SLIMMER: a randomised controlled trial of diabetes prevention in Dutch primary health care: design and methods for process, effect, and economic evaluation

For this study, a logic model of change is composed to link intervention activities, their mechanisms of change (i.e. behavioural determinants), expected behaviours, and intervention outcomes in a logical order. A logic model facilitates the understanding of intervention effectiveness and provides insights for further improvements [20, 21]. Figure 1 shows the logic model of change for the SLIMMER intervention. The overall aim of the intervention is to prevent or postpone T2DM and its consequences and to increase quality of life. On the long-term, improvement in fasting insulin is taken as the primary outcome, whereas improvements in cardio-metabolic risk factors (e.g. glucose tolerance, serum lipids, body fatness, and blood pressure) are defined as secondary outcomes. Improvements in eating behaviour and physical activity behaviour are intermediate outcomes. Eating behaviour is measured as nutrient intake and food intake. Physical activity behaviour is operationalised as mode, frequency, duration, intensity, and activity score. Improvements in intention, attitude, social influences, self-efficacy, motivation, action control, and skills are formulated as initial outcomes. These outcomes are achieved if sufficient outputs are delivered in terms of recruitment, reach, dose delivered, dose received, acceptability, implementation integrity, applicability, and context.

The SLIMMER study is a randomised, controlled intervention study, carried out in the Netherlands by a consortium of Wageningen University (WU, Wageningen) and the Community Health Service Noord- en Oost-Gelderland (GGD NOG, Apeldoorn). The total duration of the study is 1.5 years with an intervention period of 10 months. Recruitment of participants took place from October 2011 to September 2012. After baseline measurements, participants are randomly allocated to the intervention or control group, using block randomisation on the level of general practitioners (GPs) and stratification for gender. Couples are allocated to the same group to avoid contamination. Randomisation was performed by an independent dietician of the division of Human Nutrition (WU, Wageningen). The SLIMMER study has been registered with ClinicalTrials.gov (Identifier NCT02094911) since March 19, 2014. The WU Medical Ethics Committee approved the study protocol and all subjects gave their written informed consent before the start of the study.

This study is carried out in Apeldoorn and Doetinchem, two average, middle-sized, Dutch cities, located in the eastern part of the Netherlands. The SLIMMER intervention is implemented in Dutch public health and primary health care, involving GPs and their practice nurses, dieticians, physiotherapists, and sports clubs. Within the study setting, GPs are organised in a formal network to deliver coordinated diabetes care. The majority of dieticians is employed by a home care organisation, only few are self-employed. No regional organisation or network for physiotherapists exists. All GPs have natural referral lines with at least one dietician and in most cases with one physiotherapy practice in the neighbourhood. This existing structure is used for implementation of the SLIMMER intervention. Furthermore, the project is coordinated by the community health service in close collaboration with both municipalities. Sports clubs are organised in a municipal sports stimulation organisation, which has an important role in the maintenance programme.

The sample size calculation for this study is estimated based on changes in fasting insulin, observed in SLIM after one year [22]. In the SLIM study, mean difference in fasting insulin between groups was 2.9 mU/l with a standard deviation of 5.3 mU/l [22]. Because SLIMMER is conducted in a real-life setting instead of a controlled setting, it is estimated to achieve 75% of this result, that is an expected difference in fasting insulin between intervention and control group of 2.175 mU/l. Because we expect a larger SD in a real-life setting, we use 6 mU/l. To adjust for clusters (i.e. general practices), an intra-cluster correlation of 0.055 is used [23]. Based on results of SLIM [22] and the SLIMMER pilot study [19], we expect a drop-out rate of 10%. Assuming an alpha of 0.05, power of 80%, and two-sided test, a sample size of 145 subjects per group is required.

GPs and practice nurses have selected patients aged 40 through 70 years suffering from impaired fasting glucose (IFG: i.e. fasting plasma glucose concentration 6.1-6.9 mmol/l [24]) in the past five years from their patient registration database. Patients are recruited using either laboratory glucose test or the Dutch Diabetes Risk Test [25]. Patients are considered for participation if they still suffer from IFG or if the test score indicates an elevated or high risk of T2DM (i.e. a score of ≥7 points). Inclusion and exclusion criteria (Table 1) are checked by GPs using electronic medical records. GPs have invited eligible patients to participate in the SLIMMER study. A short non-response survey is conducted in case patients are not willing to participate.

The SLIMMER intervention resembles the SLIM intervention [5], which was based on the Finnish Diabetes Prevention Study [4]. The SLIM intervention used a combination of theories, such as Stages of Change model [26] and Theory of Planned Behaviour [27], and tools, such as motivational interviewing [28] and goal setting. SLIMMER is a 10-month combined lifestyle intervention consisting of a dietary and physical activity component, including case management and a maintenance programme. The SLIMMER intervention conforms regular functioning and professional performance of Dutch GPs, practice nurses, dieticians, and physiotherapists. Minimal training and a detailed manual are provided during a two-hour SLIMMER kick-off training for health care professionals. In total, 25 general practices, 11 dieticians, nine physiotherapy practices, and 15 sports clubs are participating in the SLIMMER study. An overview of core tasks and competences of these professionals is given in Additional file 1. Details of the lifestyle intervention programme are described below.

The dietary intervention is consisting of tailored dietary advice during individual consultations and one group session and is aimed to adopt, step by step, a sustainable healthy dietary pattern according to the Dutch dietary guidelines [29]. Furthermore, it is aimed to help participants to achieve 5-10% weight loss. Dietary recommendations are based on Dutch dietary guidelines [29], focussed on people at risk of developing diabetes. Dietary advice is given by a dietician from primary health care, trained in motivational interviewing [28] and using positive feedback. The number of consultations is flexible, ranging from five to eight (30–60 minutes per consultation; maximum of 4 hours per participant), and dependent on needs of participants. If desired, spouses could join consultations. In addition, the dietician organises one group session aimed at sharing experiences, motivating each other, and discussing the topic of label reading. Subjects are encouraged to drink less alcohol, quit smoking if necessary, increase daily physical activity, and to participate in the physical activity intervention. To stimulate self-management of participants, goals for behaviour change are set each consultation, evaluated in the next consultation, and if necessary adjusted. Halfway and at the end of the intervention, behaviour change is more extensively evaluated by dieticians to motivate participants, prevent drop-out, and discuss progression and goals.

The physical activity intervention is consisting of a combined aerobic and resistance exercise programme at the physiotherapist’s practice and is aimed to obtain and maintain an active lifestyle, that is moderate-intensity physical activity for at least 30 minutes per day on at least five days a week. Physical activity recommendations are based on Dutch guidelines for physical activity and type 2 diabetes [30]. Participants have free access to group-based training sessions and are stimulated to participate for at least one hour per week (maximum of two hours per week). Training sessions are given by a physiotherapist from primary health care and tailored to individual needs, desires, and opportunities. In addition, physiotherapists give tailored advice on how to increase physical activity in daily life (e.g. bicycling, walking) and goals are set. After three, six and ten months, behaviour change is monitored by physiotherapists (e.g. weight, waist circumference, and body fat percentage) aimed to motivate participants, prevent drop-out, and discuss progression and goals.

Practice nurses are appointed as case managers of the intervention programme to enhance participant compliance and the feasibility of the implementation. They refer participants to the dietician and physiotherapist at the start of the intervention. Furthermore, they have the overview of the programme and work together with dieticians and physiotherapists. Four weeks after the start of the intervention and halfway the intervention, practice nurses contact dieticians, physiotherapists, and participants of the intervention group to facilitate contact among health care professionals, to detect and solve problems, and to motivate and support participants.

A maintenance programme is added to the combined lifestyle intervention to guide participants in the process of maintaining lifestyle behaviour change in an independent and sustainable manner. This maintenance programme includes 1) intermediate evaluations (e.g. measurement of weight, waist circumference, and body fat percentage) by dieticians and physiotherapists to provide feedback and stimulate self-management; 2) sports clinics at local sports clubs to introduce participants to different sports activities; 3) final interviews with dieticians and physiotherapists at the end of the intervention to give positive feedback, discuss behaviour maintenance, and to set goals; 4) return visit with dieticians and physiotherapists to motivate and support participants in maintaining a healthy lifestyle; and 5) monitoring by practice nurses (i.e. discuss and monitor behaviour change during consultations at the general practice).

Subjects in the control group receive usual health care as provided by GPs and practice nurses. Furthermore, they receive a minimal intervention at the start of the study, consisting of written information about beneficial effects of a healthy diet and increased physical activity, whereas no individual advice or programme is provided. No additional appointments are scheduled, apart from visits for follow-up measurements.

Clinical assessments are performed by trained research assistants in research centres in Apeldoorn and Doetinchem. Furthermore, process and economic data are collected. Participants are measured at baseline (T0), after the intervention (12 months, T1), and six months after ending the intervention (18 months, T2). At each time point, participants are invited for two sessions on different days: one in the morning and one in the afternoon. Additional file 2 gives an overview of indicators, methods, and time points of the data collection.

To assess how the SLIMMER intervention is delivered and received in a real-life setting, data from both participants and health care professionals are collected. A process evaluation plan is designed based on strategies of Steckler and Linnan [52], Saunders et al.[53], Nutbeam [54], and Wang et al.[55]. Process measures include recruitment, reach, dose delivered, dose received, acceptability, implementation integrity, applicability, and context. These process measures are assessed using the project logbook, non-response surveys, participant questionnaires, registration forms, attendance lists, and semi-structured interviews with health care professionals.

Costs and effects of the SLIMMER intervention are compared with those of the usual care. Economic evaluation is performed from a societal perspective, taking all costs and benefits into account. In addition, a health care perspective is considered, in which only direct medical costs are taken into account. As in the effect evaluation, a time horizon of 1.5 years is used. Both a cost-effectiveness analysis (CEA) and a cost-utility analysis (CUA) are performed. The CEA presents clinical outcomes in terms of reduction of fasting insulin. The CUA presents outcomes in terms of quality-adjusted life years (QALYs), determined by the SF-6D health state classification, a preference-based single index derived from the SF-36 [56, 57]. Intervention costs, health care costs, medication costs, patient costs, as well as productivity losses are assessed. In order to estimate intervention costs, time spent by different types of staff involved (practice nurses, dieticians, physiotherapists, providers of sports clinics, and project coordinator) and materials are identified by means of the project logbook, attendance lists, and registration forms. Volumes of health care use, medication use, absence from work, and other expenses are identified by means of participant questionnaires and registration forms. Costs associated with resources use are valued following Dutch guidelines for costing research within health economic evaluations [58, 59]. If no standard cost prices are available, cost estimates from literature are used. All costs are expressed as year 2012 Euros. Where necessary, costs are indexed to the baseline year, as suggested in the Dutch manual [58, 59]. Costs and effects in the second year are discounted at Dutch standard discounting rates of 4% (costs) and 1.5% (effects).

Quantitative data analyses are performed following the intention-to-treat procedure. If necessary, data are transformed and analyses are adjusted for baseline measurements and possible differences between groups at baseline. To adjust for clustering on GP level, multilevel analyses are performed. To determine differences in effects between groups, multivariate analysis techniques are performed. Two-sided P values are calculated and a significance level of 0.05 is applied.

Qualitative data analyses are performed using an inductive approach [60]. Interviews with health care professionals are audiotaped and transcribed verbatim. All transcripts are read by two researchers individually to identify frequently emerging themes. These themes are used to create a coding scheme for analysis of data. Quotes are used to describe aspects of how the intervention is delivered and received.

Differences in costs and effects between intervention and control group are expressed as incremental cost-effectiveness ratios (ICERs). ICERs are plotted on a cost-effectiveness plane, a four quadrant diagram with a horizontal axis representing effect differences between the intervention and control group and the vertical axis representing costs differences between groups. In addition, a cost-effectiveness acceptability curve is constructed, which shows the probability that the SLIMMER intervention is cost-effective for a range of cost-effectiveness thresholds. Sensitivity analyses are conducted to assess robustness of results.