DiGEM is a four-year study with an open, randomised controlled pragmatic parallel group trial design (Figure ) with sequential recruitment from two centres. The trial is managed from the Department of Primary Health Care, University of Oxford following NHS R&D Health Technology Assessment Programme guidelines. The study protocol was approved by the Oxfordshire B Research Ethics Committee.
Consort diagram showing planned study numbers.
The study design is shown in Figure . Participants are randomised to three groups consisting of
(i) a control group receiving standardised usual care and three monthly measurement of HbA1c,
(ii) a self-testing group who, in addition to the above, are carrying out blood glucose self testing with the results interpreted by the study nurse,
(iii) a self monitoring group who, in addition to both of the above, are given support in interpreting and applying the results of blood testing to enhance motivation and maintain adherence to diet physical activity and medication regimens.
Following randomisation, patients receive the allocated education and training appropriate for their group, with follow up to maintain the interventions at four, 13, 26, and 39 weeks with a final assessment at 52 weeks.
Setting and recruitment
Non-insulin using patients with type 2 diabetes have been recruited from general practices in Oxfordshire and are being recruited from South Yorkshire to take part in the study. The 48 recruited practices represent a geographical spread of rural/suburban centres and cover a wide socio-economic range of patients. The mean number of patients recruited in the 24 practices in Oxfordshire was 10.2, standard deviation (SD) 5.6.
Patients suitable for trial inclusion are identified from practice generated computer lists. Eligible patients are sent an invitation to participate signed by their general practitioner accompanied by an information sheet and a reply paid envelope to facilitate response. One further letter is sent if no response is received in one month.
Study population: inclusion and exclusion criteria
Inclusion criteria are type 2 diabetes, aged 25 years or more at diagnosis, managed with lifestyle or oral hypoglycaemic agents, independent for activities of daily living. Exclusion criteria are use of blood glucose monitor twice a week or more often over the previous three months, current use of insulin, co-morbidity or limited life expectancy that would make intensive glycaemic control inappropriate, last clinic HbA1c or HbA1c at the assessment visit less than 6.2%, or unable to follow trial procedures.
Participants are randomly allocated to one of the three groups using a partial minimisation procedure to adjust the randomisation probabilities between groups to balance important covariates including duration of diabetes, HbA1c, and prior medication using a computer programme (Minim, http://www.sghms.ac.uk/depts/phs/guide/randser.htm
Baseline measures and follow up
The primary outcome measure is change in HbA1c between the baseline measurement at the assessment visit and 12-month visit. Secondary outcome measures include change in systolic and diastolic blood pressure, weight, serum cholesterol and HDL, self-reported smoking status, dietary intake and physical activity (the Diabetes Self Care Activities Questionnaire) [22
], medication adherence (The Medication Adherence Rating Scale) [23
], and the scores in the Diabetes Treatment Satisfaction Questionnaire [24
], and the Well-being Questionnaire (12 item) [25
Beliefs about diabetes and its management are assessed using the Illness perceptions Questionnaire [26
], Beliefs about Medicines Questionnaire [27
], and a questionnaire developed for the study about the effectiveness of changes in eating and physical activity on the course of diabetes, and attitudes to blood glucose self-monitoring. Table summarises the measures and their timing.
Blood glucose monitoring resources are measured by counting recorded entries in diaries. Medication use, episodes of hospitalisation and their duration will be noted and non-hospital health care resource use will be recorded by a questionnaire administered to all patients at each visit.
Participant eligibility for the study and willingness to be randomised to a group in which they might be required to test their own blood glucose six times a week or more is confirmed at the assessment visit. Following consent, beliefs about diabetes, the role of eating, physical activity and medication are discussed with all participants. A goal setting approach to lifestyle change is introduced and continued in subsequent visits. Baseline blood tests and clinical measurements are taken and questionnaires completed at this visit.
Following the assessment visit and confirmation of eligibility on the basis of HbA1c measurement, patients are randomised to one of three groups: control group, self-testing and self-monitoring.
At a visit two weeks after the assessment visit participants receive training and education appropriate to their allocated study group. The control group receives 3-monthly HbA1c measurements and identifies behavioural goals to improve glycaemic control. The self-testing group, in addition, is asked to use a blood glucose meter to record three fasting, pre-meal or 2-hour post meal readings on two days during the week. Treatment targets of fasting and pre-meal levels of 4–6 mmol/l and post meal levels of 6 to 8 mmol/l and advice about using these readings to identify high (>15 mmol/l) and low (<4 mmol/l) blood glucose readings are given. The self-monitoring group, in addition, is provided with training and support to encourage interpretation of readings and application to goals for lifestyle change based on the CSM in order to reach treatment targets.
Follow up visits
Subsequent follow up includes a telephone call 2 weeks after randomisation (after the post-randomisation visit rather than randomisation), and further visits at 4, 13, 26 and 39 weeks. The follow up visits differ according to the allocated group. Those allocated to the control group have a HbA1c measure two weeks before their scheduled visit in order for their glycaemic control to be discussed. The two groups using a meter are managed on the results of their blood glucose self-monitoring. The GP is notified of all HbA1c results and asked to consider changes in medication in line with the National Institute for Clinical Excellence diabetes guidelines. The GP is also notified if blood glucose readings are consistently above 15 mmol/l.
Study measures (see Table and Figure )
Baseline self-report measures and measures of belief are completed at the assessment visit. Baseline biochemical measures and clinical measurements are also made. Repeated measurements are made at the 52-week follow up visit. Data on adverse reactions or complications are collected at each study visit along with information about use of medication and health services.
Quality assurance and fidelity of interventions
Patients are supplied with a blood glucose meter calibrated to provide plasma equivalent results. Meters are checked at the beginning of the study and at 26 weeks by study nurses with a test aliquot.
A script outlining the topics to be covered, and explanation of the theoretical basis of the intervention are used to support the nurses in their adherence to study protocol. Study nurses attended a six day phased course in psychological theory, behaviour change techniques and skills training in the intervention. Additional measures to ensure fidelity include self-review of taped interventions by the study nurses and external review by a researcher using a checklist to ensure delivery of the intervention according to protocol. Prompts built into the patient diaries help participants adhere to their allocated group intervention.
In the absence of data relating to change in HbA1c we have calculated sample size conservatively based on the absolute level of HbA1c at follow up. We set out to detect a difference in HbA1c of 0.5% between any two groups. At the outset we estimated the SD of HbA1c as 1.5, based on data from the UKPDS. In practice, the SD of baseline HbA1c among the 245 patients recruited in Oxford was 0.9, but we assumed that the SD among additional patients recruited elsewhere could be as high as 1.5. With 205 further patients, the overall SD would be 1.2, and there would be 150 patients in each group (135 allowing for 10% attrition). These numbers would give 93% power to detect a difference of 0.5% in HbA1c between any two trial groups (2-tailed alpha = 0.05). Figure gives estimates of likely numbers in each group and attrition.
We propose to conduct a single analysis of main trial endpoints at the end of the study. The proposed intention to treat analysis will compare mean levels of HbA1c at follow up between the three study groups, with baseline HbA1c as a covariate, using analysis of covariance. Post-hoc t-tests between groups will be conducted in the event of a statistically significant result. Subsequent analysis will include comparing the two intervention groups against the control group.
We will estimate the intervention effect in sub-groups defined by duration of diabetes (above or below median duration), current management (oral hypoglycaemic dugs or dietary management only), health status at baseline (above or below median EQ-5D score) and co-morbidity (presence or absence of diabetes related complications). We will also explore the extent to which the measures of beliefs included in the study can explain changes in behaviour; firstly by comparing mean levels of beliefs e.g. about controllability of diabetes between experimental groups and secondly by a more formal mediation analysis [28
]. Within group analyses will be used to determine fidelity to protocol and conformity to the theoretical model, between group analyses will be used to assess impact of the intervention on key variables proposed by the theoretical model. Additional exploratory analysis will include changes in behaviour in relation to perceived threat and changes in perceived thereat from diabetes.
The economic evaluation will be on an intention-to-treat basis. A cost-effectiveness/cost-utility analysis will be performed in which the difference in effectiveness will be compared to the difference in total costs between each study intervention group, and the results will be expressed as incremental cost-effectiveness ratios. Effectiveness will be measured in terms of change in HbA1c, and modelled for life years gained and quality adjusted life years gained. Unit costs will be attached to the resource items collected from the healthcare resource utilisation assessment using published national average costs and tariff averages for procedures to calculate costs. Mean values and 95% confidence intervals will be reported for each component of resource use and cost and for total costs and effectiveness. Sensitivity analyses will be performed on all aspects of the economic evaluation that are subject to uncertainty.