysis including randomised trials comparing a patient group using self-monitoring of blood glucose with a control group.

Objectives

  • (i)

    To estimate the effect of SMBG on glycaemic control for patients with non-insulin treated type 2 diabetes;

  • (ii)

    to estimate the effect of self-monitoring on other available study outcomes including cardiovascular risk factors, quality of life measures, treatment satisfaction or related measures and scales;

  • (iii)

    to estimate the effect of self-monitoring of blood glucose in pre-defined sub-groups of patients by baseline HbA1c, age and duration of disease at the outset;

  • (iv)

    to explore the interaction between

Criteria for trial inclusion

All published and unpublished randomised controlled trials will be eligible for inclusion where: (i) the main intended effect of at least one of the trial interventions was to improve disease outcomes through incorporating SMBG into self-management regimen of non-insulin treated patients with type 2 diabetes, (ii) the trial was unconfounded with respect to this intervention, no other differences in modification of risk factors or behaviour between the relevant treatment groups were intended and

Search strategy for identification of studies

Trials meeting the inclusion criteria will be identified using a search strategy for relevant trials adapted from the collaborative Cochrane Metabolic and Endocrine Disorders Review Group and used in previous systematic reviews [5]. As with other recent systematic reviews [4], [5], databases will be searched from 1996 onwards and include MEDLINE, EMBASE and the Cochrane Library. Additional trials will be identified from other systematic reviews of SMBG, conference proceedings, abstracts, the

Trial eligibility and methodological quality assessment

Two trained researchers will independently assess the articles for inclusion at both the title and abstract phase and the full-text review phase of searching for eligible trials. Disagreements will be resolved by discussion with a third reviewer if unsolved after contacting authors. Trials will be assessed for methodological quality by assessing method of randomisation, allocation concealment, use of masked outcome assessment and follow up rates. Investigators of trials identified as eligible

Data collection

The following data will be requested from participating trialists for each participant that meet the inclusion criteria: date of randomisation, age, ethnicity, demographic, physical, and biochemical measures at randomisation, allocated treatment and outcomes recorded including measures of glycaemia, details of the methods of HbA1c analysis, other cardiovascular risk factors, psychosocial factors and quality of life and health status measures. Compliance with intended frequency of SMBG will be

Definitions of interventions and treatments

The interventions used in the included trials will be reported using an approach in which factors likely to promote self-management are identified, where possible based on a model informed by behavioural theory [3]. In addition, the frequency of recommended SMBG testing is likely to vary between trials and this may contribute to the effect. Inclusion of trials will therefore not be restricted by minimum frequency of testing. Treatment strategies within the trials will be categorised by

Data transfer

Data will be accepted in electronic or paper form. Optimal format will be agreed, but trialists will be able to provide data in the most convenient way open to them provided details of coding are supplied. To ensure accurate pooling of data, items will be consistently derived, labelled and coded.

Individual patient data from each trial will be transferred to a secure database held at Oxford University. Data from each trial will remain the property of each individual group.

Data validation strategy

Original data will be archived with copies made to be used in the analyses. Trial details and summary measures will be cross-checked against published articles by two reviewers. Consistency checks will be applied with errors and inconsistencies discussed with the original trialist.

Outcome measures

The primary outcome will be mean difference in HbA1c from baseline to final measure between intervention and control groups, or, where not available, differences between mean level of HbA1c in intervention and control groups at follow up. Secondary outcomes will include alternative measures of glycaemia, cardiovascular risk factors, persistence with monitoring and measures of health status, quality of life, and psychosocial factors including anxiety and depression.

Reporting of these outcomes in

Data analysis

Where possible, data on all patients will be included in order to conduct an intention-to-treat analysis. Aggregate analyses will initially be undertaken to estimate treatment effects using a random effects model. Regression models, stratified by trial, will be used to explore the effects of treatment, psychosocial, demographic and treatment covariate interactions on the outcomes of interest. Prespecified factors, other than treatment, are age at randomisation, gender, duration of diabetes,

Study management

A steering committee consisting of the principal investigator, project statistician, an independent chair and two trialists representing collaborators will have formal oversight of the study with a management committee including all collaborators ensuring that arrangements for secure handling of data are in place, and that data are only used within the terms of this protocol. Publications will be made on behalf of the collaboration (SMBG Trialists’ Collaboration).

Discussion

Predictive models from this individual patient pooled data set should lead to the identification of those non-insulin treated patients with type 2 diabetes most likely to benefit from self-monitoring of blood glucose. This will guide future work to develop a more targeted and cost-effective use of the intervention. In addition, the study will provide a systematic assessment of the individual components comprising the interventions delivered in the trials, which will inform future development of

Conflicts of interest statement

A.J. Farmer, C. Heneghan, M.B. Davidson, B. Guerci, M O’Kane, U. Schwedes, R. Perera have no conflicts of interest. AH Barnett has received lecture fees and advisory payments from LifeScan Scotland.

Acknowledgements

Writing Group: A.J. Farmer, C. Heneghan, A.H. Barnett, M.B. Davidson, B. Guerci, M O’Kane, U. Schwedes, R. Perera.

Self-monitoring of Blood Glucose International Trialists Collaboration: A.H. Barnett, M.B. Davidson, A.J Farmer, B Guerci, C. Heneghan, M O’Kane, R. Perera, U. Schwedes.

This work is supported by the NIHR Health Technology Assessment Programme. The opinions expressed in this paper are not necessarily those of the Department of Health.

AF, CH, RP are members of the NIHR School of

References (15)

There are more references available in the full text version of this article.
View full text