Digital Health’s Hidden Costs: Weighing ESG Trade-Offs in Healthcare Innovation

Digital health technologies are widely promoted as cost-efficient, scalable tools for improving healthcare delivery and accelerating progress toward the Sustainable Development Goals. But a new narrative review argues that these benefits come with environmental, social, and governance costs that are rarely evaluated together — and that assessing any one of them in isolation can produce incomplete or misleading conclusions. The review, published in Frontiers in Digital Health, integrates digital health within an Environmental, Social, and Governance (ESG) framework and proposes a preliminary conceptual model for weighing the trade-offs involved in implementing these technologies.

The international team, led by Yacine Hadjiat of Dubai Health Innovations and Mohammed Bin Rashid University of Medicine and Health Sciences, with colleagues from institutions in the United States, Singapore, Australia, Malaysia, and France, searched PubMed/MEDLINE, Scopus, Web of Science, and Google Scholar for literature on digital health, sustainability, and ESG-relevant risk, with searches last updated in May 2026. Eligible sources had to address digital health technologies — including artificial intelligence applied to health, digital therapeutics, telemedicine, wearables, and health data infrastructure — and engage with at least one of the three ESG domains. Policy documents and institutional reports were included where they contributed evidence unavailable in the peer-reviewed record.

The authors deliberately chose a narrative rather than systematic design, reflecting the emerging and interdisciplinary nature of the topic. Notably, they did not use ESG terminology as a search term, because most relevant digital health literature does not adopt the vocabulary of sustainable finance; instead, the search was built around the substantive concepts that populate each domain, such as data-centre energy demand, digital exclusion, and algorithmic accountability. The authors are candid about the limitations of this approach: sources were selected purposively, screening counts were not maintained, no formal appraisal of methodological quality was undertaken, and the search is not reproducible in the sense required of a systematic review.

On the environmental side, the review highlights substantial costs associated with digital health technologies, including energy consumption by data centres, carbon emissions from computational processes, and environmental degradation linked to hardware production and disposal. The authors point to blockchain as an example of a digital technology with high energy costs, associated with between 22 and 29 million metric tonnes of carbon dioxide emissions annually. They also flag the growing energy demands of artificial intelligence and large language models, whose implications are rarely considered explicitly when AI solutions are chosen for health applications. The review emphasizes that the net environmental effect of any digital health intervention is deployment-specific rather than intrinsic to the technology: telemedicine may reduce travel emissions while adding computational burden, and the direction of that net effect remains unresolved and rarely quantified.

The social domain presents a parallel set of concerns. While digital health solutions can expand access to care, the review finds they may also worsen existing health inequities. Requirements for internet access, digital competence, and compatible devices can systematically exclude socioeconomically vulnerable groups. In one review of digital therapeutics cited in the paper, 11 of 18 studies required participants to have internet access, a smartphone or tablet, computer skills, medical insurance, or employment in order to take part, and only 3 of those studies addressed the ethical and distributional issues inherent in implementing digital technologies. The authors argue that interventions expanding access in aggregate may nonetheless widen relative inequity, setting the health goal of SDG 3 against SDG 10 on reduced inequalities.

A particularly stark trade-off emerges at the intersection of environmental and social domains. Low- and middle-income countries bear a disproportionate share of the environmental costs of digital infrastructure — through mineral extraction for devices and toxic electronic waste — while receiving fewer of the benefits, which accrue predominantly to high-income settings. The review also identifies a dependency risk: when external vendors or donors supply digital health capabilities, continuity of care becomes contingent on arrangements over which the local health system has limited control, and health data may be governed by terms set elsewhere. The authors caution, however, that little primary evidence was identified from low- and middle-income settings, and that conclusions concerning those contexts draw substantially on policy and intergovernmental sources rather than empirical studies.

Governance concerns form the third pillar of the analysis, and the authors note that evidence in this domain is more fragmented than for the environmental and social domains. Four recurring concerns are distinguished. First, data stewardship: questions of ownership, consent, secondary use, and cross-border transfer remain unresolved in most implementation contexts, and where governance is weak, the infrastructure enabling population-level insight is the same infrastructure enabling surveillance and commercial exploitation of health data. Second, algorithmic accountability: models trained on non-representative data can reproduce and amplify existing disparities, and opacity makes responsibility for clinical decisions difficult to locate. Third, regulatory and procurement architecture: environmental criteria are largely absent from health technology assessment and procurement processes, even though procurement is the point at which most digital health technologies enter health systems. Fourth, policy coherence: health, digital, and environmental policies are typically formulated by separate institutions, so trade-offs between them are seldom adjudicated explicitly.

Building on these findings, the review proposes an ESG-informed framework for evaluating digital health innovations, adapted from the Digital Health Equity Framework originally developed in the context of COVID-19. The framework treats ESG factors as integrated rather than parallel, on the premise that each domain conditions the others. It is explicitly a conceptual model rather than a measurement instrument, intended to structure deliberation when interventions are designed, procured, or evaluated. At the environmental level, factors such as resource availability, scarcity, pollution, and the presence or absence of a circular economy should be considered; at the social level, suitability for socio-cultural context, beliefs, and social norms; and at the governance level, ethics, data sharing, privacy, AI bias, and the alignment of health and environmental policy.

The authors also suggest practical mechanisms for making these trade-offs visible at the point of decision. Because digital health technologies are overwhelmingly procured by hospitals, health systems, insurers, and governments rather than patients, they propose that public tenders for AI-enabled health technologies could require disclosure of training and inference energy consumption, the carbon intensity of the compute on which the technology depends, hardware lifecycle and take-back arrangements, and energy efficiency per unit of clinical output. Much as food labelling enables informed consumer choice, energy labelling for AI-enabled health technologies would make the environmental trade-off visible to procurement bodies, who could weight such disclosures alongside clinical performance and cost.

The review concludes that no comprehensive ESG framework for digital health — encompassing a detailed taxonomy, standardized measures, and indicators — yet exists. The authors identify four priorities for future work: developing metrics and defensible weightings for each category in the model; prospectively validating the framework alongside real procurement or implementation decisions; conducting empirical work on the environmental footprint of specific digital health interventions measured at the level of the deployment rather than the technology class; and generating primary research from low- and middle-income settings so that frameworks guiding global digital health investment are not derived exclusively from high-income evidence. Until such work is done, the review argues, digital health innovations should be evaluated by weighing their benefits against potential environmental impacts, ethical considerations, and social implications — because digital health, in the authors’ framing, is not neutral, and the Sustainable Development Goals cannot be effectively addressed in isolation.

Subject of Research: Environmental, social, and governance trade-offs in digital health innovation

Article Title: Considering the trade-offs of environmental, social, and governance risk in digital health

Article References: Hadjiat, Y., McManus, D. D., Montana, C., Ng, Y. Y., Rezayan, L. A., Scharoun, L., Sapanel, Y. S., & Berbenyuk, A. (2026). Considering the trade-offs of environmental, social, and governance risk in digital health. Frontiers in Digital Health, 8, Article 1947743. https://doi.org/10.3389/fdgth.2026.1947743

Image Credits: AI Generated

DOI: 10.3389/fdgth.2026.1947743

Keywords: digital health, ESG, sustainability, health equity, artificial intelligence, carbon footprint, data governance, telemedicine, e-waste, Sustainable Development Goals, procurement, narrative review