In total, six Investigational Medicinal Product (IMP) trials were footprinted [in breast cancer, gestational diabetes, COVID-19, Intracerebral Haemorrhage (ICH) (n = 2) and Human Immunodeficiency Virus (HIV)], one Nutritional trial (lung disease), one surgical trial (benign prostate enlargement), one health surveillance trial (dental) and one complex intervention trial (behavioural). Six trials were completed at the time of inclusion and four were ongoing (two recruiting, two in follow up). Seven trials included UK participation only, two trials were run within the Republic of Ireland, and one trial was international (participation from 10 countries, regional trial management and a Sponsor CTU based in Australia). Table 3 provides more details of the trial designs.
Our initial guidance and method included the majority of clinical trial activities and corresponding emission factors required to calculate the carbon footprint of the 10 selected trials. Where new activities were identified, emission factors were sourced from publications, Life Cycle Analysis (LCA) databases and articles, and all new activity data and emission factors have been added to the Guidance to create V0.5. All sources are cited and referenced in the guidance. Where emission factors were unavailable, proxy emission factors were used.
The results of carbon footprinting are presented in Table 3, including the total carbon footprint (tonnes CO2e) and the three modules which had the largest contributions to the footprint. Figure 1 demonstrates the proportion of greenhouse gas emissions attributed to each module in the 10 trials.
Total carbon footprint
The estimated trial carbon footprints ranged from 16 tonnes CO2e in a single site study with 102 participants, to 765 tonnes CO2e in an international trial which recruited 7064 participants from 122 sites across 10 countries.
16 tonnes CO2e is equivalent to the ghg emissions from 9.7 return flights from London to New York5,6, the annual footprint of 1.26 UK citizens4, or the electricity used by 3.2 homes for one year7. Conversely, 765 tonnes CO2e is equivalent to the ghg emissions from 465.2 return flights from London to New York, the annual footprint of 60 UK citizens, or the electricity used by 151 homes for one year.
Carbon hotspots
In nine of the 10 trials, CTU emissions featured in the top three hotspots. Typically, this becomes more of a hotspot as the CTU staff Full Time Equivalent (FTE) increases with increased number of sites and participants and a longer trial duration. Contribution from commuting was likely higher pre-pandemic when most CTU staff were 100% office based. Some of the trials conducted during COVID-19 (MAVMET and HEAL-COVID) also had lower commuting emissions due to staff working from home 100% of their time during lockdowns. The CTU location can also affect commuting emissions; the carbon footprint of commuting was much lower in the CTUs located in London (Imperial and UCL) where public transport is used more in comparison to Edinburgh CTU where over 70% of commuting was by car. CTU emissions also significantly contribute to the carbon footprint of large international trials such as INTERACT-3 due to there being multiple trial coordination centres in multiple countries, some of which may have higher intensity national grids than the UK.
In eight of the 10 trials, trial-specific patient assessments were a hotspot. Patient travel to hospital for visits that were in addition to standard-of-care was frequently a large contributor to this. The absolute contribution in terms of carbon emissions could depend on the location/spread of the trial participant population and the mode of transport generally used. For example, in MAVMET, which was based in London, public transport use was assumed compared to the SHAMROCK trial in Ireland where 100% of participants were assumed to travel by car over larger distances; although trial-specific patient assessments was a hotspot in both trials which were similar in terms of the number of sites and participants, the total carbon emissions was much higher in SHAMROCK.
Staff meetings and travel was a hotspot in four of the trials. This was the largest contributor to emissions in the INTERVAL trial due to travel for site initiation visits (SIVs), regional recruitment events, monitoring at a portion of the sites, in person Trial Management Group (TMG) and Trial Steering Committee (TSC) meetings and conferences.
Laboratory activity was a hotspot in three of the trials. This is mostly attributed to international shipment of samples/sample kits, or storage of samples in ULT freezers, sometimes for up to 10 years.
Trial supplies and equipment was also a hotspot in three of the trials. In the UK Stand Together trial, trial supplies and equipment had the largest contribution to the trial carbon footprint due to provision of 360 tablets to sites for completion of questionnaires. Similarly, in EMERGE, this hotspot was attributed to provision and use of 535 glucometers, and in RESTART, this was related to purchase of IT equipment.
Trial close-out was a hotspot in two of the trials. In MAVMET, this was attributed to storage of 28 archive boxes for 25 years, whereas in HEAL-COVID (in which only six of the 10 footprinting modules were relevant as there was a standard of care, locally prescribed intervention and no samples or patient assessments in addition to standard of care), this was attributed to data storage.
In addition to the more general and frequently seen hotspots, these results also illustrated trials with hotspots that were specific to the trial design or intervention.
For example, in HEAL-COVID, data collection and exchange had the largest contribution to emissions due to the considerable cost attributed to accessing and linking data from NHS England (formerly NHSDigital) and purchase of software to operationalise the decentralised trial design. However, in the absence of sufficient activity data and published emission factors, emissions attributed to these activities were calculated using a spend-based emission factor, which are known to be less accurate than activity-based emission factors.A spend based approach involves multiplying the cost of an activity or service by an emission factor representing the average emissions per pound spent in that particular industry.
Feedback on carbon calculations
The time reported to collate trial activity data and complete the carbon footprinting calculations ranged from 5 hours to 60 hours, largely depending on trial size and complexity and the extent to which the individual performing the footprinting was familiar with the trial and could easily locate the required information. Collaborators who went on to footprint more than one trial anecdotally noted that it took approximately 50% less time on repeat application of guidance.
Previously, our guidance was applied retrospectively to two completed trials, and we anticipated that application to trials which are currently active or in development would take less time and be less resource intensive2. To assess this, both ongoing and completed trials were footprinted. In four of the completed trials footprinted by trial teams, the time required to retrospectively collate the trial data alone ranged from 10 to 25 hours, whereas the information was much more readily available in the ongoing trials. For SHAMROCK, the trial in set up, most of the anticipated activity was gathered via the protocol, email correspondence and a 1-hour meeting. This is because prospective application of the guidance requires the user to make more assumptions which can speed up data collection. For example, the number of planned trial meetings and patient visits can be taken directly from the funding application of a trial in set up, whereas identifying the number of visits or meetings that actually took place in a trial can require review of multiple folders and databases. However, attempting to make more accurate assumptions can also be more time consuming. For example, instead of counting the number of boxes stored in an office or looking at the GB of storage used by a trial folder, to estimate this for a trial in set up you would first need to identify a trial with similar number of sites and participants and then use that to estimate the activity data.
The majority of users required very little clarification or help to use the guidance and there were few corrections made to calculations by the project team. However, in some instances, calculation of the trial carbon footprint was iterative which helped to establish and inform where guidance was ambiguous and required clarification.
New emission factors and activities added to guidance
The guidance from our initial publication has been updated during this application phase to include the following new activities involved in the PREMISE, ON-PACE, UK Stand Together and INTERVAL trials: blood pressure monitoring, saline use, oxygen use, business travel by car, commuting using activity data, dental examinations, laptop usage and telephony.
Existing emission factors have been updated in line with 2023 data from GOV.UK. Calculations using electricity and natural gas emission factors were updated, along with freight, business travel, building energy benchmarks and other clinical activities e.g. radiotherapy.
Additional assumptions have been included to aid the user with the calculations, for example the number of samples that can be stored in a freezer, the number of working hours in one full time equivalent (FTE), the number of folders that can be stored in 1m2 and the carbon footprint of common sample kit supplies. The updated “Detailed Guidance and method to calculate the carbon footprint of a clinical trial guidance (V0.5)” and associated “Data collation quick guide and worksheet” are included as Appendices B and C respectively.