Abstract
Aims/hypothesis
The aim of this RCT was to evaluate the effectiveness of a digital health programme (BetaMe/Melon) vs usual care in improving the control of type 2 diabetes and prediabetes in a primary care population.
Methods
We conducted a randomised parallel-group two-arm single-blinded superiority trial in the primary care setting in two regions of New Zealand. Eligible participants were identified through Primary Health Organisations and participating practices. Eligibility criteria were as follows: age 18–75 years, HbA1c 41–70 mmol/mol (5.9–8.6%), not taking insulin, and daily access to the internet. BetaMe/Melon is a 12 month mobile-device and web-based programme with four components: health coaching; evidence-based resources; peer support; and goal tracking. Participants were randomised into the intervention or control arm (1:1 allocation) based upon baseline HbA1c (prediabetes or diabetes range), stratified by practice and ethnicity. Research nurses and the study biostatistician were blind to study arm. Primary outcomes of the study were changes in HbA1c and weight at 12 months, using an intention-to-treat analysis.
Results
Four hundred and twenty-nine individuals were recruited between 20 June 2017 and 11 May 2018 (n = 215 intervention arm, n = 214 control arm), most of whom were included in analyses of co-primary outcomes (n = 210/215, 97.7% and n = 213/214, 99.5%). HbA1c levels at 12 months did not differ between study arms: mean difference was −0.9 mmol/mol (95% CI −2.9, 1.1) (−0.1% [95% CI −0.3, 0.1]) for the diabetes group and was 0.0 mmol/mol (95% CI −0.9, 0.9) (0.0% [95% CI −0.1, 0.1]) for the prediabetes group. Weight reduced slightly at 12 months for participants in both study arms, with no difference between arms (mean difference −0.4 kg [95% CI −1.3, 0.5]).
Conclusions/interpretation
This study did not demonstrate clinical effectiveness for this particular programme. Given their high costs, technology-assisted self-management programmes need to be individually assessed for their effectiveness in improving clinical outcomes for people with diabetes.
Trial registration
www.anzctr.org.au ACTRN12617000549325 (universal trial number U1111–1189-9094)
Funding
This study was funded by the Health Research Council of New Zealand, the Ministry of Health New Zealand and the Healthier Lives National Science Challenge.
Similar content being viewed by others
Introduction
Type 2 diabetes mellitus is a common long-term condition affecting adults in New Zealand, with a prevalence of 6% in 2018/2019 [1]. Diabetes is associated with multiple long-term complications, higher mortality and substantial healthcare costs [2, 3]. The burden of type 2 diabetes is frequently unequal, with those in socially marginalised populations having higher rates. In New Zealand, the burden of diabetes is particularly high among Māori, Pacific and Indian people, who have up to three times the prevalence of those who self-identify with NZ European ethnicity [1].
Prediabetes is a precursor stage to type 2 diabetes, with up to 70% of affected individuals eventually progressing to type 2 diabetes; many with prediabetes already have complications seen with type 2 diabetes [2]. Self-management measures including appropriate diet, regular exercise and weight management are key to preventing the progression of prediabetes to type 2 diabetes, controlling existing type 2 diabetes and reducing complications [4,5,6,7,8]. While people with diabetes who undergo lifestyle interventions tend to have better glycaemic control and improved long-term outcomes [8], many people struggle to initiate and maintain these strategies [9, 10].
Technology-assisted self-management programmes are increasingly recommended to people with type 2 diabetes for delivery and support of lifestyle interventions. These scalable programmes offer the potential to reduce the burden of type 2 diabetes for populations, and improve health equity by addressing barriers to accessing healthcare, such as time and financial cost [11]. However, evidence is mixed as to whether these technology-based programmes are successful at improving glycaemic control, improving wellbeing or reducing complications [12,13,14,15]. A recent meta-review of type 2 diabetes self-management interventions found a diverse range of technology-assisted self-management programmes, ranging from simple text reminder to multi-model programmes [12]. Overall, technology-assisted programmes perform similarly to other self-management programmes. However, long-term adherence is problematic and observed short-term improvements tend to attenuate over time [12]. Other systematic reviews of technology-assisted interventions for diabetes have found that while improvements in glycaemic control are commonly achieved, effectiveness is sensitive to the amount of healthcare provider input, the degree of personalisation of the programme, and participant age (with greater gains seen in younger people with type 2 diabetes) [13, 15]. Similarly, potential improvements in wellbeing and quality of life are not consistently realised [14].
Given the excess burden among underserved populations, there is a critical lack of evidence relating to the usefulness of these programmes among these groups. In the absence of this evidence, poorly considered large-scale use of these tools may exacerbate inequities in diabetes-related outcomes or miss opportunities to reduce inequalities. The current RCT was intended to contribute to the evidence base for self-management of a comprehensive (multi-modal), patient-centred, technology-assisted self-management programme for individuals with prediabetes and diabetes (the BetaMe/Melon programme) compared with usual care within primary care services in New Zealand. We investigated the effect of this intervention on changes in HbA1c and weight, as well as other secondary outcomes, for both a general population and for the minority Māori and Pacific population.
Methods
Study design
This study was a randomised parallel-group two-arm single-blinded superiority trial. Participants were recruited from 25 general practices (primary care centres) in the greater Wellington and Waikato regions of the North Island of New Zealand, covering both urban and rural settings. The study was designed as an effectiveness trial [16], with an emphasis on recruitment that matched the real-world target population of people with diabetes or prediabetes, delivery of the intervention within the wider context of diabetes care in a primary care setting, and a long follow-up to examine clinically relevant outcomes in a sufficiently large sample of participants.
Ethical approval was obtained from the NZ Health and Disability Ethics Central Committee (HDEC: reference 17/CEN/49). Informed consent was obtained from all participants at the baseline assessment. The trial methods have been published as a protocol [17].
Participants
All patients enrolled with participating practices were screened for eligibility for invitation into the study (for criteria that could be identified at screening), based on review of clinical database information.
Eligibility criteria were as follows: HbA1c 41–70 mmol/mol inclusive (5.9–8.6%) at study enrolment (current or tested in preceding 3 months); not currently receiving insulin treatment; aged 18–75 years (inclusive); having daily access to the internet; and able to provide informed consent.
Exclusion criteria were as follows: pregnant at baseline; cognitive impairment that might make participating in the programme difficult; unable to read/write in English; and inability to use a phone or computer due to physical disability (e.g. poor eyesight).
Participants were recruited into either the diabetes-range group (baseline HbA1c 50–70 mmol/mol inclusive; 6.7–8.6%) or prediabetes-range group (baseline HbA1c 41–49 mmol/mol inclusive; 5.9–6.6%)), as defined by the New Zealand criteria [18]. Eligibility for the prediabetes-range group required that the participant had no previous diagnosis of type 2 diabetes (in clinical notes or reported by the participant).
Randomisation and masking
Full details of randomisation are provided in the protocol [17]. Participants were individually randomised to the intervention (BetaMe/Melon programme) or control arm, using a pre-allocated computer-generated sequence stratified by key clinical and demographic variables (practice, prediabetes- or diabetes-range group, and ethnicity [Māori, Pacific and non-Māori/non-Pacific]). Allocation was conducted by the trial manager (MM) from within the REDCap system [19, 20], following confirmation of eligibility and consent. Participants could not be blinded to study arm following randomisation. Research nurses conducting follow-up assessments were blinded to study arm, although participants may have revealed their study arms during assessments.
The study statistician was blinded during data cleaning and statistical analysis. Results were unblinded following completion of main analyses; per-protocol and sensitivity analyses (noted below) could only be conducted following unblinding.
Procedures
Participants in the control and intervention arms received usual care, which at a minimum included annual checks of glycaemic control and, for those with type 2 diabetes, a review of treatment and checks for complications. Usual care varied across practices and could include education and advice on lifestyle factors.
In addition to usual care the intervention arm received the BetaMe/Melon programme over 12 months, delivered through mobile devices and web-based platforms. BetaMe/Melon was developed and delivered by Melon Health (New Zealand), a company with experience in developing and delivering evidence-based, innovative mobile health solutions in partnership with primary care clinicians. The BetaMe/Melon programme has foundations in behavioural change theory, using cognitive behavioural theory, motivation interviewing, goal setting, health tracking, reminders and intrinsic rewards to support and encourage positive behaviour change [21,22,23]. The BetaMe/Melon programme was piloted in 2015 [17]. Of the 108/117 individuals with prediabetes that completed the programme, 91% reduced their baseline HbA1c, 94% lost weight (mean loss 4.2 kg), 87% had reduced waist circumference (mean reduction 4.2 cm) and 78% achieved HbA1c levels in the normal range.
BetaMe/Melon incorporates four evidence-based components (see Table 1): (1) individual health coaching [24,25,26]; (2) fortnightly provision of evidence-based resources [27,28,29,30,31]; (3) online peer support through a closed forum [32,33,34,35,36]; and (4) online goal tracking [37, 38]. The core part of the programme lasted 16 weeks (all four programme components), with the remaining 36 weeks comprising maintenance activities (web-based peer support and goal tracking only). Melon Health played no role in the RCT design, data collection or analysis.
Follow-up assessments were at 4 and 12 months following baseline assessment. Study data were collected and managed using REDCap electronic data capture tools [19, 20].
Data on participant engagement with the programme were provided to the Otago research team by Melon Health. From the data provided, we identified the number of completed initial health coaching sessions and measured ‘active engagement’ with the online programme (diary entries and posting or ‘likes’ on the peer support forum). We were unable to obtain data from Melon Health on the average number of health coach sessions completed per participant or on passive engagement with the online tool such as the opening of e-mailed fortnightly educational resources or browsing online. Participant engagement and satisfaction with the BetaMe/Melon programme are reported fully elsewhere.
Outcomes
Co-primary outcomes were HbA1c (analysed separately for diabetes- and prediabetes-range participants) and weight in kg (analysed across both groups combined). Standardised measurement procedures are described in the protocol [17]. The primary endpoint was the follow-up at 12 months.
Secondary outcomes included the following variables: HbA1c and weight at interim follow-up (4 months); other physical measurements at both follow-up times (BMI, waist circumference, systolic and diastolic BP); and quality of life and self-management scales at both follow-up times (Partners in Health scale [39], Diabetes Distress Scale [40, 41], Diabetes Self-Care Activities [42] and EQ-5D [43]). Finally, diabetes medication receipt was reported at baseline and at each follow-up: participants were classified (relative to baseline) as to whether they had started insulin, started metformin or changed total daily dose, or started/changed daily dose for other glycaemic control agents. Baseline characteristics included New Zealand Index of Deprivation 2018 [44] as an area-based measure of socioeconomic deprivation, reported as quintiles.
Statistical analysis
A statistical analysis plan was finalised before the trial statistician accessed the data (see electronic supplementary material [ESM] Methods). Main analyses were conducted following an intention-to-treat framework (see Sensitivity analyses below for per-protocol approach). Participants were included in a given analysis if they had at least one follow-up measurement for that outcome.
Comparisons of continuous outcomes (all primary and most secondary outcomes) used linear mixed models, looking at outcomes at follow-up (4 and 12 months) adjusted for baseline. These models adjusted for age group, sex and ethnicity (as a stratifying element in randomisation). All analyses examining the total group (diabetes- and prediabetes-range participants) further adjusted for HbA1c range (stratifying element in randomisation). For the co-primary outcomes, HbA1c was further adjusted for weight at baseline and weight analyses were adjusted for HbA1c at baseline (pre-specified in protocol).
Assessment of baseline balance showed some differences in baseline diabetes medication use by study arm, so primary outcome analysis was adjusted for baseline diabetes medication (binary variable indicating whether a participant was taking hypoglycaemic control agent at baseline).
Results are presented as means at each follow-up (adjusted for baseline) and mean differences by study arm at each follow-up time, with 95% CI.
All statistical analysis was conducted using R 3.5 (R Institute, Austria). Linear mixed models were conducted using the nlme package [45].
Sample size calculations are given in full in the protocol, with the HbA1c outcome having distinct minimal clinically important differences (MCIDs) for diabetes-range participants (5.5 mmol/mol, 0.5%; assumed SD of 15 mmol/mol, 1.4% [22]) and prediabetes-range participants (2.5 mmol/mol, 0.2%; assumed SD 5.5 mmol/mol, 0.5% [46]). Sample size was n = 234 for diabetes-range participants and n = 152 for prediabetes-range participants. This was inflated by 10% to allow for loss of information from incomplete follow-up, giving a final target of n = 430 participants.
For the weight outcome, the sample size required for 80% power to detect an MCID of 5 kg (n = 284, assuming SD 15 kg [47]) was well below the combined sample size calculated for HbA1c.
Sensitivity analyses
One pre-specified sensitivity analysis (ESM Methods/Statistical analysis plan) examined primary outcomes when restricting intervention participants to those with at least one health coaching session recorded. The first health coaching session was the only element of the programme that was required; all other components were optional.
Examination of raw data for objective measurements showed one participant with extreme weight gain of 60 kg between 4 months and 12 months. A post hoc analysis removed this participant from the analysis set before re-assessing the weight outcome.
Trial registration and data monitoring committee
The trial was registered with the Australian New Zealand Clinical Trials Registry on 19 April 2017 (www.anzctr.org.au registration number ACTRN12617000549325 [universal trial number U1111–1189-9094]).
A data monitoring committee oversaw the study, with a remit to assess the effectiveness of study procedures, review any arising adverse events, and review and approve any amendments to study protocols.
Changes to methods after trial commencement
A minor amendment to inclusion criteria was made after the trial commenced. The upper limit of baseline HbA1c was increased for diabetic participants from 64 to 70 mmol/mol (from 8.0% to 8.6%). The rationale for this change was as follows: (1) recruitment of diabetic individuals had been slower than expected; (2) some diabetic participants had baseline measures in the range but were keen to be included in the study; and (3) participants in this range had not been started immediately on insulin. The rationale and decision were discussed and agreed by the study data monitoring committee, as well as clinical team members (AD, JK). The Clinical Trials Registry (ANZCTR) and Ethics Committee (HEDC) were advised of the change (date of approval by ANZCTR: 31 July 2017).
Results
Figure 1 shows participant flow from assessment for eligibility through to analysis. A total of 429 individuals were recruited between 20 June 2017 and 11 May 2018 (n = 215 and n = 214 for intervention and control arms, respectively); almost all were included in analysis of co-primary outcomes (n = 423 [98.6%] for HbA1c; n = 421 [98.1%] for weight).
Balance at baseline
The study arms were balanced on sociodemographic and clinical characteristics, and on baseline values of outcomes (Table 2, ESM Figs 1 and 2, and ESM Tables 1 and 2). There were some exceptions: the control arm had a higher proportion of participants taking metformin or another oral hypoglycaemic agent than the intervention arm (45.3% vs 39.5%; Table 2); the intervention arm had a higher proportion of current smokers than the control arm (12.6% vs 6.6%; Table 2); and a higher proportion of individuals in the intervention arm reported ‘lung disease’ comorbidity than in the control arm (16.3% vs. 12.6%; ESM Table 2).
Primary outcomes
HbA1c in the diabetes-range group (Fig. 2a, Table 3) was not different between study arms during follow-up: at the 12 month endpoint, the mean intervention effect was −0.9 mmol/mol (95% CI −2.9, 1.1) (−0.1% [95% CI −0.3, 0.1]). For the prediabetes-range group (Fig. 2b, Table 3), there was no clinically important difference between study arms at 12 months: the mean difference in HbA1c was 0.0 mmol/mol (95% CI −0.9, 0.9) (0.0% [95% CI −0.1, 0.1]).
Weight (analysed for the combined diabetes- and prediabetes-range groups; Fig. 3, Table 3) reduced slightly in both study arms but at 12 months there was minimal evidence of an intervention effect (mean difference −0.4 kg [95% CI −1.3, 0.5]).
Measured secondary outcomes
At 4 months there was no strong evidence for a clinically important intervention effect on HbA1c relative to the MCID of 5.5 mmol/mol (0.5%): difference in diabetes-range group −1.3 mmol/mol (95% CI −3.1, 0.5) (−0.1% [95% CI −0.3, 0.0]); difference in prediabetes-range group −0.2 mmol/mol (95% CI −0.8, 0.5) (0.0% [95% CI −0.1, 0.0]) (Table 3).
For weight, differences by study arm at 4 months were slightly larger than at 12 months (Fig. 3, Table 3) but the CI excluded the MCID of 5 kg (difference for all participants −0.7 kg [95% CI −1.2, −0.3]). Table 3 also reports results for BMI (calculated using height at baseline).
Separate analyses of weight for the diabetes- and prediabetes-range groups (ESM Table 3 and ESM Fig. 3) showed slightly greater weight loss at 4 months for the intervention arm and this was sustained for the prediabetes group at 12 months. Neither group achieved a mean weight loss at a clinically important level.
Intervention effects on weight for Māori/Pacific participants (ESM Table 4; n = 79 analysed) had wide CIs around point estimates, with slightly higher weight in the intervention arm at both follow-up measurements (mean intervention effect 0.8 kg [95% CI −0.3, 1.8] at 4 months and 1.9 kg [95% CI −1.9, 5.7] at 12 months). The estimate at 12 months was strongly influenced by a single observation in the intervention arm (see Sensitivity analyses, above).
Other objective secondary outcomes are summarised in Table 3 (ESM Fig. 4). While systolic and diastolic BP both decreased over follow-up, there were no observed differences by between the study arms. Waist circumference decreased slightly more for the intervention arm than for the control arm participants.
Self-reported secondary outcomes
Self-reported questionnaire measures are summarised in Table 4 (see ESM Figs 5–9). While the Partners in Health scores increased over follow-up in both study arms (indicating improved self-management), differences were small and slightly favoured the control arm (difference at 12 months −1.1 [95% CI −2.7, 0.5]). A similar pattern was seen for the Diabetes Distress Scale score, EQ-5D and EQ-5D Visual Analogue Scale (Table 4). For Diabetes Self-Care Activities, diet and exercise followed similar trajectories for intervention and control arms (excepting a slightly higher mean number of days of planned exercise at 4 months for the intervention arm).
Medication uptake/change following baseline (see ESM Table 5) showed few participants started insulin during follow-up (n = 4 [2.0%] in intervention arm; n = 1 [0.5%] in control arm). Starting metformin was slightly more common in the intervention arm (n = 11 [5.4%]) than in the control arm (n = 7 [3.4%]) and more people stopped metformin in the control than intervention arm (n = 6 [2.9%] in control arm; n = 0 [0.0%] in intervention arm). Uptake of other oral glucose-lowering agents was slightly higher in the control arm (n = 7 [3.4%]) than in the intervention arm (n = 4 [2.0%]) and discontinuation of these agents was similar in both groups (n = 2 [1.0%] in intervention arm; n = 3 [1.5%] in control arm).
There were no reported adverse events.
Sensitivity analyses
Post hoc analysis (see ESM Table 6) included accounting for one participant with extreme weight gain between 4 and 12 months (weight gain of 60 kg: diabetes-range participant in intervention arm, Māori ethnicity). Removing this participant suggested the slight weight loss seen in the intervention group at 4 months (−0.7 kg [95% CI −1.2, −0.3]) was maintained to 12 months (−0.7 kg [95% CI −1.4, 0.0]). However the confidence intervals for these differences still excluded the MCID of 5 kg. Removing this participant’s 12 month weight measurement had a similar impact on estimates for diabetes-range participants and the combined Māori/Pacific participant group.
Pre-specified per-protocol analysis for primary outcomes (ESM Table 7) restricted the intervention group to the 92% of participants that completed a first health coaching session. Among diabetes-range participants HbA1c showed slightly greater differences favouring the intervention arm over the intention-to-treat analysis: −1.6 mmol/mol (95% CI −3.5, 0.2) (−0.1% [95% CI −0.3, 0.0]) at 4 months; and −1.0 mmol/mol (95% CI −3.0, 1.1) (−0.1% [95% CI −0.3, 0.1]) at 12 months. For prediabetes-range participants, intervention effects were still minimal: −0.3 mmol/mol (95% CI −0.9, 0.4) (0.0% [95% CI −0.1, 0.0]) at 4 months; and 0.0 mmol/mol (95% CI −1.0, 0.9) (0.0% [95% CI −0.1, 0.1]) at 12 months. In all these per-protocol analyses, point estimates and CIs excluded the MCIDs (5.5 mmol/mol for diabetes range, 0.5%; 2.5 mmol/mol for prediabetes range, 0.2%).
For weight (combined participant group), per-protocol analysis showed larger differences, favouring the intervention arm over the intention-to-treat analysis: −0.8 kg (95% CI −1.3, −0.3) at 4 months; and −0.4 kg (95% CI −1.4, 0.5) at 12 months. Removing the one participant with extreme weight gain suggested this difference was maintained across follow-up: −0.9 kg (95% CI −1.3, −0.4) at 4 months; and −0.8 kg (95% CI −1.5, −0.1) at 12 months.
Measures of engagement
Initial engagement with the health coach was high, with 92% receiving an initial health coaching session. Seventy-four per cent of participants had any active engagement with the online programme in the 16 weeks of the core programme, primarily through diary completion (40% reducing to 20% during active phase), and a lower-level of engagement with the support forum (20% reducing to 5%). Overall, online engagement was lower for participants identifying as Māori, starting at 40% and dropping to only 3% at 4 months.
Discussion
This RCT assessed an online self-management programme developed for type 2 diabetes and prediabetes (BetaMe/Melon) and found small improvements in HbA1c and weight at 4 months that had largely attenuated by 12 months, a pattern seen in previous studies. Changes in HbA1c and weight were greater for diabetes-range participants than for prediabetes-range participants, and greater for non-Māori non-Pacific people than for Māori and Pacific people. No group achieved clinically important mean improvements in glycaemic control (HbA1c) or weight. Based upon these findings, the BetaMe/Melon programme in its current form cannot be recommended for use in the management of diabetes or prediabetes.
There is strong evidence for the role of self-management strategies in the management of type 2 diabetes [4,5,6,7,8] but mixed evidence as to whether web-based programmes are a useful platform to deliver these programmes [12,13,14,15]. The lack of effect of the BetaMe/Melon programme likely results from a combination of delivery of an inadequate dose of the programme, insufficient engagement of Māori and Pacific people in the programme design, a lack of connection to primary care, and the limited benefits of providing additional support to the relatively well controlled type 2 diabetes and prediabetes populations.
Engagement with technology-supported self-management programmes is critical to their success [11]. In this RCT, initial engagement with the health coach was high (92%) and 74% of participants had any active online engagement with the programme. However, it is likely that the high degree of programme flexibility (considering the intervention modules as optional) resulted in an inadequate overall ‘dose’ of the intervention and the loss of the potential benefits of a multi-modal programme. The BetaMe/Melon programme was developed by a multidisciplinary team with Māori input [17] but achieved lower engagement from Māori people. Improved engagement and outcomes from technology-based healthcare programmes requires increased involvement of target populations (in this case Māori and Pacific) in the design and development phases [11].
Our recruited RCT population was chosen to represent the real population that this intervention targets: those with relatively well-managed type 2 diabetes (HbA1c 50–70 mmol/mol, 6.7–8.6%) and prediabetes. The lack of an effect may reflect the limited additional benefit of self-management advice and support to that already provided ‘as usual care’ in primary care in this relatively well controlled population. In addition, previous studies have shown that active engagement of health providers with a programme increases the likelihood of patient improvements in glycaemic control [13]. The lack of a direct connection to the participant’s primary care provider was a limitation of the assessed programme.
A major strength of this trial was the extremely high retention of participants (over 98% for primary outcomes). This was largely due to research nurse flexibility around appointments (including home visits) and multiple follow-up attempts. While there may have been changes in participant self-management driven by upcoming nurse assessments, we believe the impact of this on outcomes was likely to be small and non-differential. Over the course of the study the control group had little to no change in HbA1c levels but did achieve some reduction in weight. Relatedly, while research nurses were blinded to participant study arm, participants could not be blinded (given the nature of the intervention) and in some cases chose to reveal their allocation to the nurses. Given the objective nature of the primary outcome measures (HbA1c, weight) any potential arising bias is likely to be minimal.
Another strength was the range of secondary outcomes assessed in this study, including self-reported measures of quality of life. One concerning study finding was a reduction in quality of life for the intervention arm at 12 months (relative to the control arm): as a secondary outcome taken in the context of the entire trial, this result should be considered as an important outcome for follow-up in future research (including qualitative studies of participant experience). Previous studies of technology-assisted programmes have found either no impact or a small positive impact on measures of wellbeing [14]. It is possible that reports of lower quality of life may have been due to frustration or disappointment at the lack of progress with weight loss and glycaemic control while on the intervention. Finally, we assessed the impact of several design and analysis decisions on the results: these sensitivity analyses returned results broadly consistent with the main results.
Technology-assisted self-management tools, such as the programme assessed in this RCT, are promoted and used in healthcare settings based on mixed evidence of their effectiveness for the total population and scant evidence regarding subgroups within populations. In our study, even the small benefits that we detected appeared smaller in the groups most affected by diabetes. In the absence of demonstrated effectiveness, such programmes are problematic due to their costs and potentially negative effects on equity; the potential for unintended negative impacts for all or part of the population (e.g. negative impact on quality of life) needs to be examined further.
Conclusion
This study did not support rollout of this self-management programme in its current state, based upon the lack of demonstrated effectiveness. Further research needs to identify key elements of successful self-management programmes and investigate strategies to improve programme adherence. These lessons could help to improve future development or refinement of technology-assisted self-management programmes.
Data availability
Data and analysis code are available on request from the authors.
Abbreviations
- MCID:
-
Minimal clinically important difference
References
Ministry of Health (2019) Annual data explorer 2018/2019: New Zealand Health Survey. Available from: https://minhealthnz.shinyapps.io/nz-health-survey-2018-19-annual-data-explorer/ (accessed 10 Jan 2020)
Ministry of Health (2015) Living well with diabetes: a plan for people at high risk of or living with diabetes 2015–2020. Ministry of Health, Wellington
World Health Organization (2006) Definition and diagnosis of diabetes mellitus and intermediate hyperglycemia: report of a WHO/IDF consultation. World Health Organization, Geneva
Diabetes Prevention Program Research Group, Knowler WC, Fowler SE et al (2009) 10-year follow-up of diabetes incidence and weight loss in the Diabetes Prevention Program Outcomes Study. Lancet 374(9702):1677–1686. https://doi.org/10.1016/S0140-6736(09)61457-4
Knowler WC, Barrett-Connor E, Fowler SE et al (2002) Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med 346(6):393–403. https://doi.org/10.1056/NEJMoa012512
Lindstrom J, Ilanne-Parikka P, Peltonen M et al (2006) Sustained reduction in the incidence of type 2 diabetes by lifestyle intervention: follow-up of the Finnish Diabetes Prevention Study. Lancet 368(9548):1673–1679. https://doi.org/10.1016/S0140-6736(06)69701-8
Perreault L, Pan Q, Mather KJ et al (2012) Effect of regression from prediabetes to normal glucose regulation on long-term reduction in diabetes risk: results from the Diabetes Prevention Program Outcomes Study. Lancet 379(9833):2243–2251. https://doi.org/10.1016/S0140-6736(12)60525-X
Vinik AI, Richardson DW (1997) Implications of the diabetes control and complications trial for persons with non-insulin-dependent diabetes mellitus. South Med J 90(3):268–282. https://doi.org/10.1097/00007611-199703000-00001
Rivellese AA, Boemi M, Cavalot F et al (2008) Dietary habits in type II diabetes mellitus: how is adherence to dietary recommendations? Eur J Clin Nutr 62(5):660–664. https://doi.org/10.1038/sj.ejcn.1602755
Ruggiero L, Glasgow R, Dryfoos JM et al (1997) Diabetes self-management. Self-reported recommendations and patterns in a large population. Diabetes Care 20(4):568–576
Shan R, Sarkar S, Martin SS (2019) Digital health technology and mobile devices for the management of diabetes mellitus: state of the art. Diabetologia 62(6):877–887. https://doi.org/10.1007/s00125-019-4864-7
Captieux M, Pearce G, Parke HL et al (2018) Supported self-management for people with type 2 diabetes: a meta-review of quantitative systematic reviews. BMJ Open 8(12):e024262. https://doi.org/10.1136/bmjopen-2018-024262
Greenwood DA, Gee PM, Fatkin KJ, Peeples M (2017) A systematic review of reviews evaluating technology-enabled diabetes self-management education and support. J Diabetes Sci Technol 11(5):1015–1027. https://doi.org/10.1177/1932296817713506
Hadjiconstantinou M, Byrne J, Bodicoat DH et al (2016) Do web-based interventions improve well-being in type 2 diabetes? A systematic review and meta-analysis. J Med Internet Res 18(10):e270. https://doi.org/10.2196/jmir.5991
Hou C, Xu Q, Diao S, Hewitt J, Li J, Carter B (2018) Mobile phone applications and self-management of diabetes: a systematic review with meta-analysis, meta-regression of 21 randomized trials and GRADE. Diabetes Obes Metab 20(8):2009–2013. https://doi.org/10.1111/dom.13307
Singal AG, Higgins PD, Waljee AK (2014) A primer on effectiveness and efficacy trials. Clin Transl Gastroenterol 5(1):e45. https://doi.org/10.1038/ctg.2013.13
Sarfati D, McLeod M, Stanley J et al (2018) BetaMe: impact of a comprehensive digital health programme on HbA1c and weight at 12 months for people with diabetes and pre-diabetes: study protocol for a randomised controlled trial. Trials 19(1):161. https://doi.org/10.1186/s13063-018-2528-4
New Zealand Guidelines Group (2011) Guidance on the management of type 2 diabetes 2011. New Zealand Ministry of Health, Wellington
Harris PA, Taylor R, Minor BL et al (2019) The REDCap consortium: Building an international community of software platform partners. J Biomed Inform 95:103208. https://doi.org/10.1016/j.jbi.2019.103208
Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG (2009) Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 42(2):377–381. https://doi.org/10.1016/j.jbi.2008.08.010
Abraham C, Michie S (2008) A taxonomy of behavior change techniques used in interventions. Health Psychol 27(3):379–387. https://doi.org/10.1037/0278-6133.27.3.379
Liang X, Wang Q, Yang X et al (2011) Effect of mobile phone intervention for diabetes on glycaemic control: a meta-analysis. Diabet Med 28(4):455–463. https://doi.org/10.1111/j.1464-5491.2010.03180.x
Ryan P (2009) Integrated theory of health behavior change: background and intervention development. Clin Nurse Spec 23(3):161–170; quiz 171-162. https://doi.org/10.1097/NUR.0b013e3181a42373
Clarke DM, Baird DE, Perera DN, Hagger VL, Teede HJ (2014) The INSPIRED study: a randomised controlled trial of the Whole Person Model of disease self-management for people with type 2 diabetes. BMC Public Health 14(1):1. https://doi.org/10.1186/1471-2458-14-134
Funnell MM, Brown TL, Childs BP et al (2009) National standards for diabetes self-management education. Diabetes Care 32(Suppl 1):S87–S94
Jordan JE, Osborne RH (2007) Chronic disease self-management education programs: challenges ahead. Med J Aust 186(2):84–87. https://doi.org/10.5694/j.1326-5377.2007.tb00807.x
Cole-Lewis H, Kershaw T (2010) Text messaging as a tool for behavior change in disease prevention and management. Epidemiol Rev 32:56–69. https://doi.org/10.1093/epirev/mxq004
Franklin VL, Waller A, Pagliari C, Greene SA (2006) A randomized controlled trial of Sweet Talk, a text-messaging system to support young people with diabetes. Diabet Med 23(12):1332–1338. https://doi.org/10.1111/j.1464-5491.2006.01989.x
Nundy S, Dick JJ, Chou CH, Nocon RS, Chin MH, Peek ME (2014) Mobile phone diabetes project led to improved glycemic control and net savings for Chicago plan participants. Health affairs (Project Hope) 33(2):265–272. https://doi.org/10.1377/hlthaff.2013.0589
Wolever RQ, Dreusicke MH (2016) Integrative health coaching: a behavior skills approach that improves HbA1c and pharmacy claims-derived medication adherence. BMJ Open Diabetes Res Care 4(1):e000201. https://doi.org/10.1136/bmjdrc-2016-000201
Fischer HH, Fischer IP, Pereira RI et al (2016) Text message support for weight loss in patients with prediabetes: a randomized clinical trial. Diabetes Care 39(8):1364–1370. https://doi.org/10.2337/dc15-2137
Fisher EB, Boothroyd RI, Coufal MM et al (2012) Peer support for self-management of diabetes improved outcomes in international settings. Health Aff 31(1):130–139. https://doi.org/10.1377/hlthaff.2011.0914
Jennings A, Powell J, Armstrong N, Sturt J, Dale J (2009) A virtual clinic for diabetes self-management: pilot study. J Med Internet Res 11(1):e10
Thom DH, Ghorob A, Hessler D, De Vore D, Chen E, Bodenheimer TA (2013) Impact of peer health coaching on glycemic control in low-income patients with diabetes: a randomized controlled trial. Ann Fam Med 11(2):137–144. https://doi.org/10.1370/afm.1443
Vissenberg C, Stronks K, Nijpels G et al (2016) Impact of a social network-based intervention promoting diabetes self-management in socioeconomically deprived patients: a qualitative evaluation of the intervention strategies. BMJ Open 6(4):e010254. https://doi.org/10.1136/bmjopen-2015-010254
Yin J, Wong R, Au S et al (2015) Effects of providing peer support on diabetes management in people with type 2 diabetes. Ann Fam Med 13(Suppl 1):S42–S49. https://doi.org/10.1370/afm.1853
Glasgow RE, Funnell MM, Bonomi AE, Davis C, Beckham V, Wagner EH (2002) Self-management aspects of the improving chronic illness care breakthrough series: implementation with diabetes and heart failure teams. Ann Behav Med 24(2):80–87. https://doi.org/10.1207/S15324796ABM2402_04
Yamada T, Kiuchi Y, Nemoto M, Yamashita S (2014) Charting weight four times daily as an effective behavioural approach to obesity in patients with type 2 diabetes. Diab Vasc Dis Res 11(2):118–120. https://doi.org/10.1177/1479164113508141
Smith D, Harvey P, Lawn S, Harris M, Battersby M (2017) Measuring chronic condition self-management in an Australian community: factor structure of the revised Partners in Health (PIH) scale. Qual Life Res 26(1):149–159. https://doi.org/10.1007/s11136-016-1368-5
Fisher L, Glasgow RE, Mullan JT, Skaff MM, Polonsky WH (2008) Development of a brief diabetes distress screening instrument. Ann Fam Med 6(3):246–252. https://doi.org/10.1370/afm.842
Fisher L, Hessler DM, Polonsky WH, Mullan J (2012) When is diabetes distress clinically meaningful?: establishing cut points for the Diabetes Distress Scale. Diabetes Care 35(2):259–264. https://doi.org/10.2337/dc11-1572
Toobert DJ, Hampson SE, Glasgow RE (2000) The summary of diabetes self-care activities measure: results from 7 studies and a revised scale. Diabetes Care 23(7):943–950. https://doi.org/10.2337/diacare.23.7.943
Eurocare Working Group (1990) EuroQol - a new facility for the measurement of health related quality of life. The EuroQol group. Health Policy 16:199–208
Atkinson JSC, Crampton P (2019) NZDep2018 Index of Deprivation, Interim Research Report, December 2019. University of Otago, Wellington
Pinheiro JBD, DebRoy S, Sarkar D, R Core Team (2018) _nlme: Linear and Nonlinear Mixed Effects Models_. R package version 3. Available from https://CRAN.R-project.org/package=nlme
Garvey WT, Ryan DH, Henry R et al (2014) Prevention of type 2 diabetes in subjects with prediabetes and metabolic syndrome treated with phentermine and topiramate extended release. Diabetes Care 37(4):912–921. https://doi.org/10.2337/dc13-1518
Heymsfield SB, Segal KR, Hauptman J et al (2000) Effects of weight loss with orlistat on glucose tolerance and progression to type 2 diabetes in obese adults. Arch Intern Med 160(9):1321. https://doi.org/10.1001/archinte.160.9.1321
Acknowledgments
The authors thank I. Osborne (Auckland, New Zealand) who designed the graphical abstract.
Authors’ relationships and activities
The authors declare that there are no relationships or activities that might bias, or be perceived to bias, their work.
Funding
This study was funded by the Health Research Council of New Zealand, the Ministry of Health New Zealand and the Healthier Lives National Science Challenge, 16/727. The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.
Author information
Authors and Affiliations
Contributions
DS, MM and JaS led the study concept and design with input from VS, JK, RG and AD. MM managed the study. VS developed the study database with support from MM and JaS. MM, VS, DT and KH acquired and managed the data with support from JeS and CD. MM, JaS and DS led the analysis and interpretation of data. VS, JeS, JK, AD and RG contributed to data analysis and interpretation. MM and JaS wrote the first draft the manuscript with input from DS. MM, JaS, DS, VS, JeS, DT, KH, CD, JK, AD and RG provided critical revision of the manuscript draft. All authors gave final approval of the version to be published. MM is the guarantor of this work and had full access to all the data in the study and final responsibility for the decision to submit for publication.
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
ESM
(PDF 2.00 MB)
Rights and permissions
About this article
Cite this article
McLeod, M., Stanley, J., Signal, V. et al. Impact of a comprehensive digital health programme on HbA1c and weight after 12 months for people with diabetes and prediabetes: a randomised controlled trial. Diabetologia 63, 2559–2570 (2020). https://doi.org/10.1007/s00125-020-05261-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00125-020-05261-x