Digital Weight-Loss Programmes: Which Components Actually Work — and Which Add Burden

A Comprehensive Evidence Review

Prepared: 2026-10-28 Style: Vancouver numbered citations; every PMID mechanically verified via NCBI E-utilities on the date of preparation Evidence cut-off: 28 October 2026 Note: This review was prepared for personal knowledge. All factual claims carry citation numbers resolved in the reference list. Inferences are labelled as such.


Abstract

Digital behavioural weight-management programmes — commercial apps, app-plus-coach services, and digital translations of the Diabetes Prevention Program (DPP) — are now a routine referral destination in primary care, yet the field’s central question has shifted. It is no longer whether digital delivery can produce weight loss; the average effect is small, reproducible, and well characterised. The live question is which components inside a programme carry that effect, which add burden without benefit, and for whom. The seed trial for this review, a 2⁴ factorial cluster randomised trial embedded in a commercial digital programme (n = 1335 adults, BMI ≥ 21 kg/m², 90.8% ≥ 25 kg/m²), tested four added components: a health-coach introductory call, coaching webchat sessions, goal-setting statements, and food-diary reviews with feedback [1,2]. The introductory call produced no effect at 16 weeks but about 1 kg greater weight loss at 24 weeks (p = 0.002), which did not survive per-protocol analysis; the food-diary review-and-feedback component was associated with less weight loss (+0.52 kg at 16 weeks, p = 0.07; +0.74 kg at 24 weeks, p = 0.02) [1]. Neither finding is straightforwardly generalisable, and the trial’s full component estimates are not available in the abstract. Read against the wider literature, the result is best understood as a marginal-effect experiment layered on a programme that already contains self-monitoring: adding a review-and-feedback loop to an already-monitored programme can add burden without adding benefit, whereas low-intensity human contact appears to improve engagement and modestly improve weight outcomes in several — though not all — settings [3–8].

This review synthesises 152 sources: randomised trials of digital DPP translations, app-only and coach-supported programmes, meta-analyses of mHealth and text-messaging interventions, component-isolation and multiphase optimisation studies, engagement and attrition cohorts, equity analyses, safety studies, clinical guidelines, and the incretin-therapy literature. Five conclusions are defensible. First, digitally delivered behavioural programmes produce mean weight loss of roughly 1–3 kg beyond control at 6–12 months, with the better-designed coach-supported and DPP-derived programmes reaching 3–6 kg and 5–7% in completers [3,9–14]. Second, human coaching is the most consistent single additive component, worth roughly 2 kg in direct comparisons, but it is not uniformly superior to automated alternatives and its effect is partly a retention effect [5,15–18]. Third, self-monitoring is associational bedrock but experimentally fragile: logging frequency predicts loss across dozens of studies, yet the incremental value of adding a monitoring layer to an existing programme is small and can be negative, and adherence decays steeply within 10–12 weeks [2,6,19–23]. Fourth, the binding constraint is engagement, not efficacy: completion rates of 45% in a national digital programme, 27% 12-month adherence in a commercial GLP-1-supported service, and 12% persistence in another are typical of the genre, and they dominate any population-level effect estimate [24–27]. Fifth, the incretin era changes the role of behavioural programmes from primary weight-loss vehicle to adjunct, maintenance scaffold, and triage partner; the trial evidence supports combination (semaglutide plus intensive behavioural therapy, tirzepatide after lifestyle induction), but the maintenance-after-discontinuation literature shows that behavioural support alone rarely holds the loss once pharmacotherapy stops [28–33].

The practical message for clinicians is narrower than the marketing: recommend digital programmes as a low-burden, low-risk option for motivated adults with BMI ≥ 27–30, prefer programmes with human contact and structured self-monitoring over app-only products, set expectations at 3–5% rather than 10–15%, screen for disordered eating and weight-stigma sensitivity, and do not substitute a digital referral for clinical follow-up in people with severe obesity, comorbidities, or a history of eating disorder. The seed trial’s most useful contribution is methodological: it shows that commercial programmes can run component-level optimisation trials at scale, and that some plausibly useful features are, at the margin, not useful at all.


1. Introduction and Framing: Why Component-Level Questions Now Matter

Digital weight-management programmes entered clinical practice on the back of a simple proposition: behavioural weight loss works, face-to-face delivery does not scale, and software can scale it. That proposition is largely correct at the level of average effect. The US Preventive Services Task Force recommends intensive multicomponent behavioural interventions for adults with obesity, with moderate net benefit and small-to-no harms [34]; national programmes in England and the United States have delivered DPP-derived curricula to hundreds of thousands of people [14,35]; and meta-analyses of app-based and mHealth interventions consistently find small but statistically significant weight differences favouring intervention [12,36–39]. The argument about whether digital delivery “works” is, in that narrow sense, settled.

What is not settled is the interior of the black box. A digital programme is not one intervention but a bundle: a curriculum, a self-monitoring system, a feedback engine, a coaching layer, goal-setting prompts, notifications, social features, and increasingly a pharmacological service. Payers are asked to fund bundles; clinicians are asked to recommend them; vendors compete on feature lists. Yet the evidence base has mostly tested bundles against usual care, which cannot tell a commissioner whether the coach is worth its cost, whether the food diary is worth the user’s daily 15 minutes, or whether the goal-setting module is decoration. The Multiphase Optimisation Strategy (MOST) was developed precisely to answer such questions by randomising components rather than packages [40,41]. Its application to digital weight management has been slow, partly because commercial providers have little incentive to discover that half their feature set is inert, and partly because component trials require large samples and stable delivery infrastructure.

The seed trial is important because it does exactly that inside a live commercial product. Wren and colleagues randomised 1335 adults enrolling in Second Nature, a commercial digital weight-management programme, to one of 16 conditions in a 2⁴ factorial cluster design, testing four added components over 24 weeks [1,2]. The design is a pragmatic test of the MOST logic in an industry setting, using routinely collected data. It is also, inevitably, a test of marginal effects: every participant received the core programme, so the trial asks what each component adds on top of an already structured digital intervention, not whether the core works. That distinction is essential to interpreting the counter-intuitive food-diary result and is returned to repeatedly in this review.

Three contextual shifts make component-level questions more urgent now than when the trial was designed in 2024. The first is financial: health systems are moving from pilot funding to procurement, and procurement requires a defensible specification — what must a programme contain, what may it omit, and what should be paid for [42–44]. The second is therapeutic: incretin-based pharmacotherapy produces weight losses of 15–21% in trials, an order of magnitude beyond digital behavioural programmes, which forces a re-examination of what behavioural programmes are for [28,29,45,46]. The third is equity: digital delivery changes who is reached, and the early evidence suggests it reaches a different population from face-to-face services — younger, heavier, more often white, more often male, and more often already digitally literate [14,47–49]. A component that improves average weight loss but widens disparities is not obviously a public-health gain.

This review therefore asks five questions. What did the seed trial actually find, and what can a factorial design of this kind support? What does the broader trial literature say about digitally delivered behavioural weight management? What is the component-level evidence for coaching, self-monitoring, feedback, goal setting, self-weighing, and text messaging? What do engagement, clinical significance, equity, and safety imply for recommendation? And how should guidelines, commissioners, and clinicians position digital programmes in an era of effective pharmacotherapy? The review is deliberately hedged where the evidence is weak, and it distinguishes throughout between association (who does well) and causation (what makes people do well).


2. The Seed Trial: Design, Population, Interventions, Outcomes, and What It Did and Did Not Find

2.1 Design and population

The trial was a 2⁴ factorial cluster randomised controlled trial conducted within a commercial digital behavioural weight-loss programme delivered by Second Nature, with academic leadership from the University of Oxford’s Nuffield Department of Primary Health Care Sciences [1,2]. Adults enrolling in the programme were eligible if their BMI was at least 21 kg/m²; 1335 were randomised, and 90.8% had a BMI of at least 25 kg/m². The lower eligibility threshold is unusual for a weight-loss trial and reflects the programme’s positioning as a general healthy-weight service rather than a clinical obesity treatment; it also means that roughly one in eleven participants was in the normal-weight range at baseline, which dilutes expected weight change and complicates comparison with obesity-treatment trials.

Randomisation was to one of 16 conditions formed by crossing four binary components. The primary outcome was weight change at 16 weeks. Secondary and exploratory outcomes included weight change at 4 and 24 weeks, programme drop-out, and engagement measured as interactions with the three main app functions. The protocol specified a decision threshold: components would be considered worth retaining if they contributed at least 0.75 kg of additional weight loss, alone or in combination [2]. That threshold is itself a statement about clinical ambition — 0.75 kg is well below the conventional 5% clinically meaningful threshold — and it framed the trial as a service-improvement exercise rather than a regulatory efficacy trial.

The trial was pre-registered on 5 January 2024 (ISRCTN14407868) and reported in npj Digital Medicine in September 2026 [1,2]. The commercial relationship is material and is discussed in the limitations: Second Nature delivered the intervention and covered delivery costs; the company was the industrial partner on the lead author’s Medical Research Council iCASE studentship; and several co-authors were Second Nature employees [1]. That does not invalidate the findings, but it shapes what was tested (components the company was considering), what was measured (routinely collected service data), and how the results should be read.

2.2 The four components

The four components were chosen because they were plausible, deliverable, and separable within the existing product:

  1. Health-coach introductory call. A one-off video or telephone call with a health coach at enrolment, intended to establish rapport, clarify expectations, and personalise the programme.
  2. Coaching webchat sessions. Drop-in, text-based group or individual webchat sessions with a coach, intended to provide ongoing human contact at low marginal cost.
  3. Goal-setting statements. Structured prompts asking participants to articulate specific behavioural goals, intended to convert the curriculum’s general advice into concrete intentions.
  4. Food-diary reviews plus feedback. A human or semi-automated review of the participant’s food diary with personalised feedback, intended to close the loop between self-monitoring and behaviour change.

Each component was added to the core digital programme, which itself included the usual apparatus of lessons, self-monitoring, and app-based interaction. This is the crucial design feature: the trial estimated the incremental effect of each component in the presence of the others, averaged over all combinations. Factorial designs of this kind are efficient — all participants contribute to every component’s estimate — but they estimate marginal effects at the doses and in the context tested. A null result for the food-diary review does not mean that food diaries are useless; it means that adding a review-and-feedback layer to a programme that already encourages self-monitoring did not improve weight loss in this population over 16–24 weeks.

2.3 What the trial found

The published abstract reports two principal signals [1]. The introductory call showed no effect at 16 weeks but was associated with approximately 1 kg greater weight loss at 24 weeks (p = 0.002). That effect did not persist in the per-protocol analysis, which considered only participants who engaged with the component. The food-diary review-and-feedback component was associated with less weight loss: +0.52 kg at 16 weeks (p = 0.07) and +0.74 kg at 24 weeks (p = 0.02), where positive values indicate less loss than the comparator.

The abstract does not report the estimated effects of the webchat or goal-setting components. It also does not report the primary outcome result for the trial as a whole, the distribution of weight change, or the engagement and drop-out findings in numeric form. This review therefore cannot state the direction or magnitude of those components’ effects, and any claim that they were “null” would be an inference from silence rather than a reported finding. What can be said is that the trial’s headline conclusions emphasise low-intensity human-supported components and caution about the burden of more complex features [1].

Table 1. Seed trial components and reported effects (from the published abstract).

Component What was added Reported effect at 16 weeks Reported effect at 24 weeks Persistence
Health-coach introductory call One-off call at enrolment No effect ≈1 kg greater weight loss (p = 0.002) Did not persist in per-protocol analysis
Coaching webchat sessions Drop-in coach webchats Not reported in abstract Not reported in abstract Unknown
Goal-setting statements Structured goal prompts Not reported in abstract Not reported in abstract Unknown
Food-diary reviews + feedback Diary review with personalised feedback +0.52 kg (less loss), p = 0.07 +0.74 kg (less loss), p = 0.02 Direction consistent at both timepoints

Source: [1]. The trial’s full factorial estimates, engagement analyses, and drop-out results are not available in the abstract; the protocol provides design detail [2].

2.4 Interpreting the food-diary finding

The food-diary result is the most cited and most misread finding in the trial. Three interpretations are available, and they are not mutually exclusive.

Burden and substitution. A review-and-feedback layer imposes costs on the participant (more detailed logging, waiting for feedback, responding to comments) and on the service (coach time). If the core programme already prompts self-monitoring, the marginal diary review may displace other activities or create a sense of surveillance that reduces autonomous motivation. The trial authors’ own framing — “considering the potential burden of more complex features” — is consistent with this [1]. The effect size (+0.74 kg at 24 weeks) is small but not trivial in a population whose mean loss is likely modest.

Reverse causation within a factorial design. Factorial randomisation protects the assignment from confounding, but the per-protocol and engagement analyses are observational. Participants who struggled may have been more likely to receive, or to attract, diary feedback; coaches may have prioritised review for those who were not losing weight. If the feedback component was delivered with greater intensity to non-responders, the intention-to-treat estimate would be biased toward showing harm. The trial reports that the effect did not survive per-protocol analysis for the introductory call; it does not report an equivalent analysis for the diary component in the abstract. This is a plausible mechanism, not a demonstrated one.

Measurement and population. The primary outcome was weight at 16 weeks in a population with a mean BMI only modestly above 25. In such a population, the absolute weight changes are small, and a 0.5–0.7 kg difference can be generated by differential engagement with weighing rather than by differential fat loss. The trial used service-collected weights, which are vulnerable to differential missingness.

The broader literature makes the third interpretation less likely to be the whole story and the first two more plausible. Observational studies consistently find that more frequent food logging predicts greater weight loss [19,20,50–52]. But those studies compare people who log more with people who log less, and both logging and weight loss are driven by motivation, self-regulation, and coach attention. When logging is randomised or added, the effect is much smaller and sometimes absent: the SMARTER trial found no difference between self-monitoring with and without tailored daily feedback [6]; a pilot trial of simplified versus standard calorie monitoring found no difference in weight loss [53]; and the Cochrane review of smartphone apps found little to no difference in weight at 12 months [12]. The seed trial’s result is therefore best read as a marginal-effect finding consistent with a literature in which the core of self-monitoring matters but additional monitoring infrastructure often does not.

2.5 Interpreting the coaching finding

The introductory call’s 1 kg effect at 24 weeks, lost in per-protocol analysis, is a fragile signal. It is nonetheless consistent with a broader pattern: human contact improves engagement and produces small weight differences, particularly when it is personalised and when it occurs early. Direct comparisons support this. In a 400-person randomised trial, a wireless feedback system plus 12 weekly 10–15 minute coaching calls produced 4.8 kg loss at 6 months versus 2.8 kg with the feedback system alone — a 2.0 kg difference, and the coaching arm was not non-inferior [15]. In the ConTxt trial, text messaging plus brief monthly coaching calls produced 3.63% weight loss at 12 months versus 0.61% in controls, while text messaging alone produced 1.68% and was not significantly different from control [16]. In a Danish primary-care-anchored eHealth coaching trial, 12-month weight loss was 4.5 kg versus 1.5 kg with usual care [17]. Against this, a fully automated AI-led DPP was non-inferior to a human-led DPP on a composite outcome (31.7% vs 31.9% achieving the composite), suggesting that the function of coaching — structure, accountability, tailoring — can sometimes be replicated by software [18].

The seed trial’s per-protocol attenuation is important. If the call only works among people who take it up and engage, its population effect depends on uptake, and its cost-effectiveness depends on targeting. That is a different claim from “coaching works,” and it is the claim the data actually support.

2.6 What the trial did not find, and what it could not show

The trial did not show that any component produced clinically meaningful weight loss on its own. The largest reported effect (1 kg at 24 weeks) is below the 5% threshold that guidelines use to define clinically significant loss for most participants. It did not show a durable effect beyond 24 weeks; the follow-up ends there. It did not show that the programme works in people with severe obesity, type 2 diabetes, or eating disorders, because those groups were not the target population and were not analysed separately in the abstract. It did not show that the components are equivalent across delivery modes, coaches, or fidelity levels. And it could not show whether the food-diary review was harmful in any clinically meaningful sense, because the outcome was weight, not eating behaviour, psychological distress, or dropout.

These limits are not criticisms of the design; they are the boundaries of a service-embedded optimisation trial. They do, however, mean that the trial should be cited for what it is: a demonstration that component-level experimentation is feasible inside a commercial digital programme, and a cautionary signal that adding features is not the same as adding value.


3. The Broader Evidence Base for Digitally Delivered Behavioural Weight Management

3.1 Digital translations of the Diabetes Prevention Program

The DPP is the most influential behavioural weight-loss curriculum in preventive medicine, and its digital translations are the closest analogue to a “standard” digital programme. The best-known randomised translation is the Omada digital DPP trial reported by Katula and colleagues: 599 adults with prediabetes were randomised to a 52-week digital programme with lifestyle coaching, virtual peer support, and behaviour tracking, or to a single-session small-group education class [9]. At 12 months, the digital programme produced a 5.5% weight reduction versus 2.1% in the comparison group; 43% versus 21% achieved ≥5% loss; HbA1c fell by 0.08 percentage points more; and 58% versus 48% shifted from prediabetes to normoglycaemia. Engagement was related to both HbA1c and weight improvement. The trial is registered as NCT03312764 [54].

Other translations are more mixed. A fully mobile, coach-guided DPP (Noom) in 202 adults with prediabetes produced a 2.64 kg greater weight loss at 6 months than usual care, but no significant HbA1c difference at 6 months (mean difference 0.004%), and a non-significant 0.23% versus 0.16% HbA1c reduction at 1 year [10]. A UK feasibility RCT of a web-based DPP with step-dependent SMS feedback found no weight effect at 6 or 12 months (mean difference 0.07 kg at 12 months) and lower retention in the intervention arm (70% vs 85%) [11]. A secondary analysis of the Omada trial found that predicted 10-year atherosclerotic cardiovascular risk fell by 0.96 percentage points more in the digital arm at 4 months, but the difference was not sustained at 12 months [55]. A single-arm analysis of the Transform digital DPP reported a 6.5% mean weight loss at 4 months among 273 completers, but the uncontrolled design limits inference [56].

The English NHS Digital Diabetes Prevention Programme provides the largest real-world test. In a prospective cohort of 3623 registered participants across nine pilot areas, with 50% follow-up at 12 months, mean weight change was −3.1 kg and HbA1c fell by 1.6 mmol/mol [57]. The associated qualitative work found that service users valued coach support, self-monitoring, and social support, and that human contact alongside the digital programme improved engagement [58]. The national DPP more broadly has been evaluated with quasi-experimental methods: a regression-discontinuity analysis using data from about one-fifth of English primary care practices found that referral improved HbA1c, excess body weight, and lipid levels [59], and the DIPLOMA multimethod assessment concluded that the programme was associated with significant reductions in type 2 diabetes incidence and was highly likely to be cost-effective [60]. These are population-level findings, not component-level ones, but they establish that digital and mixed-mode DPP delivery can produce measurable benefit at scale.

In the United States, the National DPP enrolled 455,954 participants between 2012 and 2019, of whom 269,004 (59%) enrolled in online organisations [14]. Online enrolment was younger (41.5% aged 18–44), more male (27.1% vs 19.4%), and more white (70.3%) than in-person enrolment, which enrolled more Black participants (16.1%) and more people aged 65+ (28.2%). A multistate comparison found that online programmes had higher enrolment (56% vs 51%) but lower engagement (46% vs 66% attending ≥9 sessions) and lower effectiveness (23% vs 35% achieving ≥5% loss) than in-person programmes [47]. That trade-off — broader reach, shallower engagement — is a recurring theme.

Table 2. Selected randomised and large observational evaluations of digital DPP delivery.

Study (design) Population / n Intervention Weight outcome Other outcomes Notes
Katula 2022 RCT [9] 599 adults with prediabetes 52-week digital DPP with coaching and peer support −5.5% vs −2.1% at 12 mo; ≥5% loss 43% vs 21% HbA1c −0.08 pp; normoglycaemia 58% vs 48% NCT03312764
Toro-Ramos 2020 RCT [10] 202 adults with prediabetes Mobile DPP (Noom) with coach −2.64 kg greater loss at 6 mo No HbA1c difference at 6 mo 1-year HbA1c 0.23% vs 0.16%
Staite 2020 RCT [11] 200 adults with prediabetes Web DPP + wearable + SMS No effect (0.07 kg at 12 mo) No activity effect Retention 70% vs 85%
NHS DDPP cohort [57] 3623 registered; 50% follow-up Digital DPP (five providers) −3.1 kg at 12 mo HbA1c −1.6 mmol/mol Uncontrolled cohort
Golovaty 2021 [47] 26,743 referred Online vs in-person DPP — ≥5% loss 23% vs 35% Online enrolment higher, engagement lower
National DPP 2012–2019 [14] 455,954 enrolled In-person, online, distance, combination — — Online 59% of enrolment; different demographics

3.2 App-only versus coach-supported programmes

The app-only literature is larger and less impressive than the coach-supported literature, and the difference is instructive. The 2024 Cochrane review of smartphone interventions included 18 randomised trials with 2703 participants. At 12 months, smartphone apps compared with no or minimal intervention produced a mean weight difference of −2.5 kg (95% CI −6.8 to 1.7; low-certainty evidence), and at 24 months +0.7 kg (−1.2 to 2.6; moderate-certainty). The review concluded that the available evidence “does not demonstrate a clear benefit” and that comparisons with personal coaching showed minor, mostly non-clinically-significant effects [12]. An earlier meta-analysis of 12 app trials found −1.07 kg (−1.92 to −0.21) [36]; a 2022 meta-analysis found peak weight loss of 2.18 kg at 3 months tapering to 1.63 kg at 12 months [37]; and a 2025 meta-analysis of 11 app-only trials in adults without comorbidities found a standardised mean difference of −0.33 (−0.48 to −0.17) at 4–6 months, with no effect on waist circumference [38]. A 2020 meta-analysis of mHealth self-monitoring interventions found moderate weight loss and higher adherence than paper-based or usual-care comparators, with effects concentrated in the short term [39].

Coach-supported programmes do better, though the evidence is heterogeneous. A 12-month digital programme with one-on-one video coaching produced −7.2% (overweight) and −7.6% (obesity) weight change in a single-arm analysis of 683 participants — comparable to in-person programmes, but uncontrolled [61]. A Danish RCT of eHealth lifestyle coaching anchored in primary care produced −4.5 kg versus −1.5 kg at 12 months [17]. A German app-based multimodal programme (zanadio) produced −7.75% at 12 months versus 0.00% in a wait-list control [13]. A Japanese app (CALO mama Plus) produced a −1.60 kg between-group difference at 3 months [62]. A Singaporean app with dietitian messaging produced −3.6 kg versus −1.2 kg and a 0.4 percentage-point greater HbA1c reduction in adults with type 2 diabetes [63]. A UK pragmatic trial of a self-regulation app with no human contact (daily self-weighing, action planning, weekly reflections) produced −1.85 kg (−2.53 to −1.17) and an odds ratio of 2.11 for ≥5% loss at 6 months, with a reduction in disordered-eating symptoms [64]. The last result is important: a well-designed app-only intervention can produce a real, if small, effect, and the effect is not necessarily purchased at the cost of psychological harm.

The most direct test of the coaching increment is the Spring trial: 400 adults were randomised to a wireless feedback system alone or the same system plus 12 weekly coaching calls. Six-month weight loss was 2.8 kg versus 4.8 kg, a 2.0 kg difference, and the coaching arm was not non-inferior [15]. That is a clean estimate of the marginal value of low-intensity coaching on top of digital self-monitoring, and it is larger than the seed trial’s 1 kg signal. The difference may reflect dose (12 calls vs one call), population (BMI 27–45 vs BMI ≥21), or outcome timing.

3.3 What the meta-analytic average hides

The summary estimate for digital behavioural weight management — roughly 1–3 kg beyond control at 6–12 months — is robust but not very informative. Three sources of heterogeneity matter clinically.

Baseline weight and comorbidity. Trials in adults with prediabetes or type 2 diabetes often show larger absolute effects because the population is heavier and more motivated; trials in general-adult or lower-BMI populations show smaller effects. The seed trial’s BMI ≥21 eligibility is at the low end. A digital programme that produces 5% loss in a 35 kg/m² population may produce 2% in a 27 kg/m² population with the same behavioural dose.

Comparator. Usual care, wait-list, minimal advice, and “enhanced standard care” are not equivalent. The Katula trial’s comparator was a single education session, which is weaker than usual primary care in some systems; the Cochrane review’s “no or minimal intervention” comparator includes handouts and food diaries. Effect sizes shrink as comparators strengthen.

Delivery and dose. Programmes with human coaching, longer duration, and structured curricula outperform short app-only interventions. The 2025 app meta-analysis restricted to app-only trials in people without comorbidities found a small effect; the coach-supported trials find larger ones. The dose-response is not linear, but the direction is consistent.

The practical implication is that a commissioner should not accept “digital weight management works” as a specification. The relevant questions are: in whom, compared with what, at what dose, and with what support?


4. Component-Level Evidence: The Heart of the Review

4.1 Coaching and human support

Human support is the component with the most consistent positive signal and the most ambiguous mechanism. Four lines of evidence converge.

Direct randomised comparisons. The Spring trial’s 2.0 kg coaching increment is the cleanest [15]. The ConTxt trial’s pattern — text plus brief monthly calls −3.63%, text alone −1.68%, control −0.61% — suggests that a small amount of human contact converts a null text intervention into a modestly effective one [16]. The Danish eHealth trial’s −4.5 kg versus −1.5 kg at 12 months, with an initial face-to-face motivational interview and ongoing digital coaching, points the same way [17].

Coach access to self-monitoring data. A randomised experiment within a 12-week group behavioural programme found that giving coaches access to participants’ digital self-monitoring data (scale, activity sensor, food record) increased weight loss from 5.3% to 6.2% (p = 0.04), increased self-monitoring of eating and activity, and increased perceived supportive accountability, with no evidence of iatrogenic effects [8]. This is a component-level finding of considerable practical importance: the value of a coach may depend on whether the coach can see the data and act on it, not merely on whether a coach exists.

Primary-care counselling. In a digital obesity treatment embedded in community health centres, participants whose providers documented intervention-specific counselling lost 3.1 kg more than those whose providers documented only general weight counselling, and 4.0 kg more than those with no documented counselling; perceived provider empathy was associated with greater loss from 6 to 12 months (0.8 kg per unit) [65]. This is observational, but it suggests that clinical endorsement and empathy are active ingredients in referral-based programmes.

Automated alternatives. The AI-led DPP trial is the strongest challenge to the coaching consensus: a fully automated intervention was non-inferior to a human-led DPP on a composite of weight, activity, and HbA1c at 12 months (31.7% vs 31.9%), with higher initiation in the AI arm (93.4% vs 82.7%) [18]. Non-inferiority on a composite does not mean equivalence on every component, and the composite’s weight criterion was ≥5% loss, which both arms achieved at modest rates. But it does suggest that the informational and structural functions of coaching can be automated, while the relational functions — empathy, accountability, tailoring to social context — may remain distinctively human.

Table 3. Selected evidence on coaching as an additive component.

Study Design Coaching dose Comparison Weight result Interpretation
Spring 2024 [15] RCT, n = 400 12 weekly 10–15 min calls Same digital system without calls −4.8 vs −2.8 kg at 6 mo Coaching added ≈2 kg
Godino 2019 [16] RCT, n = 298 Monthly 5–10 min calls + SMS SMS alone; print control −3.63% vs −1.68% vs −0.61% at 12 mo Calls converted null SMS into effect
Hesseldal 2022 [17] RCT, n = 340 Initial interview + digital coaching Usual care −4.5 vs −1.5 kg at 12 mo Coach-supported eHealth effective
Butryn 2024 [8] Randomised experiment, n = 322 Coach access to self-monitoring data No access 6.2% vs 5.3% at 12 wk Data-sharing adds ≈0.9 pp
Mathioudakis 2025 [18] Non-inferiority RCT, n = 368 AI-led vs human-led DPP Human coach 31.7% vs 31.9% composite Automation can be non-inferior
Seed trial [1] Factorial cluster RCT, n = 1335 One introductory call No call ≈1 kg at 24 wk; not per-protocol Small, fragile signal

4.2 Self-monitoring and the food diary

Self-monitoring is the most studied component in behavioural weight management and the one with the widest gap between association and causation.

Association. A systematic review of 39 randomised trials with digital self-monitoring found that weight was tracked in 72% of interventions, diet in 81%, and physical activity in 82%; few interventions achieved engagement on ≥75% of days; and greater digital self-monitoring was linked to weight loss in 74% of reported associations [19]. A separate review of 59 dietary self-monitoring studies found that both high- and low-intensity monitoring protocols supported weight loss, but that variability in adherence measurement limited comparison [20]. In the Livongo digital DPP cohort, each submitted food log was associated with 0.23 kg of weight loss, each completed lesson with 0.14 kg, and each week of 150 active minutes with 0.1 kg [50]. In a commercial app cohort, consistent trackers (≥6 of 7 days for ≥75% of weeks) were 25% of participants but achieved 54% versus 15% ≥5% loss at 6 months [51]. In the SMARTER trial’s parent study, logging frequency predicted loss, and the association was bidirectional: poorer loss predicted poorer subsequent monitoring in group treatment [52].

Causation. When monitoring is manipulated, the picture is murkier. The SMARTER trial randomised 502 adults to self-monitoring with or without three daily tailored feedback messages and found no difference in 6-month weight change (−3.16% vs −3.20%) [6]. A pilot trial of simplified (red-zone foods only) versus standard calorie monitoring found no difference in weight loss (5.7% vs 4.0%, not significant) or tracking adherence [53]. A pilot trial of photo-based versus calorie-based monitoring found no difference in weight loss, with tracking on fewer than 30% of days in both arms [66]. A secondary analysis of a commercial app trial found that 4-month weight loss was partly mediated by app use (60.3%) and weigh-ins (55.8%), but 12-month loss was almost fully mediated by 4-month loss (94.8%) — suggesting that early engagement sets a trajectory rather than that monitoring has a continuous independent effect [67].

Dose and consistency. The observational literature suggests that consistency matters more than total volume. In a 220-woman behavioural treatment cohort, high frequency of self-monitoring reduced weight regain only when coupled with high consistency (>3 days/week), and the effect was partly mediated by adherence to calorie goals [68]. In a 24-week online programme, participants who lost ≥5% logged in 2.4 times per day versus 1.6 times per day for those who lost <5%, and time spent monitoring fell from 23.2 to 14.6 minutes per day between months 1 and 6 [69]. In a mobile-monitoring study, tracking at least two eating occasions per day was the best adherence marker, and fewer than half of participants were still tracking after week 10 [70]. A Delphi consensus recommended transitioning to abbreviated monitoring (self-weighing only, or tracking only high-calorie foods, or fewer days per week) after 2 weeks of no monitoring in struggling participants and after 5–10% loss in successful ones [71]. A maintenance study found that self-monitoring at least 3 days/week was associated with less regain, with the largest benefit at 5–6 days/week [72].

The seed trial’s contribution. The seed trial’s food-diary review-and-feedback component was associated with less loss, which is consistent with the experimental literature’s weak marginal effects and with the possibility of burden. It is not consistent with a strong causal benefit of adding diary review to an already-monitoring programme. The most defensible synthesis is: self-monitoring is a correlate and probably a facilitator of weight loss, especially when it is consistent, brief, and linked to actionable goals; but adding more monitoring infrastructure to a programme that already encourages monitoring has small and unpredictable marginal effects, and can backfire.

4.3 Feedback

Feedback is conceptually distinct from monitoring but usually bundled with it. The evidence suggests that feedback content and framing matter more than feedback frequency.

The SMARTER trial found no benefit of three daily tailored feedback messages over self-monitoring alone [6]. The seed trial found a negative association for diary review plus feedback [1]. But a factorial experiment on feedback composition found that emails with more reinforcing comments and more areas for change were better received, with the best combination for several outcomes being 8 reinforcing comments and 4 areas for change; 4 reinforcing comments and 4 areas for change produced the highest probability of intending to self-monitor [7]. That is a study of message design, not weight outcomes, but it suggests that feedback can be optimised on dimensions other than frequency.

Coach access to data is a form of feedback that appears to work: the Butryn experiment’s 0.9 percentage-point difference at 12 weeks was achieved by enabling coaches to see and respond to monitoring data [8]. The mechanism — supportive accountability — is plausible and measurable.

The practical implication is that “feedback” is not a single intervention. Automated daily messages, weekly coach reviews, and data-driven accountability calls are different products with different evidence. Commissioners should specify which one they are buying.

4.4 Goal setting

Goal setting is a core behaviour-change technique and appears in most effective digital interventions. A systematic review of technology-driven diabetes prevention interventions found that goal setting (outcome/behaviour), feedback on behaviour, self-monitoring of outcomes, and social support were present in over 90% of effective interventions; effective interventions used on average 5.6 more behaviour-change techniques than ineffective ones in the short term and 3.7 more in the long term [73]. A meta-regression of automated digital interventions found that goal setting, feedback, and instruction on behaviour were the most common clusters, but that only reward and threat techniques were associated with higher between-group efficacy, and that this was not robust to sensitivity analysis [74].

The seed trial’s goal-setting component was not reported in the abstract, so its effect is unknown. A small pilot trial (n = 32) is testing different goal intensities across calorie, step, eating-window, and red-zone food domains, but results are not yet available [75]. A systematic review of commitment-making — a close relative of goal setting — found that commitment devices increased short-term weight loss by 1.5 kg (0.7–2.4) and that benefits were sustained at 12 months in two trials (1.7 kg; 0.0–3.4), with public commitments and dietary targets most successful [76]. The evidence is limited but suggests that specific, public, consequential goals may work better than generic goal prompts.

4.5 Self-weighing

Self-weighing is the most scalable self-monitoring behaviour and the one with the clearest experimental signal, albeit a modest one.

A meta-analysis of randomised trials found that self-weighing as a single strategy had no significant effect (−0.5 kg; −1.3 to 0.3), but adding self-weighing/self-regulation techniques to multicomponent programmes produced −1.7 kg (−2.6 to −0.8), and multicomponent programmes including self-weighing versus minimal intervention produced −3.4 kg (−4.2 to −2.6). There was no difference between daily and weekly weighing instructions, and trials with accountability elements had greater loss [77]. Systematic reviews of longitudinal studies found regular self-weighing associated with more loss and no adverse psychological outcomes in adults [78,79].

Trial evidence is consistent but small. A two-year RCT of daily self-weighing with visual feedback produced 2.6 kg versus 0.5 kg loss in year 1, with maintenance in year 2 [80]. A six-month RCT in 54 men found additional loss with self-weighing [81]. A pilot RCT in an online commercial programme found that a smart scale plus weekly tailored feedback increased self-weighing from 6.7% to 44.7% of days and increased the proportion achieving ≥5% loss at 3 months (52.2% vs 28.3%), though mean weight differences were not significant [82]. A cohort of 9768 smart-scale users found that 2.8 weighs per week was typical, that weighing frequency correlated inversely with weight change (r = −0.111), and that 72.5% had at least one ≥30-day break during which weight increased [83]. A maintenance study found that consistency (weighing ≥6–7 days/week) predicted less regain, while total frequency did not [84].

Safety is the caveat. A prospective study of 762 college students found that more frequent self-weighing predicted greater weight gain (+0.8 BMI), especially in those with binge eating (+1.6 BMI), and that frequent weighers had higher onset of compensatory behaviours (OR 3.90; 1.76–8.75) [85]. A population-based adolescent study found that frequent self-weighing predicted disordered eating behaviours in females but not males, and did not predict weight loss [86]. These findings do not contraindicate self-weighing in adults in structured programmes, but they argue for screening and for avoiding prescriptive daily weighing in adolescents and people with binge-eating pathology.

4.6 Text messaging

Text messaging is cheap, scalable, and modestly effective. A meta-analysis of 15 trials (2705 participants) found a mean difference of −2.28 kg (−3.17 to −1.36) for weight-loss interventions and −0.68 kg (−1.31 to −0.05) for maintenance interventions, with no subgroup differences by duration, frequency, theory use, interactivity, or tailoring [87]. Individual trials show a similar range: a 12-month RCT of personalized interactive texts in English- and Spanish-speaking adults found no effect of texts alone but a −3.0% difference when texts were combined with brief monthly coaching calls [16]; a prediabetes trial found −2.6 lb versus −0.6 lb and 38.5% versus 21.5% achieving 3% loss [88]; a Latvian trial found −3.4 kg at 1 year with biweekly texts [89]; and the Game of Stones trial in 585 men with obesity found −4.8% with texts plus financial incentives, −2.7% with texts alone, and −1.3% in controls at 12 months, with the text-only comparison falling just short of conventional significance (p = 0.053) and no significant between-group differences at 24 months [90]. A maintenance trial found no difference between intensive and minimal text support after group CBT [91], and a single-group extended-contact study found that texts helped sustain a 4.2 kg loss at 18 months [92].

The Game of Stones economic evaluation is instructive: texts alone cost £110 per participant and texts plus incentives £243; the lifetime ICER for texts plus incentives versus control was £9,748 per QALY, and for texts alone £628 per QALY [93]. The intervention was equally effective across socioeconomic and health strata, but the trial was restricted to men and excluded people without mobile phone access [90,93].

4.7 Component interactions and the limits of additive thinking

The seed trial’s factorial design allows, in principle, the estimation of interactions, but the abstract does not report them. The broader literature suggests that components are not additive in a simple way. Coaching may work partly by improving adherence to self-monitoring, which then produces weight loss; self-monitoring may work only when feedback is actionable; goal setting may work only when goals are specific and reviewed. A mediation analysis of a commercial app found that 4-month weight loss was partly mediated by app use and weigh-ins, but that 12-month loss was almost entirely explained by 4-month loss [67]. A meta-review of self-regulatory behaviour-change techniques found that no single technique was consistently related to improved outcomes across 30 meta-analyses, and argued that the meta-analytic design was mismatched to the question [94]. That is a warning against interpreting component main effects as stable causal parameters.

Table 4. Component-level evidence summary.

Component Best evidence Typical incremental effect Certainty Key caveat
Human coaching Multiple RCTs and experiments [8,15–17] ≈1–2 kg; improved retention Moderate Dose and function vary; automation can substitute [18]
Coach data access Randomised experiment [8] ≈0.9 pp weight loss at 12 wk Low–moderate Single trial
Self-monitoring (diet) Large observational literature; weak experimental marginal effects [6,19,20,51,53] Association strong; added layer ≈0 Moderate (association), low (causal) Burden; reverse causation
Feedback RCTs and factorial experiments [6–8] 0 to ≈1 pp depending on design Low Content matters more than frequency
Goal setting BCT meta-regression and commitment review [73,74,76] ≈1.5 kg short-term for commitment devices Low–moderate Specificity and consequence matter
Self-weighing Meta-analysis and RCTs [77,80,82,84] ≈1.7 kg added to multicomponent; 2.6 kg vs 0.5 kg in one RCT Moderate Harms in adolescents/binge eating [85,86]
Text messaging Meta-analysis and RCTs [16,87,90] ≈2.3 kg vs control Moderate Effect largely disappears at 24 mo without incentives

5. Engagement, Adherence, and Attrition: The Binding Constraint

5.1 The scale of the problem

If efficacy trials ask whether a programme can work, engagement data ask whether it does work in the population offered it. The answer is sobering.

The English NHS Digital Weight Management Programme is the best population-level example. Of 63,937 referrals from general practices in 2021–22, 31,861 (50%) took up the offer; of 31,718 who had time to finish, 14,268 (45%) completed (defined as attending ≥60%) [24]. Mean weight change was −2.2 kg in those who had time to finish, −3.9 kg in completers, and −0.74 kg in non-completers. The programme was judged cost-effective at £4,202 per QALY, but that estimate is built on completers’ weight loss and a modelled lifetime horizon [42]. The gap between the intention-to-treat effect and the completer effect is the engagement penalty, and it is larger than most component effects.

Commercial digital services show similar or worse persistence. A UK tirzepatide-supported digital weight-loss service reported 27% 12-month adherence among 19,693 patients [25]. A German semaglutide-supported service reported 12% persistence and 88% attrition, with mean 12-month loss of 9.47% under last-observation-carried-forward but only 2.40% under a worst-case assumption that all dropouts lost nothing [26]. An Australian digital weight-loss service reported mean adherence of 171 days, with 43.7% of discontinuations attributed to inadequate supply of the desired GLP-1 medication and 26.2% to cost [95]. In the Game of Stones trial, follow-up was 73% at 12 months and 64% at 24 months [90]. In the Doody app trial, weight was reported by only 39.3% of participants at 6 months [64].

5.2 Why people disengage

Qualitative and mixed-methods work identifies a consistent set of barriers: time burden, privacy concerns, embarrassment about self-disclosure, technical problems, lack of access to technology or internet, and low digital confidence [96]. Facilitators include digital competency, convenience, self-monitoring, tailored feedback, accountability, regular check-ins, professional support, social support, and ease of use [96]. A systematic review of adherence to mHealth apps for noncommunicable diseases found that intervention-related factors (usability, personalisation, feedback) and patient-related factors (motivation, digital literacy, mental health) both matter, and that adherence declines over time across domains [97].

Psychological factors are increasingly recognised. Weight-related information avoidance — the tendency to avoid weighing or logging after perceived setbacks — prospectively predicted poorer attendance and poorer self-monitoring of physical activity and weight in a 12-week behavioural programme [98]. Binge eating, counter-intuitively, predicted better adherence to digital self-monitoring in one study, perhaps because treatment-seeking participants with binge eating were more motivated or rigid [99]. Internalised weight bias predicted clinical attrition in a medical weight-loss clinic: 34% of patients did not return after the initial consultation, and dropouts had higher weight-bias internalisation scores [100].

5.3 Engagement is not monolithic

Engagement is often measured as logins, sessions, or days monitored, but these are proxies. A study of engagement with individual components in a young-adult trial found that overall engagement declined 33.6% from week 1 to week 26 and was related to weight loss; each 10% increase in feedback-report engagement over 6 months was associated with 0.27 kg loss [101]. A digital DPP study identified five engagement personas with different demographic and weight-loss profiles, suggesting that a single engagement metric obscures distinct user trajectories [102]. A machine-learning analysis of digital CBT for obesity found that engagement rate was associated with 8-week weight loss (r = −0.59) and that digital phenotypes predicted both engagement and outcomes [103].

The practical lesson is that “engagement” is a bundle of behaviours — opening the app, logging food, weighing, reading lessons, messaging a coach — and that different components may have different engagement curves. A programme that optimises logins may not optimise weight loss.

5.4 Retention interventions

The evidence on improving retention is thinner than the evidence on why people leave. Coach access to data improved self-monitoring and weight but did not significantly change attendance [8]. Text messaging can sustain contact at low cost, but effects on retention are inconsistent [92]. Financial incentives improve engagement and weight loss in the short term, but the effect fades when incentives stop [90]. A systematic review of digital health equity for adolescent obesity found that 82% of equity criteria were not addressed across 27 studies, which limits inference about what works for disadvantaged groups [104].

The most robust retention finding may be the simplest: programmes that offer human contact and a clear structure retain better than app-only products, and programmes that impose high monitoring burden retain worse. The seed trial’s negative food-diary signal and the broader burden literature point in the same direction [1,71,96].

Table 5. Retention and engagement benchmarks across digital programmes.

Programme / study Population Measure Result
NHS Digital Weight Management [24] 63,937 referrals Uptake / completion 50% uptake; 45% completion
NHS DWP economic evaluation [42] 31,753 completers Mean BMI change −1.18 to −1.20 kg/m²
UK tirzepatide-supported service [25] 19,693 12-month adherence 27%
German semaglutide-supported service [26] 4535 12-month persistence 12%
Australian digital weight-loss service [95] 5604 Mean adherence 171 days
Game of Stones [90] 585 men Follow-up 73% at 12 mo; 64% at 24 mo
Doody app trial [64] 1607 Weight reported at 6 mo 39.3%
Omada digital DPP [9] 599 12-month completion 80%

6. Clinical Significance and the Weight-Loss Magnitude Problem

6.1 What 1–3 kg means

The average effect of digital behavioural programmes — 1–3 kg beyond control at 6–12 months — sits below the threshold that guidelines and clinicians usually treat as clinically meaningful. The conventional target is 5–10% of body weight. In the Look AHEAD cohort, participants who lost 5–10% had substantially higher odds of clinically significant improvements in HbA1c (OR 3.52), systolic blood pressure (OR 1.56), diastolic blood pressure (OR 1.48), HDL cholesterol (OR 1.69), and triglycerides (OR 2.20) than weight-stable participants; larger losses produced greater benefits [105]. A narrative review of weight-loss thresholds concluded that 5–10% improves glycaemia, blood pressure, and HDL, while 10–15% is needed for obstructive sleep apnoea and steatohepatitis, and that mortality reduction may require more than 10% [106].

For a person weighing 100 kg, 1–3 kg is 1–3%, which is unlikely to change clinical risk factors materially. For a person weighing 80 kg with a BMI of 29, 2 kg may move them below a treatment threshold or improve motivation. The individual-level interpretation depends on baseline weight, comorbidity, and what the alternative is. The population-level interpretation is different: a 1 kg mean shift across a large population can produce measurable changes in diabetes incidence, as the DPP and its translations suggest [59,60]. The NHS DWP modelling estimated 0.096–0.110 QALYs gained per completer and a cost per QALY of £4,202, which is highly cost-effective by UK standards [42]. Small individual effects can be worthwhile at population scale if costs are low and harms are minimal.

6.2 The measurement problem

Weight change in digital programmes is measured with variable rigour. Trials use research scales; service evaluations use clinic or self-reported weights; commercial programmes use connected scales with unknown coverage. Differential missingness is the norm: participants who disengage are less likely to be weighed, and their absence is usually non-random. The German service’s contrast between 9.47% (LOCF) and 2.40% (worst-case) is a vivid illustration of how imputation choices can change the apparent effectiveness by a factor of four [26]. The NHS DWP’s intention-to-treat estimate (−2.2 kg) and completer estimate (−3.9 kg) differ by nearly 80% [24]. Any claim that a digital programme produces “5–7% weight loss” should be interrogated for whether it is a completer analysis, a per-protocol analysis, or an intention-to-treat analysis with a defensible imputation.

6.3 The cardiovascular outcome gap

Weight loss is a surrogate. The Look AHEAD trial randomised 5145 adults with type 2 diabetes to intensive lifestyle intervention or diabetes support and education; weight loss was 8.6% versus 0.7% at 1 year and 6.0% versus 3.5% at study end, yet the primary cardiovascular composite was not reduced (HR 0.95; 0.83–1.09) [107]. The trial was stopped for futility. That result does not mean behavioural weight loss is worthless — it improved risk factors, fitness, and quality of life — but it is a caution against claiming that a digital programme’s 2 kg weight loss will translate into fewer heart attacks. The evidence for hard outcomes from digital programmes specifically is essentially absent; the DPP translations show diabetes incidence reductions, but those are largely driven by weight loss and are observed in prediabetic populations [59,60].

6.4 The “small effect” defence

Advocates of digital programmes often argue that small effects are acceptable because the intervention is cheap and scalable. That argument is strongest when the programme is truly low-cost and low-burden, and weakest when it is a commercial product with per-user fees and high dropout. The NHS DWP’s £4,202/QALY is a good deal; a commercial app costing several hundred pounds per user with 60% dropout may not be. The relevant metric is not weight loss alone but cost per QALY, cost per completer, and equity of reach. The Game of Stones economic analysis — £628/QALY for texts alone, £9,748/QALY for texts plus incentives — shows that the cheapest component can be the most cost-effective even when its effect is modest [93].


7. Equity, Access, and the Digital Divide

7.1 Who digital programmes reach

Digital delivery changes the composition of the population reached. In the US National DPP, online organisations enrolled a higher proportion of men (27.1% vs 19.4%), younger adults (41.5% vs 28.2% aged 18–44), and non-Hispanic White participants (70.3%) than in-person organisations, which enrolled more Black participants (16.1%) and older adults (28.2% aged 65+) [14]. A fully digital DPP in health professional shortage areas reached 24.4% rural members and 30.8% in whole-county shortage areas, and only 7.6% lived in a zip code with an in-person DPP; enrollees were younger and had higher BMI than national comparators [48]. That is a genuine access gain for rural and younger populations. But the same programme reached a population that was more white and more educated than the in-person alternative, which is a different kind of access loss.

7.2 The inverse care law in digital form

The evidence on socioeconomic gradients is mixed but leans toward widening. A systematic review of eHealth weight-loss interventions for adults with low income found only nine studies (1606 participants), of which four reported significant weight reductions of small-to-moderate magnitude (mean −2.2 kg); many studies did not describe tailoring, and the evidence that these interventions are effective in this population was limited [108]. A review of digital health equity for adolescent obesity found that only eight of 27 studies targeted disadvantaged populations and that 82% of equity criteria were not addressed [104]. A review of eHealth in obesity care concluded that evidence is limited to short-term effects and is lacking by socioeconomic status and ethnic group, raising concerns about widening disparities [109].

Access is also geographically uneven. Only 27.9% of US counties had a National DPP site, and rural counties were much less likely than urban counties to have one (14.6% vs 48.4%) [49]. The Medicare DPP had 126 suppliers and 601 sites in 2019 — one site per 100,000 beneficiaries — and 75% of states had no sites or fewer than one site per 100,000 [110]. Reimbursement was $138.52 per beneficiary against delivery costs of $800, a gap of $661 that is likely to deter suppliers serving low-income populations [111].

7.3 Digital literacy and the “consistency” gradient

Even within a programme, engagement is socially patterned. In a commercial app trial, consistent trackers were more likely to be married or living with a partner, non-Hispanic White, and have higher health literacy than inconsistent trackers [51]. In the NHS DPP, uptake was associated with older age, perceived vulnerability to diabetes, self-efficacy, and belief in the programme’s efficacy; demographic and health factors played a minor role after these were accounted for [112]. That is a hopeful finding — beliefs are modifiable — but it also means that the people who most need help may be the least likely to believe that a digital programme will help them.

7.4 What might reduce disparities

The evidence on tailoring is thin but suggestive. The low-income eHealth review found that studies achieving significant results tended to use more tailoring strategies [108]. The NHS DWP triages participants to three intensity levels based on age, sex, ethnicity, and deprivation [113], which is an explicit attempt to match support to need. Qualitative work on the NHS digital DPP found that service users valued coach support and multiple app features, and that human contact improved engagement [58]. The NDPP’s online programmes reached more men and younger adults, which is a equity gain on those dimensions even as it is a loss on ethnicity.

A realistic equity position is that digital programmes are neither inherently equitable nor inherently inequitable; their distributional effect depends on referral, design, and support. A programme offered only through a smartphone app to a population with low digital literacy will widen gaps. A programme with telephone support, translated materials, community partnerships, and proactive outreach may narrow them. The evidence base does not yet tell us which specific adaptations work, and the absence of standard equity reporting is a barrier to learning [104].


8. Safety: Disordered Eating, Weight Stigma, Very-Low-Calorie Approaches, and Who Should Not Be in a Self-Directed Programme

8.1 Disordered eating

The relationship between weight-loss interventions and disordered eating is bidirectional and non-trivial. On one side, behavioural weight loss can improve binge-eating symptoms: in a 2×2 factorial trial of 136 patients with binge-eating disorder and obesity, behavioural weight loss produced significantly higher remission than no behavioural weight loss, and the combination of behavioural weight loss with naltrexone-bupropion produced 57.1% remission versus 17.7% with placebo [114]. On the other side, frequent self-weighing and app-based monitoring can reinforce restrictive or compensatory behaviours in vulnerable people. A prospective study of 762 college students found that frequent self-weighing predicted greater weight gain and higher onset of compensatory behaviours (OR 3.90), especially in those with binge eating [85]. A population-based adolescent study found that frequent self-weighing predicted disordered eating in females [86]. A survey of 106 young people found that almost half reported some negative experience or behaviour associated with healthy eating and fitness apps, and a review of the top 100 apps found that many lacked safeguards against maladaptive use [115].

The app ecosystem itself can be a risk environment. An analysis of a weight-loss app community found that 6.78% of 18,601 users had a goal BMI in the underweight range, and some users described the app as exacerbating disordered eating [116]. A proof-of-concept trial of self-monitoring in binge eating and weight management found that consumer-selected metrics did not improve satisfaction or compliance, but the study was small and the manipulation failed [117].

The counterpoint is important: a large pragmatic trial of a self-regulation app with no human contact found that the proportion of participants scoring above threshold for disordered-eating symptoms declined in the intervention arm relative to control (adjusted OR 0.51; 0.29–0.91) [64]. That suggests that well-designed behavioural support — focused on regular eating, self-weighing, and action planning rather than calorie restriction — need not be harmful and may be protective. The difference between the app-community findings and the trial findings may lie in design: apps that gamify restriction and thinness versus apps that promote routine and self-regulation.

For clinicians, the practical implications are: screen for binge eating, purging, and weight/shape overvaluation before referral; avoid prescriptive daily weighing in adolescents and in people with active binge eating; prefer programmes that emphasise regular eating and non-weight goals; and monitor for compensatory behaviours during follow-up. GLP-1 use in eating-disorder populations is not supported by sufficient evidence, and the literature explicitly cautions against assuming benefit [118].

8.2 Weight stigma

Weight stigma is not a side issue; it is a mechanism of harm and a barrier to engagement. Healthcare providers hold negative attitudes and stereotypes about people with obesity, and these attitudes influence judgement, interpersonal behaviour, and decision-making; experiences or expectations of poor treatment cause stress, avoidance of care, mistrust, and poor adherence [119]. A qualitative evidence synthesis of 32 studies identified three themes: verbal and non-verbal communication of stigma, stigma’s impact on care provision, and systemic barriers to healthcare [120]. Weight stigma is prospectively associated with mortality and chronic disease, and it can drive weight gain through behavioural and physiological pathways [121]. Internalised weight bias predicts clinical attrition [100]. Weight-discriminatory policies such as BMI cutoffs and weight-loss requirements limit access to care [122].

Digital programmes can amplify stigma in at least three ways: through the language of referral (“obesity” as a label, “failed diet” framing), through app content that moralises food and weight, and through the visibility of weight data to coaches or peers. The seed trial’s programme used a commercial service with a health-coach model; the trial did not report stigma-related outcomes, which is a gap. Guidelines increasingly emphasise non-stigmatising care: the Swiss guidance mandates a non-stigmatising, empathetic approach and explicitly aims to reduce weight bias and internalised stigma [123]; Obesity Canada’s guideline is framed around chronic disease and patient-centred care [124]; and reviews of stigma-reduction strategies identify education, causal information, empathy evocation, weight-inclusive approaches, and mixed methods as promising [125].

8.3 Very-low-calorie diets and total diet replacement

Very-low-calorie diets (VLCDs) and total diet replacement (TDR) are not the same as digital behavioural programmes, but they are increasingly delivered through digital or remote services, and clinicians need to understand their risk-benefit profile.

The DiRECT trial is the strongest evidence for TDR in type 2 diabetes: 306 adults in UK primary care were randomised to a structured programme of 825–853 kcal/day formula diet for 3–5 months, stepped food reintroduction, and structured maintenance support, or to guideline-based care. At 12 months, 46% of the intervention group achieved remission and 24% lost ≥15 kg [126]. At 2 years, 36% were in remission and mean weight loss was 7.6 kg [127]. At 5 years, continued low-intensity support was associated with maintained weight loss in a subset, and 29 of 36 participants who maintained >10 kg loss at 2 years were in remission [128]. The programme requires medical supervision, deprescribing of antihypertensive and antidiabetic drugs, and nutritional monitoring [129].

Long-term observational data are less encouraging. A 15-year follow-up of 325 patients treated with a VLCD in UK community clinics found that weight had returned to baseline (97.9 kg to 100 kg), only 5.9% maintained >10% loss, and diabetes remission was rare (2 of 10 at 5 years) [130]. A systematic review of VLCDs in children and adolescents found substantial short-term weight loss but concluded that safety could not be established from the available literature [131]. A review of preconception VLCDs raised concerns about maternal nutrition and ketosis exposure in early pregnancy [132].

The safety issues specific to digital delivery are supervision and screening. A self-directed 800 kcal/day programme purchased online, without medical assessment or deprescribing, is a different risk profile from DiRECT. Clinicians should ask whether a digital programme includes medical oversight, contraindication screening (eating disorders, pregnancy, type 1 diabetes, renal or hepatic disease, gallstones), and a structured reintroduction and maintenance phase.

8.4 Who should not be in a self-directed programme

The evidence supports a conservative list of exclusions or precautions for self-directed digital programmes: active or past eating disorder (especially binge eating, purging, or anorexia nervosa), pregnancy or planning pregnancy, type 1 diabetes, severe obesity with uncontrolled comorbidities, recent bariatric surgery, significant psychiatric instability, and inability to use the digital interface without support. These are clinical judgements, not trial-derived criteria; the trials rarely enrolled such patients, so the evidence of harm is indirect [85,86,116,118,132]. The safer approach is to offer supported programmes with clinical oversight, or to refer to specialist care.


9. Guidelines and Health-System Positioning

9.1 What the guidelines actually say

Guidelines converge on behavioural intervention as the foundation of obesity care, but they differ in how explicitly they address digital delivery.

The US Preventive Services Task Force recommends that clinicians offer or refer adults with a BMI of ≥30 to intensive, multicomponent behavioural interventions (B recommendation). It found adequate evidence that these interventions produce clinically significant improvements in weight status and reduce type 2 diabetes incidence in adults with elevated glucose, that behaviour-based maintenance interventions are of moderate benefit, and that harms are small to none [34]. The recommendation is for intensive, multicomponent, in-person or referred programmes; it does not specifically endorse app-only products.

NICE updated its obesity guidance in 2025 (NG246), replacing earlier clinical guidelines. The updated guidance retains a tiered, person-centred approach in which multicomponent behavioural interventions are the foundation, with pharmacotherapy and bariatric surgery for eligible people, and it recognises digital and remote delivery as delivery modes within the pathway [133]. NICE’s health-economic standards are demanding, and the NHS Digital Weight Management Programme was commissioned against them.

The American Diabetes Association Standards of Care 2026 include a dedicated section on obesity and weight management for the prevention and treatment of diabetes [134]. The Standards recommend intensive behavioural lifestyle intervention as a foundation, with pharmacotherapy and metabolic surgery for eligible patients, and emphasise sustained weight maintenance. The 2026 Standards are consistent with the 2025 edition in treating obesity as a chronic, relapsing condition requiring long-term care.

Obesity Canada published a comprehensive adult clinical practice guideline in 2020 that framed obesity as a chronic disease and recommended individualized, multi-component management, with behavioural intervention as the foundation and pharmacotherapy and bariatric surgery as options for eligible patients [124]. The guideline is notable for its explicit attention to weight bias and patient-centred care.

The European Association for the Study of Obesity published a new framework for diagnosis, staging, and management in 2024, moving beyond BMI alone to incorporate adiposity and comorbidity, and positioning behavioural, pharmacological, and surgical treatments within a chronic-care model [135]. EASO’s 2025 pharmacological framework extends this to incretin-based therapy [136].

The World Health Organization issued a 2026 guideline on GLP-1 therapies for obesity that recommends long-term GLP-1 treatment combined with intensive behavioural therapy, graded conditional because of limited long-term data, cost, system readiness, and equity considerations [137].

A scoping review of 38 guidelines from 18 countries found that nutrition and physical activity were included in 100%, psychology in 97%, pharmacotherapy in 84%, and bariatric surgery in 82%; only 21% included recommendations for people with eating disorders, and consumer representation was limited [138]. The review concluded that future guidelines would benefit from consumer involvement and explicit attention to weight bias and stigma.

Table 6. Guideline positions relevant to digital behavioural weight management.

Guideline Core recommendation Digital delivery Strength / caveat
USPSTF 2018 [34] Offer/refer adults with BMI ≥30 to intensive multicomponent behavioural intervention Not specified; referral may be to digital providers Grade B; harms small to none
NICE NG246 2025 [133] Tiered, person-centred multicomponent behavioural care; pharmacotherapy/surgery for eligible Digital/remote delivery recognised within pathway Updated 2025; UK commissioning context
ADA Standards 2026 [134] Intensive behavioural lifestyle intervention as foundation; pharmacotherapy/surgery for eligible Not specified Annual update; diabetes-focused
Obesity Canada 2020 [124] Obesity as chronic disease; individualised multicomponent care Not specified Strong emphasis on weight bias
EASO 2024 framework [135] Diagnosis/staging by adiposity and comorbidity; chronic-care model Not specified European; staging approach
WHO 2026 [137] Long-term GLP-1 plus intensive behavioural therapy Not specified Conditional recommendation; equity concerns

9.2 The gap between guideline and procurement

Guidelines describe clinical care; procurement describes what a health system will pay for. The two are not aligned. A guideline may recommend “intensive multicomponent behavioural intervention” without specifying the dose, duration, or delivery mode; a procurement specification must. The NHS DWP’s design — a 12-week behavioural programme with three intensity levels, triaged by age, sex, ethnicity, and deprivation — is an attempt to operationalise the guideline at population scale [113]. Its early outcomes (50% uptake, 45% completion, −2.2 kg intention-to-treat) show both the promise and the limits of that operationalisation [24].

The seed trial’s most policy-relevant finding may be that component-level testing is feasible within a commercial service. If commissioners required vendors to run factorial or SMART optimisation studies as a condition of contract, the field would accumulate the evidence it currently lacks. That is a governance proposal, not an evidence-based recommendation, and it would need to be balanced against procurement timelines and vendor capacity.


10. The GLP-1 Era: How Incretin Pharmacotherapy Changes the Role of Behavioural Digital Programmes

10.1 The efficacy gap

Incretin-based therapies have changed the therapeutic landscape. In STEP 1, once-weekly semaglutide 2.4 mg produced −14.9% weight loss at 68 weeks versus −2.4% with placebo, both with lifestyle intervention [28]. In SURMOUNT-1, tirzepatide produced −15.0% to −20.9% depending on dose versus −3.1% with placebo [45]. These effects are an order of magnitude larger than those of digital behavioural programmes, and they have shifted the clinical question from “can behavioural support produce weight loss?” to “what is behavioural support for when pharmacotherapy is available?”

10.2 Combination evidence

The combination literature is more encouraging than either modality alone. In STEP 3, semaglutide plus intensive behavioural therapy (30 counselling visits) produced −16.0% weight loss versus −5.7% with placebo plus the same behavioural therapy — a 10.3 percentage-point difference [29]. In SURMOUNT-3, participants who achieved ≥5% loss during a 12-week intensive lifestyle intervention and then received tirzepatide lost an additional 18.4% by week 72, versus a 2.5% gain with placebo [30]. In the DiRECT extension and other TDR studies, structured behavioural support was essential to maintaining weight loss [128]. In a randomised post-treatment analysis, participants who had received exercise plus liraglutide maintained better weight loss one year after stopping treatment than those who had received liraglutide alone (−5.1 kg difference), suggesting that behavioural components can improve the durability of pharmacological weight loss [31].

10.3 Discontinuation and maintenance

The discontinuation literature is the strongest argument for behavioural programmes as maintenance infrastructure. In the STEP 1 extension, participants regained 11.6 of the 17.3 percentage points lost after treatment withdrawal, leaving a net loss of 5.6% at week 120 [32]. A 2026 review reported that up to 65% of GLP-1 users discontinue within a year and that two-thirds of lost weight is typically regained within a year of discontinuation [33]. In a real-world cohort of 7938 patients who discontinued semaglutide or tirzepatide, 19.6% restarted the index medication and 35.2% received an alternative obesity treatment within a year; mean weight change after discontinuation was small on average but highly variable [139]. The SURMOUNT-MAINTAIN and ATTAIN-MAINTAIN trials show that continuing or switching pharmacotherapy maintains weight loss far better than placebo [140,141].

The inference for behavioural programmes is clear but should be labelled as inference: if most people cannot or will not stay on pharmacotherapy indefinitely, then a scalable behavioural maintenance programme is a rational complement. The evidence that digital behavioural programmes prevent post-GLP-1 regain is thin. The most relevant data come from small trials and service evaluations: a remote programme combining GLP-1 therapy with app-based behavioural support reported 12-month losses of 22.1% (tirzepatide) and 17.1% (semaglutide) in completers, but the design was retrospective and uncontrolled [142]; a UK service evaluation of a tirzepatide-plus-digital programme reported 22.7% loss at 12 months in 60 participants [143]. These are promising but not causal evidence that the behavioural component adds value.

10.4 The cost-offset question

Health systems facing GLP-1 costs have an obvious interest in cheaper behavioural alternatives and in behavioural programmes that improve persistence or reduce discontinuation. The economic case is plausible but unproven. The NHS DWP costs £133.49 per completer and produces 0.096–0.110 QALYs at £4,202/QALY [42]. Semaglutide and tirzepatide cost orders of magnitude more per patient-year, and their cost-effectiveness depends on price, discontinuation, and cardiovascular benefit [144]. A behavioural programme that keeps a patient on GLP-1 longer, or that maintains loss after discontinuation, could be cost-saving; a behavioural programme that substitutes for GLP-1 in a patient who needs pharmacotherapy could be harmful. The distinction between adjunct, maintenance, and substitute is the key policy question, and the evidence supports the first two more than the third.

10.5 What this means for digital programmes

Three roles are defensible:

  1. Adjunct to pharmacotherapy. Deliver the behavioural curriculum, self-monitoring, and coaching that STEP 3 and SURMOUNT-3 show can improve outcomes when combined with medication. The digital modality is attractive because it can provide frequent contact at low cost.
  2. Maintenance after discontinuation. Provide structured self-monitoring, self-weighing, and coach contact during and after medication tapering. The evidence for efficacy is indirect, but the mechanism is plausible and the risk is low.
  3. Triage and supportive care. Identify people who need pharmacotherapy or surgery, manage expectations, address weight stigma, and support nutrition and physical activity. This is a clinical service, not a weight-loss product.

What is not defensible is marketing a digital programme as an alternative to pharmacotherapy for people with severe obesity or weight-related complications. The efficacy gap is too large, and the guidelines are clear that pharmacotherapy and surgery are options for eligible patients [134,135,137].

Table 7. Selected incretin combination and maintenance trials.

Trial Design Intervention Weight result Relevance to behavioural programmes
STEP 1 [28] RCT, n = 1961 Semaglutide 2.4 mg + lifestyle −14.9% vs −2.4% at 68 wk Lifestyle was background, not tested additively
STEP 3 [29] RCT, n = 611 Semaglutide + intensive behavioural therapy −16.0% vs −5.7% Behavioural therapy added to drug; both arms received it
SURMOUNT-3 [30] RCT, n = 579 Tirzepatide after ≥5% lifestyle-induced loss −18.4% additional vs +2.5% Lifestyle induction followed by drug
STEP 1 extension [32] Extension, n = 327 Withdrawal of semaglutide and lifestyle Regained 11.6 of 17.3 pp; net −5.6% Maintenance is the weak point
Jensen 2024 [31] Post-treatment RCT analysis Exercise ± liraglutide, then withdrawal Combination maintained better at 1 yr Behavioural component may improve durability
SURMOUNT-MAINTAIN [140] RCT Continue tirzepatide vs reduce vs placebo Continuing maintained loss Pharmacotherapy, not behaviour, drives maintenance

11. Implementation and Procurement Lessons: What Health Systems and Clinicians Should Ask a Vendor

11.1 The implementation evidence

Implementation research on digital weight programmes identifies consistent barriers and facilitators. A qualitative study of the NHS National Enhanced Service, which paid practices £11.50 per eligible referral, found that the incentive solidified the position of staff already supportive of referral and reduced dissonance among those less supportive; facilitators included a coherent national policy, a local champion, and framing the incentive as benefiting the practice; barriers included workload, complex referral systems, and scepticism about service benefit [145]. An interrupted time-series analysis found that the incentive increased referrals from about 1 to 4 per 1,000 patients per month, with no evidence of differential effect by age, sex, ethnicity, or socioeconomic status; weight-management advice was unchanged and was at least three times more common than referral [146]. A mixed-methods evaluation of patient experience found that about half of respondents felt the programme helped them change diet, activity, or wellbeing, and that participants from minority ethnic groups were less likely to describe a positive experience [113].

Primary-care referral studies identify similar themes: perceived patient barriers, doubts about programme effectiveness for disadvantaged groups, limited clinician self-efficacy, and a need for feedback loops [147]. A cluster RCT of clinician education, a prediabetes champion, and an EHR report increased referral from 1.5% to 6.9% of eligible patients [148]. Rural EHR-based referral programmes emphasised starting small with a provider champion and designing a process that does not exceed resources [149]. A CFIR-based qualitative study found that structural characteristics, compatibility, goals and feedback, and leadership engagement distinguished high-reach from low-reach organisations [150]. The American Medical Association’s best-practice recommendations organise implementation into maturity phases and six categories: organisational support, workforce and funding, promotion and dissemination, clinical integration and support, evaluation and outcomes, and programme [151].

11.2 What procurement should specify

The evidence supports a due-diligence checklist that goes beyond feature lists.

Clinical content and dose. Does the programme deliver a structured behavioural curriculum consistent with DPP or equivalent? What is the dose (number of sessions, duration, coach contact hours)? Is there a maintenance phase? The evidence favours programmes with human contact, structured self-monitoring, and a maintenance component [3,15,17,29].

Engagement and retention. What are the observed uptake, completion, and 12-month retention rates in comparable populations? Are they reported with a denominator that includes non-initiators? The NHS DWP’s 50% uptake and 45% completion are a useful benchmark [24]; a vendor reporting 80% completion among app users is not reporting the same metric.

Outcome measurement. Is weight measured objectively or self-reported? What is the missing-data handling? Is there a pre-specified primary outcome and a published analysis plan? The difference between LOCF and worst-case imputation in the German service (9.47% vs 2.40%) shows why this matters [26].

Safety and stigma. Does the programme screen for eating disorders and contraindications? Does it avoid prescriptive daily weighing in adolescents and people with binge eating? Does it use non-stigmatising language and imagery? Does it have a pathway for participants who deteriorate? The evidence on self-weighing harms and app-related disordered eating supports explicit safeguards [85,86,116].

Equity. Who is reached? Are uptake and outcomes reported by age, sex, ethnicity, deprivation, and digital literacy? What adaptations exist for low-literacy, low-income, and non-smartphone populations? The evidence that digital programmes can widen disparities is sufficient to require monitoring [14,47,104,108].

Data governance and coaching model. Who sees the data? Are coaches clinically trained or lay? What is the escalation pathway for safeguarding concerns? Is the coach’s access to self-monitoring data used to drive accountability, as the Butryn experiment suggests it should be [8]?

Cost and cost-effectiveness. What is the cost per participant, per completer, and per QALY? Are claims based on a modelled lifetime horizon or observed outcomes? The NHS DWP’s £4,202/QALY and Game of Stones’ £628–£9,748/QALY provide benchmarks [42,93].

Regulatory status. Is the product a wellness app or a regulated medical device/digital therapeutic? The distinction matters for evidence standards, advertising claims, and liability [152]. Many commercial weight-management apps are wellness products with limited clinical evidence; a smaller number have regulatory clearance or reimbursement as digital therapeutics.

Table 8. Vendor due-diligence questions mapped to evidence.

Domain Question Evidence anchor
Clinical content Is the curriculum DPP-equivalent? What is the dose? [9,29]
Coaching Is there human contact? Can coaches see self-monitoring data? [8,15,17]
Self-monitoring What is logged, how often, and is it simplified for burden? [6,53,71]
Feedback Is feedback tailored, actionable, and non-judgemental? [6,7]
Retention What are uptake, completion, and 12-month retention? [24,25]
Outcomes Objective or self-reported? How is missing data handled? [26]
Safety Eating-disorder screening? Weight-stigma safeguards? [85,86,116]
Equity Reporting by ethnicity, deprivation, digital literacy? [14,104]
Economics Cost per completer and per QALY? [42,93]
Regulation Wellness app or regulated digital therapeutic? [152]

11.3 The commissioning dilemma

Commissioners face a genuine dilemma. The evidence for digital programmes is strongest at the level of average effect and weakest at the level of component specification. A commissioner who demands component-level evidence will find little; a commissioner who accepts average-effect evidence will fund bundles that may contain inert or harmful features. The seed trial points to a way out: require vendors to participate in component-level evaluation as a condition of contract, and use the results to iterate the specification. That is a governance innovation, not an established practice, and it would need to be tested.


12. Practical Bottom Line for Clinicians

  1. Recommend digital programmes as a low-burden option, not a treatment for severe obesity. The average effect is 1–3 kg beyond control at 6–12 months; coach-supported and DPP-derived programmes do better, app-only programmes do worse [9,12,17,38]. Set expectations at 3–5% and be explicit that this is a starting point, not a cure.

  2. Prefer programmes with human contact and structured self-monitoring. The coaching increment is about 2 kg in direct comparisons, and coach access to self-monitoring data adds about 0.9 percentage points at 12 weeks [8,15]. App-only products have not demonstrated durable benefit in meta-analysis [12].

  3. Do not overvalue the food diary. Logging correlates with loss, but adding a review-and-feedback layer to an already-monitoring programme did not help and may have hurt in the seed trial [1]. Ask what the programme does with the data, not just whether it collects it.

  4. Use self-weighing judiciously. It is a cheap, modestly effective component in adults, especially when consistent (≥5–6 days/week), but it is associated with harm in adolescents and people with binge eating [77,84–86]. Screen before recommending.

  5. Screen for eating disorders and weight stigma. Ask about binge eating, purging, and weight/shape overvaluation; use non-stigmatising language; and avoid programmes that moralise food or thinness [116,118–120].

  6. Position digital programmes correctly in the GLP-1 era. They are an adjunct to pharmacotherapy, a maintenance scaffold after discontinuation, and a triage partner — not a substitute for medication in people with severe obesity or weight-related complications [29,30,32,33,137].

  7. Follow up. A digital referral is not a handoff. The primary-care counselling literature suggests that clinician engagement and empathy are associated with better outcomes [65]. A brief check-in at 4–6 weeks, and again at 3 months, is reasonable.

  8. Be sceptical of completer analyses and “percentage weight loss” claims. Ask for the denominator, the missing-data method, and the 12-month retention rate [24,26].


13. Limitations

13.1 Limitations of the seed trial

The seed trial has several limitations that bear on interpretation. First, it was conducted inside a single commercial programme (Second Nature), with the company as delivery partner and industrial partner on the lead author’s studentship, and several authors employed by the company [1]. This creates a conflict of interest and limits generalisability to other programmes, populations, and delivery models. Second, the population was restricted to adults with BMI ≥21 kg/m², 90.8% of whom had BMI ≥25; people with severe obesity, type 2 diabetes, and eating disorders were not the target population, and the results may not apply to them. Third, the primary outcome was weight at 16 weeks, a short horizon for a chronic condition; the 24-week secondary outcome is better but still short. Fourth, the trial used service-collected data, which may be vulnerable to differential missingness and measurement error. Fifth, the component effects reported in the abstract are limited: the webchat and goal-setting results are not described, and the full factorial estimates, interaction tests, engagement analyses, and drop-out results are not available in the abstract. This review could not access the full text, so it cannot state those results. Sixth, the per-protocol analysis of the introductory call attenuated the effect, suggesting that uptake and engagement moderate the benefit; the trial was not powered to detect interactions of that kind. Seventh, the trial did not measure eating behaviour, psychological outcomes, weight stigma, or adverse events, so it cannot speak to safety. Eighth, the 0.75 kg decision threshold is a service-improvement threshold, not a clinical-significance threshold, and the trial’s conclusions should be read in that frame.

13.2 Limitations of this review

This review has methodological limitations. It relied on PubMed abstracts and structured records for most sources; full texts were not retrieved for the majority of included studies, so numbers are taken from abstracts unless otherwise stated. The seed trial’s full text and the full texts of NICE NG246, ADA Standards of Care 2026, Obesity Canada 2020, and EASO 2024 were not accessible in this environment; their positions are described at a general level from indexed metadata and publicly available summaries, and specific recommendations should be checked against the source documents before clinical use. The review is not a systematic review: searches were purposive rather than exhaustive, no formal risk-of-bias assessment was performed, and publication bias was not assessed. The reference list is large but not comprehensive; important trials and reviews may have been missed. The component-level evidence is heterogeneous in dose, population, comparator, and outcome timing, and the summary effects in Table 4 are indicative rather than pooled estimates. Several claims about the GLP-1 era are inferences from indirect evidence, and they are labelled as such in the text. Finally, the review was prepared for personal knowledge, not for guideline development or regulatory submission.

13.3 Conflicting interests and funding

The seed trial’s competing interests are material and are restated here: Second Nature was the commercial delivery partner and covered delivery costs; the company was the industrial partner on the lead author’s MRC iCASE studentship; and multiple co-authors were Second Nature employees. The trial’s funders had no role in design, data collection, analysis, interpretation, or writing, according to the published statement [1]. This review has no competing interests and received no funding.



References

  1. Wren GM, Koutoukidis DA, Scragg J, et al. Optimising a digitally delivered behavioural weight loss programme: a factorial cluster randomised controlled trial. npj Digit Med. 2026 (published 19 September 2026). doi:10.1038/s41746-026-03231-y

  2. Wren GM, Koutoukidis DA, Scragg J, Preston G, Hennessy M, Estephane D, et al. Optimising a digitally delivered behavioural weight loss programme: study protocol for a factorial cluster randomised controlled trial. Trials. 2024;25(1):477. PMID: 39003481.

  3. Chatterjee A, Prinz A, Gerdes M, Martinez S. Digital Interventions on Healthy Lifestyle Management: Systematic Review. J Med Internet Res. 2021;23(11):e26931. PMID: 34787575.

  4. Sniehotta FF, Evans EH, Sainsbury K, Adamson A, Batterham A, Becker F, et al. Behavioural intervention for weight loss maintenance versus standard weight advice in adults with obesity: A randomised controlled trial in the UK (NULevel Trial). PLoS Med. 2019;16(5):e1002793. PMID: 31063507.

  5. Moulaei K, Moulaei R, Bahaadinbeigy K, Bastaminejad S. Effect of telemedicine interventions on weight loss: A systematic review and meta-analysis. J Telemed Telecare. 2025;31(9):1216-1227. PMID: 39370913.

  6. Burke LE, Sereika SM, Parmanto B, Bizhanova Z, Kariuki JK, Cheng J, et al. Effect of tailored, daily feedback with lifestyle self-monitoring on weight loss: The SMARTER randomized clinical trial. Obesity (Silver Spring). 2022;30(1):75-84. PMID: 34898011.

  7. Krukowski R, Kim H, Stansbury M, Li Q, Sen S, Farage G, et al. Importance of Multiple Reinforcing Comments and Areas for Change in Optimizing Dietary and Exercise Self-Monitoring Feedback in Behavioral Weight Loss Programs: Factorial Design. J Med Internet Res. 2020;22(11):e18104. PMID: 33226348.

  8. Butryn ML, Miller NA, Hagerman CJ, Arigo D, LaFata E, Zhang F, et al. Coach access to digital self-monitoring data: an experimental test of short-term effects in behavioral weight-loss treatment. Obesity (Silver Spring). 2024;32(11):2111-2119. PMID: 39358838.

  9. Katula JA, Dressler EV, Kittel CA, Harvin LN, Almeida FA, Wilson KE, et al. Effects of a Digital Diabetes Prevention Program: An RCT. Am J Prev Med. 2022;62(4):567-577. PMID: 35151522.

  10. Toro-Ramos T, Michaelides A, Anton M, Karim Z, Kang-Oh L, Argyrou C, et al. Mobile Delivery of the Diabetes Prevention Program in People With Prediabetes: Randomized Controlled Trial. JMIR Mhealth Uhealth. 2020;8(7):e17842. PMID: 32459631.

  11. Staite E, Bayley A, Al-Ozairi E, Stewart K, Hopkins D, Rundle J, et al. A Wearable Technology Delivering a Web-Based Diabetes Prevention Program to People at High Risk of Type 2 Diabetes: Randomized Controlled Trial. JMIR Mhealth Uhealth. 2020;8(7):e15448. PMID: 32459651.

  12. Metzendorf MI, Wieland LS, Richter B. Mobile health (m-health) smartphone interventions for adolescents and adults with overweight or obesity. Cochrane Database Syst Rev. 2024;2(2):CD013591. PMID: 38375882.

  13. Roth L, Ordnung M, Forkmann K, Mehl N, Horstmann A. A randomized-controlled trial to evaluate the app-based multimodal weight loss program zanadio for patients with obesity. Obesity (Silver Spring). 2023;31(5):1300-1310. PMID: 37140392.

  14. Cannon MJ, Ng BP, Lloyd K, Reynolds J, Ely EK. Delivering the National Diabetes Prevention Program: Assessment of Enrollment in In-Person and Virtual Organizations. J Diabetes Res. 2022;2022:2942918. PMID: 35118160.

  15. Spring B, Pfammatter AF, Scanlan L, Daly E, Reading J, Battalio S, et al. An Adaptive Behavioral Intervention for Weight Loss Management: A Randomized Clinical Trial. JAMA. 2024;332(1):21-30. PMID: 38744428.

  16. Godino JG, Golaszewski NM, Norman GJ, Rock CL, Griswold WG, Arredondo E, et al. Text messaging and brief phone calls for weight loss in overweight and obese English- and Spanish-speaking adults: A 1-year, parallel-group, randomized controlled trial. PLoS Med. 2019;16(9):e1002917. PMID: 31553725.

  17. Hesseldal L, Christensen JR, Olesen TB, Olsen MH, Jakobsen PR, Laursen DH, et al. Long-term Weight Loss in a Primary Care-Anchored eHealth Lifestyle Coaching Program: Randomized Controlled Trial. J Med Internet Res. 2022;24(9):e39741. PMID: 36149735.

  18. Mathioudakis N, Lalani B, Abusamaan MS, Alderfer M, Alver D, Dobs A, et al. An AI-Powered Lifestyle Intervention vs Human Coaching in the Diabetes Prevention Program: A Randomized Clinical Trial. JAMA. 2025;334(23):2079-2089. PMID: 41144242.

  19. Patel ML, Wakayama LN, Bennett GG. Self-Monitoring via Digital Health in Weight Loss Interventions: A Systematic Review Among Adults with Overweight or Obesity. Obesity (Silver Spring). 2021;29(3):478-499. PMID: 33624440.

  20. Raber M, Liao Y, Rara A, Schembre SM, Krause KJ, Strong L, et al. A systematic review of the use of dietary self-monitoring in behavioural weight loss interventions: delivery, intensity and effectiveness. Public Health Nutr. 2021;24(17):5885-5913. PMID: 34412727.

  21. Muralidharan S, Ranjani H, Mohan Anjana R, Jena S, Tandon N, Gupta Y, et al. Engagement and Weight Loss: Results from the Mobile Health and Diabetes Trial. Diabetes Technol Ther. 2019;21(9):507-513. PMID: 31184922.

  22. Stansbury ML, Harvey J, Krukowski RA, Pellegrini CA, Wang X, West DS. Describing Transitions in Adherence to Physical Activity Self-monitoring and Goal Attainment in an Online Behavioral Weight Loss Program: Secondary Analysis of a Randomized Controlled Trial. J Med Internet Res. 2022;24(1):e30673. PMID: 35089159.

  23. Carpenter CA, Eastman A, Ross KM. Consistency With and Disengagement From Self-monitoring of Weight, Dietary Intake, and Physical Activity in a Technology-Based Weight Loss Program: Exploratory Study. JMIR Form Res. 2022;6(2):e33603. PMID: 35179513.

  24. Taylor K, Indulkar T, Thompson B, Pinkard C, Barron E, Frost T, et al. Early outcomes of referrals to the English National Health Service Digital Weight Management Programme. Obesity (Silver Spring). 2024;32(6):1083-1092. PMID: 38644161.

  25. Talay L, Hom J, Scott T, Ahuja N. 12-Month Weight Loss and Adherence Predictors in a Real-World UK Tirzepatide-Supported Digital Obesity Service: A Retrospective Cohort Analysis. Healthcare (Basel). 2025;14(1). PMID: 41516991.

  26. Talay L, Hom J, Tan M, Ahuja N. Real-World Effectiveness and 12-Month Persistence of a Semaglutide-Supported Digital Weight-Loss Service: A Retrospective Cohort Study in Germany. Diabetes Obes Metab. 2026;28(8):6985-6996. PMID: 42186215.

  27. Behr H, Earl S, Ho AS, Lee J, Mitchell ES, McCallum M, et al. Changes in Health-Promoting Behaviors and Their Association with Weight Loss, Retention, and Engagement on a Digital Program: Prospective Study. Nutrients. 2022;14(22). PMID: 36432498.

  28. Wilding JPH, Batterham RL, Calanna S, Davies M, Van Gaal LF, Lingvay I, et al. Once-Weekly Semaglutide in Adults with Overweight or Obesity. N Engl J Med. 2021;384(11):989-1002. PMID: 33567185.

  29. Wadden TA, Bailey TS, Billings LK, Davies M, Frias JP, Koroleva A, et al. Effect of Subcutaneous Semaglutide vs Placebo as an Adjunct to Intensive Behavioral Therapy on Body Weight in Adults With Overweight or Obesity: The STEP 3 Randomized Clinical Trial. JAMA. 2021;325(14):1403-1413. PMID: 33625476.

  30. Wadden TA, Chao AM, Machineni S, Kushner R, Ard J, Srivastava G, et al. Tirzepatide after intensive lifestyle intervention in adults with overweight or obesity: the SURMOUNT-3 phase 3 trial. Nat Med. 2023;29(11):2909-2918. PMID: 37840095.

  31. Jensen SBK, Blond MB, Sandsdal RM, Olsen LM, Juhl CR, Lundgren JR, et al. Healthy weight loss maintenance with exercise, GLP-1 receptor agonist, or both combined followed by one year without treatment: a post-treatment analysis of a randomised placebo-controlled trial. EClinicalMedicine. 2024;69:102475. PMID: 38544798.

  32. Wilding JPH, Batterham RL, Davies M, Van Gaal LF, Kandler K, Konakli K, et al. Weight regain and cardiometabolic effects after withdrawal of semaglutide: The STEP 1 trial extension. Diabetes Obes Metab. 2022;24(8):1553-1564. PMID: 35441470.

  33. Moiz A, Filion KB, Knäuper B, Tsoukas MA, Brazeau AS, Zolotarova T, et al. Weight maintenance after discontinuation of GLP-1 therapies. EClinicalMedicine. 2026;96:103992. PMID: 42256676.

  34. US Preventive Services Task Force, Curry SJ, Krist AH, Owens DK, Barry MJ, Caughey AB, et al. Behavioral Weight Loss Interventions to Prevent Obesity-Related Morbidity and Mortality in Adults: US Preventive Services Task Force Recommendation Statement. JAMA. 2018;320(11):1163-1171. PMID: 30326502.

  35. Lalani B, Mathioudakis N. The U.S. National Diabetes Prevention Program Demonstrates Effectiveness With Insufficient Reach. Diabetes Care. 2026;49(9):1551-1553. PMID: 42623546.

  36. Islam MM, Poly TN, Walther BA, Jack Li YC. Use of Mobile Phone App Interventions to Promote Weight Loss: Meta-Analysis. JMIR Mhealth Uhealth. 2020;8(7):e17039. PMID: 32706724.

  37. Chew HSJ, Koh WL, Ng JSHY, Tan KK. Sustainability of Weight Loss Through Smartphone Apps: Systematic Review and Meta-analysis on Anthropometric, Metabolic, and Dietary Outcomes. J Med Internet Res. 2022;24(9):e40141. PMID: 36129739.

  38. Pujia C, Ferro Y, Mazza E, Maurotti S, Montalcini T, Pujia A. The Role of Mobile Apps in Obesity Management: Systematic Review and Meta-Analysis. J Med Internet Res. 2025;27:e66887. PMID: 40327853.

  39. Cavero-Redondo I, Martinez-Vizcaino V, Fernandez-Rodriguez R, Saz-Lara A, Pascual-Morena C, Álvarez-Bueno C. Effect of Behavioral Weight Management Interventions Using Lifestyle mHealth Self-Monitoring on Weight Loss: A Systematic Review and Meta-Analysis. Nutrients. 2020;12(7). PMID: 32635174.

  40. Nahum-Shani I, Dziak JJ, Venera H, Pfammatter AF, Spring B, Dempsey W. Design of experiments with sequential randomizations on multiple timescales: the hybrid experimental design. Behav Res Methods. 2024;56(3):1770-1792. PMID: 37156958.

  41. Tate DF, Lytle LA, Sherwood NE, Haire-Joshu D, Matheson D, Moore SM, et al. Deconstructing interventions: approaches to studying behavior change techniques across obesity interventions. Transl Behav Med. 2016;6(2):236-43. PMID: 27356994.

  42. Amies-Cull B, Haffner S, Scarborough P, Mounsey S, Aveyard P, Jebb S. Estimating the cost-effectiveness of the National Weight Management Enhanced Service and NHS Digital Weight Management Programme in England: a modelling study. BMJ Open. 2026;16(10):e117143. PMID: 42864125.

  43. Smith DH, O’Keeffe-Rosetti M, Fitzpatrick SL, Mayhew M, Firemark AJ, Gruß I, et al. Costs and Cost-Effectiveness of Implementing a Digital Diabetes Prevention Program in a Large, Integrated Health System. Perm J. 2022;26(3):74-82. PMID: 36038966.

  44. Park S, Ballreich J, Ward T, Shi L. Cost-effectiveness analysis of a digital diabetes-prevention programme versus an in-person diabetes-prevention programme in people with prediabetes in the United States. Diabetes Obes Metab. 2024;26(10):4522-4534. PMID: 39056211.

  45. Jastreboff AM, Aronne LJ, Ahmad NN, Wharton S, Connery L, Alves B, et al. Tirzepatide Once Weekly for the Treatment of Obesity. N Engl J Med. 2022;387(3):205-216. PMID: 35658024.

  46. Rubino D, Abrahamsson N, Davies M, Hesse D, Greenway FL, Jensen C, et al. Effect of Continued Weekly Subcutaneous Semaglutide vs Placebo on Weight Loss Maintenance in Adults With Overweight or Obesity: The STEP 4 Randomized Clinical Trial. JAMA. 2021;325(14):1414-1425. PMID: 33755728.

  47. Golovaty I, Wadhwa S, Fisher L, Lobach I, Crowe B, Levi R, et al. Reach, engagement and effectiveness of in-person and online lifestyle change programs to prevent diabetes. BMC Public Health. 2021;21(1):1314. PMID: 34225674.

  48. Auster-Gussman LA, Lockwood KG, Graham SA, Stein N, Branch OH. Reach of a Fully Digital Diabetes Prevention Program in Health Professional Shortage Areas. Popul Health Manag. 2022;25(4):441-448. PMID: 35200043.

  49. Ariel-Donges AH, Gordon EL, Dixon BN, Eastman AJ, Bauman V, Ross KM, et al. Rural/urban disparities in access to the National Diabetes Prevention Program. Transl Behav Med. 2020;10(6):1554-1558. PMID: 31228199.

  50. Painter SL, Lu W, Schneider J, James R, Shah B. Drivers of weight loss in a CDC-recognized digital diabetes prevention program. BMJ Open Diabetes Res Care. 2020;8(1). PMID: 32624481.

  51. Patel ML, Brooks TL, Bennett GG. Consistent self-monitoring in a commercial app-based intervention for weight loss: results from a randomized trial. J Behav Med. 2020;43(3):391-401. PMID: 31396820.

  52. Goldstein SP, Goldstein CM, Bond DS, Raynor HA, Wing RR, Thomas JG. Associations between self-monitoring and weight change in behavioral weight loss interventions. Health Psychol. 2019;38(12):1128-1136. PMID: 31556659.

  53. Nezami BT, Hurley L, Power J, Valle CG, Tate DF. A pilot randomized trial of simplified versus standard calorie dietary self-monitoring in a mobile weight loss intervention. Obesity (Silver Spring). 2022;30(3):628-638. PMID: 35146942.

  54. ClinicalTrials.gov. Digital Diabetes Prevention Program trial (Omada Health). Identifier NCT03312764. Available from: https://clinicaltrials.gov/study/NCT03312764

  55. Michaud TL, Almeida FA, Porter GC, Kittel CA, Schwab RJ, Brito FA, et al. Effects of a digital diabetes prevention program on cardiovascular risk among individuals with prediabetes. Prim Care Diabetes. 2023;17(2):148-154. PMID: 36697280.

  56. Alwashmi MF, Mugford G, Abu-Ashour W, Nuccio M. A Digital Diabetes Prevention Program (Transform) for Adults With Prediabetes: Secondary Analysis. JMIR Diabetes. 2019;4(3):e13904. PMID: 31350833.

  57. Ross JAD, Barron E, McGough B, Valabhji J, Daff K, Irwin J, et al. Uptake and impact of the English National Health Service digital diabetes prevention programme: observational study. BMJ Open Diabetes Res Care. 2022;10(3). PMID: 35504697.

  58. Moussas-Alvarez D, Hawkes RE, Miles LM, Dack C, French DP. Service users’ experiences of, and engagement with, a nationally implemented digital diabetes prevention programme. Br J Health Psychol. 2025;30(1):e12787. PMID: 39968862.

  59. Lemp JM, Bommer C, Xie M, Michalik F, Jani A, Davies JI, et al. Quasi-experimental evaluation of a nationwide diabetes prevention programme. Nature. 2023;624(7990):138-144. PMID: 37968391.

  60. Bower P, Soiland-Reyes C, Bennett C, Brunton L, Burch P, Cameron E, et al. The effectiveness and cost-effectiveness of the NHS Diabetes Prevention Programme (NHS-DPP): the DIPLOMA long-term multimethod assessment. Health Soc Care Deliv Res. 2025;13(19):1-48. PMID: 40323644.

  61. Silberman JM, Kaur M, Sletteland J, Venkatesan A. Outcomes in a digital weight management intervention with one-on-one health coaching. PLoS One. 2020;15(4):e0232221. PMID: 32353035.

  62. Nakata Y, Sasai H, Gosho M, Kobayashi H, Shi Y, Ohigashi T, et al. A Smartphone Healthcare Application, CALO mama Plus, to Promote Weight Loss: A Randomized Controlled Trial. Nutrients. 2022;14(21). PMID: 36364870.

  63. Lim SL, Ong KW, Johal J, Han CY, Yap QV, Chan YH, et al. Effect of a Smartphone App on Weight Change and Metabolic Outcomes in Asian Adults With Type 2 Diabetes: A Randomized Clinical Trial. JAMA Netw Open. 2021;4(6):e2112417. PMID: 34081137.

  64. Doody P, Wren GM, Mounsey S, Haasova S, Stewart C, Haffner SJP, et al. The effectiveness and safety of a mobile application-based self-regulation intervention to support weight loss among adults living with obesity: a large-scale pragmatic randomised controlled trial. BMC Med. 2025;24(1):6. PMID: 41316243.

  65. McVay M, Steinberg D, Askew S, Bennett GG. Provider Counseling and Weight Loss Outcomes in a Primary Care-Based Digital Obesity Treatment. J Gen Intern Med. 2019;34(6):992-998. PMID: 30891688.

  66. Dunn CG, Turner-McGrievy GM, Wilcox S, Hutto B. Dietary Self-Monitoring Through Calorie Tracking but Not Through a Digital Photography App Is Associated with Significant Weight Loss: The 2SMART Pilot Study-A 6-Month Randomized Trial. J Acad Nutr Diet. 2019;119(9):1525-1532. PMID: 31155474.

  67. Farage G, Simmons C, Kocak M, Klesges RC, Talcott GW, Richey P, et al. Assessing the Contribution of Self-Monitoring Through a Commercial Weight Loss App: Mediation and Predictive Modeling Study. JMIR Mhealth Uhealth. 2021;9(7):e18741. PMID: 34259635.

  68. Peterson ND, Middleton KR, Nackers LM, Medina KE, Milsom VA, Perri MG. Dietary self-monitoring and long-term success with weight management. Obesity (Silver Spring). 2014;22(9):1962-7. PMID: 24931055.

  69. Harvey J, Krukowski R, Priest J, West D. Log Often, Lose More: Electronic Dietary Self-Monitoring for Weight Loss. Obesity (Silver Spring). 2019;27(3):380-384. PMID: 30801989.

  70. Turner-McGrievy GM, Dunn CG, Wilcox S, Boutté AK, Hutto B, Hoover A, et al. Defining Adherence to Mobile Dietary Self-Monitoring and Assessing Tracking Over Time: Tracking at Least Two Eating Occasions per Day Is Best Marker of Adherence within Two Different Mobile Health Randomized Weight Loss Interventions. J Acad Nutr Diet. 2019;119(9):1516-1524. PMID: 31155473.

  71. Krukowski RA, Harvey J, Borden J, Stansbury ML, West DS. Expert opinions on reducing dietary self-monitoring burden and maintaining efficacy in weight loss programs: A Delphi study. Obes Sci Pract. 2022;8(4):401-410. PMID: 35949285.

  72. Arroyo KM, Carpenter CA, Krukowski RA, Ross KM. Identification of minimum thresholds for dietary self-monitoring to promote weight-loss maintenance. Obesity (Silver Spring). 2024;32(4):655-659. PMID: 38529540.

  73. Van Rhoon L, Byrne M, Morrissey E, Murphy J, McSharry J. A systematic review of the behaviour change techniques and digital features in technology-driven type 2 diabetes prevention interventions. Digit Health. 2020;6:2055207620914427. PMID: 32269830.

  74. Berry MP, Chwyl C, Metzler AL, Sun JH, Dart H, Forman EM. Associations between behaviour change technique clusters and weight loss outcomes of automated digital interventions: a systematic review and meta-regression. Health Psychol Rev. 2023;17(4):521-549. PMID: 36102170.

  75. Patel ML, King AC, Rosas LG, Bennett GG, Collins LM, Gallis JA, et al. Optimizing Self-Monitoring in a Digital Weight Loss Intervention (Spark): Protocol for a Factorial Randomized Trial. JMIR Res Protoc. 2025;14:e75629. PMID: 40986860.

  76. Coupe N, Peters S, Rhodes S, Cotterill S. The effect of commitment-making on weight loss and behaviour change in adults with obesity/overweight; a systematic review. BMC Public Health. 2019;19(1):816. PMID: 31234818.

  77. Madigan CD, Daley AJ, Lewis AL, Aveyard P, Jolly K. Is self-weighing an effective tool for weight loss: a systematic literature review and meta-analysis. Int J Behav Nutr Phys Act. 2015;12:104. PMID: 26293454.

  78. Zheng Y, Klem ML, Sereika SM, Danford CA, Ewing LJ, Burke LE. Self-weighing in weight management: a systematic literature review. Obesity (Silver Spring). 2015;23(2):256-65. PMID: 25521523.

  79. Shieh C, Knisely MR, Clark D, Carpenter JS. Self-weighing in weight management interventions: A systematic review of literature. Obes Res Clin Pract. 2016;10(5):493-519. PMID: 26896865.

  80. Pacanowski CR, Levitsky DA. Frequent Self-Weighing and Visual Feedback for Weight Loss in Overweight Adults. J Obes. 2015;2015:763680. PMID: 26064677.

  81. Hernández-Reyes A, Cámara-Martos F, Vidal Á, Molina-Luque R, Moreno-Rojas R. Effects of Self-Weighing During Weight Loss Treatment: A 6-Month Randomized Controlled Trial. Front Psychol. 2020;11:397. PMID: 32210897.

  82. Thomas JG, Raynor HA, Bond DS, Luke AK, Cardoso CC, Wojtanowski AC, et al. Weight loss and frequency of body-weight self-monitoring in an online commercial weight management program with and without a cellular-connected ‘smart’ scale: a randomized pilot study. Obes Sci Pract. 2017;3(4):365-372. PMID: 29259794.

  83. Vuorinen AL, Helander E, Pietilä J, Korhonen I. Frequency of Self-Weighing and Weight Change: Cohort Study With 10,000 Smart Scale Users. J Med Internet Res. 2021;23(6):e25529. PMID: 34075879.

  84. Brockmann AN, Eastman A, Ross KM. Frequency and Consistency of Self-Weighing to Promote Weight-Loss Maintenance. Obesity (Silver Spring). 2020;28(7):1215-1218. PMID: 32437055.

  85. Rohde P, Arigo D, Shaw H, Stice E. Relation of self-weighing to future weight gain and onset of disordered eating symptoms. J Consult Clin Psychol. 2018;86(8):677-687. PMID: 30035584.

  86. Neumark-Sztainer D, van den Berg P, Hannan PJ, Story M. Self-weighing in adolescents: helpful or harmful? Longitudinal associations with body weight changes and disordered eating. J Adolesc Health. 2006;39(6):811-8. PMID: 17116510.

  87. Skinner R, Gonet V, Currie S, Hoddinott P, Dombrowski SU. A systematic review with meta-analyses of text message-delivered behaviour change interventions for weight loss and weight loss maintenance. Obes Rev. 2020;21(6):e12999. PMID: 32043809.

  88. Fischer HH, Fischer IP, Pereira RI, Furniss AL, Rozwadowski JM, Moore SL, et al. Text Message Support for Weight Loss in Patients With Prediabetes: A Randomized Clinical Trial. Diabetes Care. 2016;39(8):1364-70. PMID: 26861922.

  89. Silina V, Tessma MK, Senkane S, Krievina G, Bahs G. Text messaging (SMS) as a tool to facilitate weight loss and prevent metabolic deterioration in clinically healthy overweight and obese subjects: a randomised controlled trial. Scand J Prim Health Care. 2017;35(3):262-270. PMID: 28812403.

  90. Macaulay L, O’Dolan C, Dombrowski S, Swingler J, Cotton S, Avenell A, et al. Effectiveness and cost-effectiveness of text message and financial incentives for weight management in men with obesity: The Game of Stones RCT. Public Health Res (Southampt). 2026;14(10):1-46. PMID: 42104569.

  91. Zwickert K, Rieger E, Swinbourne J, Manns C, McAulay C, Gibson AA, et al. High or low intensity text-messaging combined with group treatment equally promote weight loss maintenance in obese adults. Obes Res Clin Pract. 2016;10(6):680-691. PMID: 26992569.

  92. Spark LC, Fjeldsoe BS, Eakin EG, Reeves MM. Efficacy of a Text Message-Delivered Extended Contact Intervention on Maintenance of Weight Loss, Physical Activity, and Dietary Behavior Change. JMIR Mhealth Uhealth. 2015;3(3):e88. PMID: 26373696.

  93. Getaneh AM, van der Pol M, Boyers D, Avenell A, Cotton S, Dombrowski SU, et al. Text messaging with or without financial incentives versus a waitlist control for weight loss in men: cost-effectiveness analysis of the Game of Stones randomised controlled trial. Lancet Reg Health Eur. 2025;54:101328. PMID: 40496185.

  94. Spring B, Champion KE, Acabchuk R, Hennessy EA. Self-regulatory behaviour change techniques in interventions to promote healthy eating, physical activity, or weight loss: a meta-review. Health Psychol Rev. 2021;15(4):508-539. PMID: 31973666.

  95. Talay L, Vickers M. Patient Adherence to a Real-World Digital, Asynchronous Weight Loss Program in Australia That Combines Behavioural and GLP-1 RA Therapy: A Mixed Methods Study. Behav Sci (Basel). 2024;14(6). PMID: 38920813.

  96. Whitehall JM, Cook EJ, Vseteckova J, Jones K, Pappas Y, Donald L, et al. A systematic review of influences on engagement with remote health interventions targeting weight management for individuals living with excess weight. Int J Obes (Lond). 2025;49(8):1427-1468. PMID: 40483377.

  97. Jakob R, Harperink S, Rudolf AM, Fleisch E, Haug S, Mair JL, et al. Factors Influencing Adherence to mHealth Apps for Prevention or Management of Noncommunicable Diseases: Systematic Review. J Med Internet Res. 2022;24(5):e35371. PMID: 35612886.

  98. Schumacher LM, Martinelli MK, Convertino AD, Forman EM, Butryn ML. Weight-Related Information Avoidance Prospectively Predicts Poorer Self-Monitoring and Engagement in a Behavioral Weight Loss Intervention. Ann Behav Med. 2021;55(2):103-111. PMID: 32491152.

  99. Martinelli MK, D’Adamo L, Butryn ML. Binge eating predicts adherence to digital self-monitoring during behavioral weight loss. Eat Behav. 2020;39:101448. PMID: 33157520.

  100. Verhaak AMS, Ferrand J, Puhl RM, Tishler DS, Papasavas PK, Umashanker D. Experienced weight stigma, internalized weight bias, and clinical attrition in a medical weight loss patient sample. Int J Obes (Lond). 2022;46(6):1241-1243. PMID: 35173281.

  101. Whiteley JA, Tjaden AH, Bailey CP, Faro J, DiPietro L, Hayman LL, et al. Engagement with Digital Weight Loss Intervention Components and Weight Outcomes. J Technol Behav Sci. 2024;9(4):714-723. PMID: 41122737.

  102. Hori JH, Sia EX, Lockwood KG, Auster-Gussman LA, Rapoport S, Branch OH, et al. Discovering Engagement Personas in a Digital Diabetes Prevention Program. Behav Sci (Basel). 2022;12(6). PMID: 35735369.

  103. Kim M, Yang J, Ahn WY, Choi HJ. Machine Learning Analysis to Identify Digital Behavioral Phenotypes for Engagement and Health Outcome Efficacy of an mHealth Intervention for Obesity: Randomized Controlled Trial. J Med Internet Res. 2021;23(6):e27218. PMID: 34184991.

  104. Partridge SR, Knight A, Todd A, McGill B, Wardak S, Alston L, et al. Addressing disparities: A systematic review of digital health equity for adolescent obesity prevention and management interventions. Obes Rev. 2024;25(12):e13821. PMID: 39256958.

  105. Wing RR, Lang W, Wadden TA, Safford M, Knowler WC, Bertoni AG, et al. Benefits of modest weight loss in improving cardiovascular risk factors in overweight and obese individuals with type 2 diabetes. Diabetes Care. 2011;34(7):1481-6. PMID: 21593294.

  106. Ryan DH, Yockey SR. Weight Loss and Improvement in Comorbidity: Differences at 5%, 10%, 15%, and Over. Curr Obes Rep. 2017;6(2):187-194. PMID: 28455679.

  107. Look AHEAD Research Group, Wing RR, Bolin P, Brancati FL, Bray GA, Clark JM, et al. Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes. N Engl J Med. 2013;369(2):145-54. PMID: 23796131.

  108. Clark TL, Savin KL, Perez-Ramirez P, Valdez T, Toba G, Gallo LC. eHealth weight loss interventions for adults with low income: A systematic review. Health Psychol. 2023;42(6):353-367. PMID: 37053431.

  109. Sithole BR, Pappas Y, Randhawa G. eHealth in obesity care. Clin Med (Lond). 2023;23(4):347-352. PMID: 37524418.

  110. Ritchie ND, Sauder KA, Gritz RM. Medicare Diabetes Prevention Program: where are the suppliers?. Am J Manag Care. 2020;26(6):e198-e201. PMID: 32549070.

  111. Ritchie ND, Gritz RM. New Medicare Diabetes Prevention Coverage May Limit Beneficiary Access and Widen Health Disparities. Med Care. 2018;56(11):908-911. PMID: 30312286.

  112. Reeves D, Woodham AA, French D, Bower P, Holland F, Kontopantelis E, et al. The influence of demographic, health and psychosocial factors on patient uptake of the English NHS diabetes prevention programme. BMC Health Serv Res. 2023;23(1):352. PMID: 37041541.

  113. Barrett S, Haffner SJ, McSweeney L, Rothwell C, Fong M, Jebb SA, et al. Evaluation of the experience of people referred under the NHS enhanced service incentive for obesity to the NHS digital weight management programme: a mixed method study. BMC Public Health. 2026;26(1):638. PMID: 41578273.

  114. Grilo CM, Lydecker JA, Fineberg SK, Moreno JO, Ivezaj V, Gueorguieva R. Naltrexone-Bupropion and Behavior Therapy, Alone and Combined, for Binge-Eating Disorder: Randomized Double-Blind Placebo-Controlled Trial. Am J Psychiatry. 2022;179(12):927-937. PMID: 36285406.

  115. Honary M, Bell BT, Clinch S, Wild SE, McNaney R. Understanding the Role of Healthy Eating and Fitness Mobile Apps in the Formation of Maladaptive Eating and Exercise Behaviors in Young People. JMIR Mhealth Uhealth. 2019;7(6):e14239. PMID: 31215514.

  116. Eikey EV, Reddy MC, Booth KM, Kvasny L, Blair JL, Li V, et al. Desire to Be Underweight: Exploratory Study on a Weight Loss App Community and User Perceptions of the Impact on Disordered Eating Behaviors. JMIR Mhealth Uhealth. 2017;5(10):e150. PMID: 29025694.

  117. Liu J, Munson SA, Chang A, Voss C, Graham AK. Understanding self-monitoring to inform a mobile intervention for binge eating and weight management: A proof-of-concept randomized trial. Int J Eat Disord. 2022;55(5):642-652. PMID: 35277986.

  118. Bartel S, McElroy SL, Levangie D, Keshen A. Use of glucagon-like peptide-1 receptor agonists in eating disorder populations. Int J Eat Disord. 2024;57(2):286-293. PMID: 38135891.

  119. Phelan SM, Burgess DJ, Yeazel MW, Hellerstedt WL, Griffin JM, van Ryn M. Impact of weight bias and stigma on quality of care and outcomes for patients with obesity. Obes Rev. 2015;16(4):319-26. PMID: 25752756.

  120. Ryan L, Coyne R, Heary C, Birney S, Crotty M, Dunne R, et al. Weight stigma experienced by patients with obesity in healthcare settings: A qualitative evidence synthesis. Obes Rev. 2023;24(10):e13606. PMID: 37533183.

  121. Tomiyama AJ, Carr D, Granberg EM, Major B, Robinson E, Sutin AR, et al. How and why weight stigma drives the obesity ‘epidemic’ and harms health. BMC Med. 2018;16(1):123. PMID: 30107800.

  122. Pearl RL, Sheynblyum M. How Weight Bias and Stigma Undermine Healthcare Access and Utilization. Curr Obes Rep. 2025;14(1):11. PMID: 39832116.

  123. Gerber PA, Collet TH, Favre L, Bueter M, Ruiz Arana I, Durrer D, et al. Swiss obesity clinical practice guidance. Swiss Med Wkly. 2026;156:5415. PMID: 42312974.

  124. Wharton S, Lau DCW, Vallis M, Sharma AM, Biertho L, Campbell-Scherer D, et al. Obesity in adults: a clinical practice guideline. CMAJ. 2020;192(31):E875-E891. PMID: 32753461.

  125. Talumaa B, Brown A, Batterham RL, Kalea AZ. Effective strategies in ending weight stigma in healthcare. Obes Rev. 2022;23(10):e13494. PMID: 35934011.

  126. Lean ME, Leslie WS, Barnes AC, Brosnahan N, Thom G, McCombie L, et al. Primary care-led weight management for remission of type 2 diabetes (DiRECT): an open-label, cluster-randomised trial. Lancet. 2018;391(10120):541-551. PMID: 29221645.

  127. Lean MEJ, Leslie WS, Barnes AC, Brosnahan N, Thom G, McCombie L, et al. Durability of a primary care-led weight-management intervention for remission of type 2 diabetes: 2-year results of the DiRECT open-label, cluster-randomised trial. Lancet Diabetes Endocrinol. 2019;7(5):344-355. PMID: 30852132.

  128. Lean ME, Leslie WS, Barnes AC, Brosnahan N, Thom G, McCombie L, et al. 5-year follow-up of the randomised Diabetes Remission Clinical Trial (DiRECT) of continued support for weight loss maintenance in the UK: an extension study. Lancet Diabetes Endocrinol. 2024;12(4):233-246. PMID: 38423026.

  129. Taylor R. Very low calorie diets and total diet replacement in type 2 diabetes: where are we now?. Drug Ther Bull. 2024;62(3):39-42. PMID: 38336461.

  130. Paisey R, Daniels C, Howitt W, Greatorex D, Campbell C, Paisey C, et al. Body weight, diabetes incidence vascular events and survival 15 years after very low calorie diet in community medical clinics in the UK. BMJ Nutr Prev Health. 2022;5(1):55-61. PMID: 35814720.

  131. Andela S, Burrows TL, Baur LA, Coyle DH, Collins CE, Gow ML. Efficacy of very low-energy diet programs for weight loss: A systematic review with meta-analysis of intervention studies in children and adolescents with obesity. Obes Rev. 2019;20(6):871-882. PMID: 30734459.

  132. Price SA, Sumithran P. Using a Very Low Energy Diet to Achieve Substantial Preconception Weight Loss in Women with Obesity: A Review of the Safety and Efficacy. Nutrients. 2022;14(20). PMID: 36297107.

  133. National Institute for Health and Care Excellence. Overweight and obesity management. NICE guideline NG246. London: NICE; 2025. Available from: https://www.nice.org.uk/guidance/ng246

  134. American Diabetes Association Professional Practice Committee. Standards of Care in Diabetes—2026: obesity and weight management for the prevention and treatment of diabetes. Diabetes Care. 2026. Available from: https://diabetesjournals.org/care

  135. European Association for the Study of Obesity. A new framework for the diagnosis, staging and management of obesity in adults. EASO framework; 2024.

  136. European Association for the Study of Obesity. Pharmacological framework for obesity management. EASO framework; 2025.

  137. Celletti F, Farrar J, De Regil L. World Health Organization Guideline on the Use and Indications of Glucagon-Like Peptide-1 Therapies for the Treatment of Obesity in Adults. JAMA. 2026;335(5):434-438. PMID: 41324410.

  138. Gaskin CJ, Cooper K, Stephens LD, Peeters A, Salmon J, Porter J. Clinical practice guidelines for the management of overweight and obesity published internationally: A scoping review. Obes Rev. 2024;25(5):e13700. PMID: 38296655.

  139. Gasoyan H, Schulte R, Boyer CB, Casacchia NJ, Butsch WS, Le P, et al. Obesity Treatments and Weight Changes in Clinical Practice After Discontinuation of Semaglutide or Tirzepatide. Diabetes Obes Metab. 2026;28(6):4795-4805. PMID: 41816857.

  140. Horn DB, Aronne LJ, Wharton S, Bays HE, le Roux CW, Srinath R, et al. Tirzepatide for maintenance of bodyweight reduction in people with obesity in the USA (SURMOUNT-MAINTAIN): a multicentre, double-blind, randomised, placebo-controlled trial. Lancet. 2026;407(10545):2305-2318. PMID: 42119587.

  141. Aronne LJ, Horn DB, le Roux CW, Chao AM, Ho W, Halpern B, et al. Orforglipron for maintenance of body weight reduction: the double-blind, randomized phase 3b ATTAIN-MAINTAIN trial. Nat Med. 2026;32(7):2679-2687. PMID: 42120723.

  142. Richards R, Lunt W, Whitman M, Spaltro G, Hall R. Semaglutide and Tirzepatide in a Remote Weight Management Program: 12-Month Retrospective Observational Study. JMIR Form Res. 2025;9:e81912. PMID: 40838489.

  143. Sikorska LM, Liu VN, Johnson H, Chaudry E, Reisel D, Clift AK, et al. Beyond Weight Loss: Holistic Impacts of a Digital Weight Management Programme Integrating Tirzepatide. Cureus. 2026;18(2):e102817. PMID: 41788104.

  144. Hwang JH, Laiteerapong N, Huang ES, Kim DD. Lifetime Health Effects and Cost-Effectiveness of Tirzepatide and Semaglutide in US Adults. JAMA Health Forum. 2025;6(3):e245586. PMID: 40085108.

  145. Joyce JB, Hajizadeh A, Begh R, Jolly K, Jebb SA, Aveyard P. The implementation of a Nationally Enhanced Service incentive for weight management: A longitudinal qualitative study of the perceptions and experiences of UK primary care staff on weight management using normalisation process theory. Clin Obes. 2025;15(5):e70020. PMID: 40302173.

  146. Haffner SJP, Stevens RJ, Amies-Cull B, Heath L, Bankhead C, Aveyard P, et al. Effect of the English National Enhanced Service on weight management referral rate: an interrupted time-series analysis. BMJ Open. 2026;16(3):e109943. PMID: 41775470.

  147. Stephenson-Hunter C, Gonzalez CJ, Franco S, Hashmi M, Tisor O, Gonzalez CM. Primary care providers’ perspectives on referrals to the Diabetes Prevention Programme: a qualitative comparative study across varied referral patterns. BMJ Nutr Prev Health. 2024;7(2):e000932. PMID: 39882293.

  148. Keck JW, Roper KL, Hieronymus LB, Thomas AR, Huang Z, Westgate PM, et al. Primary Care Cluster RCT to Increase Diabetes Prevention Program Referrals. Am J Prev Med. 2020;59(1):79-87. PMID: 32418801.

  149. Olsen J, Peterson S, Stevens A. Implementing electronic health record-based National Diabetes Prevention Program referrals in a rural county. Public Health Nurs. 2021;38(3):464-469. PMID: 33368605.

  150. Madrigal L, Manders OC, Kegler M, Haardörfer R, Piper S, Blais LM, et al. Inner and outer setting factors that influence the implementation of the National Diabetes Prevention Program (National DPP) using the Consolidated Framework for Implementation Research (CFIR): a qualitative study. Implement Sci Commun. 2022;3(1):104. PMID: 36183133.

  151. Williams J, Sachdev N, Kirley K, Moin T, Duru OK, Brunisholz KD, et al. Implementation of Diabetes Prevention in Health Care Organizations: Best Practice Recommendations. Popul Health Manag. 2022;25(1):31-38. PMID: 34161148.

  152. Mayuri, Bhattacharya B, Kar K, Pawar P, Singh I, Sangnim T, et al. Transforming Obesity Care With Digital Therapeutics: Mobile Health Apps, Clinical Evidence, and Regulatory Considerations. J Obes. 2026;2026(1):e9482093. PMID: 42895797.

Leave a Comment