Carbon Footprint

Prior to the pandemic, we studied the organizational challenges associated with roll-out of video consultations across multiple clinical directorates in the UK's largest acute hospital trust (25 (link)–27 (link)), including sub-studies on physical examination by video (28 (link), 29 (link)). We also undertook contract research for the Scottish Government to evaluate the national roll-out of video consultations—an initiative that was driven partly by the policy goal of reducing carbon footprint and travel costs from remote settings (30 ). Others in our team have studied help-seeking behavior in urgent care settings, including NHS 999 and NHS111 (31 (link), 32 (link)). Insights from these studies informed our theoretical work.
Since the pandemic began, we have been involved in three separately-funded but theoretically related case studies. Details of ethics approvals are given at the end of the paper, and full empirical reports of these studies are in preparation for publication elsewhere. All studies were of mixed-methods design but predominantly qualitative, using interviews, ethnography, and documentary analysis to generate and follow an emerging story of change, using quantitative data to illustrate and enrich the story.
First, we were funded by the Scottish Government (June–October 2020) to extend our evaluation of the video consultation service (branded “Near Me”) to cover the early months of the pandemic to August 2021 (33 ). This study covered both primary and secondary care. It included 60 h of ethnographic observation; 223 interviews with healthcare staff, patients, and national-level stakeholders (policymakers, professional leaders, industry); quantitative analysis of automated activity reports on over 69,000 consultations (including over 18,000 patient assessments of consultation quality); and analysis of policy documents and implementation plans.
Second, we were funded by the UK Research and Innovation COVID-19 Emergency Fund from June 2020 to November 2021 for a study called Remote by Default, which addressed remote care in general practice. This study involves interviews (over 100 to date) with healthcare staff, patients and national-level stakeholders, as well as following four locality case studies in south London, Oxfordshire, Devon, and south Wales. Especially relevant to the development of PERCS were four online focus groups involving 19 participants (clinicians, support staff, and patients), four facilitated cross-sector workshops (held via Zoom) which brought together ~160 national policymakers, clinicians, patients, and other stakeholders, and a four-round Delphi study (described in detail below) of ethical principles and decisions relating to remote consulting.
Third, we were funded by a medical charity from June 2020 to July 2021 to study the roll-out of video consultations across the UK. The Health Foundation Video Consulting (HFVC) study involved a quantitative survey of current practice (to over 800 NHS staff), qualitative follow-up interviews with a sample of 40 of these (repeated longitudinally with a sub-sample of 20 as the pandemic unfolded), interviews with 10 patients, and two group discussions involving 15 patient and public representatives. This study also included 7 locality case studies of video consulting services—four in secondary care (in London, Norfolk, Oxfordshire, and Cumbria) and three on group video clinics in primary care (in England, Scotland, and Wales).
In each of these studies, our research question addressed the individual-, organizational-, and system-level challenges to introducing remote consultation services at pace and scale and routinizing such services. We used an embedded virtual researcher-in-residence model: each case study had an assigned member of the research team who built relationships with key informants, developed an understanding of local issues and contingencies, and coordinated data collection and feedback. An external advisory group with a lay chair and patient representation met 4-monthly.

Data on the two farming systems and other related processes were collected between 2010 and 2013 as part of the EU FP7 SEAT project (S1–S3 Tables). Additional data were retrieved from the literature and the ecoinvent v2.2 database (www.ecoinvent.org). A complete description of the data used in the present research is available as supporting information (S1 Dataset) and in SEAT deliverable D3.5 [50 ]. Unit process distributions and variances were developed using the protocol presented in Henriksson et al. [30 (link)], reflecting inherent uncertainties (inaccuracies in measurements and models), spread (variability resulting from averaging) and unrepresentativeness (mismatch between the representativeness and use of data). The Anderson-Darling goodness-of-fit test was used to identify the distributions best representing data, limited to the four available distributions and generically assumed lognormal data in ecoinvent v2.2 [30 (link)].
The inventory flows were characterized using the GWPs and uncertainty distributions (S4 Table) reported in the fifth IPCC assessment report [51 ,52 ](step 1). In introducing uncertainties to GWPs, problems arise by the fact that the GWP of CO₂ is 1 by definition (and thus has no uncertainty), while the GWPs of all other GHGs are normalized by that of CO₂. Underlying GWPs (in kg CO₂-eq. kg^-1) are the absolute GWPs (AGWPs), which express the time-integrated radiative forcing (in W m^-2 yr^-1 kg^-1) [51 ]. These AGWPs are uncertain, also for CO₂. By adopting the uncertainty distributions on the level of GWPs we assume that these GWP uncertainties are based on dependent sampling of AGWPs in the models used by IPCC, e.g. dividing the AGWP for CH₄ in each run by the AGWP for CO₂ in the same run, thus forming a distribution of GWPs for CH₄ and a point value of the GWP for CO₂. The fifth IPCC assessment report [52 ] does, to our knowledge, not specify if the uncertainty estimates in the GWP of GHGs have been obtained through dependent or independent sampling, but judging the values of the uncertainties, we believe that dependent sampling has been used, as it should have been. Based on this assumption and in order to stay close to the traditional carbon footprint, we choose to use the GWPs with related uncertainty information for our characterization calculations from the fifth IPCC report [51 ,52 ], thereby maintaining the relative units and hence calculating carbon footprints in kg CO₂-eq. The standard deviations (σ) supporting these GWPs were back calculated from the 90% uncertainty ranges (σ = (P95-P05) / (2*1.645)) presented in the fifth IPCC report [51 ,52 ]. For more details, please see S4 Table and Myhre et al. [51 ].
Results were scaled to one tonne of fish and propagated over 1 000 MC simulations using dependent sampling (step 2) and the matrix-based algebra [53 ] implemented in the CMLCA v5.2 (www.cmlca.eu) software. Statistical tests were conducted in SPSS (v.21).
Of the two groups, family-owned farms were more reliant on farm-made feeds and agricultural byproducts (31% of all feeds) than large corporate farms, which almost exclusively (94%) relied upon commercial feeds (Fig. 2). Apart from feeds, all other supporting processes differed only in quantity, meaning that they rely upon the same shared supply chain, and hence on the same drawn values in each MC run, as well as stochastic GWPs. Emissions resulting directly from the fish ponds, however, were not shared between the two farming practices and therefore resulted in independently sampled values. For a more complete list of the data used and more specific results, see the supporting information to this article.

To truly reflect the ecological footprint and ecological carrying capacity of Dongying city, according to the lifestyle and consumption of Dongying city and with reference to Shandong Province Statistical Yearbook and Dongying City Statistical Yearbook, the biologically productive land is divided into arable land, forestland, grassland, water, construction land and fossil energy land, and the main consumption items of each category are shown in Fig. 3.Figure 3

Traditional ecological footprint consumption accounts in Dongying city. This paper uses the carbon footprint to improve the fossil energy footprint of the traditional ecological footprint.

The ecological footprint of energy land reflects the degree of pressure on the surrounding ecological environment caused by the consumption of fossil fuels by human activities and economic development. The traditional method of measuring the ecological footprint of energy land mainly considers the CO₂ emitted after the combustion of fossil energy. This paper takes into account the difference in carbon emissions during the land use process, based on the traditional ecological footprint consumption account, and replaces the traditional ecological footprint of energy land with a carbon footprint, which can better reflect the change pattern of carbon emissions in the total ecological footprint during human activities and is closely integrated with the IPCC land use carbon emissions study. It is also possible to take into account the impact of carbon emission factors on the carbon sequestered land in the ecological footprint. $$E{F}_C=\frac{\left(E_g+E_j+E_w\right)}{\mathit{NP}}$$
In Eq. (10), ${\mathit{EF}}_C$ is the carbon footprint, $E_g$ , $E_j$ and $E_w$ denote the total annual CO₂ emissions from cropland, construction land and unused land respectively, and $\mathit{NP}$ is the average carbon sequestration capacity of grasslands, woodlands, gardens and watersheds, t/hm2.

GHGE values for individual foods and ready meals expressed as gCO2 equivalents (gCO2e) were obtained from a range of open-access sources, including academic studies, retailers and producers published between 2008 and 2016^{(20 ,21 )}, added to the NDNS nutrient databank^{(21 ,22 )}. GHGE values were based on the emissions of six greenhouse gases which were converted into an equivalent amount of carbon dioxide (CO2 equivalent or CO2e), based on the relative global warming impact of each gas, and the final carbon footprint was expressed as the weight of carbon dioxide^{(20 )}. The climate metric used to aggregate the GHGE measurements into CO2e were those reported by Department for Environment Food and Rural Affairs, UK^{(23 )}. GHGE values from studies using complete cradle-to-grave life cycle analysis (LCA)^{(20 )}, obtained following the international PAS 2050 standard^{(24 )}, were selected where possible. We identified CO2e for 153 food and drink items in the NDNS nutrient databank, and where a GHGE value for a specific item was not available, reasonable substitute data were discussed and imputed by a team of three nutrition scientists, based on the food type, food group and compositional similarity of the products.
To estimate the GHGE for home-cooked meals, we estimated GHGE of the raw ingredients, establishing the weight of each ingredient and the weight of the whole cooked meal using Nutritics, which is nutrition management software for recipe and menu management, food labels, diet and activity analysis, and meal planning (Nutritics Ltd). Based on BBC Good Food^{(25 )} and Sainsbury’s recipes^{(26 )}, we established cooking methods and times. For home-cooked meals requiring more than one cooking method, GHGE data for each cooking method were added together. In addition, we recorded the longest cooking time suggested for the frozen versions of ready meals. If there was more than one suggested cooking method (e.g. oven and microwave), data for both methods were recorded separately.
To estimate the full GHGE until serving the meal, we combined the GHGE from the recipes’ ingredients or ready meals (value up to the supermarket shelf), which include emissions due to land use change, farm-related emissions, animal feed, processing, transport, retail and packaging) with GHGE produced by the different cooking methods. For the latter, GHGE of cooking appliances were based on manufacturer information^{(27 )} and adjusted to the conversion factors provided by the UK government in 2021^{(28 )} and cooking time (Equation 1):
where a is the cooking time, b is the GHGE of cooking appliances based on manufacturer information and adjusted to the conversion factors given by the UK government 2021, and c is the weight of the recipe or ready meal product.