The myCOPD app: This Guide Reviews the Literature About 4 Studies
myCOPD App for Managing Chronic Obstructive Pulmonary Disease: A NICE Medical Technology Guidance for a Digital Health Technology
Heather Davies, Mary Chappell, Yingying Wang, Angaja Phalguni, Stephanie Wake, Mick Arber, Judith Shore · Applied Health Economics and Health Policy 21:689–700 · 2023
Although myCOPD has potential to help adults manage their COPD, insufficient evidence currently supports routine NHS adoption. Further comparative research is recommended to address uncertainties about clinical benefits and healthcare resource use.
Assessment Context
York Health Economics Consortium served as External Assessment Group for NICE’s Medical Technologies Evaluation Programme pilot for digital health technologies. The assessment critiqued clinical and economic evidence submitted by my mHealth Ltd for myCOPD, a digital self-management platform for people with COPD.
Clinical Evidence Base
Four published peer-reviewed studies (three RCTs: RESCUE, EARLY, TROOPER; one observational study) plus 15 real-world evaluations across seven UK settings. RCTs had small sample sizes limiting statistical power. RESCUE and EARLY showed non-significant trends toward improved CAT scores; TROOPER demonstrated non-inferiority of digital PR versus face-to-face PR.
Economic Modeling Approach
Company submitted two de novo cost models for subgroups: people discharged from hospital post-AECOPD and people eligible for PR. EAG updated parameters and adjusted structures. Both models showed cost savings but with significant uncertainty around uptake rates and clinical effectiveness parameters.
EAG-Revised Results
AECOPD model: £86,297 savings per CCG (73.5% probability cost-saving). PR model: £22,779 savings per CCG assuming existing myCOPD license (86% probability cost-saving). Key drivers were hospital readmission rates, uptake, and PR referral volumes.
Critical Uncertainties
Small RCT sample sizes limiting power to detect effects; non-significant clinical outcomes in self-management trials; short follow-up periods (3 months); declining app usage over time in all studies; unclear link between engagement level and clinical outcomes; uncertainty about long-term effectiveness and sustained engagement.
EAG Economic Model Results Comparison
| Model | Company Base Case | EAG Base Case | Probability Cost-Saving | Key Driver |
|---|---|---|---|---|
| AECOPD (per CCG) | £204,641 savings | £86,297 savings | 73.5% | 90-day readmission rate |
| PR (per CCG) | £20,269 savings | £22,779 savings | 86% | Uptake & referral volume |
| PR (per service) | £8,707 savings | £11,093 savings | 87% | Annual referrals (≥240) |
■ HTA Process ■ Study Evidence ■ Economic Modeling ■ Technology & Outcomes
HTA Process & Governance
Study Evidence & Trial Design
Economic Modeling & Parameters
Technology & Clinical Outcomes
Assessment Overview
York Health Economics Consortium served as EAG for NICE’s pilot DHT evaluation process. The team conducted de novo literature searches, critical appraisal of submitted evidence, economic model critique and revision, and consultation response analysis. Clinical experts provided advice via NICE’s published processes.
Literature Search Strategy
- Company provided no search strategy documentation in submission; EAG conducted de novo searches
- Original searches October 2019, updated January 2021 after DHT guidance selection
- 7,761 total records retrieved; 3,280 after deduplication for screening
- EAG searches identified five additional published RWE papers not in company submission
- No meta-analysis conducted by company or EAG due to heterogeneity and small sample sizes
Quality Appraisal Approach
- RCTs appraised using Centre for Reviews and Dissemination criteria (Khan et al. 2001)
- Observational studies appraised using CASP UK 2013 cohort checklist
- RCT internal and external validity judged acceptable despite small sample sizes
- Observational study: acceptable external validity but low internal validity (limited reporting, no confounding assessment)
- RWE not formally quality-assessed due to variation in methods and focus on service evaluation
Economic Model Revisions: AECOPD
- Company model structure judged appropriate (TreeAge cost calculator, 1-year horizon, 3-month outcomes)
- Key EAG change: uptake rate adjusted from 100% to 46% based on RESCUE trial enrollment
- Efficacy parameters recalculated: exacerbations (adjusted rate ratio 0.581), readmissions (OR 0.383 converted to RR)
- Cost updates: exacerbation costs (£53.59→£81.75), admissions (£1,583→£1,721), registration time doubled (15→30 min)
- PSA conducted by EAG using company model framework with updated parameter distributions
Economic Model Revisions: PR
- Company decision tree structure accepted; EAG added cost for patients starting but not completing PR
- Decision point shifted: from “referral to PR” to “expressed willingness to use myCOPD” (aligns with scope and TROOPER population)
- Treatment arm adjusted: removed face-to-face only option (scope specifies myCOPD as replacement, not add-on for PR)
- Uptake recalculated to 12% for both hybrid and myCOPD-only based on Southend study completion rates
- Key limitation acknowledged: no benefits captured for partial PR completion in model
Strengths
- Comprehensive independent literature search addressing company submission gap
- Transparent parameter adjustments with clear rationale and clinical expert input
- Probabilistic sensitivity analysis providing uncertainty quantification beyond company deterministic scenarios
- Threshold analyses identifying critical decision points for commissioners
- Clear distinction between internal validity (low) and external validity (acceptable) for RWE
Limitations
- Underlying clinical evidence limited by small RCT sample sizes and non-significant findings
- Short follow-up periods (3 months) preclude assessment of long-term effectiveness or engagement patterns
- Uptake parameter derived from RCT enrollment, not real-world adoption rates
- No economic studies identified for inclusion; reliance on company de novo models only
- CCG costing perspective outdated (CCGs abolished July 2022); implications for revised pricing structure unclear
- Effective engagement framework not operationalized in models; unclear link between usage patterns and outcomes
Six key references exemplify the evidence categories and methodological touchpoints in this HTA. Click any card to expand the full analysis.
North et al. 2020 — RESCUE Trial: E-health Application Supported Care vs Usual Care After COPD Exacerbation
Why this reference matters
RESCUE is the primary source of clinical effectiveness data for the AECOPD economic model. As a randomized controlled feasibility trial, it provides the most rigorous evidence available for myCOPD in the post-exacerbation population, directly testing the intervention in the highest-risk COPD subgroup.
Key contribution to this paper’s argument
Supplies three critical model parameters: exacerbation rate (1.88 vs 1.06 over 90 days, adjusted rate ratio 0.581), hospital readmission rate (39% vs 24%, adjusted OR 0.383), and GP appointment frequency (2.28 vs 1.85). The EAG’s recalculation of company-reported values using adjusted statistics rather than raw risks demonstrates methodological rigor.
Connection to the wider citation network
RESCUE sits alongside EARLY and TROOPER as the RCT evidence base, but uniquely addresses the AECOPD population. The trial’s acknowledgment of limited power to detect effects on all outcomes foreshadows the EAG’s central concern about sample size across the evidence base.
Bourne et al. 2017 — TROOPER Trial: Online vs Face-to-Face Pulmonary Rehabilitation
Why this reference matters
TROOPER is the only RCT comparing myCOPD PR with face-to-face PR, providing the effectiveness evidence for the PR economic model. Its non-inferiority design tests whether digital delivery can replace traditional PR, a fundamentally different question than add-on effectiveness.
Key contribution to this paper’s argument
Demonstrates non-inferiority for CAT score (adjusted mean difference -1.0, 95% CI -2.9 to 0.86) and 6-minute walk test (23.8m, 95% CI -4.5 to 52.2). However, the EAG identified concerns about power calculation validity: MCID distance higher than recommended (40.5m vs 30m) and power calculated at 58% rather than conventional 80-90%.
Connection to the wider citation network
Complements the self-management trials (RESCUE, EARLY) by addressing the PR use case. The deviation from NICE scope (replacement vs add-on) introduces ambiguity about the intended positioning of myCOPD PR, reflected in the company’s model structure showing three treatment pathways rather than two.
Yardley et al. 2016 — Understanding and Promoting Effective Engagement with Digital Behavior Change Interventions
Why this reference matters
Yardley’s effective engagement framework addresses a fundamental challenge in DHT evaluation: the relationship between usage patterns and clinical outcomes. This concept is central to interpreting the observed decline in app usage across all myCOPD trials and RWE.
Key contribution to this paper’s argument
Defines “effective engagement” as the engagement necessary to produce desired outcomes, not simply “more engagement, the more effective.” This framework appears in consultation comments defending myCOPD completion rates in TROOPER and contextualizing why adherence metrics alone don’t determine effectiveness. The EAG acknowledges this concept but notes it remains unoperationalized in the economic models.
Connection to the wider citation network
Part of a broader behavioral intervention literature (Baumel 2022, Ainsworth et al. 2017) cited in consultation responses. These references situate myCOPD within digital health behavior change theory, though the RCT evidence doesn’t demonstrate the mechanism linking engagement to outcomes.
Chartered Society of Physiotherapy 2017 — COPD PRIME Tool
Why this reference matters
COPD PRIME provides UK-specific reference data for PR pathways and exacerbation rates where trial data are unavailable. It represents the standard-of-care baseline against which myCOPD’s cost-effectiveness is assessed.
Key contribution to this paper’s argument
Supplies multiple PR model parameters: 40% COPD patients eligible for PR referral, 15% offered PR of those eligible, 59% start rate among those referred, 71% completion rate, and annual exacerbation rates stratified by completion status (2.11 completers vs 3.31 non-completers). These cascade probabilities determine the modeled population flow.
Connection to the wider citation network
Used alongside NHS Digital QOF data, NACAP audit reports, and PSSRU cost references to construct the economic model structure. Together these sources create a composite picture of COPD care pathways in the NHS, grounding the model in UK-specific practice patterns.
NACAP 2019-2020 — National Asthma and COPD Audit Programme: Pulmonary Rehabilitation Reports
Why this reference matters
NACAP provides national audit data on PR service delivery and capacity, enabling threshold analyses for service-level cost-effectiveness. The median 495 referrals per CCG per year grounds the model in real-world service volumes.
Key contribution to this paper’s argument
Establishes the 240 annual referrals threshold for PR service cost savings and the 13-day median waiting time used to cost the opportunity cost of PR assessment delays. The 84% of referrals being COPD-specific justifies the disease-focused modeling approach.
Connection to the wider citation network
Complements COPD PRIME by providing service delivery context rather than clinical pathway probabilities. Together they enable both CCG-level and PR-service-level economic perspectives, though the latter requires assuming an existing myCOPD license to avoid double-counting setup costs.
NICE 2022 — myCOPD for Managing Chronic Obstructive Pulmonary Disease: Medical Technologies Guidance 68
Why this reference matters
The final NICE guidance represents the policy outcome of the entire assessment process. Its recommendation against routine adoption despite potential benefits codifies the uncertainty threshold for DHT evidence in the NHS.
Key contribution to this paper’s argument
Provides the complete documentation including consultation responses, committee deliberations, and research recommendations. The consultation received 108 comments from 11 consultees challenging the draft recommendations on powering, engagement, economic uncertainty, and patient representation, revealing stakeholder perspectives.
Connection to the wider citation network
This Davies et al. 2023 paper is itself a synthesis of the MTG68 process, making the final guidance both a reference and the object of description. The circular reference structure reflects HTA methodology papers documenting processes that produce the primary guidance documents.
-
DHT Evidence Standards Remain Unresolved
NICE’s pilot DHT evaluation process adapted MTEP methods but revealed persistent challenges: linking engagement to outcomes, interpreting declining usage patterns, valuing behavioral endpoints, and determining adequate follow-up duration. Yardley’s “effective engagement” framework offers conceptual clarity but lacks operational metrics for economic modeling.
-
Small RCT Sample Sizes Create Compounding Uncertainty
RESCUE, EARLY, and TROOPER totaled 191 participants across three trials. Non-significant findings don’t prove no effect; they prove insufficient power to detect effects. When these underpowered trials become model parameters (via point estimates and confidence intervals), uncertainty propagates through the economic analysis. PSA captures statistical uncertainty but can’t remedy the fundamental evidence gap.
-
Subscription Pricing Makes Uptake the Critical Parameter
myCOPD’s unlimited license model (£0.25 per registered patient across entire CCG population) means cost-effectiveness depends critically on how many eligible patients actually use the service. The EAG’s uptake adjustment (100% to 46%) more than halved projected savings. This pricing structure creates opposite incentives from per-patient licensing: providers want high uptake, but evidence for population-level adoption is weakest.
-
Model Population Definitions Shape Cost-Effectiveness
The company focused on two high-value subgroups (post-AECOPD, PR-eligible) where benefits were most plausible, rather than the broader COPD population in the scope. This is methodologically sound but raises questions about generalizability and service design: would NHS implementation target these subgroups specifically, or offer the platform population-wide and accept lower uptake?
-
Non-Inferiority Evidence Serves a Replacement Strategy
TROOPER’s non-inferiority design tests whether myCOPD PR can replace face-to-face PR, not whether it adds value as a supplement. This deviation from the NICE scope (which specified add-on use) reflects a strategic ambiguity about positioning: is myCOPD a capacity-expansion tool for understaffed PR services, or a cost-saving replacement for traditional delivery? The answer determines the relevant comparator and economic model structure.
-
Threshold Analyses Identify Commissioning Decision Points
EAG threshold analyses operationalize uncertainty for commissioners: AECOPD model cost-saving requires uptake ≥26.2% or readmission rate ≤0.30; PR model requires ≥240 annual referrals. These thresholds translate complex economic models into actionable decision rules, though they assume all other parameters stay at base-case values (a strong assumption given parameter correlation).
-
HTA Process Transparency Enables Methodological Learning
This paper exemplifies a meta-genre: the EAG team publishing their assessment methodology separately from the NICE guidance itself. Documenting search strategies, quality appraisal decisions, model revisions, and consultation themes creates a knowledge base for future DHT assessments. The 108 consultation comments reveal stakeholder perspectives on evidence standards, particularly around powering, engagement metrics, and patient representation.