Real-World Two-Year Weight Loss Outcomes With Breath Biofeedback

Real-World Durability of Weight Loss With a Breath Biofeedback mHealth Program: Two-Year Outcomes in Adults With Overweight or Obesity

The goal of this retrospective observational study is to evaluate the real-world durability of weight loss among adults with overweight or obesity who used the Key to Health breath-biofeedback mHealth program. The main questions it aims to answer are:

1. What is the mean percent total body weight loss (TBWL%) at 104 weeks after program start?
2. What are the TBWL% estimates at 12, 24, and 52 weeks, and what proportion of users achieve clinically meaningful weight loss (≥5% and ≥10%) at 52 and 104 weeks (observed case)?

Users initiated the program between Nov 2018 and Apr 2023; follow-up data (weights/engagement) were collected through Apr 2025 to allow up to 104 weeks of observation.

There is no randomized comparison group. Participants will not be asked to do anything beyond normal app use; the study will analyze existing, de-identified weight and engagement data collected during routine program use.

This is a retrospective, observational real-world evidence study using de-identified backend data from commercially active users of the Key to Health digital weight loss program between November 2018 and April 2025. Eligible users are adults (≥18 years) with a valid onboarding baseline weight and at least one post-baseline weight entry; users must meet the pre-specified baseline BMI threshold for overweight/obesity. Baseline weight is defined as the self-reported onboarding weight. Follow-up weights are defined as user-entered or connected-scale weights recorded after onboarding.

Weight outcomes are assigned to prespecified follow-up windows at Weeks 12, 24, 52, and 104 using a deterministic rule: the weight entry closest to each target day within its window (earlier chosen if equidistant). The primary endpoint is TBWL% at Week 104, defined as (post-baseline weight - baseline weight) / baseline weight × 100. Secondary outcomes include TBWL% at Weeks 12, 24, and 52; absolute weight change; and observed-case responder summaries at Weeks 52 and 104.

The primary analysis will estimate longitudinal TBWL% across visits using a Mixed Model for Repeated Measures (MMRM) with visit as a categorical factor and adjustment for baseline covariates (e.g., baseline weight, age, sex, cohort year). Missing data will be handled under a Missing At Random (MAR) framework inherent to the MMRM approach (no single-value imputation in the primary analysis). Pre-specified sensitivity analyses may be conducted as feasible (e.g., conservative assumptions for missing Week-104 outcomes and weighted approaches for differential observation), with any deviations from the prespecified plan documented.

Inclusion Criteria:

• Adults aged ≥18 years at program start/onboarding
• Program start date between November 2018 and April 2023
• Valid baseline (onboarding) self-reported weight recorded
• Baseline BMI >25 kg/m²
• At least one valid post-baseline weight entry recorded after onboarding (manual and/or connected scale)

Exclusion Criteria:

• Missing or invalid baseline weight
• Biologically implausible baseline weight (per prespecified data-quality checks)
• No post-baseline weight data recorded after the baseline date/onboarding timestamp