Long-term conditions require sustained oversight, yet gaps in timely information can limit clinical decisions and opportunities to prevent deterioration. Mobile device-based active remote monitoring captures frequent patient-reported outcomes via smartphones or tablets, generating longitudinal data on symptoms, wellbeing and functioning. Economic evaluations across multiple conditions have reported costs and outcomes, including quality-adjusted life years (QALYs), using trial-based and model-based approaches. Results signal about cost-effectiveness, while highlighting substantial uncertainty and variation in costing methods that matter for commissioning and scale-up. Differences in monitoring frequency, implementation models and analytic perspective influence reported findings, reinforcing the need for transparent costing and clear logic models that specify how data trigger actions by clinicians or patients.
Populations, Designs and Monitoring Logic
Evaluations published between 2020 and 2023 covered rheumatoid arthritis, schizophrenia, older adults with complex chronic conditions, cancer, multiple sclerosis and inflammatory bowel disease. Target populations varied by disease status, including ongoing active disease, stability or recent recovery, with several analyses not specifying severity. Monitoring captured self-reported disease activity, broader quality of life, mental wellbeing and condition-specific constructs such as fatigue, stress, memory and medication adherence. Frequencies ranged from daily to three-monthly, with escalation during suspected flares in some protocols.
Interventions clustered by intended mechanism of value. One group aimed to alert staff or patients to worsening symptoms to enable earlier intervention. Another enabled patient-initiated care by extending the interval between routine appointments when disease activity remained stable, while preserving rapid contact pathways when activity rose. A further group focused on self-management after goal-setting with providers or based on patient-selected topics, with patient-facing graphs and in-app feedback.
Presentation and workflow also differed. Many provider-facing implementations used standalone dashboards displaying longitudinal data, graphs and alerts, sometimes with email notifications. Others integrated patient-reported data into electronic medical records for optional review. Several offered patients their own trend visualisations to encourage engagement and support shared decision-making. Study designs spanned feasibility, non-inferiority and superiority trials across single or multicentre settings, alongside model-based analyses that synthesised available evidence and explored plausible treatment effect sizes where definitive estimates were unavailable. Outcomes were reported from healthcare system and societal perspectives, with non-medical items included when societal perspectives were applied. QALYs were frequently the primary health outcome, with condition-specific measures reported alongside.
Costing Methods and Reported Values
Intervention costing varied markedly. Top-down apportionment divided an annual service cost by an expected user base. One rheumatoid arthritis example apportioned €17,500 per year at a single hospital, yielding €17.50 per patient. An oncology self-management application apportioned €450,000 per year across expected national uptake, giving €25 per patient. These approaches are simple to apply but can miss important resource items if the apportionment base or cost pool is incomplete.
Must Read: Tele-ICU Strengthens Critical Care Delivery
Bottom-up micro-costing itemised resources across recognised categories such as Development, Maintenance, Implementation and Health Personnel Involvement. Reported components included app content development, medical device registration, hosting and maintenance, technical support, licences, onboarding, staff training, routine monitoring of app data and ongoing supervision or coordination. Estimates were higher when itemised in this way. A community mental health intervention reported £2202 to £2447 per patient depending on perspective. A programme for older adults with complex chronic conditions estimated €1765.93 per patient using prospective cost sheets. A rheumatoid arthritis telemonitoring service itemised vendor app costs apportioned to participants, training and coordination, totalling €914.80 per participant. Other evaluations assumed fixed per-user fees without detailing derivation, including €480 per patient annually for a multiple sclerosis app and €40 per patient per year for an inflammatory bowel disease licence.
Implementation choices carry cost implications. Standalone dashboards may be simpler to deploy but can add workflow burden if staff must navigate outside core systems. Deeper electronic record integration requires upfront investment and interoperability work, which should be reflected in cost estimates. From healthcare system perspectives, the time required from personnel to review data, act on alerts or provide ongoing supervision represents a per-patient cost that merits explicit quantification. From societal perspectives, patient time engaging with monitoring and changes in travel costs when in-person visits are substituted are relevant items that can shift incremental costs.
Cost-Effectiveness Results and Decision Uncertainty
Base case point estimates indicated that mobile device-based active remote monitoring was cost-effective in most analyses. Four evaluations reported dominance, combining improved health outcomes with reduced costs. Two reported improved outcomes at acceptable additional cost. One model-based evaluation for older adults with complex chronic conditions found the remote monitoring strategy was dominated by multidisciplinary primary care with family health teams, with worse outcomes and higher costs.
Despite positive signals, uncertainty was substantial. In all trial-based evaluations, bootstrapped incremental cost and effect pairs were dispersed across all four quadrants of the cost-effectiveness plane, indicating uncertain direction of both costs and health outcomes. For feasibility and non-inferiority designs, such dispersion aligned with the design aims. For superiority designs, it may reflect minimal differences between arms or insufficient power to detect changes in costs and outcomes over available follow-up. Model-based early assessments addressed uncertainty by presenting results across a plausible range of treatment effects where definitive effectiveness estimates were absent.
Analytic perspective influenced reported results. Some evaluations reported only healthcare system costs, others only societal costs and several reported both. Inclusion of productivity, informal care and travel in societal perspectives often shifted incremental cost estimates. Monitoring frequency varied widely across applications and no clear relationship with cost-effectiveness was identified across conditions and designs.
Mobile device-based active remote monitoring shows consistent signals of cost-effectiveness across diverse long-term conditions, with several evaluations suggesting dominance through improved outcomes and reduced costs. However, decision uncertainty remains material, particularly in trial-based analyses where incremental costs and effects span all quadrants. For decision-makers, transparent and standardised costing that captures set-up, integration and personnel time, alongside clear logic models specifying provider actions, will strengthen assessments. Early model-based work that explores plausible effect sizes and cost structures can clarify conditions for value. As services consider adoption, careful attention to perspective, workflow integration and sustained patient engagement will be central to realising potential benefits without over- or under-estimating costs.
Source: npj Digital Medicine
Image Credit: iStock
