Development of a methodology for carbon efficiency in a healthcare system Authors: Brandon X Lum, Rebecca Y Lee, Clarence T Peck, Ricky Wiryopranoto, Eugene H Liu Published: 2026 ABSTRACT AbstractIntroductionHealthcare systems worldwide contribute significantly to global carbon emissions whilst facing increasing workloads from ageing populations. This creates a challenge of delivering more care while reducing environmental impact. This paper describes the development of a methodology to measure carbon efficiency in healthcare by relating carbon footprint to workload across different care settings.Case presentationThe National University Health System (NUHS) in Singapore developed a workload unit (WLU) methodology to quantify carbon efficiency. Carbon footprints per unit of care were calculated across six care settings, establishing carbon relativity ratios. A polyclinic appointment was assigned one WLU, whilst an acute hospital inpatient day equated to 14.7 WLU based on relative carbon intensity. Total system workload was summated using WLUs, enabling calculation of carbon efficiency as kg CO2e per WLU.DiscussionThe methodology revealed improving carbon efficiency despite increasing absolute emissions from 2020-2023. Carbon per WLU decreased from 16.85 to 16.17 kg CO2e, indicating effective sustainability measures. Challenges included changing emission factors and resource allocation patterns affecting carbon relativity calculations year-to-year.ConclusionThis methodology enables healthcare systems to monitor environmental sustainability progress while accounting for workload changes, supporting data-driven decarbonization efforts and benchmarking between institutions delivering different care portfolios.Keywords: Carbon efficiency, Healthcare sustainability, Workload unit methodology Introduction Healthcare systems worldwide contribute as much as 5% of the annual global carbon footprint [[1], [2], [3], [4]]. The environmental impact and pollution from healthcare can harm population health, creating a dilemma where more care leads to more harm [5]. At this critical juncture, healthcare systems must prepare for the burden of disease from climate change, while mitigating healthcare’s environmental impact and delivering sustainable care. A key enabler for environmentally sustainable healthcare is the measurement of healthcare’s carbon footprint to enable benchmarking, identification of carbon ‘hotspots’, setting of targets for decarbonization, and monitoring progress [6]. To enable comparability, measurements must account for changes in workload, shifts from inpatient to ambulatory care and from hospital to primary care. Metrics that relate workload to carbon footprint across all care settings are needed to measure carbon efficiency of care and the effectiveness of sustainability efforts. While many healthcare systems measure their carbon emissions, few have robust metrics that relate emissions to operational workload or patient outcomes. Hospitals often use bed numbers as part of a base metric, such as in the UK NHS, or tools such as the Geneva Sustainability Centre’s Sustainability Assessment Tool [3,7]. Whilst carbon per bed assesses carbon efficiency broadly and enable comparisons between hospitals, this does not factor for changes in workload and shifts between ambulatory and inpatient settings. This report describes the development of a methodology to measure the carbon efficiency of healthcare across a system’s care settings. Methods The National University Health System (NUHS) in Singapore consists of three acute hospitals, one community hospital and seven polyclinics, and delivers the full spectrum of primary, secondary and tertiary care to a population of 1.3 million living in western Singapore. NUHS embarked on its sustainability journey in 2021, with a goal to reduce 25% of its total carbon footprint across Scopes 1, 2, and 3 by 2030. To measure and track emissions, NUHS used two main carbon accounting approaches, detailed in a previous study [8]. The activity or process-based method employed utilization data of items (e.g. kilowatt hours of electricity, liters of medical gases, tonnage of waste) and their corresponding emission factors to calculate their emissions in tons of carbon dioxide equivalents (CO2e). The spend-based method used financial data, or expenditure, of items and the corresponding emission factors that were based on economic data, to calculate their footprint in CO2e. These spend-based emission factors, updated annually, were country and sector specific, retrieved from the Eora Environmentally Extended Multi-Region Input Output (EE-MRIO) economic tables (https://www.eora.org/). While spend-based foot-printing has a lower accuracy, it enables estimation of a healthcare system’s footprint and is the only viable method currently for complete measurement of its emissions. Singapore’s public healthcare system accounted for workload in these care settings: acute hospital inpatient days, specialist outpatient appointments, day surgery procedures, emergency department attendances, community hospital inpatient days and polyclinic appointments. The healthcare financial accounting system allocated costs to these care settings, according to usage of manpower and non-manpower resources. We developed a methodology to evaluate NUHS’ carbon efficiency of care, based on 2023 financial year data [8]. Utilisation of non-manpower resources and financial costs were used to calculate the annual total footprint for each care setting. Organizational data were extracted from multiple sources as indicated in Table 1. The inventory followed a data hierarchy prioritising activity-based, then supplier-specific, then spend-based factors respectively. Table 1Carbon accounting category approach.Table 1 dummy alt textScopeCategoryData TypeData SourceEmission Factor Source1Anaesthetic gasesActivity-basedInvoices from Pharmacy and Procurement departmentsCanadian Journal of Anaesthesia1Fuel combustionActivity-basedInvoices from gas retailersNational Environment Agency2Purchased electricityActivity-basedElectricity consumption data from utility billsSingapore Energy Market Authority Grid Emission Factor3Category 1: Purchased goods/servicesSpend-basedAudited spend data from NUHS' internal data management systemsEora3Category 2: Capital goodsSpend-basedAudited spend data from NUHS' internal data management systemsEora3Category 5: WasteActivity-basedWaste report from General Waste and Recycling CollectorIPCC Good Practice Guidance Chapter 53Category 6: Business travelSpend-basedAudited spend data from NUHS' internal data management systemsEora3Categories 8 & 13: Leased assetsSpend-basedAudited spend data from NUHS' internal data management systemsEora Carbon accounting category approach. Scope 1 and 2 emissions were derived using activity-based emission factors, such as Singapore’s published grid emission factor. Scope 3 emissions were mostly computed using spend-based emission factors, while Scope 3 emissions for waste management were calculated using Intergovernmental Panel on Climate Chanve (IPCC) emission factors. To prevent double counting, spend-based factors systematically excluded items measured through activity-based methods, and operational emissions from capital goods were allocated solely to Scopes 1 and 2. NUHS is continuously working towards moving more items towards activity-based data where available. Manpower costs were not included in carbon foot-printing, in line with the carbon accounting methodology that NUHS had developed. The workload in each care setting was obtained from annual operational reports. The use of energy, goods and services were allocated to the care settings according to the financial accounting system of NUHS. The emissions of one unit of care in the different care settings in NUHS was measured and used to quantify the relative carbon intensities of each care setting. Emissions were lowest for a polyclinic appointment and highest for a day surgery procedure, approximately 17 times that of one polyclinic appointment (Table 2). (See Supplementary Materials Annex A for additional data details.) Table 2Carbon and Workload units across care settings based on 2023 data.Table 2 dummy alt textCare settingLocationkgCO2e per unit of care in care settingCarbon RelativityWorkload Units (WLU) accorded to unit of careAcute Hospital Inpatient DayAcute Hospital20114.714.7Community Hospital Inpatient DayCommunity Hospital503.73.7Day Surgery ProcedureAcute Hospital24417.817.8Emergency Department AttendanceAcute Hospital413.03.0Polyclinic AppointmentPrimary Care13.71.01.0Specialist Outpatient Clinic AppointmentAcute Hospital624.54.5 Carbon and Workload units across care settings based on 2023 data. The concept and methodology of workload units (WLU) were developed to enable summation of all work across all care settings in NUHS, accounting for their relative carbon intensities (Fig. 1). WLUs function as an emissions-weighted workload index that standardises healthcare activity across different care settings based on their carbon footprints. An acute hospital’s workload comprised WLUs in the form of inpatient days, emergency department attendances, day surgery episodes, and specialist outpatient clinic appointments. WLUs were totalled for all hospitals and polyclinics to make up NUHS’s total workload. Methodology flow chart for carbon relativity calculations. As one polyclinic appointment had the lowest emissions, it was accorded one WLU. One acute hospital inpatient day’s emissions were 14.7 times that of a polyclinic appointment, and was accorded 14.7 WLU. Table 2 lists the relative carbon intensities and the WLUs accorded for a unit of care in each of the care settings. Carbon efficiency was measured in terms of kgCO2e per WLU for individual hospitals and for the entire NUHS. This metric represents the emissions required to deliver one standardised unit of healthcare work, enabling meaningful comparison of carbon efficiency over time by normalising for case-mix variations rather than simply measuring emissions against raw activity volumes. System-level carbon efficiency was calculated by dividing the total carbon footprint by the total workload in WLUs of all the constituent hospitals and polyclinics. As a complementary assessment, energy efficiency was also tracked using electricity consumption per workload unit (kWh per WLU), providing specific insight into operational energy optimization akin to how the aviation industry uses fuel consumption per passenger-kilometer or the manufacturing sector uses energy per unit of production to assess operational efficiency independent of scale variations. Results The relative carbon intensity across the different care settings and the accorded WLUs per unit of care based on carbon relativity are shown in Table 2. Emissions across NUHS’ institutions increased annually, despite ongoing efforts to reduce environmental impact (Table 4). NUHS’s workload grew between 2020 to 2023, across the hospitals and clinics, in terms of inpatient days and clinic appointments (Table 3, Table 4, Table 5). Table 4Total emissions and workload for each of the institutions over 2020 to 2023 (Data extracted from the NUH Sustainability Report FY24. JCH was calculated by using a 7% consumption across NTFGH and JCH, as they are co-located on-site). See Supplementary Materials, Annex C for additional data details.Table 4 dummy alt textFinancial YearCarbon Emissions (tons CO2e)AHNTFGHNUHNUPJCHNUHS (Total)202018,80070,485182,70823,4975,305300,795202122,09880,721195,75827,4806,076332,132202224,60780,764210,64829,1016,079351,199202325,15883,292219,69029,2106,269363,620 Total emissions and workload for each of the institutions over 2020 to 2023 (Data extracted from the NUH Sustainability Report FY24. JCH was calculated by using a 7% consumption across NTFGH and JCH, as they are co-located on-site). See Supplementary Materials, Annex C for additional data details. Table 3Workload units for each care setting are computed using the respective workloads and carbon relativity numbers for FY2023. See Supplementary Materials Annex B for additional data details.Table 3 dummy alt textCare settingAHJCHNTFGHNUHNUPTotal clinical work (A)Workload Units (WLU) accorded to unit of careTotal WLU(Carbon relativity x (A))Acute Hospital Inpatient days80,448-265,495385,513-731,45614.710,752,403Community Hospital Inpatient days-84,102---84,1023.7311,177Day Surgery Procedures10,344-26,70070,494-107,53817.81,914,176ED Attendances22,901-116,967165,005-304,8733.0914,619Polyclinic Appointments----2,075,9712,075,9711.02,075,971SOC Appointments73,200-294,4751,056,891-1,424,5664.56,410,547Notes: AH – Alexandra Hospital, JCH – Jurong Community Hospital, NTFGH – Ng Teng Fong General Hospital, NUH – National University Hospital, NUP – National University Polyclinics, SOC – Specialist Outpatient Clinic Workload units for each care setting are computed using the respective workloads and carbon relativity numbers for FY2023. See Supplementary Materials Annex B for additional data details. Notes: AH – Alexandra Hospital, JCH – Jurong Community Hospital, NTFGH – Ng Teng Fong General Hospital, NUH – National University Hospital, NUP – National University Polyclinics, SOC – Specialist Outpatient Clinic Table 5Carbon efficiency measured by emissions (kg CO2e) per Workload Unit. See Supplementary Materials Annex D for additional data details.Table 5 dummy alt textFinancial YearTotal Workload unitsTotal Carbon Emissions (tons of CO2e)Carbon Emissions (kg CO2e) / Workload Unit202017,850,573300,79516.85202119,305,012332,13217.20202220,901,693351,19916.80202322,378,894363,62016.25 Carbon efficiency measured by emissions (kg CO2e) per Workload Unit. See Supplementary Materials Annex D for additional data details. NUHS’s scope 2 emissions, total workload units, and scope 2 emissions per WLU are shown in Fig. 2A–2C. Scope 2 emissions increased from 2020 to 2023 (Fig. 2A); in the same period, the total workload in NUHS also increased (Fig. 2B). There is a decrease in the Scope 2 emissions per WLU, indicating improved energy efficiency despite increases in Scope 2 emissions (Fig. 2C). Total emissions per WLU are shown in Fig. 2D. A-D. (A) NUHS Scope 2 carbon emissions, (B) Workload units across 2020-2023, (C) Scope 2 emissions per workload unit, and (D) Total emissions per workload unit, from FY2020 to FY2023. Discussion We developed a methodology to relate workload and emissions of a healthcare system to evaluate the carbon efficiency of the system. NUHS’ workload changes yearly, generally increasing across all care settings, contributed by the opening of new facilities, increases in operating hours, and ageing of the population. Singapore has a rapidly ageing population – the proportion of residents aged 65 and above increased from 9.0% in 2010 to 18.4% in 2023. This proportion is projected to exceed 25% by 2030 and to further increase healthcare needs [9], [10]. These increases in workload, alongside changes in care provision, will impact the system’s emissions, causing difficulties in evaluating the effectiveness of decarbonisation efforts. Shifts of care from inpatient to ambulatory settings will increase the emissions of the clinics and day surgery settings. The environmental impacts of such changes are difficult to evaluate using metrics based only on hospital bed numbers. Allocating WLUs based on the carbon intensity of the care settings will recognise each setting’s contribution to total emissions and enable evaluation of the carbon efficiency of the whole system and its constituent hospitals and clinics. Accounting for workload and carbon relativity across care settings also enables benchmarking and comparisons between healthcare systems. Both NUHS’ annual workload (measured in WLU) and total emissions increased from 2020 to 2023. Environmental performance evaluation that is based solely on absolute emissions will miss improvements in resource efficiency where more patients are cared for per unit of carbon incurred. While absolute decarbonization is the goal, relative measures of emissions and workload enable better evaluation of sustainability efforts and carbon efficiency. Whilst the evaluation timeframe included the COVID-19 pandemic, the reductions in elective workload in FY2020 were reflected by correspondingly lower total emissions. Scope 2 emissions increased year to year, despite the implementation of several energy efficiency measures. Hospitals in Singapore have very high energy use intensity compared to other building types, averaging 360 kWh per m2 of floor area per year, and Scope 2 accounting for up to 20% of total emissions [9,11]. However, evaluation of NUHS’ Scope 2 emissions per WLU showed a decreasing trend, suggesting that energy efficiency was improving. This system of carbon relativity and WLU is an adaptation of workload accounting and resource planning systems in other industries for sustainability metrics. For example, in aviation, the International Civil Aviation Organization measured the workload of airports by combining passenger volume and air freight tonnage into workload units. Relative Value Guides are used in the US and Australia for billing of anesthesiology services, with relative value units accorded according to the complexity and duration of procedures [[12], [13], [14]]. Alternative methodologies were considered for carbon efficiency. Simple aggregation of workload to accord one WLU for each episode of care in any care setting would misrepresent the very different emissions between care settings. The resources required, financial cost and emissions are vastly dissimilar for one inpatient day compared to one outpatient clinic appointment. Calculating the relativity of the care settings using total financial costs may cause misrepresentation. Manpower costs account for over 60% of NUHS’ expenditure, much more than for goods and services, and the different products used have varied spend-based emission factors. Hence, manpower costs were excluded in the computation, and Eora emission factors were used for derivation of carbon relativities of the care settings. A limitation of this methodology is that the exact carbon relativity of the care settings can change each year, affecting the number of workload units accorded to each care setting. Patterns of resource use and costs allocations may shift, while decarbonization measures differ across care settings. However, maintaining the same relativity simplifies comparisons across years. This ensures a common baseline to analyse the year-on-year changes and assess the institution’s progress. Relativities can be reviewed every 5 years for major changes in care delivery and be rebased accordingly. Furthermore, spend-based emission factors change yearly. EE-MRIO databases differ in each sector and country, resulting in different spend-based emission factors to apply to the wide range of healthcare products. Spend-based emissions factors are impacted not only by environmental sustainability performance in sectors and countries, but also by macroeconomic indicators such as inflation rates and exchange rates. Other constraints of EE-MRIO databases include lag time, source data accuracy and the lack of carbon footprints of specific drugs and equipment. Despite these limitations, this methodology still enables summation of a system’s workload and normalised comparisons of carbon efficiency year to year. We focused on Scope 2 emissions per WLU, as energy use is a significant component of total emissions, is within NUHS’ operational control, and the grid emission factor is updated annually in Singapore [15]. Reducing energy use, improving energy efficiency and shifts to renewable energy are key for sustainable healthcare and their impact can be measured in this methodology. This methodology can be developed into an environmental extension of value-based care, to evaluate the quality of care delivered for the financial cost and carbon footprint incurred [16,17]. It can be applied to departments or medical specialties, for department level evaluation of emissions and progress in sustainability. It can also be applied to specific diagnostic groups or procedures to evaluate the carbon impact of changes in their models of care and care settings. Conclusion The methodology provides a framework for healthcare systems to develop their carbon relativity and efficiency measurements, to assess and benchmark their environmental sustainability while accounting for their workload. This anchors on the healthcare system’s ability to systematically allocate resource utilization and carbon emissions across its range of patient care settings. In summary, evaluating emissions per unit of care enabled our healthcare system to relate the environmental impact of all the services delivered as we strive to both care for our patients and care for the planet. This carbon efficiency approach enabled holistic evaluation of our sustainability efforts and goal of achieving value for the carbon incurred in healthcare. CRediT authorship contribution statement Brandon X. Lum: Writing – original draft, Methodology, Investigation, Formal analysis, Conceptualization. Rebecca Y. Lee: Writing – review & editing, Investigation. Clarence T. Peck: Methodology, Conceptualization. Ricky Wiryopranoto: Methodology, Formal analysis, Conceptualization. Eugene H. Liu: Writing – review & editing, Supervision, Conceptualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have influenced the work reported in this paper. Acknowledgement We thank Professor Khay Guan Yeoh and Mr Richard Lim for their guidance on this project. We thank our NUHS colleagues for their assistance in developing this methodology. Carbon Footprint across various patient settings AHNTFGHNUHNUPJCHEmergency Department Attendances1,551.884,318.816,544.05Day Surgery Procedures3,324.276,635.6616,278.15Inpatient Days12,276.4033,758.86101,291.824,237.77Specialist Outpatient Clinic Appointments7,055.5410,265.5270,524.91Polyclinic Appointments28,466.47Total24,208.0954,978.85194,638.9328,466.474,237.77 Raw Workload figures across NUHS Institutions NUH WorkloadFY20FY21FY22FY23Acute Hospital Inpatient Day331,478345,506375,792385,513Day Surgery Procedures49,74457,22765,03170,494Specialist Outpatient Clinic Appointment806,029868,982882,6451,056,891Emergency Department Attendance119,891130,026148,934165,005NTFGH WorkloadFY20FY21FY22FY23Acute Hospital Inpatient Day214,521234,254256,597265,495Day Surgery Procedures21,58425,36027,08226,700Specialist Outpatient Clinic Appointment247,771282,397289,687294,475Emergency Department Attendance99,862105,957114,762116,967AH WorkloadFY20FY21FY22FY23Acute Hospital Inpatient Day66,68268,24881,54180,448Day Surgery Procedures5,4496,7267,85510,344Specialist Outpatient Clinic Appointment43,26652,07862,74573,200Emergency Department Attendance12,46214,50520,37022,901NUP WorkloadFY20FY21FY22FY23Polyclinic appointment1,567,5661,741,8091,924,5372,075,971JCH WorkloadFY20FY21FY22FY23Community Hospital Inpatient Day74,73675,86379,16584,102 Workload Units across NUHS Institutions Calculation Methodology: Raw Workload (i.e. Annex B) multiplied by carbon relativity of respective patient setting Workload UnitsAHNTFGHNUHNUPJCHTotalFY201,309,3014,952,2099,744,9731,567,566276,52317,850,573FY211,400,8345,483,59910,398,0761,741,809280,69319,305,012FY221,681,9345,901,91311,100,3991,924,537292,91120,901,693FY231,764,8126,054,07512,172,8592,075,971311,17722,378,894 Carbon Footprint across NUHS institutions Carbon Footprint (kg) / Workload UnitAHNTFGHNUHNUPJCHFY2014.3613.7318.7414.9917.66FY2115.7614.8518.8215.7818.99FY2214.3713.5918.7714.9318.44FY2314.1213.4017.8113.8924.33 References