Supporting Weight Management during COVID-19 (SWIM-C): twelve-month follow-up of a randomised controlled trial of a web-based, ACT-based, guided self-help intervention

The study addresses the need for scalable, remote weight management interventions during the COVID-19 pandemic, when access to in-person services was limited and adults with overweight/obesity were vulnerable to weight gain and adverse changes in eating behaviours. Acceptance and commitment therapy (ACT), which targets experiential avoidance and promotes psychological flexibility, may enhance long-term weight outcomes. The research question was whether SWIM-C, a web-based, ACT-guided self-help intervention, improves 12-month weight change versus standard advice and positively affects determinants of weight management (eating behaviours, experiential avoidance/psychological flexibility, mental health, wellbeing, and physical activity).

Prior behavioural weight management interventions yield modest, often short-term weight loss. A systematic review and network meta-analysis suggested ACT-based strategies may have superior long-term outcomes versus standard behavioural treatment. Evidence on remotely delivered ACT-based weight management was scarce, with few studies not powered for weight outcomes and mostly feasibility or pre-post designs. Pandemic-related restrictions likely worsened mental health, wellbeing, and physical activity, contributing to weight gain risk, underscoring the need for scalable remote interventions like SWIM-C.

Design: Randomised, parallel, two-group trial (ISRCTN12107048), unblinded, with assessments at baseline, 4 months, and 12 months. Ethics: Cambridge Psychology Research Ethics Committee (PRE.2020.049); informed consent obtained. Participants: Adults in the UK, ≥18 years, BMI ≥25 kg/m², able to complete online assessments, own scales, English proficiency; excluded if bariatric surgery in previous 2 years. Recruitment: Mailing lists/social media via weight management services/organisations and a prior volunteer database. Randomisation: 1:1 to SWIM-C or standard advice; computer-generated block randomisation (block size 6) stratified by BMI class (25 to <30, 30 to <40, ≥40 kg/m²) and sex; sequence concealed from staff/participants. Interventions: SWIM-C (ACT-based web platform with 12 weekly sessions of psychoeducation, exercises, behavioural experiments; a 20-min coach phone call after session 4; tailored email at week 10). Standard advice: EASO leaflet on diet, physical activity, and mood during COVID-19; wait-list access to SWIM-C web platform (no coach support) after 4-month assessment. Procedures: Online questionnaires; guidance for at-home weight measurement; honoraria (£10 baseline, £20 at each follow-up). Data quality checks included plausibility screening and validation; biologically implausible weight change handled per Chen et al. Sample size: Target n=360 to detect a 1 kg between-group difference in weight change (SD 6 kg, correlation 0.9) with 90% power at 95% confidence. Outcomes: Primary—change in self-reported weight (kg) from baseline to 12 months. Secondary—TFEQ-R21 (cognitive restraint, uncontrolled eating, emotional eating; 0–100), AAQW-R (experiential avoidance; 10–70), IPAQ total physical activity (MET-min/week), PHQ-8 (0–24), GAD-7 (0–21), PSS-4 (0–16), ICECAP-A (0–1), and weight change from 4 to 12 months. Baseline covariates: age, sex, ethnicity, education, marital status, height. Statistical analysis: Random-intercepts linear regression on change scores at 4 and 12 months including group, timepoint, group×timepoint, stratifiers (sex, BMI class), and baseline weight; intention-to-treat. Same approach for secondary outcomes. Missing data assumed MAR; characteristics compared by missingness; MNAR sensitivity via pattern-mixture models using MICE with imputed weights scaled by ±0%, ±10%, ±20%, ±30%. Per-protocol analyses excluded standard-advice participants who engaged with SWIM-C; defined engagement thresholds as ≥4 sessions and ≥8 sessions (vs <4). Additional model estimated baseline-adjusted group differences in weight change from 4 to 12 months controlling for baseline and 4-month weight and stratifiers. Software: R 4.0.0, RStudio 1.0.153.

Participants: 486 screened; 388 randomised (SWIM-C n=192; standard advice n=196). Mean age 50.7 years (SD 14.3), mean BMI 34.8 kg/m² (SD 7.7); 78.1% female, 93.8% White, 62.4% with ≥university degree. Twelve-month weight data available for 266 (69.3%) after excluding 3 implausible values. Engagement: In SWIM-C, 46 completed ≥4 sessions, 103 ≥8 sessions, 62 all 12; in standard advice, after 4 months 82% accessed <4 sessions (9 completed all 12). Primary outcome: From baseline to 12 months, SWIM-C lost -2.08 kg (SD 7.30) vs standard advice -1.36 kg (SD 7.51); adjusted between-group difference -0.81 kg (95% CI -2.24 to 0.61), compatible with no effect. Per-protocol: comparing standard advice <4 sessions vs SWIM-C ≥4 sessions: -0.30 kg (95% CI -1.80 to 1.21); vs SWIM-C ≥8 sessions: -0.70 kg (95% CI -2.42 to 1.01). MNAR sensitivity: differences ranged -0.98 to -0.66 kg, all CIs including 0. From 4 to 12 months: adjusted difference 0.22 kg (95% CI -1.95 to 1.37). Secondary outcomes (baseline to 12 months): SWIM-C improved—experiential avoidance (AAQW-R) by -2.45 (95% CI -4.75 to -0.15), uncontrolled eating by -5.52 (95% CI -9.67 to -1.37), emotional eating by -4.49 (95% CI -7.57 to -1.42), and physical activity by +8.96 MET-min/week (95% CI 0.29 to 17.62). No clear effects on PHQ-8, GAD-7, PSS-4, ICECAP-A, or cognitive restraint at 12 months. From 4 to 12 months, SWIM-C showed increased anxiety vs control (GAD-7 +0.98, 95% CI 0.005 to 1.95) and decreased cognitive restraint (-5.38, 95% CI -9.61 to -1.16). No adverse events reported.

The trial investigated whether a remote ACT-based guided self-help program could improve 12-month weight outcomes and key determinants of weight management during COVID-19. Although SWIM-C did not demonstrate a definitive effect on weight at 12 months, it significantly improved psychological and behavioural determinants—reducing experiential avoidance and maladaptive eating (uncontrolled and emotional eating) and increasing physical activity—suggesting meaningful behaviour change aligned with ACT mechanisms. Some effects (emotional eating and physical activity) appeared stronger at 12 months than at 4 months, supporting a hypothesis that ACT skills may require time and practice to manifest fully. Maintenance of improvements in uncontrolled eating and experiential avoidance indicates sustained changes in determinants relevant to long-term weight management, even without clear weight differences. The attenuation of cognitive restraint from 4 to 12 months could reflect shifting goals (weight loss to maintenance) or reduced need for conscious restriction as uncontrolled/emotional eating decreases. Overall, findings highlight promise for remote ACT-informed interventions to influence behaviours and psychological processes important for weight management, but additional refinement (e.g., enhanced coach support) may be needed to achieve clinically meaningful weight loss.

SWIM-C, a scalable web-based ACT-guided self-help intervention, produced sustained improvements in experiential avoidance and uncontrolled eating and increased effects on emotional eating and physical activity at 12 months, but did not show a clear effect on body weight compared with standard advice. Results suggest ACT-based interventions may exert longer-term behavioural and psychological benefits that evolve post-intervention. Further development (potentially more intensive or prolonged support) and evaluation in larger trials are warranted to determine whether these benefits can translate into meaningful weight loss and to optimise remote delivery for broader implementation.

- Self-reported weight may reduce measurement accuracy. - Recruitment via social media and obesity/weight management networks may limit generalisability; sample largely female, White, and university-educated. - Did not assess concurrent participation in other weight management programs or treatments (e.g., pharmacotherapy). - Differential and substantial missing data at follow-up (more missing in intervention group); although sensitivity analyses under MNAR assumptions supported conclusions, bias cannot be ruled out. - Wait-list control received access to SWIM-C web platform after 4 months (without coaching), potentially diluting between-group differences at 12 months; per-protocol analyses excluding engaged controls yielded similar results. - Potential limited engagement with intervention among some participants, which may have attenuated effects.