Pour un système de santé décarboné : mieux compter, mieux prévenir, mieux soigner

1 -- 1 of 58 -- 2 Your participation in the work: proofreading and contributions Dear Reviewer, This report is an intermediate progress update: its purpose is to define the physical framework of the issue to ask the right questions. Although it is already the result of collective work, it remains an imperfect, incomplete, and evolving working document. It will be followed by a final report at the end of 2026. You may read this interim report in PDF format and send us your feedback by email at [email protected], or directly within this Google Docs document through comments and “suggestions” inserted into the text. Please note that by adding comments and suggestions in this Google Docs document, your contributions will be visible to everyone. If you wish to make edits directly in the text— which we encourage—but not on this public document, you may download the document in Word format. Some suggestions and comments may indeed give rise to debate: this is part of the process of constructive discussion and openness. We rely on your goodwill, and we will strive to ensure that the document remains pleasant to read. Any data sources you may provide will be highly valuable in finalizing this guide. We encourage you to contact us regarding any potential collaboration or data sharing at [email protected]. Looking forward to debating and progressing together, The team health of The Shift Project -- 2 of 58 -- 3 Table of Contents About The Shift Project think tank 4 Acknowledgements HYPERLINK "bookmark://_tyjcwt"6 I. Background, Purpose and Content of the Guide 7 A. A guide to calculating the carbon footprint at the level of care pathways 7 1) Carry out studies to better decarbonise 7 2) The need for a holistic vision of care pathways 8 B. A need for harmonization of methodology and data 10 1) Different methodologies make it impossible to compare results 10 2) A need to standardize the carbon data used 11 C. Objectives of the Guide 11 1) A guide that complements existing tools 12 2) The limits of this methodological guide 14 ● A guide focused on carbon issues 14 ● An evolving guide, which needs to be completed 15 ● A limit in the completeness of the data 16 II. Methodology: how to calculate the carbon footprint of care pathways? 17 A. General methodology 17 1) Define the objective of the study 17 2) Define the scope 18 3) Associate each stage of my journey with a reference emissions factor 19 4) Summing the emissions of each stage of my journey 21 5) Carry out a study of sensitivity, uncertainty, etc 22 6) Interpret the result 22 7) Highlight the limits of the calculation 22 B. How accurate is it to assess the carbon footprint of a care pathway? 23 1) A high variability in the quality of available carbon data 23 2) What level of detail in the scope? 26 3) The uncertainty calculation, a useful but incomplete tool for assessing the rigour of the calculations 28 4) Other methods to strengthen the robustness of a calculation 29 C. Finally, a methodology and precision that depend on the objective followed 31 1) Objective 1: Evaluate the emissions of a care pathway to identify the main emission items 31 2) Objective 2: Evaluate the average emissions of a care pathway to assess the carbon impact of a prevention or fair care action on the pathway in question 32 3) Objective 3: to compare the carbon footprint of two possible care pathways for the same pathology 35 4) What if I have multiple goals? 39 5) Assessment 39 III. Carbon data on care acts and care pathways 40 A. What is good carbon data? 40 1) General data 40 2) How precise should the carbon footprint of health products be taken into account? -- 3 of 58 -- 4 42 B. Construction of a carbon database to accompany this guide 47 1) Database construction principle 47 2) The case of care performed in the city 48 3) The case of care performed in multidisciplinary establishments 49 4) When and how to use the data in this database? 53 C. Towards frequent updating of the guide and the database 55 -- 4 of 58 -- 5 About the think tank The Shift Project The Shift Project is a think tank that works in favour of an economy free of carbon constraints. An association under the law of 1901 recognized as being of general interest and guided by the requirement of scientific rigour, its mission is to enlighten and influence the debate on the energy and climate transition in Europe. The Shift Project sets up working groups around the most decisive issues of the transition, produces robust and quantified analyses on these issues and develops rigorous and innovative proposals. It conducts influence campaigns to promote the recommendations of its working groups to political and economic decision-makers. It also organizes events that promote discussions between stakeholders and builds partnerships with professional and academic organizations, in France and abroad. The Shift Project was founded in 2010 by several personalities from the business world with experience in associations and the public sector. It is supported by several large French and European companies as well as by public bodies, business associations and, since 2020, by SMEs and individuals. Since its creation, the Shift Project has initiated more than 50 study projects, participated in the emergence of two international events (Business and Climate Summit, World Efficiency) and organized several hundred symposia, forums, workshops and conferences. It has been able to significantly influence several public debates and important political decisions for the energy transition, in France and within the European Union. The ambition of the Shift Project is to mobilise companies, public authorities and intermediary bodies on the risks, but also and above all on the opportunities generated by the "double carbon constraint" that tensions over energy supply and climate change represent together. Its approach is marked by a particular prism of analysis, based on the conviction that energy is a factor of development of the first order: therefore, the risks induced by climate change, intimately linked to the use of energy, are of a particular systemic and transdisciplinary complexity. Climate and energy issues condition the future of humanity; It is therefore necessary to integrate this dimension into our model of society as quickly as possible. It is supported by a network of tens of thousands of volunteers grouped within an association under the law of 1901: The Shifters, created in 2014 to provide volunteer support to the Shift Project. Initially designed as a structure to welcome anyone wishing to help the Shift through research, relay or support work, the Shifters are increasingly carrying out independent work, but always with one objective: to contribute effectively to the phase-out of fossil fuels at both the French and European levels. -- 5 of 58 -- 6 Acknowledgements Initiated in mid-2025 with the support of the French National Health Insurance Fund (CNAM), the French High Council for the Future of Health Insurance (HCAAM), and MGEN, this work was coordinated by Mathis Egnell (Project Coordinator at the Shift Project), led by Baptiste Verneuil (Project Engineer at the Shift Project) and produced with Thomas Rambaud (technical advisor and co- pilot of the Health Industries project in the Shift Project, consultant in a health service company, member of the Shifters), David Grimaldi (scientific advisor and co-pilot of the Health Industries project at the Shift Project, resuscitator, member of the Shifters), Elise Vacher (Project Manager at the Shift Project) and Laurie Marrauld (initiator of the Health-Climate-Resilience Research Program in the Shift Project, lecturer in public health at the EHESP and holder of the RESPECT Chair). They were also supported by Jean-Noël Geist (coordinator of the Health-Climate-Resilience Research Program at the Shift Project), Héloïse Lesimple (public affairs project manager at the Shift Project) and Mona Poulain (communication and events officer at the Shift Project). This project has also received the support of many professionals in the sector and more broadly in the health sector who have contributed voluntarily. By agreeing to reread this report, they took the time to share their knowledge. The team would also like to thank all the members of the Shifters Health Thematic Circle for their help, expertise and advice in the development of this publication. We would also like to thank the experts and organizations who were kind enough to share some data and expertise. Field data are essential to the development of our hypotheses and methods. Finally, we would like to thank our academic partner the RESPECT Chair (Resilience in Health, Prevention, Environment, Climate and Transition) of the EHESP, our technical partner ANAP (the National Agency for Health and Medico-Social Performance) for its contributions, and our partners on the entire research program the CNAM (French National Health Insurance Fund), the HCAAM (French High Council for the Future of Health Insurance) and MGEN. -- 6 of 58 -- 7 I. Background, purpose and content of the guide A. A guide to calculating the carbon footprint at the level of care pathways 1) Carrying out studies to better decarbonise Faced with climate change and increasingly restricted access to resources, particularly fossil fuels, the health system must both reduce its greenhouse gas emissions, develop its resilience and maintain or even improve its functioning. In any economic sector, emissions can be expressed as the product of the volume of goods produced (quantity) and the average greenhouse gas emissions per good produced (carbon intensity).1 The same applies to the health sector, where emissions are the product of the volume of healthcare (quantity) and emissions per treatment (carbon intensity). However, our initial work2 3 4 has highlighted a notable limitation: reducing the carbon intensity of healthcare alone is not enough to make the health system fully decarbonized and sustainable in the long term5, i.e., to reduce emissions by 80-90% by 2050. This is all the more true given that our previous work does not adequately account for a potential significant increase in demand for healthcare driven by population aging and the rise in chronic diseases. We can go a step further in the analysis by breaking down the emissions into a more complex mathematical equality: Thus, emissions can be expressed as the product of: 1. the carbon intensity of care: the amount of greenhouse gas emissions emitted per given care unit (corresponding overall to the emissions of all facilities and services in the healthcare system, relative to their level of activity), 2. the adequacy of care index: proportion of care provided that is not medically justified, whose benefit for the patient is insufficient or less than the risks incurred, 3. the intensity of care: level of resources mobilized (procedures, examinations, treatments, hospitalizations, care time) for the care of a patient or a given pathology, 1 For example, for transport, emissions depend on the distances travelled (volume) and greenhouse gas emissions per kilometre (carbon intensity). 2 Decarbonizing health for sustainable care, 2023, https://theshiftproject.org/publications/decarboner-sante- soigner-durablement/ 3 Let's decarbonize the autonomy sector together!, 2024, https://theshiftproject.org/publications/decarbonons-secteur-autonomie/ 4 Let's decarbonize the health industries, 2025, https://theshiftproject.org/publications/decarbonons- industries-sante-medicaments-dm/ 5 The Shift Project (2023). Decarbonizing health for sustainable care. https://theshiftproject.org/article/decarboner-sante-rapport-2023/ -- 7 of 58 -- 8 4. and their prevalence: frequency or proportion of patients concerned by a given type of care, pathology or treatment. The levers of decarbonization then appear to be the need to reduce the carbon intensity of care (1.), the limitation of irrelevant care through the right care (2.), the strengthening of secondary and tertiary prevention to reduce the number of care needed per patient (3.), and primary prevention to reduce the constraint on the care system (4.). Thus, to achieve the objective of an 80 to 90% reduction in greenhouse gas emissions from the health sector, a reduction in the overall demand on the health system appears necessary. However, assessing the decarbonisation potential of right care, disease prevention and health promotion measures requires the ability to carry out carbon footprint calculations at the scale of care pathways. Example: to assess the additional carbon cost related to the overprescription of imaging, it is necessary to know the average carbon footprint of an imagery. As such, we are publishing a first version of a methodological guide dedicated to the realization of these calculations, accompanied by operational carbon data allowing their appropriation and reuse by actors in the health sector. This publication is therefore a first step towards robust, harmonised and transparent assessments of the carbon implications of right care actions, prevention and health promotion policies, in support of a healthcare system that is both efficient, sustainable and decarbonised. 2) The need for a holistic view of care pathways The low-carbon transformation of the healthcare system cannot be limited to an approach based on sectors or isolated activities. The French Ministry of Health now advocates "stopping thinking in silos: community care, hospital care, medico-social care, ... Today, a pathway is understood as the comprehensive, structured and continuous care of patients."6 The decarbonization of patients care therefore requires the adoption of a global vision of the healthcare system. Example: Only comparing the carbon footprint of product A to the one of product B is not always enough to identify the least emitting option. It is often necessary to place the product within the overall context of the care pathway. Indeed, product A may have a higher carbon footprint than product B, but if it reduces hospitalizations, the risk of relapse or the frequency of use, then the carbon footprint of the entire care pathway may, in the end, be lower. 6 Ministry of Health. https://sante.gouv.fr/systeme-de-sante/parcours-des-patients-et-des- usagers/article/parcours-de-sante-de-soins-et-de-vie -- 8 of 58 -- 9 In this context, the care pathway constitutes a relevant scope of analysis. It describes, in a continuous and structured manner, all the stages of a patient's care, from the first resort to any follow-up phases, including all the places of care concerned (consultations, emergencies, hospitalization, follow-up care, home care, medico-social). This approach is consistent with the definition adopted by the European Pathway Association, which considers the care pathway as a complex intervention aimed at organising the care processes for a given group of patients, over a given period of time7. In this guide, we consider that care pathways can be broken down into a mesh of "elementary bricks" of healthcare (e.g. consultation, imaging, biology, surgery, etc.), juxtaposed to create a global care pathway. In the rest of the report, we will refer to them as "building blocks" or "carbon footprints of medical acts". They are thus the basic elements on the basis of which a journey can be analysed, compared or optimised. Please note that these "building blocks" already correspond to data built on the basis of other emission factors (used to assess the carbon footprint of emission items in Figure 1). For example, if treatment A is administered in a day hospital by infusion, the "day hospital for intravenous treatment" brick will include the carbon footprints of the equipment needed for the infusion, the carbon footprint of the drug but also a part of the emissions related to the hospital's heating, staff travel, patient travel, etc. Figure 1: Diagram distinguishing the different steps leading to the assessment of the carbon footprint of a care pathway Source: The Shift Project, 2026 7 European Pathway Association, https://e-p-a.org/care-pathways/ -- 9 of 58 -- 10 Carrying out carbon footprint studies at the scale of care pathways requires summing the emissions of each "elementary brick" of care involved and then allows: ● identify the dominant emission items on a real scale in patient care; ● prioritise decarbonisation levers where they are most effective (organisation of care, prevention, right care); ● compare different pathways for the same pathology. B. A need for harmonization of methodology and data 1) Different methodologies make it impossible to compare results Existing studies on the carbon footprint in health show a high variability both in terms of the methodologies chosen and the study perimeters. For example, most of the studies published today are hospital-centric and exclude care provided outside the hospital, such as consultations in town or home care. Significant differences also appear depending on the physical perimeter considered: some studies only consider medical equipment, while others include all the department's consumption, which can lead to differences of a factor of ten8 9 in the results obtained. Finally, the methodological choices for allocating emissions, whether based on expenditure10, headcount or surface area11, strongly influence the results. Some studies show that these allocation choices can lead to significantly different results12, making comparisons between studies and the reuse of certain data very complicated. In the rest of the report, we will delve into this variability in more detail. 2) A need to standardize the carbon data used The results of studies on the carbon footprint of care pathways are strongly influenced by the quality and nature of the carbon data used (this issue is described in the PEF methodology13). These studies remain largely heterogeneous, which limits the comparability and robustness of the results. This heterogeneity can be explained, for example, by the calculation methods used. Emission factor studies may be based on detailed life cycle assessments (LCAs14), 8 Pollard, A. S., Paddle, J. J., Taylor, T. J., & Tillyard, A. (2014). The carbon footprint of acute care: how energy intensive is critical care?. Public health, 128(9), 771–776. https://doi.org/10.1016/j.puhe.2014.06.015 9 McGain, F., Burnham, J. P., Lau, R., Aye, L., Kollef, M. H., & McAlister, S. (2018). The carbon footprint of treating patients with septic shock in the intensive care unit. Critical care and resuscitation : journal of the Australasian Academy of Critical Care Medicine, 20(4), 304–312. 10 Sack, F., Irwin, A., van der Zalm, R., Ho, L., Celermajer, D. J., & Celermajer, D. S. (2024). Healthcare-related carbon footprinting-lower impact of a coronary stenting compared to a coronary surgery pathway. Frontiers in public health, 12, 1386826. https://doi.org/10.3389/fpubh.2024.1386826 11 Prasad, P.A., Joshi, D., Lighter, J. et al. Environmental footprint of regular and intensive inpatient care in a large US hospital. Int J Life Cycle Assess 27, 38–49 (2022). https://doi.org/10.1007/s11367-021-01998-8 12 Ibid. 13 Product environmental footprint. Suggestions for updating the Product Environmental Footprint (PEF) method, tableau 20 page 95, https://eplca.jrc.ec.europa.eu/permalink/PEF_method.pdf 14 In scientific articles, the term LCA (Life cycle assessment) is used. -- 10 of 58 -- 11 simplified material analyses, or input-output macroeconomic approaches15. These methods do not cover the same areas and can lead to significant differences in the emission factors produced16. This topic will be discussed in more detail in the section A high variability in the quality of available carbon data. Therefore, the harmonization of methodologies and data used appears to be an essential condition for producing robust, transparent and comparable results. C. Objectives of the guide The long-term objectives of this guide are: ● provide a clear and reproducible methodology for assessing the carbon footprint of a care pathway; ● to constitute a public database of emission factors associated with the basic building blocks of care pathways; ● to provide a robust methodology for the construction of new building blocks; ● to provide an analysis grid to describe the quality of (future) studies on the carbon footprint of a basic building block (act of care) or a care pathway, ● to help adopt a rigorous and adapted methodology for comparing care pathways or scenarios. It is aimed at both those who produce studies and those who analyse or use them (decision- makers, institutions or reviewers of academic publications), by giving them the keys to interpret the results correctly, according to the objectives pursued and the methodological choices made. 1) A complementary guide to existing tools This methodological guide will provide both the basis for our work on prevention and fair care, and tools for health actors so that they can in turn carry out such studies. This guide will be accompanied by a database of emission factors for medical procedures that will be co- constructed with many other actors. It should be noted, however, that our work is in addition to many initiatives on this subject that have emerged in France in recent months and years. Although they share common points with these projects, our approach aims to be complementary above all. We are not trying to ignore the value of this work or start from scratch. On the contrary, as you will see, we want to rely on their methods and data to achieve our goals. 15 In scientific articles, the term EEIO (Environmentally extended input-output) is used. 16 Ecovamed (2025). Medical device carbon footprint assessment: comparison of different methodologies. https://cdn.prod.website- files.com/6151b650ce4cd9198b1fd7e8/6919f968449a0b63f0cbd39c_Medical%20devices%20carbon%20footpr int%20and%20life%20cycle%20assessment_Ecovamed_Sept%202025.pdf -- 11 of 58 -- 12 Among the initiatives we identify, some focus specifically on the health system and medical procedures: ● Carebone®17: is a methodological tool developed by the Assistance Publique – Hôpitaux de Paris (AP-HP) to estimate the carbon footprint of healthcare activities, health products and patient journeys. This tool, which currently focuses on the hospital, is available in open access (with transparency on the methodologies, sources and data used) and provides factors. ● Osmose stands for "Monitoring and measurement tool towards the optimization of eco-responsible care"18. It is a software for the eco-design of care that the National Association for the Continuing Education of Hospital Staff (ANFH) plans to make available to healthcare establishments. This tool will make it possible to assess the carbon impact, waste production and water consumption associated with care pathways. ● The Sustainable Care Pathways Guidance. This guide, produced by the Sustainable Healthcare Coalition19, is the result of a collaboration between private and public actors in Great Britain. It also aims to standardize the assessment of the carbon footprint associated with care pathways. It offers methods and data to estimate the impact of different phases of a care pathway: consultation, hospital room, surgery, transport, etc. ● The Methodology for Assessing the Carbon Footprint of French State Medicines20. This is a guide that should allow pharmaceutical operators to calculate the carbon footprint of their medicines more simply over their entire life cycle (from the extraction of raw materials to the end of the drug's life) and with a breakdown of these emissions according to the main items. ● The "Greenhouse Gas Accounting Sector Guidance for Pharmaceutical Products and Medical Devices"21 is a sectoral version of the "Green House Gas Protocol"22 adapted to the production of drugs and medical devices. Just like the previous methodology, the scope is not "the care pathway" but "health products". However, these methods and the data derived from them remain very valuable and necessary to assess the carbon footprint of care pathways. 17https://www.aphp.fr/careboner-un-outil-pour-decarboner-le-soin-mis-la-disposition-de-tous-les- professionnels-de-sante 18 https://achat-logistique.info/durable/osmose-logiciel-de-lanhf-pour-ecoconcevoir-les-soins/ 19 https://shcoalition.org/sustainable-care-pathways-guidance/ 20https://www.entreprises.gouv.fr/la-dge/publications/methodologie-devaluation-de-lempreinte-carbone-des- medicaments 21https://ghgprotocol.org/sites/default/files/tools/Summary-Document_Pharmaceutical-Product-and-Medical- Device-GHG-Accounting_November-2012.pdf 22 The Greenhouse Gas Protocol is an international protocol aimed at establishing a regulatory framework to better define greenhouse gas emissions, particularly those from industry, with a view to accounting for them in order to reduce them. -- 12 of 58 -- 13 Other of these initiatives correspond to work aimed at standardizing carbon footprint assessment methods, without a direct link to the health system: ● The carbon® footprint method23, initially developed by the French Environment and Energy Management Agency (ADEME), is now supported by the Association for the Low Carbon Transition (ABC). Its objective is to measure all the greenhouse gas emissions of an organization, a territory or an activity. ● The Life Cycle Assessment (LCA) method. It is based on the ISO 14040 (LCA principles and framework) and ISO 14044 (LCA requirements and guidelines) standards, which define how to set objectives, define the scope, collect data, perform impact calculations and interpret results. It is also important to note the strong link between these different works. For example, the Sustainable Care Pathways Guidance is based on: ● The ISO 14040 and ISO 14044 standards for the general Life Cycle Assessment (LCA) methodology. ● The Greenhouse Gas Protocol Sector Guidance for Pharmaceutical Products and Medical Devices ("Sector Guidance") for the specific application to the pharmaceutical and medical device sector. In this way, our work adds to a range of national and international initiatives, some providing methods and others providing data. We hope that it will enrich the reflections on the environmental footprint of care pathways. 2) The limitations of this methodological guide ● A guide focused on carbon issues This guide focuses exclusively on the carbon footprint and does not cover other environmental indicators (water consumption, toxicity, waste production, etc.). This choice reflects the Shift Project's field of expertise, which focuses on energy and climate. However, this focus does not mean that the other impacts are negligible. On the other hand, the proposed methodology is largely transposable to other environmental indicators. Example: The methodological limitations of carbon assessments can also be found in the evaluation of other indicators. For example, many studies on water consumption in the health sector lack methodological rigor, limiting themselves to the visible uses of health facilities, without integrating indirect consumption related to the production of health goods (otherwise known as the "water footprint"), thus biasing the conclusions of the 23https://abc-transitionbascarbone.fr/wp-content/uploads/2022/03/bilan-carbone-v8-guide-methodologique- final.pdf -- 13 of 58 -- 14 study24. In addition, in many cases, greenhouse gas emissions are correlated with other environmental impacts. For example, a reduction in the consumption of medical devices linearly reduces the carbon impact and waste production; "Right care" policies usually lead to a combined reduction in emissions, waste and resource use. Carbon is thus, in many cases, a relevant proxy to guide impact reduction actions. Finally, assessments of some indicators using methodologies such as life-cycle assessments are considered less robust than carbon analyses. This is the case, for example, with the evaluation of water consumption or fossil resources consumed25. ● An evolving guide, which needs to be completed The carbon footprint of medical acts, care pathways or health products is the subject of an ever-increasing number of publications. The HealthcareLCA database 26 references a large part of it. As of December 31, 2021, it had 152 studies, half of which had been published between 2018 and 202127, thus reflecting an acceleration in the published and therefore available data on these issues. This phenomenon did not stop in 2021 since, since then, more than 240 new studies have been added28. If we add to these figures the increasing quantities of Carbon® Footprints carried out and published by hospitals, theses carried out29, learned societies deepening the link between their activities and the need to reduce their greenhouse gas emissions, as well as the publication of methods 24 McGain, F., McAlister, S., McGavin, A., & Story, D. (2012). A Life Cycle Assessment of Reusable and Single-Use Central Venous Catheter Insertion Kits. Anesthesia & Analgesia, 114(5), 1073–1080. https://doi.org/10.1213/ane.0b013e31824e9b69 25Suggestions for updating the Product Environmental Footprint (PEF) method, table 2, page 35, eplca.jrc.ec.europa.eu/permalink/PEF_method.pdfc 26 HealthcareLCA is a centralized, open-access repository that lists numerous environmental impact assessments related to the healthcare sector, covering more than twenty years of research. https://healthcarelca.com/database 27 Drew et al, 2022, HealthcareLCA: an open-access living database of health-care environmental impact assessments, https://www.sciencedirect.com/science/article/pii/S2542519622002571 28 Database consulted in January 2026 29 To name a few: - Environmental impact of general practice: study of the carbon® footprint of private general practice in Lot-et-Garonne, Claire Justine Houziel, 2022, - Study of the environmental impact of general anesthesia with intravenous propofol or inhaled sevoflurane by comparative life cycle analysis after ecodynamic and ecotoxicological investigations, Soizic Dieulouard, still in progress in January 2026 - Establishing the carbon® footprint in pharmacies and reducing their environmental impact, Laurent Léger, 2025 - Definition of the principles of eco-prescription of medicines, Salomé Dupray, 2024 -- 14 of 58 -- 15 such as the State Methodology 30 or the PAS 2090 method31 to assess the carbon footprint of medicines, There is a good chance that: ● We have not managed to identify and integrate into the guide all the data available to date on the carbon footprint of care-related activities (building blocks); ● The data offered in this guide is no longer valid or no longer the most accurate a few months/years later; ● New data will be produced after publication of the guide, on medical procedures not yet covered. Therefore, we would like this guide, as well as the database that will be associated with it, to be considered as tools that can evolve over time. Some methodological elements may be updated, added or completely modified after the publication of the first version of these tools. The idea would then be to transform the guide and the database into open-access documents, updated regularly with data whose quality has been audited and deemed sufficient. This point is detailed in the section Towards frequent updating of the guide and the database. ● A limit in the completeness of the data Mainly for data availability reasons, the guide and the database that will be published at the end of November are not intended to be exhaustive, in particular for specific or highly specialized cases (for example certain radiotherapy treatments or very specific techniques). In addition, the data that will be offered will be mostly average values, which do not distinguish between the specificities of organizations (for example, a consultation carried out in a multi-professional health home versus in an individual practice). Example: This guide could offer an estimate of the average carbon footprint of a "2-hour surgery, requiring sevoflurane and with the assistance of a surgical robot", without proposing the carbon footprint of a "right hepatectomy in a regional hospital". Nevertheless, the transparency of assumptions, scopes and calculations is intended to allow users to adapt the proposed data to their specific contexts, and to gradually enrich the guide as new data become available. Example: With the existing data "consultation in general medicine in an office", a user could adapt it so that it is representative of a "consultation in a multi-professional health home". The calculation method used and the transparency of the data will make it possible to adapt certain parameters to calculate a similar building block. For example, by re-calculating only the data related to buildings and the travel of medical secretaries. 30 Methodology for assessing the carbon footprint of medicines, https://www.entreprises.gouv.fr/la- dge/publications/methodologie-devaluation-de-lempreinte-carbone-des-medicaments 31 PAS 2090 Pharmaceutical products – Product category rules for environmental lifecycle assessments – Specification, https://standardsdevelopment.bsigroup.com/projects/9025-12045 -- 15 of 58 -- 16 Moreover, as we shall see, the search for absolute precision is not always necessary to reach a conclusion. -- 16 of 58 -- 17 II. Methodology: how to calculate the carbon footprint of care pathways? A. General Methodology The general methodology to be followed to assess the greenhouse gas emissions of a care pathway is divided into a succession of standardised steps, which are explained in the document "Care Pathways: Guidance on Appraising Sustainability32": ● Step 1: Define the objective of the study, ● Step 2: Define the scope of the study: ○ Description of the care path(s) in a set of stages, ○ Definition of the unit of analysis (otherwise known as the "functional unit"), ● Step 3: Collect the data necessary to assess the greenhouse gas emissions of each stage of the care pathway and explain all the sources and assumptions used, ● Step 4: Sum the emissions of each stage of the journey, ● Step 5: Conduct a sensitivity study and an uncertainty study, ● Step 6: Interpret the results against the objective, ● Step 7: Specify the limits of the calculation performed. As described in Figure 1, the carbon footprint of a given care pathway corresponds to the sum of the footprints of the elementary building blocks that make it up (e.g. consultation, imaging, surgery, etc.). The emissions of these bricks are calculated from emission factors. 1) Define the purpose of the study The first step is to define the objective of the study. This is the step that we consider the most important because it will calibrate the rest of the study. In addition, as we will see in Part C. Finally, a methodology and precision that depend on the objective followed, the methodology to be applied and the level of rigour required depends directly on the objective set. This step aims to answer the following questions: 1. What is the care pathway(s) studied? 2. What do I want to demonstrate with my study? Am I looking at: ○ Comparing several possible treatments? ○ Understanding how to decarbonize my care pathways? ○ Estimating the carbon footprint of the management of a pathology in France? ○ Highlighting one treatment over another? 3. How will I use the results of the study? 32 Sustainable healthcare coalition, Second Edition, December 2023, https://shcoalition.org/wp- content/uploads/2024/01/Sustainable-Care-Pathways-Guidance-Main-Document-December-2023.pdf -- 17 of 58 -- 18 2) Define the scope The objective of this section is to answer the following questions: ● When/where does the care pathway studied begin and when/where does it end? ● What is the functional unit of my study? ● What are the physical flows which must be considered and translated into greenhouse gas emissions? More specifically, it is necessary to start by describing the "medical" scope of the care pathway, specifying both the acts and care included as well as the duration considered. In theory, all medical steps should be considered. In practice, this task is very complex because, for each pathology, the treatment modalities can vary from one patient to another, even infinitesimally33. Also, the description of an "average" pathway necessarily implies choices and simplifications given the multiplicity of possible trajectories and the gaps between theoretical and actual trajectories (linked in particular to differences in complications, access to care, compliance, etc.). If such simplifications are possible, the assumptions made and their reasons should be explicitly documented. We also recommend representing the route in the form of scenarios or modular bricks, where several plausible alternatives would be explored. Example: Assess the greenhouse gas emissions associated with diabetes management and assess, in the form of an alternative scenario, the case where the patient suffers a given complication. Finally, to facilitate communication and therefore understanding around the selected care pathway(s), we recommend representing them in the form of a diagram, as a sum of "elementary" bricks of care acts, retracing the patient's pathway, and the frequency of each act. The temporal perimeter must consider all the events likely to occur during the patient journey for one or more given pathology(ies) (e.g. relapses or adverse events). Should a discount rate be defined? When the scope covers a course spread over several years, it may be asked whether a medical act or treatment given in year t emits as much as the same act or treatment given 33 We note that it is "easier" to accurately describe the care pathway in the case of a retrospective study involving a single patient than in the case of a study involving an average patient pathway. -- 18 of 58 -- 19 in year t+1. This question is identical to the one we would ask ourselves in economics: how can we compare interventions at different times? Indeed, if France respects its climate commitments and in particular the trajectory of the National Low Carbon Strategy, then a trip, a heating unit consumed, the construction of a cabinet or even a food consumed are supposed to be less carbon-intensive than today. In other words, a medical procedure today would release more greenhouse gases than the same procedure performed in the future. But by how much? How can this be taken into account? Which "discount rate used"? How can a potential non-compliance with decarbonisation objectives be considered? Is it better to use the same emission factors? For this interim report, we do not have any certainties to say on this subject, but we are open to any element that will allow us to deepen this question. This information should then be summarized in the form of a functional unit34. Finally, the description of the perimeter must be accompanied by an exhaustive description of the physical flows to be taken into account and translated into greenhouse gas emissions: consumption of medicines and devices, quantity of electricity consumed, distances travelled by patients by mode of transport, purchases of goods and services, waste management, etc. construction of buildings, etc. Whether you adopt a carbon footprint or life cycle analysis approach, the principle is the same: integrate all the direct and indirect emissions associated with each brick of the pathway. 3) Associate a reference emissions factor at each stage of my journey If the previous step has been carried out correctly, this new step starts from a diagram summarizing all the "elementary bricks" of care (number of consultations with doctors, trips to pharmacies, days of hospitalization in intensive care, etc.) considered and all the physical flows mobilized at each stage. It then consists in theory of estimating the carbon footprint of each of the identified building blocks. In other words, for each stage of my career, it is now necessary to make an exact inventory of the physical flows mobilized and translate them into greenhouse gas emissions. 34 In life cycle analysis, the functional unit makes it possible to quantify the function provided by products or services studied by the LCA (this is called a service rendered). It depends on the objectives of the LCA, defined and serves as a reference. Here, it would have an equivalent function: to quantify the function performed by the care pathway. -- 19 of 58 -- 20 This step is complementary to the step to define the perimeter. Indeed, the previous step allows you to define all the elements to be integrated into the analysis. The aim here is to ensure that, for each of these elements, the data collection is sufficiently exhaustive. Indeed, in theory, we can have defined a complete scope of study and yet forget to take into account half of the consumables, part of the energy consumption (such as the consumption of gas for the production of domestic hot water) or certain types of transport. Example: For the "Consultation with the general practitioner" step. Greenhouse gas emissions associated with the following must be assessed: ● The movement of the patient, the doctor, the medical assistants, etc ● The energy consumed to heat and light the premises, and to operate the machines used during the consultation, ● The production of medicines and consumables used during the consultation, ● The construction of the practice, part of the emissions of which must be attributed to the consultation, ● etc. The first trap to avoid is to forget to take into account certain physical flows. While many activities are directly related to the care of a patient, some are considered to be supportive, invisible activities, but just as necessary to the care. Example: For a hospital consultation, the emissions related to the doctor's travel, the use of consumables or the heating of the consultation site are direct flows that seem natural to take into account. But it is also necessary to consider the waiting room which can be used for other specialties, the secretariat without which the appointment would not have been possible, the hospital's administrative department, the purchasing department, etc. It is then necessary to identify an imputation key (time spent by patients, number of staff mobilized, etc.) that allows the right fraction of the emissions of these support activities to be taken into account. The second trap to avoid is double counting. Indeed, depending on the data used and the assumptions made, it is possible that, without realizing it, an activity is counted twice. This can happen in particular for emissions related to patient transport or the prescription of drugs and medical devices. Example: If a "Surgical operation" and a "Post-operative hospitalization" brick both take into account the patient transport, then summing them would risk counting this transport twice. The last trap to avoid is considering a too large proportion of greenhouse gas emissions. This happens in particular when you want to compare two care pathways and take into account -- 20 of 58 -- 21 patient transport, the activity of health professionals or the emissions related to the construction of buildings. In these cases, it is necessary to think about imputing only a defined portion (and calculated using a proxy such as the duration of intervention for example) of the associated greenhouse gas emissions. Example: In the majority of studies, patient transport is entirely attributed to the greenhouse gas emissions of a care pathway. However, this same patient can go for a consultation, during a journey where he also intends to do shopping and pick up his children from school. Thus, the journey is shared and greenhouse gas emissions should not be entirely attributed to the emissions of the care pathway35. The same applies when a health professional has travelled to work. Over the course of a day, this health professional will see many patients. Also, only a portion of the emissions related to the home-work journey must be attributed to the emissions of the care pathway. The objective of this guide will be to simplify this step in part by making available a set of already consolidated data that can be directly associated with an act or that can be adapted to then be associated with this act. 4) Sum the emissions from each step of my care pathway Once the carbon footprint from each step of the care pathway has been estimated, all that remains is to sum these footprints to obtain an estimate of the greenhouse gas emissions of the care pathway. 5) Conduct a sensitivity study, uncertainty study, etc Having an estimate of the greenhouse gas emissions of a care pathway is not enough to conclude. In addition, an uncertainty study must be carried out. Indeed, throughout the study, assumptions will have been made and data with different degrees of uncertainty have been used. Therefore, the results should be presented in the form of a confidence interval. However, as detailed below, the uncertainty calculation covers only a portion of the potential sources of error: it only reflects the margins of error within the chosen perimeter, not the elements missing from the perimeter. Also, this uncertainty study must be accompanied by a sensitivity study to test the sensitivity of the results obtained to the sources and assumptions used. For more information on this, you can refer to the B)3. 35 This marks a major difference between the carbon footprint method and the method to be applied when you want to know the emissions of an activity or compare two activities with each other. Indeed, with the carbon footprint method, the objective is to integrate into the perimeter all the activities over which we have control (and therefore which we can influence to decarbonize). When you want to know the carbon footprint of an activity and compare it with others, the idea is rather to include the share of emissions actually attributable to the activity in question. -- 21 of 58 -- 22 6) Interpreting the result Once all the previous steps have been completed, it is possible to interpret the result. This interpretation depends first and foremost on the objective and the questioning set out in step 1. This interpretation of the results must therefore be accompanied by a reminder of the objective of the study and the scope and flows that made it possible to reach the conclusions communicated. This reminder is necessary to emphasize the framework in which the results can be used and communicated. Indeed, given the diversity of medical practices and types of establishments, some results may hold true in certain cases and prove invalid in others. In other words, the results are not verified and should only be communicated on the scope of the study. Finally, depending on the methodological rigour followed and the quality of the data used, it may not be possible to conclude as to the objective set. For more information, you can refer to Part C). 7) Highlight the limitations of the calculation The interpretation of the result must be followed by an analysis of the limitations related to the calculations performed. These limits relate to the assumptions made, the data used (medical and carbon) and the selected scope and aim to identify the points to be deepened to improve the study. These limitations can then be highlighted by the sensitivity study carried out in step 5, on a chosen hypothesis or on an emission factor. B. What level of precision is needed to assess the carbon footprint of a care pathway? 1) High variability in the quality of available carbon data As we quickly saw in the introduction, whether it concerns a medical procedure (e.g. the carbon footprint of a day in intensive care36 37 38 39 40) or a complete care pathway (e.g. the carbon footprint of the management of an anterior cruciate ligament injury41), the published data show significant variability at several levels. 36 The carbon footprint of acute care: how energy intensive is critical care?, 2014, Pollard et al 37 The carbon footprint of treating patients with septic shock in the intensive care unit, 2018, McGain et al 38 Environmental footprint of regular and intensive inpatient care in a large US hospital, 2022, Prasad et al 39 Which carbon footprint for my ICU? Benchmark, hot spots and perspectives, 2025, Bardoult et al 40 Assessing the carbon footprint of the initial 24 h post-severe trauma admission in a French ICU: a pilot study, 2025, Marion et al 41 Environmental life cycle assessment of surgical versus conservative care pathways for an anterior cruciate ligament injury, 2025, Boiko et al -- 22 of 58 -- 23 ● Variability linked to geographical scope A piece of data may have been constructed with activity data (available surface area per patient, amount of energy consumed per hospital, distance covered by medical transport, etc.) and emission factors (carbon intensity of electricity, medicines, etc.) focusing on a country that is not the one we wish to study. Example: A study comparing surgical procedures in different hospitals shows, for example, that the results can vary significantly due to local factors: electricity emission factors, surface area of the operating theatres, equipment used, choice of suppliers or number of patients treated per day42. For the same act of care, these differences inevitably lead to significantly different results. Depending on the data used to assess the carbon footprint of a care pathway, the results obtained can then lead to erroneous conclusions43 (remember, for example, that the U.S. electricity mix is 10 times more carbon-intensive than the French electricity mix44). Example: several studies have shown that the carbon footprint of single-use medical devices is less carbon-intensive than the reusable equivalent. However, when using data adapted to the French electricity mix and considering the appropriate scope, the conclusions are almost systematically the opposite. Indeed, as shown in the report "Let's decarbonize the medical device industries", 45if we adjust the calculations of the Duijndam study (2022) (which compares the carbon footprint of single-use and reusable bronchoscopes) by applying the emission factor of the French electricity mix, the carbon footprint of reusable bronchoscopes is thus reduced by 34%. ● Variability linked to the temporal perimeter Data evaluated in 2015 may no longer be representative of the same activity in 2026. This can result from both factors external to the health sector, such as changes in the national electricity mix, and internal factors, such as changes in healthcare practices or the consumption of health products. 42 MacNeill, A. J., Lillywhite, R., & Brown, C. J. (2017). The impact of surgery on global climate: a carbon footprinting study of operating theatres in three health systems. The Lancet. Planetary health, 1(9), e381–e388. https://doi.org/10.1016/S2542-5196(17)30162-6 43 Grimaldi, D., Egnell, M., Verneuil, B., & Hosten, E. (2023). The carbon footprint of ICUs depends on the electricity mix of the national or local grid. BMJ (Clinical research ed.), 382, 1773. https://doi.org/10.1136/bmj.p1773 44 ADEME fingerprint database consulted in January 2026, https://base-empreinte.ademe.fr/donnees/jeu-donnees 45 Let's decarbonize the health industries, 2025, https://theshiftproject.org/publications/decarbonons- industries-sante-medicaments-dm/ -- 23 of 58 -- 24 Example: with the gradual shutdown of nitrous oxide distribution networks in hospitals46, anesthesia-related emissions could be significantly reduced. ● Variability linked to the flows considered Carbon data may have, voluntarily or not, excluded from its scope of study, the energy consumed for heating, patient transport or consumables. This can then have more or less significant consequences on the results of a study based on this type of data, in particular depending on the relative share of forgotten flows in total emissions. Example: not taking into account the consumables used throughout the care pathway will give rise to more uncertainty than not taking into account the construction of the hospital car park. ● Variability linked to the quality of the carbon data used The carbon footprint of a care "building block" or care pathway may have been estimated on the basis of monetary emission factors, another may have been estimated on the basis of life cycle analyses47 and another on the basis of a hybrid method mixing LCA and monetary emission factors (often called a "hybrid method"). The level of uncertainty associated with the outcome can then be very different. In addition, the data differs in its level of quality and lifecycle coverage. Example: as seen above, some medical device LCAs only consider the production of raw materials, without integrating the processing, assembly or sterilization processes48. The majority of the published data do not consider industrial R&D or headquarters and field activities, which nevertheless account for a significant share of greenhouse gas emissions49 50. Thus, the data used can lead to significant underestimates of the results. Thus, two emission factors from the same method can be of heterogeneous quality (depending on the completeness of the scope, the quality of the primary data, etc.). The mere 46 SFAR calls for a definitive end to the use of N2O networks by stopping their supply https://sfar.org/la-sfar-appelle-a-larret-definitif-de-lutilisation-des-reseaux-de-n-2o-en-arretant-leur- approvisionnement/ 47 It is important to note that monetary data is not systematically less accurate than a life cycle analysis. 48 The impact of surgery on global climate: a carbon footprinting study of operating theatres in three health systems, MacNeill et al, 2017 49 Let's decarbonize the health industries, 2025, https://theshiftproject.org/publications/decarbonons- industries-sante-medicaments-dm/ 50 Piffoux et al, Carbon footprint of oral medicines using hybrid life cycle assessment, Journal of Cleaner Production 2025, https://www.sciencedirect.com/science/article/pii/S0959652624030257 -- 24 of 58 -- 25 mention of a life cycle analysis is therefore not in itself a guarantee of quality: in practice, there is a gradation in the level of confidence that can be given to the results, depending on methodological transparency, the representativeness of the assumptions, the use of recognised emission factor databases and the presence of critical reviews or independent verifications. In this context, it is essential to examine the origin of the data and the methodological choices in order to assess the reliability of the orders of magnitude put forward. This point will be clarified in the section The LCA approach: what criteria for a rigorous LCA?. ● Variability linked to the representativeness of the carbon data used51 In some cases, to assess the GHG emissions of a medical procedure, studies are based on a specific example. Example: in the study by Bardoult et al52, the researchers evaluate the carbon footprint of a day in intensive care based on data from the Saint-Brieuc hospital in Brittany. In others, the results are based on a more representative sample. Example: in Coustal's study53, the author uses data from 7 medical practices to assess the carbon footprint of a consultation with a general practitioner, distinguishing between rural and urban areas. While the size of the study's sample size is not a guarantee of the quality of a piece of data, it does impose a framework for use. If, in order to compare two care pathways, we use data based on a specific case that we extrapolate to all the pathways, we may be led to overestimate or underestimate a source of emissions and therefore draw an erroneous conclusion. Example: between assessing the carbon footprint of a practice-based consultation where all patients walk there, and a practice where all patients travel by car, the associated emissions can differ significantly. Depending on the data used, we can in one case conclude that transport is the main source of emissions and therefore that any changes in practices at the national level to reduce this transport would be beneficial. And in the other, the conclusion may be that patient transport does not represent a significant source of GHGs. 51 By primary data, we mean all activity data that has then been translated into a carbon footprint. For example, to evaluate the emissions related to patient transport, I use as primary data the average distance traveled and the modes of transport used. 52 Which carbon footprint for my ICU? Benchmark, hot spots and perspectives, 2025 53 Thesis in general medicine, Environmental impact of general medicine: 2021 carbon® footprint of 7 general practice practices in Gironde, Arnaud Coustal, 2023 -- 25 of 58 -- 26 Also, the resources published and currently available on the carbon footprint of care procedures are of heterogeneous quality. This does not mean that these available data should be excluded, but, depending on the type of data used, the accuracy of the calculation in the result may vary and may not be able to meet a previously set objective. In other words, the possibility of concluding a study depends on the type of data used. A conclusion can be true on average but false in some cases, it can also be true in one year but false in the following years, etc. We will come back to this later, sensitivity analyses aim in particular to understand this risk. 2) What level of detail in the scope? In theory, all the flows induced by an activity must be considered in a calculation of the carbon footprint of this activity. In practice, it is neither possible nor relevant to model all flows with the same degree of precision. The challenge is therefore to determine at what point a data can be simplified, aggregated or excluded without calling into question the validity of the conclusions. Several methodological frameworks recommend the use of exclusion thresholds. The PAS 205054 international standard for calculating the carbon footprint of products allows the exclusion of a flow if it represents less than 1% of total emissions, provided that the sum of the excluded flows does not exceed 5%. These criteria have also been used in the State's Carbon Footprint Assessment Methodology55 and in the European Commission's Product Environmental Footprint Category Rules Guidance56. The identification of insignificant items can be done with an initial order of magnitude calculation using proxies or, on the basis of other studies. The mass can be a seemingly alluring proxy and quite simple to use. Be careful, however, that it is not always a good indicator and therefore a good proxy : for example, for complex health products involving low masses (such as monoclonal antibodies or stents), the carbon footprint depends mainly on the manufacturing processes. In these situations, the euros spent may be a more relevant proxy than weight. In the case where two "close" proxies can be used, the most conservative proxy (the one that gives the highest result) should always be used. For those items whose emissions are below the exclusion thresholds, we can choose to simplify them (for example, for emissions induced by waste treatment, not to distinguish between incineration and landfill), to estimate them quickly using proxies or to exclude them 54 BSI, PAS 2050, 2011. https://knowledge.bsigroup.com/products/specification-for-the-assessment-of-the-life- cycle-greenhouse-gas-emissions-of-goods-and-services 55 French State, Carbon Footprint Assessment Methodology. 2025. https://www.entreprises.gouv.fr/files/files/Publications/2025/Guide/15012025-methologie-carbone- medicaments.pdf 56 European Commission, Product Environmental Footprint Category Rules Guidance, 2018. https://eplca.jrc.ec.europa.eu/permalink/PEFCR_guidance_v6.3-2.pdf -- 26 of 58 -- 27 from the perimeter, possibly by adding to the final result a conservative margin of around 5% (equal to the maximum authorised exclusion threshold). Some emissions can also be excluded if they are too difficult to estimate, such as support functions in hospitals or purchases of services in the health industry, even if these items can be significant57. In any case, transparency on the excluded data is essential. Finally, and we will come back to this in the section "Finally, a methodology and precision that depend on the following objective", when the objective of the study is to compare several care pathways with each other, the medical acts common to these acts can be excluded from the perimeter. 3) Uncertainty calculation, a useful but incomplete tool for assessing the rigour of calculations The purpose of calculating uncertainty is to assess the robustness of the results with respect to the quality of the data used and the assumptions made. In concrete terms, uncertainty is most often calculated using Monte Carlo simulations, which make it possible to estimate confidence intervals and to judge the stability of conclusions. The uncertainty calculation can also be approximated using Pedigree matrices58. However, the uncertainty calculation covers only a portion of the potential sources of error: it only reflects the margins of error within the chosen perimeter, and not the elements that are absent from the perimeter. However, in the carbon footprints of the health sector, the main differences often come from the "omitted" elements, for example: ● the production of medical devices in the clean room, ● air freight in the supply chain, ● purchases of services (sterilization, laundry, IT), ● support activities of the hospital to be allocated to medical procedures. The study by McGain et al. (2012)59 illustrates this limitation: the authors conclude that single- use kits have a lower carbon footprint than reusable kits, a result that appears robust in light of the uncertainty analysis, with an estimated margin of error of around 10%. However, this uncertainty only applies to the scope chosen in the study. It does not take into account the 57 Max Piffoux et al., Carbon footprint of oral medicines using hybrid life cycle assessment, Journal of Cleaner Production, 2024, https://doi.org/10.1016/j.jclepro.2024.143576. 58 Greenhouse Gas Prototocl, Quantitative Inventory Uncertainty, 2022. https://ghgprotocol.org/sites/default/files/2022-12/Quantitative%20Uncertainty%20Guidance.pdf 59 McGain, F., McAlister, S., McGavin, A., & Story, D. (2012). A Life Cycle Assessment of Reusable and Single-Use Central Venous Catheter Insertion Kits. Anesthesia & Analgesia, 114(5), 1073–1080. https://doi.org/10.1213/ane.0b013e31824e9b69 -- 27 of 58 -- 28 fact that the carbon footprint of the specific production processes of central venous catheter kits (which are complex, highly standardised and energy-intensive) have not been included in the calculation of the carbon footprint of single-use, which is therefore probably largely underestimated. Thus, a low level of uncertainty alone does not guarantee the reliability of a result if the scope of analysis is incomplete. This underlines the crucial importance of a transparent and comprehensive definition of the scope. Finally, it is important to consider that a LCA leads to an almost systematic underestimation of the real impact, because value chains are most often simplified in carbon modelling. The uncertainty calculation should therefore be considered as a useful complement, but not as a sufficient guarantee of methodological robustness. 4) Other methods to strengthen the robustness of a computation a) Transparency: an essential criterion for robust results Transparency, both on the methodology followed, the assumptions made and the data used, is an essential criterion of credibility in carbon footprint calculations. An approximate calculation that is transparent about its underlying assumptions is more valuable than a precise but opaque one. Without this transparency, it is impossible to correctly interpret the results or compare them with other work. Indeed, the results depend very largely on methodological choices (scope, emission factors, assumptions). Thus, access to sources, data and intermediate calculations is essential to assess the robustness of the results and to then use them appropriately. This applies particularly to the environmental communication of manufacturers on the carbon footprint of their products or activities. Depoers et al. (2016)60 show, for example, that French companies tend to report lower GHG emissions in their annual reports than in their CDP reports, using less stringent internal standards. These discrepancies underline the importance of methodological frameworks and their adherence to ensuring the reliability of the published figures. b) The sensitivity study to strengthen the conclusion of a study Beyond the uncertainties related to the data, the results of a carbon footprint calculation depend strongly on the methodological choices and assumptions made. In some cases, a limited number of key parameters can have a major influence on the conclusions, for example: 60 Depoers, F., Jeanjean, T. & Jérôme, T. (2016). Voluntary Disclosure of Greenhouse Gas Emissions: Contrasting the Carbon Disclosure Project and Corporate Reports. J Bus Ethics 134, 445–461. https://doi.org/10.1007/s10551-014-2432-0 -- 28 of 58 -- 29 ● when an emission item predominates in the result, and the emissions of this item are related to the use of an uncertain emission factor; ● when average values are used when the real situations are heterogeneous (lifespan, number of reuses, energy mix, urban/rural travel). In these situations, a sensitivity study (the term comes from the fact that we try to find out if the result is "sensitive" to the hypotheses) makes it possible to ensure the robustness of the results. It involves varying one or more key parameters, or even comparing plausible alternative scenarios, to assess the extent to which the conclusions depend on the assumptions made. Unlike the uncertainty calculation, which analyses the variability of the data according to the scope and the assumptions chosen, the sensitivity study makes it possible to question the dependence of conclusions on methodological choices. Several recent studies illustrate the interest of this approach. For example, Marion et al. (2025)61, in their study estimating the carbon footprint of the first 24 hours of trauma patient care, identified that the results were highly dependent on the emission factors associated with the health products used. To take this into account, the authors estimated maximum and minimum values of these factors, thus generating "best-case" and "worst-case" scenarios, which reinforces the robustness of the study's conclusions. Lichtnegger et al. (2023)62 demonstrated that remanufactured intermittent pneumatic compression sleeves have a lower carbon footprint than their single-use equivalent. The sensitivity study confirmed that this conclusion is robust to assumptions influencing the carbon footprint, such as distances travelled for refurbishment, or the electricity mix. Finally, the sensitivity analysis developed in the study by Birnbach et al. (2020)63 highlights the main factors influencing the carbon footprint of condom use, as well as the questions to be explored further for future research. When used to present alternative scenarios, the sensitivity study is particularly relevant to strengthen the findings of a study comparing two care pathways. Indeed, confidence in a conclusion can be strengthened by showing that the conclusions of a comparison of two paths always hold in the worst case (by overvaluing the hypotheses) and the best (by undervaluing the hypotheses). Example: If, when comparing two care pathways in order to highlight one health product over another, I show that the product in question leads to a reduction in greenhouse gas emissions, notably by reducing emissions associated with patient travel. Showing that the 61 Marion, C., Bernat, M., Hammad, E., Calvet, J. P., Roche, M., Marecal, L., Zieleskiewicz, L., & Leone, M. (2025). Assessing the carbon footprint of the initial 24 h post-severe trauma admission in a French ICU: a pilot study. Annals of intensive care, 15(1), 117. https://doi.org/10.1186/s13613-025-01536-x 62 Lichtnegger, S., Meissner, M., Paolini, F., Veloz, A., & Saunders, R. (2023). Comparative Life Cycle Assessment Between Single-Use and Reprocessed IPC Sleeves. Risk management and healthcare policy, 16, 2715–2726. https://doi.org/10.2147/RMHP.S439982 63 Birnbach, M., Lehmann, A., Naranjo, E. et al. (2020). A condom’s footprint - life cycle assessment of a natural rubber condom. Int J Life Cycle Assess 25, 964–979 . https://doi.org/10.1007/s11367-019-01701-y -- 29 of 58 -- 30 conclusion still holds with a patient moving in a carbon-free way can reinforce the value of the result. c) Independent and competent third-party review of published studies Another method inspired by the peer review of scientific studies would be to carry out a critical review of the carbon footprint of the care pathway by an independent third party. However, this method requires a significant increase in the skills of a sufficient number of actors to be able to meet the demand and also to develop a method, which is also transparent, for analysing the quality of studies. We will go into more detail about this in the section "Towards frequent updating of the guide and database". Although this review is not currently fully feasible, it could be an interesting compromise to respond to the desire of certain actors directly or indirectly involved in the care pathways (for example manufacturers) not to share certain data for fear of disseminating confidential information or betraying industrial secrets. C. Finally, a methodology and precision that depend on the objective followed For all the reasons mentioned above, we consider that the level of detail and rigour necessary to assess the emissions of a care pathway depends directly on the objective of the calculation. In other words, while the general methodology described in Part II)A) remains valid, there are some nuances to be made depending on what we are trying to show through our assessment of the carbon footprint of a care pathway. In some cases, an order of magnitude is enough and the search for absolute detail can become counterproductive; in others, a high level of precision is essential. The precision must be equal to the decision-making challenge. 1) Objective 1: Assess the emissions of a care pathway to identify the main emission items a) When should you have this goal? -- 30 of 58 -- 31 These are situations where the objective is to assess the carbon footprint of a care pathway with the aim of understanding which are the most important sources of emissions (both in terms of medical procedures 64and in terms of emission items65) and how they can be reduced. b) What level of precision is needed to draw conclusions? This type of target requires a lower level of stringency regarding the quality of the data used, particularly for certain emission sources. It then becomes possible to use monetary emission factors, average data or fairly strong approximations. On the scope of the study and data collection: It is at the level of these stages that the real challenge lies. It is necessary to define a scope of study that integrates all the stages of the care pathway (patient travel, city, hospital, treatments, etc.) but also all the main emission sources (energy consumption, production of medicines and consumables, travel by health professionals, food, waste, etc.), in accordance with the “What level of detail in the perimeter?" It is also essential to ensure that data collection is comprehensive. Indeed, in theory, we can have defined a complete scope of study and yet forget to take into account half of the consumables, part of the energy consumption (such as the consumption of gas for the production of domestic hot water) or certain types of transport. c) Which test to guarantee the robustness of the conclusion? To validate the result, it is recommended to carry out a sensitivity study around the main assumptions (concerning the main emission sources). d) When to conclude? If the scope has been properly defined (in a sufficiently exhaustive manner) and the assumptions and data underlying the estimation of the care pathway’s carbon footprint are clearly stated, then it is possible to draw conclusions with respect to the objective set. If the goal is not only to understand how to decarbonize the care pathway, but also to communicate the total emissions of this pathway, then an uncertainty and sensitivity study must be carried out and the results must be communicated in the form of a confidence interval. 64 On my background, are the shows mainly related to consultations with the general practitioner or to the emergency room? 65 On my journey, are the emissions mainly related to electricity consumption, consumables or patient transport? -- 31 of 58 -- 32 2) Objective 2: Evaluate the average emissions of a care pathway to assess the carbon impact of a prevention or right care action on the pathway in question a) When should you have this goal? When the objective is to assess the carbon implications of a right care action (limiting drug waste, reducing over-prescribing of imaging), it is necessary to assess the carbon footprint of a medical act or a care pathway. Example: To assess the additional carbon cost related to the overprescription of imaging in the city, it is necessary to know the average carbon footprint of an imagery. When the objective is to know the carbon weight of care in France, or the emissions avoided or induced by a preventive act, it is necessary to assess the average carbon footprint of a care pathway. Example: The carbon footprint related to the management of colorectal cancer, or type II diabetes. Example: To determine the carbon cost-benefit of influenza vaccination, it is first necessary to assess the average emissions associated with the management of influenza. Then we need to assess the average carbon footprint of the influenza vaccination. Finally, the estimated emissions must be multiplied by the volume of people concerned66 b) What level of precision is needed to draw conclusions? This type of objective requires a fairly high requirement regarding the quality of the data used for the activities under study. For example, if the objective is to talk about vaccination, the carbon footprint of the vaccine must be estimated accurately, and therefore not with monetary data, for example. In addition, the objective being to obtain an average data, it is possible to use activity data and average carbon data. Example: The average patient journey of a care or the average carbon footprint of an emergency room admission. 66 By comparing the number of people getting vaccinated to the decrease in the management of influenza through this vaccination (decrease in hospitalizations, complications, medication consumed, etc.) -- 32 of 58 -- 33 On the scope of the study and data collection, it is necessary to define a scope of study that integrates all the stages of the care pathway Medium (patient travel, city, hospital, treatments, etc.) but also all emission sources (energy consumption, production of medicines and consumables, travel by health professionals, food, waste, etc.) in accordance with Part "What level of detail in the perimeter?" It is also essential to ensure that data collection is comprehensive. Indeed, in theory, we can have defined a complete scope of study and yet forget to take into account half of the consumables, part of the energy consumption (such as the consumption of gas for the production of domestic hot water) or certain types of transport. As each care pathway can be unique, it is then possible to base oneself on an average/theoretical case both concerning the medical dimension (i.e. not taking into account, for example, the 1% of individuals who may potentially need "such" a medical act) and the physical flows (i.e. not taking into account, for example, the 1% of individuals who walk to the doctor). In this case, this must be made clear in the presentation of the results. Be careful, however, for some of the pathways studied, the difficulty may initially come from the ability to correctly define an average pathway, especially for care where complications are frequent. c) What tests should be carried out to guarantee the robustness of the conclusion? To validate the result, it is recommended to carry out a sensitivity study around the main assumptions (concerning the main emission items but also concerning the care pathway) and to carry out an uncertainty study in accordance with the parts (II)(B)(3) and (II)(B)4). If the calculations are based on a theoretical case, it is also interesting to model some scenarios observed in reality (for example, the case where there is an additional complication or a movement of patients on foot). d) When to conclude? If the perimeter has been correctly defined (in a sufficiently exhaustive manner), if the assumptions and data on which the estimation of the carbon footprint of the care pathway is based are explained and if the limits have been detailed, then it is possible to conclude as to the objective set. When this is the case, it is important to specify, in the discussion of the study’s limitations, that the calculation is based on a theoretical scenario and provides an order-of-magnitude estimate. However, due to factors such as imperfect adherence or difficulties in accessing care in a timely manner, real-world outcomes may differ. -- 33 of 58 -- 34 Finally, to conclude at a national level, it is necessary to rescale the calculation carried out on an average act/journey by multiplying it by the volume of activity. Example: To know the decarbonization potential linked to the overprescription of imaging, it is necessary to evaluate the average carbon footprint of an imagery. And on the other hand, multiply this result by the total quantity of useless imaging from which we subtract the total quantity of imaging not performed when it should be done (to also take into account sub-processing). This scaling up is particularly necessary if the objective is to assess the carbon implications of a prevention policy. Example: To study whether the carbon cost/benefit of vaccination is positive or negative (in other words, whether or not the carbon cost of producing a vaccine is offset by, for example, the decrease in the prevalence of the pathology in question), it is not only necessary to compare the carbon footprint of vaccination with the carbon footprint of the care pathway of an unvaccinated person. The results must also be put into perspective by multiplying them, on the one hand, by the quantity of people to be vaccinated, and on the other, by the number of people who would have had to follow a care pathway following complications, due to the lack of vaccination. 3) Objective 3: Compare the carbon footprint of two possible care pathways for the same pathology a) When should you have this goal? When the objective is to identify the optimal care pathway from a greenhouse gas point of view, it may involve comparing two care pathways with equivalent therapeutic benefits (with non-inferiority). Example: comparing dialysis with kidney transplantation when caring for patients with kidney failure67. In such cases, the aim may be to highlight a particular health product or medical practice as part of an eco-care approach. It can also be a question of comparing two treatments for the same pathology, which may have different therapeutic benefits. 67Switching to the Dry Powder Inhaler: Disease Control with a Lower Carbon Footprint, Janson et al, 2025 -- 34 of 58 -- 35 Example: Compare treatment A with certain side effects and a given relapse rate with treatment B with other side effects and relapse rates (e.g., radiation therapy vs. surgery). In the second case, it is still necessary to clearly explain the health implications associated with the different pathways. Finally, it can meet the following objective: it is not enough to compare the carbon footprint of product A with product B all the time to know that it is the least carbon-intensive alternative. It is often necessary to put the analyzed product in the context of the care pathway. Indeed, product A may be more carbon-intensive than product B, but if it reduces the number of hospitalizations, the risk of relapse, or the frequency of consumption, the carbon footprint on the entire care pathway may be lower. b) What level of precision is needed to draw conclusions? This type of objective first requires the correct definition of comparators (the elements that distinguish the care pathways that are the subject of the comparison). They correspond to credible alternatives that can be routine care practices, innovative competing treatments, an emerging practice, or simply "do nothing". In this case, a high level of rigor is required regarding the quality of the data used, particularly for the activities under study (the comparators). Example: If the objective is to demonstrate the advantage of one drug over another, conclusions can only be drawn if the product’s carbon footprint has been assessed using a robust and transparent methodology. In addition, the objective being to obtain an average data, it is possible to use activity data and average carbon data, sufficiently representative of reality. Examples: data on the average patient journey of a care, data on the average carbon footprint of an emergency admission. On the scope of the study, as always, it is necessary to define a study perimeter that integrates all the stages of the care pathway in accordance with Part "What level of detail in the perimeter?". (patient travel, city, hospital, treatments, etc.) and that is transparent about "where" and "when" the perimeter starts and ends. However, as the objective is to compare two care pathways with each other, we are only interested in the difference. It is therefore possible, to simplify the calculation, to exclude all the parts common to the two care pathways, by clearly explaining why they are indeed common. -- 35 of 58 -- 36 Example: if we compare two care pathways that both begin with a consultation with the general practitioner and the prescription of identical medications, it is possible not to try to assess the associated GHG emissions. In addition, each care pathway can be unique, so it is possible to base ourselves on an average/theoretical case both concerning the medical dimension (therefore not taking into account, for example, the 1% of individuals who may potentially need "such" a medical act) and the physical flows (therefore not taking into account, for example, the 1% of individuals who walk to the doctor). In this case, this must be made clear in the presentation of the results. Be careful, however, for some of the pathways studied, the difficulty may initially come from the ability to correctly define an average pathway, especially for care where complications are frequent. Finally, comparing two care pathways requires additional caution, namely taking into account the effectiveness of the treatments and any associated complications. Example: If one seeks to compare the care pathway of an infant who has received nirsevimab to prevent hospitalizations for respiratory syncytial virus (RSV) bronchiolitis with a scenario in which the drug is not administered (“no intervention”). It is not sufficient to compare the administration of the product with a scenario in which the infant is hospitalized. The product’s effectiveness must also be taken into account. Two French studies have demonstrated the effectiveness of nirsevimab68 69. It was estimated at 73% (61% - 84%), which corresponds to one hospitalization avoided for 39 (26 - 54) doses administered. In other words, we must compare the production of 39 doses with the carbon footprint of a hospitalization. c) What tests should be carried out to guarantee the robustness of the conclusion? To validate the result, it is recommended to carry out a sensitivity study around the main assumptions (in particular those concerning the main emission sources) and to carry out an uncertainty study in accordance with the parts (II)(B)(3) and (II)(B)(4). 68 Estimates of effectiveness and impact of nirsevimab on hospitalisations for RSV bronchiolitis in metropolitan France, 2023-2024 : a modelling study, Brault et al 69 Nirsevimab Effectiveness Against Cases of Respiratory Syncytial Virus Bronchiolitis Hospitalised in Paediatric Intensive Care Units in France, September 2023–January 2024, Paireau et al -- 36 of 58 -- 37 More specifically, it is recommended to present several scenarios to ensure that the results still hold with different assumptions concerning the main emission sources (for example, if transport represents the first emission source, we propose to present the results also with carbon-free patient and visitor transport). d) When to conclude? If the scope has been properly defined (in a sufficiently comprehensive manner), and the assumptions and data underpinning the estimation of the care pathway’s carbon footprint are clearly documented in line with the section, "Transparency: an essential criterion for robust results", then conclusions may be drawn (provided that the uncertainty intervals for the two pathways do not overlap and that the difference between them reaches at least X%). Be careful, there is still a limit to the comparison of carbon footprint between two care pathways. The objective of this type of study is above all to understand how and to what extent new treatments or new practices can contribute to reducing greenhouse gas emissions from the health system. A further consideration is whether the result remains meaningful in the context of an ambitious healthcare decarbonization target (around a 90% reduction by 2050). If the result shows a slight reduction (of less than 5% for example) in the greenhouse gas emissions of the care pathway, then this result must be compared with the decarbonization potential of each care pathway, their feasibility or even their cost (an analogy could be made with the idea of clinical relevance). Example: It is estimated that care pathway A is 5% less carbon-intensive than care pathway B. But the emissions of care pathway B are mainly related to transport, while those of care pathway A are mainly related to health products. As the levers for decarbonising transport are more "easily" activated and in the hands of health professionals and patients, than those concerning health products, we can wonder whether the 5% gain is to be favoured by moving from route B to A or whether it would not be better to try to decarbonise route B. Caution: Remember that by comparing two courses with each other, the results may be different from one establishment to another or even from one department to another70. Also, a conclusion is only valid on the scope of the study. And, in the event of extrapolation of the results to a larger scope, it is necessary to: 1) explain it in a transparent manner, 2) justify the possibility of this extrapolation 3) detail the limits of this extrapolation. 70 For example, the reuse of certain MDs may be more carbon-intensive than the single-use equivalent when the activity is not sufficient to offset the carbon cost of setting up the sterilization process. -- 37 of 58 -- 38 4) What if I have multiple goals? If multiple of the above objectives are to be met, the most stringent implementation conditions should be applied. Example: if I want to assess the carbon footprint of two care pathways in order to both compare and decarbonize them, then I must apply the precision of my calculations described in Objective 3 and retain the scope of study of Objective 1 (in other words, I must still evaluate the GHG emissions of the activities common to the two pathways). 5) Conclusion Objectives Definition of the scope Rigour in the data used Robustness test Speed of implementa tion How to decarbonize my business Exhaustive scope + Sensitivity study on the most dimensionally important hypotheses +++ Assessing national emissions associated with a care pathway Exhaustive scope ++ Sensitivity study on the most important assumptions and evaluation of uncertainties ++ Assessing the carbon gain of a fair care action Assessing the carbon cost-benefit of prevention Comparing two care pathways Possibility of disregarding the "elementary bricks" of care common to both perimeters +++ Sensitivity study on the most important assumptions, evaluation of uncertainties and realization of several scenarios + -- 38 of 58 -- 39 III. Carbon data on care acts71 and care pathways So far, we have seen how to carry out an assessment of the carbon footprint of a care pathway and why the quality of these studies was heterogeneous. This part aims to understand how the carbon data of the care acts that make up the pathways are constructed. A. What makes good carbon data? Before going into detail, let's remember that good carbon data is data that best corresponds to the context studied (geographical area, physical and temporal perimeter, etc.). 1) General data As explained above, we consider that care pathways can be broken down into a mesh of "elementary bricks" of care (e.g. consultation, imaging, biology, surgery, etc.), juxtaposed to create a global care pathway. Therefore, in order to assess greenhouse gas emissions, the emissions associated with each "building block" must be summed, as many times as necessary. What defines good carbon data for calculating the emission factor of an “elementary building block”? First, it must be understood that the "elementary brick" corresponds to data built on the basis of other emission factors (Figure 1). Example: To evaluate the carbon footprint of the basic brick "consultation with the general practitioner", it was necessary to use the emission factors of gas, fuel oil and electricity to evaluate the emissions related to energy consumption. It was necessary to use the emission factor of transport by car, public transport or bicycle to assess the emissions of patient travel, etc. The quality of the data therefore depends in part on the quality of the emission factors used to evaluate it. From a methodological point of view, the carbon intensity of an elementary building block (mainly in kgCO2e/act) can be evaluated on the basis of a Carbon Footprint® of an activity. Example: Carbon® footprint of a general practitioner's office, the result of which, when divided by the number of consultations, can give the average carbon footprint of a consultation with the general practitioner. 71 Also known as "building blocks" of care -- 39 of 58 -- 40 It can also be based on a carbon footprint study of a care pathway. Example: the study of the carbon footprint of the care pathway associated with the management of an ankle sprain can give the carbon footprint of the elementary brick "management of the ankle sprain". The latter can then be reused in another care pathway, involving a sprained ankle at some point. What this means is that an "elementary brick" ultimately also corresponds to a study of care pathways covering a defined and more restricted perimeter. In other words, any carbon footprint result of a care pathway can then become "an elementary brick". This also means that the methodology used to estimate the carbon intensity of an “elementary building block” is the same as the one used to assess the carbon footprint of a care pathway (as defined in the Methodology section : how to carry out a carbon footprint calculation of a care pathway?). The same is true for the criteria that define the quality of a study on a pathway that we described in the section What precision to assess the carbon footprint of a care pathway ?. In other words, a good emission factor of a "building block" corresponds once again to a piece of data: ● Evaluated on an exhaustive and transparent physical perimeter (taking into account all emission sources, as defined for example in the carbon® footprint methodology72). Clear system boundaries are all the more important to prevent double counting. For example, when adding together an “emergency admission,” a “surgical procedure,” and a “recovery room stay,” patient transport should not be counted three times. ● Evaluated with quality and adapted emission factors: covering the right scope, the right year, the right geographical location, etc73 72 https://www.bilancarbone-methode.com/ 73 For this, many databases are available such as: ● Empreinte Database : a public database of ADEME ● Ecoinvent : private database that covers all domains, updated regularly, >20000 items. ● Ecovamed: Database on the carbon footprint of medicines in solid oral form. > 12,000 items. . Accessible free of charge for health professionals. ● Agribalyse: is the French database of LCIs in the agricultural and food sector provided by ADEME. This updated version, published in 2023, includes the LCIs of 2,600 agricultural and food products produced and/or consumed in France, combining a production-based approach and a consumption-based approach. ● Circularity Food Package: for openLCA, is supported by Agribalyse v.3.1. It is being expanded by GreenDelta GmbH as part of the EIT-funded TRIPLELINK research project Raw Material "Connecting Matters" to be able to model the circular economy. ● GaBi: private database that concerns all domains, supposedly updated annually, >15000 items. ● Carbon Minds: database targeting only chemical molecules, more than 1000 for 190 geographical areas. -- 40 of 58 -- 41 ● Assessed using a transparent methodology tailored to its intended use. As with care pathways, the carbon footprint of a "building block" must be assessed with a precision adapted to the objective of the study. The more precise the study on the care pathway requires, the more robust the methodology for evaluating the elementary building block must be (without monetary emission factors, for example). 2) What level of precision should be used to account for the carbon footprint of healthcare products? a) Monetary emission factors: limits and use cases Monetary emission factors are now the most commonly used approach to estimate the carbon footprint of health products. Their ease of use makes it easier to carry out carbon footprint studies, and allows, for example, global comparisons between institutions. This approach is associated with high levels of uncertainty, which limit its use to certain specific cases. ● Limits on the use of monetary emission factors Monetary emission factors have several structural limitations, which we documented in particular in a technical note on drug emission factors published in 2023. First, the monetary emission factors are constructed within a given time frame. Their calculation is based on economic and environmental data from specific years and is calculated for specific consumption patterns. However, the prices, production processes and volumes consumed of health products change rapidly, which can lead to significant biases when these factors are used. Secondly, geographical and monetary variability limits their use. The factors are constructed for a given country and expressed in a specific currency, making the results sensitive to exchange rates and differences in purchasing power between territories. Finally, the lack of specificity is a major limitation for monetary emission factors: they are constructed from national average consumption and therefore cannot be used for specific health products. ● The LCA Commons: database of the United States, developed by the various American government agencies. It brings together >10000 elements. ● Environmental Footprint database: follows the guidelines of the European Commission and its Joint Research Centre in its most recent version (3.1). There is an impact method of the same name. ● IDEMAT: (abbreviation for Industrial Design & Engineering MATerials database) is a database that is of university origin. It is designed to meet the needs of designers, engineers and architects in the manufacturing and building industry. -- 41 of 58 -- 42 All of these limitations result in high uncertainties in the results obtained, which restricts the use of monetary emission factors to orders of magnitude analyses rather than precise assessments. ● Use cases of monetary emission factors Given these limitations, the use of monetary emission factors should be reserved for large and heterogeneous consumptions, sufficiently representative of the French health system, (for example, for all the medical devices of a hospital or a pharmacy group), when more detailed approaches are not available. The results obtained from these factors should be interpreted with caution and limited to order-of-magnitude analyses, for example to identify the main emission sources within a healthcare facility, and not to accurately estimate the carbon footprint of specific products, or to compare two products or care pathways with each other. In addition, it is necessary to ensure the year of validity of the monetary emission factor used (an EF for the year 2017 will have to be adapted to be used for a previous year, in accordance with the technical note on drug emission factors). ● Towards more specific monetary emission factors Developing more specific monetary emission factors could help reduce the uncertainty associated with their use. Data from the U.S. Environmental Protection Agency74 suggest that biologics have a monetary emission factor about half that of chemical drugs. This trend seems to be confirmed by the work of Piffoux et al. (2025), 75who show that medicines with a high unit value ("expensive medicines") are associated with lower monetary emission factors than conventional medicines. 74 U.S. Environmental Protection Agency, USEEIO v2.0.1-411, 2022. https://catalog.data.gov/dataset/useeio-v2- 0-1-411 75 Max Piffoux et al., Carbon footprint of oral medicines using hybrid life cycle assessment, Journal of Cleaner Production, 2024, https://doi.org/10.1016/j.jclepro.2024.143576. -- 42 of 58 -- 43 Figure 2: Emission factor as a function of the price of boxes of medicines, for solid oral forms consumed in France Source: Piffoux et al. (2025) Call for participation: construction of specific monetary emission factors One possible avenue for improvement would be to develop monetary emission factors by broad categories of healthcare products. This intermediate approach reduces some of the uncertainties while maintaining operational applicability. For example, it would be relevant to distinguish: ● chemical and biologic drugs (work in progress); ● Medicines according to their pharmaceutical form (tablets, liquids, powders, etc.) ● major categories of medical devices, such as: ○ personal protective equipment; ○ simple plastic consumables; ○ medical devices with a high unit value (implants, catheters, complex devices, etc.). ○ etc. b) The LCA approach: what are the criteria for a rigorous LCA? The Life Cycle Assessment (LCA) method is the preferred method when you want to be precise in assessing the carbon footprint of a care pathway. It must be rigorous and take into account all stages of the value chain. More specifically, a rigorous LCA is one that meets the criteria of ISO 14040 and ISO 14044 and for which the description of the scope of the study has been correctly carried out (leave nothing important outside the scope of study). However, in -- 43 of 58 -- 44 practice, the scope taken into account is often incomplete, which leads to a systematic underestimation of emissions. For example, the production of plastic medical devices is often equated with the production of raw plastic, without integrating the raw material processing steps or logistics. The published figures are then much lower than reality: MacNeill et al. (2017)76 and Rizan et al. (2023),77 for example, use values in their studies that are respectively half the values found by The Shift Project, taking into account raw materials, production, sterilization, transport and end-of-life78. Figure 3: Emission factor for plastic medical devices used in different studies Source : Rizan (2023), MacNeill (2017), The Shift Project (2025) Even when studies consider a greater number of stages in the value chain, the perimeters within the positions themselves often remain incomplete. For example, plastic injection for medical devices emits about twice as much greenhouse gas emissions as for ordinary products, in particular because of the specific constraints of clean rooms. 76 MacNeill AJ, Lillywhite R, Brown CJ. The impact of surgery on global climate: a carbon footprinting study of operating theatres in three health systems. Lancet Planet Health, 2017. https://www.thelancet.com/journals/lanplh/article/PIIS2542-5196(17)30162-6/fulltext 77 Rizan, C., Lillywhite, R., Reed, M., & Bhutta, M. F. (2023). The carbon footprint of products used in five common surgical operations: identifying contributing products and processes. Journal of the Royal Society of Medicine, 116(6), 199–213. https://doi.org/10.1177/01410768231166135 78 Data calculated for plastic consumables, for the report "Decarbonizing the Medical Device Industries" (The Shift Project, 2025). -- 44 of 58 -- 45 Figure 4: Emission factors for plastic injection, medical devices and regular products Source : The Shift Project (2025) Thus, the mere inclusion of an emission source in a life cycle assessment (LCA) does not necessarily mean it is accurately accounted for. A rigorous LCA requires a detailed and comprehensive analysis of each stage of the life cycle, especially for health products. c) Special cases requiring strong vigilance As specified in Part II)(C), particular vigilance is required when LCA is used to compare health products (drugs or medical devices) or care pathways, and more particularly when it comes to highlighting one product or option over another. In these situations, methodological choices can have a decisive impact on the results and lead to biased conclusions. In 2025, we showed that several studies comparing single-use instruments to their reusable equivalents significantly underestimated the carbon footprint of single-use79. In particular, in a comparison between single-use and reusable fiberscopes, we showed that the assumptions made by Davis et al. (2023)80 and Sørensen et al. (2018)81 lead to an underestimation of a factor of 1.5 to 3.6 in the carbon footprint associated with single-use fiberscopes, distorting the conclusions of both studies in favor of single-use. Such biases in the 79 The Shift Project. Let's decarbonize the medical device sector. 2025. https://theshiftproject.org/app/uploads/2025/11/TSP-Decarbonons-les-industries-des-DM-RF.pdf 80 Davis, N. F., McGrath, S., Quinlan, M., Jack, G., Lawrentschuk, N., & Bolton, D. M. (2018). Carbon Footprint in Flexible Ureteroscopy: A Comparative Study on the Environmental Impact of Reusable and Single-Use Ureteroscopes. Journal of endourology, 32(3), 214–217. https://doi.org/10.1089/end.2018.0001 81 Birgitte Lilholt Sørensen, Henrik Grüttner. Comparative Study on Environmental Impacts of Reusable and Single-Use Bronchoscopes. American Journal of Environmental Protection. Vol. 7, No. 4, 2018, pp. 55-62. https://www.sciencepublishinggroup.com/article/10.11648/j.ajep.20180704.11 -- 45 of 58 -- 46 results are all the more problematic as they have subsequently been used for the promotion of the corresponding health products82. Another example: Bischofberger et al. (2023)83 estimated the carbon footprint associated with anastomotic leakage by attributing a significant share of emissions to the use of ostomy bags. However, this estimate is based on the assumption of a unit weight of 250 g per bag, whereas the weight of an ostomy bag can be much lower, in the order of 14 g for the models commonly used. By retaining values that are more representative of existing devices for this parameter alone, we obtain results that are about 44% lower than those reported in the study. Such skewed values are problematic when they are mobilized to support or promote alternative solutions, as was the case in this study84. A conservative approach seems essential: it is better to under-conclude than to over-affirm, especially when the results can influence both medical practices and industrial choices. B. Development of a carbon database to support this guide As you can see, access to quality data is crucial. Also, to facilitate this access, we will produce a freely accessible database with all the carbon intensities of "elementary bricks" that we have been able to obtain. 1) Database construction principle To build this carbon database of care "building blocks", we will apply the methodological principles listed in this guide as well as in the section Towards frequent updating of the guide and the database to: ● Identify in the literature the data that we consider to be of sufficient and appropriate quality, ● Build, when technically feasible, our own data when it is not already available. We assume that for each activity there is not just one possible carbon footprint value, but a spectrum of values that depend on the situation being studied. This is why, for as many "building blocks" as possible, we will try to offer a different set of data in order to cover as many use cases as possible. The database would then take the following form: 82 Ambu, https://www.ambu.fr/endoscopie/endoscope-a-usage-unique/environnement 83 Bischofberger, S., Adshead, F., Moore, K., Kocaman, M., Casali, G., Tong, C., Roy, S., Collins, M., & Brunner, W. (2023). Assessing the environmental impact of an anastomotic leak care pathway. Surgery open science, 14, 81–86. https://doi.org/10.1016/j.sopen.2023.07.001 84 Johnson & Johnson. A J&J MedTech Research Reveals Environmental Benefits of Reducing Anastomotic Leaks. https://www.jnjmedtech.com/en-EMEA/news-events/jj-medtech-research-reveals-environmental- benefits-reducing-anastomotic-leaks -- 46 of 58 -- 47 Finally, an essential principle will guide the construction of the database: that of avoiding double counting. To do this, we will try to: 1. to explain precisely the perimeter on which each data is concentrated, 2. to give priority to the production of data on a "small perimeter". Example: Instead of producing a "Surgical operation with home hospitalization" data and a "A surgical operation without home hospitalization" data, we will produce a "Surgical operation" data and a "Home hospitalization" data that can be summed together if necessary. 2) The case of care performed in the city The care provided in the city can be structured into several basic bricks: ● Consultations or home visits carried out by general practitioners and specialist doctors; ● Consultations or home visits carried out by paramedical professionals (physiotherapists, psychologists, dental care, etc.); ● Nursing; ● Biology acts; ● Imaging procedures; ● Home services (insulin therapy, infusions, oxygen therapy, etc.) ● Sales in pharmacies. We will work on the development of carbon data associated with each building block listed above. Pending the complete consolidation of this work, several technical appendices accompany (or will accompany) this guide. They will present: ● the data currently available; ● the methodological assumptions used for the current estimates; ● the main missing data identified at this stage. -- 47 of 58 -- 48 They are intended to feed exchanges, refine methodological choices and enrich the collection of data. If you would like to participate in this work, please do not hesitate to contact us at [email protected]. 3) The case of care performed in multidisciplinary establishments Assessing the carbon intensity of “elementary care building blocks” delivered in multidisciplinary facilities (acute care hospitals, psychiatric facilities, rehabilitation centers, primary care centers, etc.) requires greater precision and vigilance than in community-based care.nIndeed, "simply" carrying out the Carbon® Footprint of the structure and then dividing the result obtained by a quantity of activity carried out would either give an imprecise result or an erroneous result. Example : Several hospital Carbon Footprint® publications give an estimate of the carbon footprint of a day of hospitalization in their establishment. This is the case, for example, of the AP-HP, which estimates the average emissions of a day of hospitalization at 312 kgCO2e85. While this type of result provides an order of magnitude of a facility’s emissions per average activity, it should not be diverted from its primary purpose: tracking changes in the Bilan Carbone® over time while accounting for activity levels. In other words, it cannot really be reused by other actors in the context of the evaluation of the carbon footprint of a care pathway, because in this case, the scope of the data is too exhaustive. Indeed, it is possible to go to the hospital to have an operation, for a simple consultation, to carry out biological procedures or to be hospitalized for several days in 85https://www.aphp.fr/espace-medias/liste-ressources-presse/lap-hp-publie-son-bilan-carbone-pour-2022-et- sengage-dans-une -- 48 of 58 -- 49 intensive care. To faithfully reflect the diversity of care, it is therefore essential to go down to a more detailed level of analysis, act by act, course by path. Moreover, this result gives the erroneous impression that one less day of hospitalization will lead to a decrease in emissions of this value, whereas a proportion of these emissions are "attributional". In other words, they are not directly linked to the flow of activity: for example, it is a fraction of the hospital's heating, the impact of its construction or the movement of staff, which exists whether the bed is occupied or not. For the data produced to be useful, a methodology similar to that followed in the "Sustainable Care Pathways" guide86 or the one implemented in AP-HP's "Carebone®" tool should be applied87. Also, in the case of hospitals, our approach will consist of breaking down the facility into a set of "elementary" activities, with no overlap between them, corresponding to the main clinical services and specialties (as represented in blue in Figure 5). Each activity (surgery, emergency, imaging, biology, consultation, resuscitation or dialysis) constitutes a separate unit of analysis, for which we will seek to estimate an average carbon footprint per act or per stay. Without intersection? However, everything comes together in the hospital, doesn't it? It is indeed complicated to separate the different activities of the hospital. A patient can be admitted to the emergency room, then to have an imaging, a biological act, then to be operated on and finally monitored for several days in hospitalization in a department. From a methodological standpoint, our goal is to generate carbon data in such a way that adding together the carbon intensities of the “emergency admission,” “imaging,” “laboratory testing,” “surgical procedure,” and “postoperative hospitalization” building blocks does not result in any double counting. It is therefore necessary to define the perimeter of these elementary bricks in such a way that no activity is taken into account twice. For example: exclude from the elementary brick "admission to the emergency room" all activities related to imaging, even if imaging can be carried out from the emergency room. However, this average is not sufficient to capture the internal variability of practices. Greenhouse gas emissions can vary from surgery to surgery, from intensive care hospitalization to intensive care hospitalization or from hospital to hospital. 86 Sustainable Care Pathways Guidance, https://shcoalition.org/sustainable-care-pathways-guidance/ 87 The Carebone® tool makes it possible to estimate the carbon footprint of health products, care procedures and patient journeys. https://www.aphp.fr/careboner-un-outil-pour-decarboner-le-soin-mis-la-disposition-de- tous-les-professionnels-de-sante -- 49 of 58 -- 50 Also, for each "elementary brick" of care, we will identify a set of criteria to differentiate sub-activities within the same category. Example: In surgery, for example, emissions may depend on the duration of the procedure, the number of full-time equivalents mobilized, the use of surgical robots, the use of cardiopulmonary circulation devices, or the type of medical gases used. Proposing emission factors according to these parameters will make it possible to distinguish surgeries from each other, and to avoid smoothing out very contrasting realities behind a single average. Our idea is to build data such as "average carbon footprint of a surgery" as well as combinations of data "Average carbon footprint of a surgery requiring a surgical robot and whose anesthesia is based on isoflurane (kgCO2e/hour)". In the next phase of the work, we will aim to identify, for each blue box in Figure 5, the key parameters that drive variations in the activity’s greenhouse gas emissions. In addition, for these medical acts taking place in multidisciplinary establishments, there are activities, services and spaces that are common to all these acts. Also, the emissions associated with support activities (administrative functions, internal logistics, pharmacy, maintenance, purchasing, waste management, digital, common areas, etc., represented in orange figure 5) will be evaluated separately. A transparent and justified distribution key will then be defined in order to attribute these emissions to the basic acts mentioned above. Figure 5 - Breakdown of a hospital's activities and services into a set of "elementary bricks". The blue boxes represent the care activities for which we will seek to assess the associated greenhouse gas emissions (avoiding double counting). The orange boxes correspond to activities that support care activities. -- 50 of 58 -- 51 Finally, emissions related to both internal and external transport (patients, visitors, professionals) will be addressed within a separate boundary, in order to prevent double counting and to maintain a clear view of the impacts directly attributable to care activities. Once again, pending the complete consolidation of this work, several technical appendices accompany (or will accompany) this guide. They will present: ● the data currently available; ● the methodological assumptions used for the current estimates; ● the main missing data identified at this stage. They are intended to feed exchanges, refine methodological choices and enrich the collection of data. If you would like to participate in this work, please do not hesitate to contact us at [email protected]. 4) When and how to use the data in this database? As you can see, our objective is to consolidate a first version of an open and accessible database, bringing together a maximum of emission factors for medical acts ("elementary bricks" of care). This will be a first version that will hopefully be updated frequently in a collaborative manner. Each piece of data will be produced in a transparent manner, according to an explicit methodology or from the scientific literature when possible and selected on the basis of clear criteria. In addition, for the same act, we will offer several levels of information, in order to adapt to the different contexts of use: an average value, specific values according to the place of practice, or refined data according to certain characteristics of the act of care. As explained above, the choice of data to be used will always depend on the objective of the study: depending on the level of precision sought, it will be necessary to use average data, detailed data, extrapolated estimation or data associated with a higher uncertainty. -- 51 of 58 -- 52 Also, on the one hand, we have carbon data of varying quality, more or less available, and on the other, a rigor of analysis and calculation that depends on the objective followed. We then distinguish several cases: Possibility 1: The data I was looking for and adapted to my objective is already present in the database. In this case, no additional work is necessary. Possibility 2: The data I am looking for is present in the database but it is not sufficiently precise and adapted to my context. In this case, it is possible to retrieve the data available in the guide by applying the necessary modifications so that it is adapted to my objective. Example: The guide offers me data for the carbon footprint of a physiotherapy consultation88. However, I would like to be more precise and know the emissions of my activity in particular (45m2 gas-heated practice, in a peri-urban environment). I can then take the data from the guide, remove the emissions related to transport and energy consumption, and then add them with the data from my practice. This still allows you to start from an existing base and not have to start from scratch. It is also possible to use a less precise data, but in this case it is necessary to ensure that the objective of the study allows it (in accordance with the section Finally, a methodology and precision that depend on the objective followed). Example: The guide offers me a data for the average carbon footprint of a surgical operation. Since my objective is to understand how to decarbonize my patients’ care pathway, which at some point includes spinal surgery, I can rely on this average data without needing to adjust it. However, if the objective is to compare two care pathways, one involving spinal surgery and the other not (with the surgery being the key differentiating factor), then a more granular, operation-specific dataset is required. Possibility 3: The data you are looking for is not available in the database. In this case, it is necessary to carry out a specific study to assess this missing data, following the methodology defined in the Methodology section : how to calculate the carbon footprint of the care pathway?. Again, the level of precision to assess this data depends on the objective of my study. 88 Note that for physiotherapy practices, a module exists to calculate the carbon footprint of a physiotherapy practice and understand how to reduce it: https://www.myco2.com/kineCO2 Aggregate data from this module will be published in 2026. -- 52 of 58 -- 53 C. Towards frequent updating of the guide and database For the reasons outlined in the section An evolving guide, which needs to be completed we aim to develop both a guide and a database as tools that can adapt over time. Certain methodological components may be updated, supplemented, or entirely revised. This intention to evolve these tools in line with advances on the subject raises two major questions: ● How can we ensure that the quality, accuracy and meaning of the new data available make it usable in an exercise to assess the carbon footprint of a care pathway? Indeed, as we have seen, many studies published today contain methodological flaws or exhibit characteristics that may render them insufficiently robust. ● And given that the required methodology and level of rigor depend on the study’s objective, how can one determine when, how, and where a given dataset should be used? A dataset might be considered unsuitable for comparing two care pathways, yet still be relevant for assessing the carbon footprint of a healthcare service with a view to reducing it. Therefore, we propose a first version of an analysis grid aimed at evaluating the quality of a study (scientific or not) producing carbon data of a medical course or procedure. This framework (Figure 6) draws on methods used to compare scientific articles in meta-analysis89 90 and should be complemented by an approach to “score” the quality of studies based on a set of criteria. These criteria will need to be defined and adapted according to the topics covered91. 89 The carbon footprint of surgical operations: a systematic review update, Robinson et al, 2023 90 The carbon footprint of critical care: a systematic review, Gaetani et al, 2024 91 An article on the carbon footprint of surgery that does not take into account the consumption of anesthetic gases will not have the same quality as an article on the carbon footprint of a consultation with a physiotherapist that does not take into account the consumption of anesthetic gases. -- 53 of 58 -- 54 Figure 6 - analysis grid aimed at evaluating the quality of a study (scientific or not) producing carbon data on a medical pathway or procedure. This analysis grid first makes it possible to compare the articles with each other. It is currently only an incomplete proposal that requires feedback. Ultimately, the aim would be for it to enable a sufficiently transparent and objective assessment of data quality, in order to answer the question: Can it be incorporated into the methodological guide? And if so, under what conditions? Indeed, the audit of data is not limited to a binary decision consisting of integrating it as it is or excluding it on the grounds of insufficient quality. Several intermediate situations can arise: ● The data is indeed of insufficient quality to be integrated. Example: to assess the carbon footprint of a surgical operation, the data was constructed by taking the Carbon® Footprint of a hospital, and attributing part of these greenhouse gas emissions to surgery in proportion to the surface area dedicated to this service. ● The data is of insufficient quality to compare two care pathways with each other, but sufficient to understand how to decarbonize my activity. Example: a study based solely on monetary data to assess the carbon footprint of a surgery. -- 54 of 58 -- 55 ● Only part of the evaluation method is considered to be of insufficient quality. In this case, the data can still be retained but on a more limited scope. Example: to assess the carbon footprint of a surgery, a study applies an LCA method to all emission sources, except for drug consumption, for which monetary data is used. In addition, for consumables, the emission factor used does not take into account the production of MD in the clean room or transport. In this case, the data could be presented as a "factor in the emission of surgery excluding consumables and medicines". The missing data can then be added by hand. ● The data concerns a geographical scope and must therefore be adapted before being integrated. Example: A study evaluating the carbon footprint of surgery in the United States has just been published. It is possible to start from the basic data of the study and to use French emission factors to re-evaluate the data. This requires that the primary data and the hypotheses that made it possible to obtain this data are available (which is rarely the case). -- 55 of 58 -- 56 Project Team Mathis Egnell - Health Project Coordinator at The Shift Project, pilot of the report Mathis Egnell is the Health Project Coordinator at The Shift Project. He has led the work on Autonomy and on the pharmaceutical industries and is co-author of the work on health and on the medical device industries. He also co-pilots the major health consultation. A graduate engineer from Mines Paris and trained in economics at AgroParisTech, he also worked as a biomechanical engineer at Pasteur Hospital in Nice and as a consultant for the WHO with P4H, the global network dedicated to social health protection and health financing systems. Baptiste Verneuil - Health Project Engineer at the Shift Project, pilot of the report Baptiste Verneuil is a Health Project Engineer at Shift Project. He led the work on the medical device industries and is a co-author of studies on health, autonomy and pharmaceutical industries. A graduate engineer from École Polytechnique with a master’s degree in environmental engineering from the Technical University of Munich, he also conducted research on climate models at the Leipzig meteorology laboratory and contributed, as an engineer, to the activities of the RESPECT Chair (Resilience in Health, Prevention, Environment, Climate and Transition) at EHESP. Mona Poulain - Communications Officer After earning a Master's degree ("Magistère, management, cultures and communication strategies") from CELSA Sorbonne University, Mona Poulain joined the Shift team as a communication and events manager. She works in particular on the health program and supports the organization of events. David Grimaldi - Scientific advisor to the Shift Project and intensive care physician David Grimaldi is an intensive care physician trained at Pierre and Marie Curie University and holds a PhD in immunology from Université Paris Cité. A member of the Shifters' Health Thematic Circle and The Shifters Belgium, he has been contributing to the Shift's work on health since 2021. He was a member of the Epidemiology and Clinical Research Commission of the SRLF (Société de Réanimation de Langue Française), of which he is currently a member of the REAGIR group (which aims to promote sustainability within French-language intensive care units). For several years he was a professor of intensive care at the Erasmus Hospital of the Université Libre de Bruxelles, and now works at the Medical Directorate of the Belgian Health Insurance. Scientific advisor to the Health, Climate and Resilience Programme of the Shift Project, he has been co-piloting the Health Industries report from mid-2024. Laurie Marrauld - Initiator of the Health-Climate-Resilience Research Program at the Shift Project, Lecturer in Public Health at the EHESP and holder of the RESPECT Chair Laurie Marrauld initiated and led the work on the health system, climate and energy at the Shift in 2019, which is now grouped together in a dedicated research program. She holds a PhD in management sciences from Télécom ParisTech and joined the LGI at École Centrale de Paris and the CRG at École Polytechnique before becoming a lecturer at the École des Hautes Études en Santé Publique (EHESP) where she focuses her research on the consequences of the introduction of ICT in health in a context of epidemiological transition. demographic and -- 56 of 58 -- 57 socio-technical issues, as well as the resilience and decarbonization of the health system. Holder of the RESPECT Chair – Resilience in Health, Prevention, Environment, Climate and Transition, she is also a sustainable health expert at ANAP and scientific advisor to the HCAAM. Thomas Rambaud - Health Technical Advisor at the Shift Project Thomas Rambaud assists Laurie Marrauld in the Shift on health. He co-piloted the major health consultation and co-piloted the work on the health industries as a technical advisor. He contributed to the work of the Shift on higher education. A graduate of Polytech Nantes and holder of an MBA from the International Institute of Management (CNAM), he has a 25-year career in large service companies in the health sector, first in IT, then as a manager in operational excellence and finally as Director of programs on compliance and transparency of the links of interest between health professionals and the pharmaceutical industry. Team Supports Héloïse Lesimple - Project Manager, Public Affairs and Health Héloïse Lesimple has joined the Shift team as Public Affairs Project Manager and is following the work of the French Economy Transformation Plan for the cultural and health sector. A graduate of EDHEC, she followed a ten-year career as a health consultant, then as a production manager in culture. She recently obtained a Master's degree in Environment from AgroParisTech. Jean-Noël Geist - Coordinator of the Health-Climate-Resilience Research Program A graduate of the IEPs of Strasbourg and Toulouse and the University of Thessaloniki, this science fiction reader and inveterate cyclist joins the Shift to combine two passions: scientific progress and public policy. He coordinates the public affairs of the Shift, relations with the volunteer association The Shifters and from the PTEF several sectoral works (public administration, defence, culture, health, sport). -- 57 of 58 -- 58 -- 58 of 58 --