Key messagesWhat is the key question?

Under conditions close to routine clinical practice, what are the effects of switching from a pressurised metered dose inhaler (pMDI)-based to a dry powder inhaler (DPI)-based maintenance therapy versus continued usual care on greenhouse gas emissions (carbon dioxide equivalents) and asthma control?

Introduction

The healthcare sector is a large contributor of greenhouse gas emissions and there is a growing interest in reducing the carbon footprint of healthcare organisations.1–3 Chlorofluorocarbons (CFCs) were banned from being used as aerosol propellants under the 1987 Montreal Protocol agreement due to their harmful effects on global warming and ozone depletion.4 While the hydrofluorocarbons (HFCs) that replaced CFCs in pressurised metered dose inhalers (pMDIs) do not deplete the ozone layer, they are potent greenhouse gases,5 6 which are now planned to be phased down under the Kigali Amendment to the Montreal Protocol in 2016,4 and through national F-gas regulations in Europe7 and the USA.8 HFC pMDIs account for 3%–4% of the total carbon footprint related to healthcare in the UK, with the majority of emissions associated with use and disposal of pMDIs rather than their manufacture.5 9–11 To put this in context, a single dose from an HFC-134a pMDI is approximately equivalent to driving 1 mile in a family car.12

In recognition of the impact of HFCs on the environment, in the UK, the government, the National Health Service (NHS) and the British Thoracic Society (BTS) all highlight the desirability of switching from pMDI to low carbon-impact alternatives,9 10 13 supported by derived data that the carbon footprint of a dry powder inhaler (DPI) is in the range of 20–200 times less than an HFC pMDI.5 11 14 Producers of respiratory medicines are also developing alternative, sustainable inhalers, including greener propellants for pMDIs. While most patients can generate an adequate peak inspiratory flow for dose delivery from a range of DPIs,15–17 a few patients (eg, children or the elderly) prefer or require a pMDI with spacer, and patients’ capabilities and preferences are key components in determining inhaler choice.13 18 19

The cost of treatment is another consideration when choosing between treatment options.18 19 In the UK, for controller medications, there is a range of DPI alternatives at comparable prices to pMDIs.5 However, salbutamol pMDIs are very inexpensive and less than half the cheapest, commonly available DPI salbutamol.5 The cost of goods in switching to an alternative low-carbon inhaler to the healthcare provider or individual patient should be considered in the context of the comparative cost-effectiveness of the two treatments,13 as well as the long-term financial cost of their environmental impact.5 A more effective inhaler which reduced exacerbations and rescue use could reduce overall cost.

The aim of this post-hoc analysis was to evaluate the effects of patients switching maintenance therapy from a pMDI to a DPI compared with those who continued pMDI-based treatment according to usual care on carbon footprint and asthma control. We used data from the Salford Lung Study in Asthma20 because it included a broad asthma population in a primary care setting with minimal intervention/supervision during the treatment period and a large proportion of patients who were using a pMDI prior to recruitment. During the study patients were allowed to switch to maintenance therapy, making it representative of routine UK clinical practice.

MethodsStudy design and study population

Full details of the study design and results for the primary study have been published previously.20 21 In brief, the study was a 12-month, open-label, primary care study in which adults with symptomatic asthma who were taking regular maintenance therapy (inhaled corticosteroids (ICS) or ICS/long-acting beta-agonist) were randomised to either a combination of fluticasone furoate/vilanterol (FF/VI) via the ELLIPTA DPI or to continue their usual care as prescribed by their general practitioner. For this post-hoc analysis, all patients who used a pMDI-based controller therapy prior to screening and randomisation were the subset of interest. At the randomisation visit, the study staff trained patients in both treatment groups to follow the correct inhaler techniques. During the 1-year treatment period, patients could have their maintenance treatment adjusted, including dose or inhaler type, at the physician’s discretion; that is, they were managed in conditions close to normal everyday clinical practice with no additional prompts or reminders to encourage adherence. The groups included in this analysis were based on the treatment at randomisation.

The study also measured the clinical outcomes of asthma control and rescue medication use. Asthma control was assessed using the Asthma Control Test (ACT)22 at baseline and at 12, 24, 40 and 52 weeks of treatment. Rescue medication use was measured as the total number of salbutamol inhalers prescribed over the whole treatment period.

Carbon footprint analysis

Total carbon footprint over 12 months was calculated separately for the total number of maintenance and rescue therapy inhalers prescribed during the study, using previously published carbon footprint data expressed as carbon dioxide equivalents (CO2e kg),5 11 summarised in table 1. For each subject, the total CO2e of each inhaler was calculated as the following: (number of inhalers prescribed) × (CO2e per inhaler), where the CO2e per inhaler was adjusted for the number of actuations in each inhaler, according to inhaler category (table 1). Within the primary study, there were a wealth of inhalers prescribed in the usual care arm, and given the data came from patients’ medical records it could not always be made certain which inhaler type the treatment was contained in. A medical expert review was performed on the prescription data to determine which inhaler the treatments were associated with based on dose, frequency of inhalation, total number of actuations, etc, and they determined which were in DPI, metered dose inhaler (MDI) or undeterminable. The inhalers were then associated to one of the CO2e categories listed in table 1 and adjusted based on the total number of actuations. For example, if the inhaler had 60 actuations and the category it was associated with was one based on 200 actuations, the inhaler was adjusted to a lower level of carbon emission, as less actuations result in lower levels of emission.

Table 1

Summary of CO2e (kg) categories by inhaler type

There was no imputation for missing data. Instead, this emission was adjusted for the subjects’ time on study medication, so it corresponded to 12 months on treatment. This approach was taken due to the real-world nature of the study as participants could have ended treatment before or after 12 months of study duration. Treatment groups were compared using a generalised linear model (GLM) assuming a log-normal distribution and by implementing an identity link function adjusting for randomised treatment, age, gender, and both ACT total score and asthma maintenance therapy at baseline per randomisation stratification. The carbon footprint data were analysed on the log scale, and the least square (LS) means of the treatment effects were computed from the GLM. The LS treatment means and 95% CI results on the log scale were back-transformed via the exponential to obtain least squares geometric means (LS GM) and 95% CI for the results to be interpretable on the original scale. Similarly, for the log difference between the LS treatment means and 95% CI, the results were back-transformed via the exponential to obtain the ratio of the LS GM and 95% CI.

Clinical outcomes analysis

Treatment differences in the percentage of patients defined as an ACT responder (an ACT total score ≥20 and/or an increase from baseline in ACT total score of ≥323 24) were analysed using a logistic regression model for all assessment time points (weeks 12, 24, 40 and 52). The model was adjusted for randomised treatment, asthma maintenance therapy at baseline (per randomisation stratification), baseline ACT total score, baseline ACT total score squared, gender and age. The time point for the primary analysis was 24 weeks. Changes from baseline in ACT total score were analysed via a mixed model repeated measures assuming an unstructured covariance matrix and adjusted for randomised treatment, asthma maintenance therapy at baseline per randomisation stratification, baseline ACT total score, randomised treatment by baseline ACT total score interaction, gender, age, visit, and randomised treatment by visit interaction, with visit included as a repeated measures factor. The number of salbutamol inhalers (adjusted to an equivalence of 200 actuations) prescribed per patient over the 12-month treatment period was analysed using an analysis of covariance model adjusted for randomised treatment, asthma maintenance therapy at baseline per randomisation stratification, ACT total score at baseline per randomisation stratification, gender, age and number of salbutamol inhalers prescribed in the year prior to randomisation.

Patient and public involvement

For this post-hoc analysis, the authors had no direct contact information of the study participants because anonymised data were used in accordance with strict confidentiality guidelines. Therefore, it was not appropriate for any patient to have involvement in developing the hypothesis, specific aims, research questions or plans for the study’s design or implementation. No patient was involved in the interpretation or writing of the results. There are no plans to disseminate the results of this study to the individual study participants.

ResultsAnalysis groups and treatment pathways

We studied patients from the Salford Lung Study in Asthma taking MDI-based controller therapy prior to randomisation. This subset of patients (2236 of 4233, 53%) comprised 1155 and 1081 patients randomised to the usual care and FF/VI DPI groups, respectively. The groups were well matched for demographic characteristics and baseline ACT total score (mean age: 49 years; mean ACT score: usual care, 16.6; FF/VI, 16.5) (table 2). During this real-world study, patients were able to shift from FF/VI DPI to usual therapy as per normal clinical practice, and on usual therapy some patients switched to DPI controller medications (not FF/VI which was not permitted). Over 1 year, the majority of patients in each group remained on the same inhaler type, that is, 80% remained on a pMDI in the usual care group and 85% remained on a DPI in the FF/VI group (table 3). The analysis was based on the prescribed treatment.

Table 2

Baseline characteristics of patients included in the analysis

Table 3

Treatment pathways during the study

Carbon footprint

Over the 12-month treatment period, the LS GM total carbon footprint (maintenance plus rescue therapy) was more than halved in patients taking FF/VI (108 kg, 95% CI 102 to 114) (‘switch’ group) compared with those continuing usual care treatment (240 kg, 95% CI 229 to 252; LS GM ratio usual care vs FF/VI 2.23, 95% CI 2.08 to 2.39, p<0.001) (figure 1 and table 4). The carbon footprint of maintenance therapy alone was 10-fold lower for FF/VI compared with usual care (LS GM 118 kg (95% CI 112 to 125) vs 11 kg (95% CI 10 to 12) for usual care and FF/VI groups, respectively; LS GM ratio of 10.65 (95% CI 9.86 to 11.50), p<0.001). The amount of rescue medication (almost all salbutamol MDIs) was also lower in the FF/VI group (LS GM 88 kg (95% CI 82 to 93) vs 109 kg (95% CI 103 to 116), respectively; LS GM ratio usual care vs FF/VI of 1.25 (95% CI 1.15 to 1.35), p<0.001).

Figure 1

Adjusted annual carbon footprint following maintenance and rescue medication in the usual care and FF/VI groups. CO2e, carbon dioxide equivalent; FF/VI, fluticasone furoate/vilanterol.

Table 4

Summary of carbon footprint data

The data for the carbon footprint analyses demonstrated variable levels of skewness and kurtosis. The quantile-quantile plots of the residuals showed adequate evidence of normality for the combined and rescue inhaler analyses; however, the maintenance alone analysis showed some evidence of kurtosis, indicating inferences should be interpreted with caution in this group.

Clinical outcomes

Asthma control improved in both treatment arms (table 5). At week 24, the odds of being an ACT responder (ACT total score of ≥20 and/or an increase from baseline of ≥3) in the FF/VI group (76%) were approximately twice that of being a responder in the usual care group (63%) (adjusted OR: 1.91 (95% CI 1.57 to 2.33), p<0.001), and this difference was consistent over the 12-month treatment period (figure 2). Patients who switched to FF/VI were prescribed approximately one fewer salbutamol inhalers over the 12-month treatment period compared with those who continued on usual care (LS mean of 7.2 vs 8, respectively).

Figure 2

Responders according to ACT score. The analysis method at each visit was logistic regression adjusted for randomised treatment, asthma maintenance therapy at baseline per randomisation stratification, baseline ACT total score, baseline ACT total score squared, gender and age. ACT, Asthma Control Test; FF/VI, fluticasone furoate/vilanterol.

Table 5

Summary of change from baseline in ACT total score and number of prescribed salbutamol inhalers

Discussion

This post-hoc analysis of data from the Salford Lung Study in Asthma showed that switching from a pMDI to a DPI maintenance therapy, in conditions representing everyday clinical practice, more than halved the per-patient annual carbon footprint for those on FF/VI DPI compared with those continuing on predominantly pMDI maintenance therapy. The remaining carbon footprint was almost all due to salbutamol MDI rescue, which was also significantly lower in the FF/VI DPI arm. There was no loss of asthma control in either group. Indeed, asthma control was consistently better throughout the 12 months in those patients randomised to FF/VI DPI, consistent with the findings of the primary study.20

The approximate per-patient annual carbon footprint saving demonstrated in this study was 130 kg CO2e. If these carbon savings were scaled up to include all adult patients with asthma in the UK who use a pMDI (n≈3 million13 25), this could result in an annual reduction of approximately 390 CO2e kilotons. This saving would represent approximately 40% of the total carbon footprint due to MDIs in the UK, which is reported to be one megaton of CO2e.5 The potential for carbon saving could be much greater if patients also switched to a rescue medication administered via a DPI as, in this study, the majority of patients used pMDI rescue medication inhalers, typical of usual care in the UK.5 11 Nonetheless, there was still a small carbon saving observed with respect to rescue medication in the group switching to FF/VI treatment, as a result of patients using less rescue medication in this group due to their improved asthma control.

A previous study on SERETIDE and VENTOLIN given via pMDI, based on an assumption of full adherence, calculated the annual per-patient total carbon footprint to be 439 kg CO2e.11 The annual per-patient total carbon footprint of 240 kg CO2e in the usual care group (using pMDI as maintenance as well as rescue medication) in the current study is lower due to lower adherence in real-world data. It was calculated using actual prescribed data and, although it could not be determined if patients picked up the pharmacy prescription and took the medication, it is much more likely to represent the real picture in the UK. In addition, the real-world nature of the Salford Lung Study in Asthma design allowed patients to switch inhaler type during the study (ie, patients in the usual care group could potentially switch to a DPI), which although small in number may have resulted in the carbon footprint estimation not being totally representative of the designated treatment group. This shows that substantial carbon savings can be made in a real clinical setting while improving asthma control. We acknowledge that any study with some degree of intervention cannot exactly reflect routine clinical practice; however, the Salford Lung Study in Asthma was managed to represent conditions as close to routine clinical practice as possible.

The study design was also in keeping with position statements from the UK government, NHS and BTS that highlight the need to implement switches from pMDIs to low carbon-impact alternatives,9 10 13 and the ability to switch back from a DPI to a pMDI in the usual care arm increases the external validity of our results. An encouraging message from this study is that the majority of patients who switched from a pMDI-based treatment to FF/VI did not switch back during the 1-year treatment period, suggesting that they tolerated DPI therapy well and without any detriment to their asthma control. The quality of the data collected was good overall, with only a relatively small number of prescriptions being undeterminable (table 3). The distribution of the greenhouse gas inhaler data exhibited heavy skewness due to the extremities in CO2e between MDI and DPI categories used in this analysis.11 The log-normal distribution was assessed and selected as the best fitting for these data, which is also in line with previous literature that describes the log-normal distribution to be a good fit for greenhouse emission data.26 27 Although the maintenance alone analysis showed some evidence of kurtosis, indicating a degree of caution should be taken in interpreting these results, the combined (maintenance plus rescue use) and rescue alone analyses showed adequate evidence of normality to provide reassurance that these results are robust.

The UK is an outlier compared with the rest of Europe in its high continued use of pMDIs.13 The results of this analysis support the growing calls from official bodies that, where possible, switches from pMDI to low carbon-impact alternatives should be sought, for example from pMDI to DPI or from high-volume pMDI to low-volume pMDI.5 9 10 13 Together with the role of pharmaceutical companies in producing accessible alternatives, prescribers, pharmacists and patients should be made aware of the significant differences in the global warming potential of different inhalers.13 The carbon footprint data used in our analysis were based on the cradle-to-grave emissions analysis previously described.11 We do recognise, however, that there are current shortcomings in the disposal of inhalers in terms of the level of recycling.

DPIs are an effective and cost-effective alternative to pMDIs13 and, as these results suggest, one that appears is well tolerated by patients with asthma. This latter point is also supported by real-life, primary care data that showed patients made fewer errors using DPIs compared with pMDIs.28 The ELLIPTA DPI, used in the Salford Lung in Asthma study, has also been shown to result in fewer critical errors compared with other inhalers, including pMDIs.29

In conclusion, in the subset of patients of the Salford Lung Study in Asthma who at enrolment were on a pMDI-based asthma maintenance therapy and who switched to a DPI therapy (FF/VI DPI), there was a substantial reduction in carbon footprint without loss of asthma control. The remaining inhaler carbon footprint could be further reduced through switches from pMDI to rescue medications in DPIs or to alternative lower-carbon footprint rescue inhalers if and when they become available. Both groups showed improvements in asthma control, with greater control demonstrated in those initiated on FF/VI. These data indicate that switching from a pMDI to a DPI is an acceptable and worthwhile option for most patients managed in normal everyday practice.

Ethics statementsPatient consent for publication

Not required.

Ethics approval

The Salford Lung Study in Asthma involved human participants and was approved by the National Research Ethics Service Committee North West, Greater Manchester South (reference 12/NW/0455; IRAS ID 106072). Participants gave informed consent to participate in the study before taking part.

Acknowledgments

ACCUHALER/DISKUS, ELLIPTA, RELVAR, SERETIDE and VENTOLIN are owned by/licensed to the GSK group of companies. Salamol is owned by Norton Healthcare. Atrovent is owned by Boehringer Ingelheim Pharma. Clenil is owned by Chiesi Farmaceutici. Editorial support was provided by Kate Hollingworth of Continuous Improvement Ltd.