Measuring the environmental footprint of the Dutch healthcare sector

Since 2015, under the coordination of the Ministry of Health, Welfare and Sport (VWS), representatives of Dutch care providers have collaborated on the Green Deal ‘’Samen werken aan Duurzame Zorg’’ (GDDZ). In November 2022 the third phase (GDDZ 3.0) was launched, setting sector-wide targets for the period up to 2050, among which to achieve a 55 percent reduction in CO₂‑equivalent emissions by 2030 (relative to 2018) and a milestone of 30 percent reduction by the end of 2026.

At the 2021 COP26 in Glasgow, the Netherlands joined over 50 countries in pledging to develop and scale up ‘’resilient and sustainable health systems’’ under the COP26 Health Programme. Key Dutch commitments included:

• Regular monitoring of the health sector’s environmental footprint
• Support for low-carbon, sustainable supply chains in health care
• Facilitation of global collaboration on climate-smart procurement and manufacturing of health-care goods

This international pledge underpins national efforts like the Green Deal, and underscores the importances of a robust, repeatable methodology for measuring, monitoring – and ultimately reducing – the carbon intensity of health care in the Netherlands.

Institutional arrangements

The purpose of the emission assessment is to clarify the sector’s contribution to the overall environmental footprint and, by extension, to show that it can significantly contribute to achieving national climate targets. To this end, it is essential to identify environmental impact hotspots, prioritize mitigation measures, and monitor progress. The assessment also helps to communicate the urgency of mitigation efforts and provides a foundation for existing and future sustainability initiatives. The intention is to periodically review the methodology and perform up to date assessments.

The assessment of the Dutch healthcare sector’s GHG emissions is therefore organized as a national effort that builds on an institutional framework. A key aspect of this framework is the collaboration between the national statistical agency, environmental research institutes, the Ministry of Health, Welfare and Sports (VWS), and sectoral stakeholders. For example, Statistics Netherlands (CBS) plays a pivotal role by supplying high-quality, up-to-date financial data - including detailed expenditure records and sector classifications based on internationally recognized standards like the System of Health Accounts (SHA). The financial data is then coupled with an environmentally-extended multi-regional input-output (EE-MRIO) database such as Exiobase, ensuring that monetary flows are converted into environmental impact estimates, including greenhouse gas emissions and natural resources use. In addition, partnerships exist between RIVM and other stakeholders, including VWS, ‘National Healthcare Institute’ and healthcare providers, to ensure that both the data collection and interpretation processes are traceable, transparent and meet technical requirements.

Overall process

The overall process to measure the GHG emissions of the healthcare sector involved several key steps (Steenmeijer et al., 2022):

1. Defining the system and establishing the collaborative framework

• Sector Boundary & Economic Classification: The health system is defined using the System of Health Accounts (SHA) as well as other national classification schemes. CBS provides detailed health expenditure data, which sets the economic boundaries of the healthcare sector. The assessment spans the main SHA categories (e.g., medical care, rehabilitation care, home care), but also the non-SHA categories welfare care, social services, childcare, and youth care. The system boundaries are defined to include emission sources across scope 1 (direct emissions from healthcare facilities such as on-site fuel use and anesthetic gases), purchased electricity and steam and hot water supply (scope 2), and the bulk of the footprint through supply‑chain emissions, pharmaceuticals, medical equipment, waste treatment, water distribution (scope 3), and private travel by patient and visitors (non-GHG protocol).
• Inter-agency Collaboration: RIVM collaborates with CBS and is supported by ministries that set the policy and sustainability targets. This collaboration ensures that the data used in the analysis is both accurate and relevant for policy decisions. In addition, data from CBS is complemented by environmentally-extended input–output databases (e.g., Exiobase) that are managed or aligned with the work of nationally recognized research institutions.

2. Method selection

• Top-Down Analysis: A major part of the sector’s GHG emissions is estimated using an environmentally-extended (multi-regional) input–output analysis (EE-IOA). This method uses CBS (financial) data to quantify healthcare expenditure and links it with environmental impact factors from databases such as Exiobase.
• Bottom-Up Life Cycle Assessments (LCAs): For specific products - for instance, the emissions associated with the use of anesthetic gases or pressurized metered dose inhalers (pMDIs) - an LCA results are used. The results of these LCAs are then integrated with the broader top-down evaluation to produce a footprint.

3. Data collection, and harmonization.

Data are gathered from multiple sources:

• CBS: national IO tables and national emissions inventories
• Exiobase v3: multi-regional IO satellite for cross-border emissions flows
• OECD, WHO, Eurostat SHA: healthcare activity classification and expenditure data via CBS
• GIP databank: publicly available data on outpatient pharmaceutical and medical/therapeutic device volumes and spending

Harmonisation is achieved by aligning classification systems (linking CBS spending with Exiobase sector codes (NACE)) and applying a common impact assessment framework (e.g., using ReCiPe 2016 (Huijbregts, et al., 2017) characterization factors) so that all emissions are reported converted into CO₂‑equivalents. This ensures that top‑down and bottom‑up data are directly comparable and can be aggregated.

4. Analytical workflow

• Model assembly: conducting the EE-IOA by mapping health-sector outputs to IO sectors and environmental stressors (greenhouse gas emissions, raw material extractions, blue water consumptions, land use waste production, all converted into impact categories using ReCiPe 2016 and DESIRE FP7 factors)
• Bottom-up modules: developing methodologies to estimate impacts from additional sources, e.g., from pMDIs, anesthetic gases, and private travel.
• Characterization: converting stressors to mid-point indicators, i.e., CO₂-eq for climate change (ReCiPe 2016).
• Uncertainty analysis: Limitations in resolution and assumptions (e.g., aggregated product categories) are discussed. Sensitivity to data sources and methodological choices is evaluated.
• Hotspot and contribution analysis: Ranking product groups and services, as well as regions (e.g., Asia, Europe excl., Netherlands) by their contribution to GHG emissions. The regional breakdown refers to the province of products and services in the global supply chain of the Dutch healthcare system, not to a comparison of national health systems. It shows where along the supply chains (domestically and abroad) the emissions embodied in Dutch healthcare consumption occur.

5. Reporting

Results and methodologies are published open access on the website of the National Institute for Public Health and the Environment, i.e., RIVM. Examples of assessment results are presented in the figures below, with full detail in the report by Steenmeijer et al., 2022.

In the Dutch experience, the following approaches were identified as important for emissions measurement in the health sector and will be further explained in sections below:

• Data harmonization and gap-filling: Ensure use of harmonized data (scope, timeframe, classification and terminology) and support further harmonization if non-existing. Use complementary data to fill critical gaps (e.g., bespoke LCAs for pharmaceuticals) early on.
• Applying a hybrid approach: Combining top-down EE-IOA with targeted bottom-up LCAs to capture system-wide emissions and, and other environmental stressors, add impacts from stressors not captured in the EE-IOA, such as unreported gases (i.e,. anesthetics), household-level use-phase gases (i.e., pMDIs), and sector-adjacent emissions (i.e., private travel).
• Having clear boundaries and governance: Defining Scope 1–3 boundaries from the start.
• Data governance: Secure data-sharing agreements with suppliers.
• Multi-stakeholder collaboration: Maintaining ongoing engagement (webinars, working groups) across government, providers, insurers and industry to align methods and share data. Furthermore, securing executive endorsement to ensure resourcing, data access and integration of sustainability goals.
• Transparent reporting and uncertainty management: Accompany all estimates with confidence intervals, document assumptions, and map out priority areas for future work.
• Iterative monitoring & improvement: Setting a clear baseline, update regularly, and progressively deepen granularity where the largest uncertainties remain.

• Data gaps & temporal mismatches: Key datasets were misaligned (e.g., waste statistics from 2011 (note: 2022 data will not be updated anymore) vs., 2016 footprint), leading to uncertainty in waste‐related emissions. Furthermore, many LCA studies for pharmaceuticals and chemicals are scarce, not harmonized and proprietary, hampering further disaggregation.
• Aggregation & methodological limits: The EE-IOA aggregates at sector level; conversely, bottom-up inventories (e.g., anesthetic gases) require bespoke studies. This necessitates complex hybrid integrations and introduces methodological uncertainty.
• Governance & transparency: Lack of publicly available data from pharmaceutical and medical-device producers impeded precise impact estimation. Data sharing agreements had to be negotiated and remained largely unavailable, slowing progress.
• Defining system boundaries: The health sector is highly diverse, ranging from direct care services (e.g., hospital operations) to indirect activities (e.g., procurement, energy use, and staff/patient travel). This diversity makes it difficult to define clear, consistent boundaries across scopes (Scope 1, 2, and 3).
• Data collection and harmonization: Integrating detailed economic and environmental data with consistent granularity is difficult. Inconsistent data quality and gaps - especially for certain supply-chain (Scope 3) components - pose significant hurdles. An example is that spending on pharmaceuticals and medical consumables is aggregated under broad categories like ‘’Chemicals not elsewhere classified’’ in Exiobase. This aggregation makes it hard to pinpoint specific supply chain components responsible for emissions since detailed product-level data is lacking and quality varies across sources.

• First step towards hybrid analytical approach: it was helpful to implement a hybrid method that combines a top-down EE-IOA with detailed bottom-up LCAs for specific products and processes that are not captured by the EE-IOA. This combination allowed for broad, system-wide estimates while also capturing the nuances of high-impact components (such as emissions from anesthetic gases and pressurized metered dose inhalers), thereby increasing accuracy and resolution.
• Collaboration: establishing a coordination framework among the Netherlands’ National Institute for Public Health and the Environment (RIVM), Statics Bureau (CBS), Ministry of Health, Welfare and Sports (VWS), and key healthcare stakeholders fostered a shared approach to emissions measurement. Regular consultations and webinars helped clarify methodologies and align expectations. This multi-stakeholder engagement not only increased data accessibility and methodological transparency but also ensured that the assessment met the needs of policymakers. The Green Deal Covenant (ref GDDZ v1-3) and commissioned project to measure the environmental footprint (by the ministry to RIVM), was crucial to help set up and guide the necessary alignments and collaboration to measure the national footprint of the healthcare sector.
• Transparent reporting and uncertainty management: various estimates were accompanied by confidence intervals and explicit notes on data gaps and assumptions. By mapping out confidence intervals around each estimate and highlighting areas of greatest uncertainty, the analysis not only bolstered its credibility but also created a clear, prioritized roadmap for future refinements; whether that means deeper LCAs, improved data-sharing standards or even machine-learning proxies to narrow the remaining blind spots.

The following approaches for measuring emissions in health systems can be drawn from the Dutch experience:

• Design a modular framework: begin with a basic model that uses existing national datasets to estimate GHG emissions broadly across healthcare subsectors. Then, incrementally develop a detailed model through sector-specific collaboration. See also van Bodegraven et al., (2025)
• Further development of the hybrid approach: continue integrating top-down and bottom-up methods to leverage the strength of each.
• Secure institutional buy-in early: engage healthcare providers, procurement networks, and government agencies in the design process. Shared ownership facilitates access to granular financial and material flow data needed for deeper analysis.
• Plan for repeatability and updates: structure the methods to allow regular recalculations (e.g., biennial) using consistent data flows and tools. While this enables long-term tracking of sector-wide trends (hence, not product-level or supplier-specific data), it may be less suitable for assessing the direct impact of specific sustainability measures due to the average-based nature of EEIOA.
• Balance detail with feasibility: where full detail is not yet possible, for instance due to data fragmentation, focus first on the highest-emitting sector or activities (e.g., energy use, pharmaceuticals, transport) and iterate.
• Facilitate international comparability: use internationally recognized environmental accounting systems to allow for benchmarking and learning across borders.
• Invest in data infrastructure: improve data collection systems (e.g., real-time procurement data, digital inventory).
• Explore machine learning: it has been explored (Steenmeijer et al., 2022) whether AI can provide support in this area. Currently, AI does not offer direct added value when it comes to analyzing data on the environmental impact of pharmaceuticals and medicines. However, further exploration in whether AI can help improve the harmonization of datasets is needed. In addition, the availability of high-quality and FAIR data remains essential.

• Huijbregts, M. A., Steinmann, Z. J., Elshout, P. M., Stam, G., Verones, F., Vieira, M., ... & Van Zelm, R. (2017). ReCiPe2016: a harmonised life cycle impact assessment method at midpoint and endpoint level. The International Journal of Life Cycle Assessment, 22(2), 138-147.
• Steenmeijer, M.A., Pieters, L.I., Warmenhoven, N., Huiberts, E.H.W., Stoelinga, M., Zijp, M.C., van Zelm, R., & Waaijers-van der Loop, S.L. (2022). The impact of Dutch Healthcare on the environment. Environmental footprint method, and examples for a health-promoting healthcare environment.
• Steenmeijer, M.A., Rodriques, J.F., Zijp, M.C., & Waaijers-van der Loop, S.L. (2022). The environmental impact of the Dutch health-care sector beyond climate change: an input-output analysis. The Lancet Planetary Health, 6(12), e949-e957.
• Van Bodegraven, M., de Bruycker, T., Coenen, J., & Waaijers-van der Loop, S.L. (2025). Calculating the environmental impact of the Dutch healthcare sector. Method report. https://doi.org/10.21945/RIVM-2025-0096
• Environmental Extension Databases Databases such as Exiobase, ICIO, Gloria, WIOD and Eora provide environmental extension data necessary for a top-down assessment and can be paired with national statistics to generate estimates.
• Open-source python package Pymrio This Python package is specifically designed to work with EE-MRIO databases. It facilitates the downloading, parsing, and integration for MRIO datasets (e.g., Exiobase). The tool allows users to aggregate data by country, sector, or product, thereby supporting both contribution and hotspot analyses.

For more information, please contact Martijn van Bodegraven (RIVM) at martijn.van.bodegraven@rivm.nl, Tinia De Bruycker (RIVM) at tinia.de.bruycker@rivm.nl, Jannie Coenen (RIVM) at jannie.coenen@rivm.nl and Susanne Waaijers – van der Loop (RIVM) at susanne.waaijers@rivm.nl.

This case study is part of the work of the ATACH Task Team for health systems GHG emissions measurement, complementing the WHO guidance for measuring greenhouse gas emissions in health systems.