The carbon footprint of surgical operations: a systematic review update

Abstract

Introduction

Sustainability in healthcare is a rapidly developing area of research with recent formal recognition from institutions around the world. We completed an update of a systematic review published in 2020. The aims of this review were to determine the reported carbon footprints of surgical operations in hospitals worldwide, identify variations in reported carbon footprints and highlight carbon hotspots associated with surgery.

Methods

A systematic review was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines. The MEDLINE^®, Embase^® and Cochrane Library databases were searched, and eligibility criteria applied. The study characteristics, scope of product inventory and results were extracted and synthesised. A quality assessment of each study was completed to inform the reliability of the research.

Results

1,308 articles were identified and 7 met the inclusion criteria for the review. The carbon footprint ranged from 28.49kg to 505.1kg carbon dioxide equivalents (CO₂e). Medical devices and consumables were the greatest contributor to emissions, with material production and manufacture representing the majority of this carbon hotspot. There were significant methodological limitations and a lack of consistency in carbon footprint calculations between studies.

Conclusions

This systematic review identifies medical devices and consumables as the largest carbon hotspot where healthcare providers should target their sustainability initiatives. Nevertheless, the number of studies was limited and the quality of the evidence was weak. We recommend that researchers in healthcare sustainability develop international standards for conducting and reporting such studies. This would allow for comparison of individual studies and facilitate meta-analysis of cumulative evidence. A reliable evidence base is a prerequisite for identifying optimal interventions to ensure societal benefits.

Introduction

Globally, the healthcare sector accounts for approximately 5% of greenhouse gas emissions, which are the main driver of climate change.¹ In England, the National Health Service (NHS) accounts for 25% of all public sector greenhouse gases, with two-thirds attributable to the use of medicines, medical equipment and supply chains.² The Sustainable Healthcare Coalition’s care pathways guidance, published in 2015, provides recommendations for assessing the impact of healthcare on the environment.³ Significantly, operating theatres are “three to six times more energy-intense than the hospital as a whole” and are a major contributor to emissions associated with healthcare delivery.⁴ Identifying surgery-specific carbon hotspots would facilitate the formulation of targeted interventions.

The assessment of greenhouse gases associated with surgery is an emerging research area. A systematic review by Rizan et al, published in 2020, searched over 50 years of records but identified only 8 studies evaluating the carbon footprint of surgical operations, all of which were published after 2011.⁵ Importantly, since the final search date of the study by Rizan et al, there have been commitments by both professional and healthcare organisations to address healthcare-associated emissions.

In 2019, the Royal College of Surgeons of England established a working group on sustainability in surgery to identify environmental impacts of surgical practice, with its Sustainability in Surgery Strategy being published in 2021.^6,7 The following year, the Intercollegiate Green Theatre Checklist outlined actions required to minimise emissions associated with operations.⁸ The checklist covers four domains: anaesthetic care, surgery preparation, intraoperative practice and postoperative activities. In 2020, the NHS became the first national health system to commit to becoming a “carbon net zero” organisation, and this has been embedded in legislation through the Health and Care Act 2022.^9,10 The White House/US Department of Health and Human Services health sector climate pledge, launched in 2022, is a voluntary commitment to “reduce organizational emissions by 50% by 2030”.¹¹ Over 100 organisations have signed the pledge, including healthcare providers and pharmaceutical organisations.

Although sustainability in healthcare is clearly a rapidly developing area of research, there are only two systematic reviews assessing the carbon footprint of surgical operations.^5,12 Both of these highlight the paucity of high-quality data.

Methods

This systematic review provides an update on research calculating the carbon footprint of surgical operations. Utilising guidance on the typology for systematic reviewers, it was determined that this is a systematic review of “costs”.¹³ The cost in the context of this review is the carbon footprint associated with performing surgical operations. Consequently, the review question was formulated using the PICOC framework:¹⁴

• • P (population) – surgical operations performed worldwide;
• • I (intervention) – surgical operations as defined by the Organisation for Economic Co-operation and Development;¹⁵
• • C (comparators) – other surgical operations within the population;
• • O (outcome) – reported carbon footprint;
• • C (context) – carbon footprint calculations of surgical operations that include the evaluation of at least two scopes outlined in the Greenhouse Gas (GHG) Protocol.¹⁶

The review question was: What is the variation in reported carbon footprints of surgical operations worldwide?

Protocol and registration

The protocol was registered in the Open Science Framework repository.¹⁷ The review has been reported in accordance with the 2020 PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines.¹⁸

Eligibility criteria

Records were deemed to be eligible for inclusion if they contained any evaluation of the carbon footprint of surgical operations. Records were excluded based on publication type (case reports, congress abstracts, literature reviews, systematic reviews, meta-analyses), language (not written in English) and if studies focused exclusively on:

• • only one of the scopes outlined in the GHG Protocol (e.g. anaesthetic components of operations, pharmaceuticals delivered intraoperatively);¹⁶
• • unused anaesthetic gases and/or surgical equipment;
• • processes outside the theatre itself (e.g. sterilisation);
• • pre or postoperative care;
• • the whole hospital system.

Information sources and search strategy

The MEDLINE^®, Embase^® and Cochrane Library databases were utilised. A modified search strategy from Rizan et al comprising two search domains, “surgery specific terms” and “carbon footprint specific terms”, was employed.⁵ Terms in each domain were combined with “OR” and the two domains were combined with “AND”. The full search strategies for each database are available in the review protocol.¹⁷ Search results were limited by publication date between 4 August 2019 and 31 October 2022. All databases were searched on 21 November 2022. A forward citation search of all articles included in the review was performed on 28 February 2023, using the “cited by” function on Google Scholar™. The additional unique references identified were screened by title and abstract, with review of full texts where applicable.

Selection process

Records from the database searches were imported into EndNote™ (Clarivate, Philadelphia, PA, US) and duplicate records removed. Two reviewers (PR and KS) independently screened the title and abstract of each record to determine eligibility. Inter-rater agreement between the two reviewers was measured with Cohen’s kappa (κ), and the strength of agreement was interpreted according to Landis and Koch.¹⁹ The reviewers compared results and disagreements were resolved by consensus. Where a consensus could not be reached, disagreements were arbitrated by a third reviewer (MR). Articles that satisfied the inclusion criteria were subject to full-text analysis. Records with ambiguous titles and/or no abstracts were also subject to full-text analysis. Articles not meeting the eligibility criteria after full-text analysis were excluded, with reasons provided.

Data extraction and data items

Two broad domains of data were collected: study characteristics and scope of product inventory. If data points were not clear in the study manuscripts, the supplementary material was consulted. For studies that utilised carbon footprinting methodologies from previously published studies, these studies were analysed to determine which guidelines and assessment methods were used. Data points that could not be determined were marked as “not assessable”.

The specific data items listed in Table 1 for each domain were extracted independently in duplicate using pre-piloted spreadsheets (Excel^®; Microsoft, Redmond, WA, US). This was followed by checks for concordance with resolution by consensus or with a third reviewer.

Table 1 .

Data items extracted

Study characteristics • Study setting: type of hospital, country of origin • Focus of study: operation type, surgical specialty • Carbon footprinting approach: process-based, input–output or hybrid and the carbon footprinting guideline(s) • Sample size: number of operations observed • Data collection timeframe: time period over which data were collected

Scope of product inventory • Functional unit: the operation under investigation • Functional unit start and end point: for example, preoperative appointments to postoperative care or timepoints of patient entering and leaving theatre • Greenhouse gases included: determined where explicitly stated, or deduced from carbon footprinting guidelines or databases used • Inventory boundary: descriptive data categorised based on the adapted Greenhouse Gas Protocol scopes17 • Type of data collected: data underpinning carbon footprinting calculations categorised as primary activity data, secondary activity data, primary monetary value data or secondary monetary value data • Carbon footprinting results: numerical values for carbon footprinting results for overall operations and sub-processes or descriptive data (e.g. percentages and graphic summaries) were collected if actual values were not recorded

Data synthesis and analysis

Methodological heterogeneity between studies was anticipated due to variations in functional units and assessment methods reported by Rizan et al.⁵ Consequently, results were not combined through meta-analysis. Quantitative data extracted were expressed as percentage stacked column charts to visualise hotspots between studies. Key effect modifiers (e.g. operation type, country setting and life cycle assessment method) and quality scores were recorded to provide context to the results. The quality scores were assigned using a quality assessment adapted from Rizan et al.⁵ The scores were used to compare variation in methodological quality across resources as well as to inform judgement on the level of confidence in the evidence contributing to the final conclusions. A thematic analysis was performed of recurrent themes identified from similarities and/or differences in carbon footprint results and common hotspots.

Results

Study selection

Database searching identified 1,308 records (Figure 1). After the removal of duplicate records, the titles and abstracts of 1,256 records were screened. This led to a further 1,210 records being excluded (κ=0.73, substantial agreement). From the 46 records subject to full-text analysis, 7 were included in the review (κ=0.85, almost perfect agreement). The other 39 full-text records were excluded, with reasons stated in Figure 1 and Supplementary Table 1. As the number of studies included was limited, a forward citation search was performed, which identified 64 unique records. After removing duplicates, 50 titles and abstracts were screened but none satisfied the eligibility criteria (κ=0.98, almost perfect agreement).

Figure 1 .

Study selection

Supplementary Table 1 .

Excluded studies

Focused exclusively on medical devices and/or consumables (n=8):

1. Baxter NB, Yoon AP, Chung KC. Variability in the use of disposable surgical supplies: a surgeon survey and life cycle analysis. J Hand Surg Am 2021; 46: 1,071–1,078.

2. Boberg L, Singh J, Montgomery A, Bentzer P. Environmental impact of single-use, reusable, and mixed trocar systems used for laparoscopic cholecystectomies. PLoS One 2022; 17: e0271601.

3. Jabouri H, Abbott RA. Sustainability in skin cancer surgery. Br J Dermatol 2022; 186: 735–736.

4. Kooner S, Hewison C, Sridharan S et al. Waste and recycling among orthopedic subspecialties. Can J Surg 2020; 63: E278–E283.

5. Misrai V, Rijo E, Cottenceau JB et al. A standardized method for estimating the carbon footprint of disposable minimally invasive surgical devices. Ann Surg 2021; 2: e094.

6. Rizan C, Bhutta MF. Environmental impact and life cycle financial cost of hybrid (reusable/single-use) instruments versus single-use equivalents in laparoscopic cholecystectomy. Surg Endosc 2022; 36: 4,067–4,078.

7. Ryan MT, Malmrose J, Riley CA, Tolisano AM. Operating room waste generated across otolaryngology cases. Mil Med 2021; usab548.

8. Shum PL, Kok HK, Maingard J et al. Environmental sustainability in neurointerventional procedures: a waste audit. J Neurointerv Surg 2020; 12: 1,053–1,057.

Focused exclusively on unused components of operation (n=2):

1. Bhatter P, Cypen SG, Carter SL, Tao JP. Pharmaceutical and supply waste in oculofacial plastic surgery at a hospital-based outpatient surgery center. Ophthalmic Plast Reconstr Surg 2021; 37: 435–438.

2. Bravo D, Thiel C, Bello R et al. What a waste! The impact of unused surgical supplies in hand surgery and how we can improve. Hand 2022; 15589447221084011.

Study characteristics

The characteristics of the seven included studies are shown in Table 2.^20–26 Four surgical specialties were represented: cardiology, dermatology, interventional radiology and ophthalmology. Two of the studies considered a “conventional” operation, for which the average emissions were calculated from a range of related operations.^21,23 All studies assessed the carbon footprint at public institutions and two studies included additional assessment in the private sector.^20,22 Four studies used a process-based methodology^20,23,25,26 and three used a hybrid approach.^21,22,24 The majority of the studies utilised the International Organization for Standardization guidelines; however, only two studies referenced these standards formally.^21,22 The study designs were variable. For example, the number of operations observed ranged from 2 to 142 and the period of data collection ranged from 1 day to 3 months.

Table 2 .

Study characteristics, scope of product inventory and results

Study characteristics						Scope		Results
Study	Study setting; country	Focus of study; surgical specialty	Carbon footprinting approach; guideline	Sample size	Sampling duration	Functional unit; start point – end point	Included GHGs (in brackets where deduced); guideline / database	Carbon footprint (CO2e)
Tan, 202120	1 private clinic, 1 public hospital; Australia	Conventional surgery for 1 skin cancer excision; dermatology	Process-based; ISO 14040 and 14044	2 skin cancer excisions	2 days	1 skin cancer excision; patient and staff travel to theatre – patient and staff travel home	Not assessable; IMPACT World+™	28.49kg
Chua, 202121	1 university hospital; US	Typical interventional radiology procedures (drainages, placement and removal of venous access, computed tomography guided biopsies); interventional radiology	Hybrid; ISO 14040 and 14044, GHG Protocol	98 interventional radiology procedures	1 week	1 typical interventional radiology procedure; staff commute and patient enters department – patient leaves department	(91); TRACI 2.1	243kg
Latta, 202122	2 public hospitals, 2 private hospitals; New Zealand	Cataract surgery; ophthalmology	Hybrid; ISO 14040 and 14044	142 cataract procedures	3 months	Cataract surgery (1 eye); patient and staff travel to theatre – patient and staff travel home	(7); DEFRA 2011	151.9kg
Grinberg, 202123	1 university hospital; France	Conventional cardiac surgery (single valve repair, single valve replacement, isolated on-pump CABG); cardiology	Process-based; ISO 14040	28 cardiac procedures: anaesthetic component (5), surgical field component (17), cardiopulmonary component (6)	4 weeks	1 conventional cardiac procedure; patient enters theatre – patient leaves theatre	(CO2); Ansys Granta EduPack	124.3kg
Ferrero, 202224	1 university hospital; France	Cataract surgery; ophthalmology	Hybrid; PAS 2050, GHG Protocol, Bilan Carbone®	12 cataract procedures	1 day	Cataract surgery (1 eye); patient and staff travel to theatre – patient and staff travel home	7; Kyoto Protocol, PAS 2050, GHG Protocol, Bilan Carbone®, DEFRA 2012	81.13kg
Hubert, 202225	1 university hospital; US	CABG; cardiology	Process-based; nil	18 CABG procedures	57 days	1 CABG procedure; patient enters theatre – patient leaves theatre	(7); DEFRA 2011	505.1kg
Ditac, 202226	1 public hospital; France	Atrial fibrillation catheter ablation; cardiology	Process-based; ISO 14040	30 atrial fibrillation catheter ablations	8 weeks	1 atrial fibrillation catheter ablation; patient enters theatre – patient leaves theatre	(CO2); Ansys Granta EduPack	76.9kg
CABG = coronary artery bypass graft; CO2e = carbon dioxide equivalents; DEFRA = Department for Environment, Food and Rural Affairs; GHG = greenhouse gas; ISO = International Organization for Standardization; PAS = Publicly Available Specification; TRACI = Tool for Reduction and Assessment of Chemicals and Other Environmental Impacts

Scope of product inventory

None of the studies included perioperative care in the defined functional unit (Table 2). Four studies accounted for patient and/or staff travel to the operation site,^20–22,24 with the remaining studies only considering processes occurring in the time between the patient entering and exiting the theatre.

Due to ambiguity between the categorisation of processes included in the inventory boundary, processes were assigned to ten broad categories (Table 3). All the processes recorded were explicitly stated; ambiguous or inferred processes were excluded from data collection. All studies accounted for electricity, medical devices and consumables. Four studies considered scope 1 emissions, all of which were volatile anaesthetic gases.^21,23,25,26 Two studies directly measured the amount of anaesthetic gas used for each individual operation,^21,25 one study calculated the average anaesthetic gas per patient using the total consumption of anaesthetic gas for the institution²³ and for one study, it could not be determined how volatile anaesthetic gas emissions were calculated.²⁶ The majority of the data collected were primary process data (Supplementary Table 2), with only three studies utilising the monetary value of items^21,22,24 and all using site-specific monetary values.

Table 3 .

Inventory boundary

Process		Tan, 202120	Chua, 202121	Latta, 202122	Grinberg, 202123	Ferrero, 202224	Hubert, 202225	Ditac, 202226
Travel	Patients	✓		✓		✓
	Staff	✓	✓	✓		✓
Electricity	Principle of proportionality	✓		✓	✓	✓
	Wattage of equipment		✓				✓	✓
Heating, ventilation and air conditioning	Heating		✓			✓
	Ventilation	✓	✓	✓	✓	✓
	Air conditioning	✓	✓	✓	✓	✓
Water		✓
Consumables (e.g. linen, packaging)	Manufacturing and procurement	✓	✓	✓	✓	✓	✓	✓
	Transport	✓	✓	✓	✓	✓		✓
Medical devices (e.g. instruments, gauze)	Manufacturing and procurement	✓	✓	✓	✓	✓	✓	✓
	Transport	✓	✓	✓	✓	✓		✓
Sterilisation of reusables		✓				✓
Non-volatile pharmaceuticals	Manufacturing and procurement			✓	✓	✓		✓
	Transport			✓	✓	✓
Direct volatile anaesthetics			✓		✓		✓	✓
Waste processing and disposal	Transport	✓		✓		✓	✓
	Sterilisation		✓				✓
	Incineration	✓				✓
	Landfill	✓	✓	✓			✓

Supplementary Table 2 .

Scope of product inventory

Study	Inventory boundary (scope classified based on surgical operation as the functional unit)				Data collection		Calculating inventory results
	Process/item	Scope 1	Scope 2	Scope 3	Data source: data collected	Data type	Emission factor/global warming potential source	Impact assessment database
Tan, 202120	Transportation of staff			X	Literature: the Australian Bureau of Statistics estimates 16km as the average work commute, with light personal vehicle the most common form of transportation. More than half of all car trips in Australia <5km; used 10km return trip for dermatologist. Light passenger vehicles account for 74% of registered vehicles in Australia with average petrol consumption of 11 litres per 100km. Petrol makes up 51.4% of all fuel consumption, consequently petrol used for calculations.	Secondary process data	Evah OzLCI2019	IMPACT World+™
	Transportation of patients			X	Literature: assumed that most patients would be from a metropolitan area, utilising same methodology for transportation of staff	Secondary process data
	Transportation of consumables			X	Clinic data: deliveries carried out by light commercial vehicle consuming 13 litres of petrol per 100km	Primary process data
	Transportation of waste			X	Literature: assumed transportation of waste to landfill by a typical waste truck weighing 20 tonnes with a diesel consumption of 78 litres per 100km	Secondary process data
	Electricity consumption private clinic		X		Literature: based on premise that private clinical rooms in Australia classified as small/medium enterprises by Australian Bureau of Statistics which have average annual electricity consumption of 25,000kWh. Assumed that skin cancer excision takes 30 minutes and the average working hours of a dermatologist equal to 34.5 hours.	Secondary process data
	Electricity consumption public hospital		X		Literature: average consumption of hospital buildings, proportioned to average operating theatre space	Secondary process data
	Sterilisation of surgical equipment		X		Clinic data: steam steriliser using 2.3kWh of electricity with cycle duration of 40–60 minutes	Primary process data
	Water supply and treatment		X		Literature: estimated 10 litres of water required, and supply and treatment powered by electricity	Secondary process data
	Surgical instruments			X	Clinic data: weight of instruments constituent materials, transportation	Primary process data
	Consumables			X	Clinic data: weight of consumables constituent materials	Primary process data
	Repackaging			X	Clinic data: weight of packaging constituent materials	Primary process data
	Waste disposal			X	Waste audit: waste divided into non-hazardous, biohazardous and sharps waste, then weighed. Non-hazardous sent to landfill; hazardous and sharps waste processed.	Primary process data
Chua, 202121	Transportation of staff			X	Literature: assumed staff distribution as reported by census data and same mix of transportation used by people in that area	Secondary process data	Ecoinvent 3.3 emissions database, USEEIO database	TRACI 2.1
	Electricity consumption: units, lighting, electrical equipment		X		Hospital data: average power specifications of each device and typical number of hours equipment in use	Primary process data
	Heating, ventilation, air conditioning		X		Hospital data: Thiel (2013) bin model based on floor area, room temperature, outdoor air ratio, humidity set points, number of air changes per hour	Primary process data
	Anaesthetic gas	X			Direct measurement: total volume of gas administered during procedure	Primary process data
	Disposable instruments and supplies			X	Hospital data: price centre paid for items collected from financial records	Primary monetary value data
	Waste disposal			X	Waste audit: waste divided into soiled linen and municipal, solid biohazard and sharps waste, then weighed. Retrospective data on biohazardous fluid collected.	Primary process data
Latta, 202122	Transportation of staff			X	Survey: travel methods collected from all staff; driving distances calculated from suburbs; calculations based on fuel consumption of average New Zealand car	Primary process data	New Zealand Ministry for the Environment 2020	N/A
	Transportation of patients			X	Survey: travel methods collected from first 10 patients; driving distances calculated from suburbs; calculations based on fuel consumption of average New Zealand car	Primary process data	New Zealand Ministry for the Environment 2020
	Electricity consumption		X		Hospital data: average monthly power consumption of each hospital/surgical unit; proportion based on floor space of ophthalmic department and time scheduled for cataract surgery. Assumed 1m2 of operating theatre twice as energy intensive as average floor space of hospital.	Primary process data	New Zealand Ministry for the Environment 2020
	Disposable items			X	Hospital data: procurement costs collected from theatre managers	Primary monetary value data	DEFRA 2011
	Pharmaceuticals			X	Hospital data: procurement costs collected from theatre managers	Primary monetary value data	DEFRA 2011
	Waste disposal			X	Waste audit: weight and composition	Primary monetary value data	New Zealand Ministry for the Environment 2020, DEFRA 2011
Grinberg, 202123	Energy consumption		X		Hospital data: data related to the surgical platform at hospital provided by healthcare system administration	Primary process data	Consumable analysis and carbon production were analysed by an independent team of scientists from the LGEF Laboratory at INSA, Université de Lyon	Ansys Granta EduPack
	Disposable medical products			X	Hospital data: all disposable items used from time patient entered until patient exited weighed and analysed	Primary process data
	Injectable pharmaceuticals			X	Hospital data: considered a "standard injectable drug tray" in preliminary phase of the study. Researched synthesis methods and modelled reaction using "SciFinder".	Primary process data
	Volatile pharmaceuticals	X			Hospital data: annual average consumption of sevoflurane per patient at hospital	Primary process data
Ferrero, 202224	Transportation of staff			X	Survey: addresses and mode of transport for all staff to and from hospital	Primary process data	Bilan Carbone®, DEFRA 2012	N/A
	Transportation of patients			X	Survey: addresses and mode of transport for all patients to and from hospital	Primary process data
	Energy consumption		X		Hospital data: annual consumption of electricity and steam of building per m2. Based on calculations by hospital engineering department, operating room 1.4 times more energy intensive than rest of building.	Primary process data
	Surgical instruments			X	Hospital data: price paid	Primary monetary value data
	Transportation of surgical instruments			X	Hospital data: recorded place of manufacture, mode of transport, storage and distribution methods	Primary process data
	Pharmaceutical products			X	Hospital data: price paid	Primary monetary value data
	Transportation of pharmaceutical products			X	Hospital data: recorded place of manufacture, mode of transport, storage and distribution methods	Primary process data
	Single-use device sterilisation		X	X	Hospital data: estimated emissions from sterilisation department; proportioned emissions based on percentage related to ophthalmology and sterilisation of phacoemulsifier hand piece Survey: collected addresses and mode of transport for sterilisation staff	Primary process data
	Waste disposal			X	Waste audit: waste collected, divided into unregulated and regulated waste, then weighed	Primary process data
Hubert, 202225	Electricity consumption		X		Hospital data: kWh of each device recorded; local energy production profile applied	Primary process data	Andersen et al (2012), DEFRA 2011	N/A
	Volatile anaesthesia	X			Direct measurement: total volume of gas administered during procedure	Primary process data
	Waste generation		X	X	Waste audit: all waste counted and weighed. All waste in cardiac suite processed through autoclaving.	Primary process data
Ditac, 202226	Energy consumption		X		Hospital data: electricity consumption of each equipment collected; data related to surgical platform of hospital collected by the administration	Primary process data	Consumable analysis and carbon production estimate were made by an independent team of scientists and engineers (LGEF Laboratory at INSA, Université de Lyon)	Ansys Granta EduPack
	Disposable materials			X	Hospital data: weight of item’s constituent materials	Primary process data
	Pharmaceuticals (including anaesthetic gases)	X		X	Not assessable	Not assessable

Carbon footprints

The reported carbon footprint ranged from 28.49kg to 505.1kg carbon dioxide equivalents (CO₂e) (Table 2). In some instances, there were minor discrepancies in the sum of data points within individual studies, likely due to early rounding errors. Consequently, the reported total carbon footprint for each study was used.

Interestingly, there were notable differences between carbon footprints of the same operation. Two studies analysed cataract surgery with one site calculating 81.13kg CO₂e²⁴ and the other reporting 151.8kg CO₂e.²² Carbon footprints also varied between operations performed in the public and private sectors.^20,22 The carbon footprint of one cataract procedure in a New Zealand private hospital was calculated to be 9.2% greater than one conducted in a public hospital (145.2kg CO₂e vs 158.6kg CO₂e).²² Conversely, the carbon footprint of one skin cancer excision in an Australian public hospital was calculated to be 2.5 times greater than one performed in a private clinic (28.49kg CO₂e vs 72.15kg CO₂e).²⁰

Ditac et al analysed the carbon footprint associated with three different types of catheter ablation techniques: radiofrequency pulmonary vein isolation, radiofrequency Marshall vein isolation and cryoballoon catheter pulmonary vein isolation.²⁶ The study found that radiofrequency Marshall procedures had the highest emissions (87.9kg CO₂e) and that cryoballoon catheter pulmonary vein isolation procedures had the lowest (71.2kg CO₂e).

Two other studies calculated emissions for cardiac surgery.^23,25 Grinberg et al considered conventional cardiac procedures, which included valvular and coronary artery bypass, and calculated average emissions to be 124.3kg CO₂e.²³ Hubert et al determined one coronary artery bypass procedure to be equivalent to 505.1kg CO₂e.²⁵ In both studies, the majority (80.1–86.8%) of emissions were attributed to medical devices and consumables. Hubert et al²⁵ compared their results with those of Grinberg et al,²³ and suggested that their larger calculated carbon footprint was a result of the high volume of plastic waste and single-use sharps, and the fact that all waste at their institution is managed as biohazardous, requiring energy-intensive disposal.

Carbon hotspots

The breakdown of the total carbon footprint is shown in Figure 2. In five of the studies, medical devices and consumables were identified as the greatest contributor to the carbon footprint, ranging from 73.3% to 86.8% of the total carbon footprint.^22–26 The study investigating interventional radiology determined electricity use to be the greatest contributor (53.8%), and medical devices and consumables to be the second largest contributor (42.1%).²¹ The skin cancer excision study determined patient and staff transport to be the largest contributor (44.9%).²⁰

Figure 2 .

Carbon hotspots HVAC = heating, ventilation and air conditioning

In all studies, medical devices and consumables were described as “disposable” or “single-use” items. Three studies provided a breakdown of this category.^23,24,26 All three studies found material production and manufacture of supplies to be the greatest contributor to this category, ranging from 73.3% to 88.3%. Two of the studies provided separate figures for material production and manufacture, and both found emissions from material production to be greater than emissions from material manufacture (45.2% vs 38.0% and 71.3% vs 17.0%).^23,26 All three found transport of devices and consumables to be the smallest contributor, ranging from 6.71% to 16.8%.^23,24,26

Chua et al investigated interventional radiology, and determined that heating, ventilation and air conditioning (HVAC) was accountable for the majority (54%) of emissions.²¹ Interestingly, more than half (57%) of the energy use associated with HVAC occurred outside scheduled working hours when few procedures were performed and the unit was largely unoccupied.

Quality assessment (risk of bias)

The quality scores for each individual study, expressed as percentages, are provided in Supplementary Table 3. There were several factors affecting the internal and external validity of studies. Internal validity was limited due to study methodologies not utilising data specific to the site under investigation, such as application of emission factors not specific to the country setting or utilisation of secondary data. Variation in nomenclature of processes in the inventory boundary (Supplementary Table 2) and limited breakdown of individual items within each process made determining the completeness of studies challenging. All of the studies failed to include scenario or parameter uncertainty, meaning that the impact of methodological limitations on carbon footprinting results is unknown. Further examples of limitations and assumptions of the inventory boundary are available in Supplementary Table 4.

Supplementary Table 3 .

Quality assessment

Supplementary Table 4 .

Stated exclusions, assumptions and other limitations

Study	Stated exclusions of boundary: functional unit, processes	Stated assumptions in data collection	Other limitations
Tan, 202120	Mohs micrographic surgery	Transportation of staff and patients was a 10km return trip and utilised the average petrol consumption of all registered vehicles in Australia Surgical team of 7 staff in public hospital Waste transportation carried out by 20 tonne truck with diesel consumption of 78 litres per 100km Dermatology private clinic had the average annual electricity consumption of a small/medium enterprise classified by the Australian Bureau of Statistics Electricity consumption of public hospital based on the averages outlined in Baseline Energy Consumption and Greenhouse Gas Emissions in Commercial Buildings in Australia published by Australian Government Department of Climate Change and Energy Efficiency Local production of consumables	Stated: • Addition of tungsten carbide to instruments, local anaesthetic and sutures not assessed owing to paucity of data within life cycle inventory databases • Top-down approach may not have captured full extent of impact and likely underestimates true impact Not stated: • Data collection for public hospital ambiguous, unclear what processes are included and lack of breakdown for the total carbon footprint • No parameter uncertainty or scenario uncertainty modelling • Unclear how many observations made at each site
Chua, 202121	Procedures performed at night between 7pm and 7am, procedures performed on weekend, postoperative management Patient gowns Patient travel Housekeeping, patient escorts and other staff members not primarily based in interventional radiology suites	Assumed staff had same residential distribution as adult New York City area residents from 2010 census data, staff lived in the population centre of each borough or suburban area and staff used the same mix of transportation modes as those of other workers commuting to Manhattan Electricity use based on typical number of hours each piece of hardware was either in use, off or in idle mode Applied the average electricity mix for the entire US Assumed each piece of linen had average lifespan of 20 uses	Stated: • Use of environmentally extended input–output methodology for single-use supplies does not account for all processes underpinning items • Likely overestimated emissions associated with sevoflurane use as an unquantifiable portion of exhaled gas recirculated into anaesthesia machine Not stated: • No parameter uncertainty or scenario uncertainty modelling
Latta, 202122	Larger combined procedures (e.g. cataract surgery with planned vitrectomy, or trabeculectomy, or iris suturing), perioperative clinic appointments Waste generated for local or general anaesthesia Construction and provisioning of hospital Information technology Stationery Linen Laundry Capital items (e.g. operating microscope, phacoemulsification device) Food for staff and patients Security and reception staff Scientific activities, background knowledge and training underlying the operation Sterilisation of instruments	Survey of all staff and first 10 patients showed that all interviewed drove; assumed all subsequent patients also drove All staff and patients lived in suburb area and used the average fuel performance of the average New Zealand car Power consumption of operating theatre used twice the energy as average floor space of hospital (based on previously published methods by Thiel et al [2017] and Morris et al [2013]) All hospitals studied had equivalent energy sources form the national grid Landfill waste for hospitals that recycled was all plastic; landfill waste for hospitals not recycling was as a mix of paper and plastic (used a general waste coefficient)	Stated: • Used UK emission coefficients for procurement of pharmaceuticals and medical equipment, and plastic in landfill (not specific to study site) • Used top-down approach for procurement • Possibility of selection bias due to incomplete sampling of cataract procedures in one hospital site Not stated: • No parameter uncertainty or scenario uncertainty modelling
Grinberg, 202123	Emergency procedures, thoracic aortic procedures and beating heart coronary revascularisations, pre and postoperative care Transport of staff and patients Water filtration system Instrument sterilisation Prosthetic heart valves Product use and disposal (collection and sterilisation)	Data collection performed by one investigator; only one "workstation" was analysed per procedure. Consequently, data collected were not complete for each single procedure but an average across all procedures. Anaesthetic gas emissions calculated based on yearly mean sevoflurane consumption per patient at institution For pharmaceutical products considered a "standard injectable drug tray" (i.e. did not consider additional consumption related to occurrence of adverse events during surgery) based on 5 procedures in pilot study	Stated: • N/A Not stated: • No parameter uncertainty or scenario uncertainty modelling
Ferrero, 202224	Pre and postoperative care Construction of building Manufacturing and delivery of phacoemulsification machine Surgical microscope Food and beverages of staff Activities related to medical research Water consumption Disposable linen	Estimated energy consumption of operating room 1.4 times that of rest of hospital (based on calculations made by the hospital’s engineering department)	Stated: • DEFRA 2012 used for medical device production and transportation (not specific to study site) • Unable to account for transportation of raw material to manufacturing sites Not stated: • No parameter uncertainty or scenario uncertainty modelling
Hubert, 202225	Emergency procedures, combined procedures, cases with prolonged surgical times due to complications, pre and postoperative care Heating, ventilation, air conditioning Reprocessing of surgical instruments	N/A	Stated: • N/A Not stated: • Emission factors associated with waste streams taken from MacNeil et al (2017), which used DEFRA emission factors (not specific to study site) • Ambiguity on autoclaving emission calculation • No parameter uncertainty or scenario uncertainty modelling • Does not follow a carbon footprinting guideline • No assumptions in data collection outlined
Ditac, 202226	Pre and postoperative phases Building construction Transportation of staff and patients Recycling or reprocessing	Approximations used for missing data for modes of transport for material and products	Stated: • N/A Not stated: • Unable to determine how “;pharmaceutical products (including anaesthetic gases)” was calculated • No parameter uncertainty or scenario uncertainty modelling

External validity of studies (the extent to which carbon footprinting results could be generalised to other settings) was limited because of inherent variations in hospital settings.²⁷ For example, emissions associated with electricity use can vary even if other factors (such as operation duration and equipment used) are similar, due to differences in a country’s electricity generation. Operations conducted in France, which has 67% of electricity generation from nuclear sources, would have lower emissions than similar operations in the US, where 68% of electricity generation is from gas.^25,28 Although studies were not excluded based on their quality assessment scores, there is an inherent risk of reporting bias.

Heterogeneity across studies

Heterogeneity between studies was anticipated based on previously published systematic reviews.^5,12 The included studies displayed “clinical diversity” as a range of operations were studied in a variety of clinical settings.²⁹ Additionally, studies demonstrated “methodological diversity” with variations in inventory boundaries, included processes and methodologies implemented. Variations in reported carbon footprints between studies evaluating the same intervention (e.g. cataract surgery) displayed “statistical heterogeneity” due to both clinical and methodological diversity. The input–output methodology can lead to artificial differences in emissions for the same items owing to the purchasing power of institutions. Latta et al highlighted that hospitals demanding more sustainable supply chains may pay a greater price for those supplies and emission coefficients would incorrectly estimate greater emissions.²²

Discussion

The most consistent carbon hotspot across studies was medical devices and consumables, with material production and manufacture of items the greatest contributor to emissions. Other hotspots identified were electricity, and transport for staff and patients. The hotspots identified in this review are consistent with the findings in a previous systematic review by Rizan et al, which identified eight studies, all published between 2011 and 2019.⁵ Our update, using similar inclusion criteria over a three-year period, identified a comparable number of relevant articles, suggesting that there is increased interest in the sustainable delivery of surgery.

The carbon hotspots identified underscore the adverse effects on the environment that can be attributable to medical devices and consumables. On further examination, it is apparent that transportation of these goods represents only a small proportion of the carbon footprint and that logistics interventions are unlikely to have a significant effect. Reducing emissions associated with this hotspot can be achieved through the substitution of single-use items with reusable or hybrid types.^30,31 Additionally, adequate sorting of waste could reduce the volume destined for autoclaving or incineration and maximise the potential for reprocessing of materials.

Reducing emissions associated with electricity use can include practices such as shutting down electrical equipment and lighting, and permitting climate control systems to drift within a wider range of temperatures outside clinical hours.²¹ Interestingly, a meta-analysis concluded that there was no benefit of laminar airflow over conventional turbulent ventilation in reducing risk of surgical site infections, which is another potential energy saving strategy in the operating theatre.³²

A recurrent theme emerging from the current study and other systematic reviews^5,12 is the heterogeneity of the studies that makes comparisons challenging. Although there are existing guidelines for calculating carbon footprints, there are none that are specific to surgical operations. When considering the inventory boundary of a surgical operation, the included processes are variable and diverse between studies. It is important for researchers in this emerging field to develop international guidelines that define inventory boundaries to facilitate comparison between studies and increase the generalisability of results. Notwithstanding the standardisation of inventory boundaries, variation in methodologies to calculate individual processes also lead to differences in results and conclusions. Consequently, it is important to control these confounding factors so that the most efficient hospitals and surgical protocols can be identified, to guide us towards more sustainable, “greener” surgical practice.

The methodologies of some studies were unclear, which made determining what items were accounted for problematic, particularly under categories such as “disposable items” and “surgical instruments”. There were difficulties in interpreting the breakdown figures of the total carbon footprint due to minor discrepancies between addition of these figures and the reported total carbon footprint. In order to improve interpretation of carbon footprinting results, a detailed protocol should be made available alongside the raw data. Medical supply companies should be encouraged to publish the carbon impact of their products using a similar format to the EU energy labelling standards.³³

This systematic review has been transparently reported in accordance with the PRISMA 2020 statement.¹⁸ Selection bias of resources was minimised by using two reviewers who independently screened and selected studies for inclusion, and by not excluding articles based on quality assessment. The strict eligibility criteria, which excluded articles not written in English, could have led to selection bias during identification of resources. Publication bias was mitigated by searching several databases but could have been reduced further by searching the grey literature. The quality of studies included in this review varied. Incomplete reporting of carbon footprinting methodologies, lack of detail in processes accounted for and study assumptions could have hampered synthesis as well as interpretation of carbon footprinting results. There is a need to develop a validated quality assessment tool for medical sustainability research, similar to those produced by the Joanna Briggs Institute for other fields of research.³⁴

Conclusions

This systematic review identifies medical devices and consumables as the largest carbon hotspot where healthcare providers should target their sustainability initiatives. Nevertheless, the number of studies was limited and the quality of the evidence was weak. We recommend that researchers in healthcare sustainability develop international standards for conducting and reporting such studies. This would allow for comparison of individual studies and facilitate meta-analysis of cumulative evidence. A reliable evidence base is a prerequisite for identifying optimal interventions to ensure societal benefits.