ICS delivered by low climate impact inhalers was not associated to a higher risk of exacerbations requiring hospitalisation and all-cause mortality than ICS delivered by high climate impact inhalers. ICS delivered by low climate impact inhalers was also not associated to a higher risk of all-cause mortality, exacerbations requiring hospitalisation, or all-cause admission. ICS delivered by low climate impact inhalers was neither associated to a lower or a higher risk of admission with pneumonia. Finally, ICS delivered by high climate impact inhalers was associated to a slightly higher risk of exacerbations requiring a prescription of prednisolone without hospitalisation.

This is the first study comparing COPD exacerbation and death of ICS delivered by low and high climate impact inhalers in patients with both asthma and COPD. Our study is large, has complete follow-up, and available data for controlling several important confounders such at spirometry measures, tobacco history, MRC and BMI. Furthermore, we found no difference in safety profile depending on ICS delivered by low and high climate impact inhalers in patients with asthma and COPD, despite extensive theories on different anatomical deposition patterns, pharmacokinetic properties and device managing [4,5,6, 31]. This finding should be interpreted in the light of the extensive damage to the climate from high climate impact inhalers [1,2,3].

The patients in our primary analysis were very closely matched by nine clinical characteristics (age, gender, tobacco exposure, exacerbations requiring hospitalization, MRC, BMI, FEV1%, ICS dose and entry date) in order to minimize bias by indication.

Analyses of secondary outcomes similarly revealed no difference in safety profile for ICS delivered by low vs. high climate impact inhalers, with comparable HRs for pneumonia. The analysis on the effect of ICS delivered by high and low climate impact inhalers was elaborated by secondary outcome analyses on all-cause mortality, exacerbations requiring hospitalization or all-cause admission.

One outcome analysis did show an increased risk in patients treated with ICS delivered by high climate impact inhalers: exacerbations requiring prescriptions of prednisolone, but not hospitalization. However, the clinical relevance remains uncertain with a CI almost crossing one. We analysed hospitalisations with exacerbations and hospitalisations with pneumonia as well as exacerbations treated with prednisolone, but we did not analyse exacerbations of concomitant asthma and COPD treated only with antibiotics, as this is not a standard recommended treatment regimen when a patient also suffers from asthma, however, we may have missed some exacerbations / pneumonias not requiring hospitalisation in this context and a difference between groups cannot be ruled out in our study set-up.

A hypothesis has claimed that inhalation medicine delivered by high climate impact inhalers may be a better choice for patients with a low FEV1% [32,33,34,35,36,37]. Therefore, we conducted a sensitivity analysis in which we tested if there was an effect modulation between low FEV1% (GOLD FEV1% stage 4 with FEV1% < 30%) and effect on ICS delivered by low vs. high climate impact inhalers towards the primary outcome. However, this was not the case.

No studies have been performed on treatment with ICS delivered by different types inhaler in patients with both asthma and COPD, however our results are in line with RCTs performed on patients with asthma without COPD and COPD without asthma, which all showed no difference in disease control and safety profile in ICS delivered by high and low climate impact inhalers [7,8,9,10,11,12,13,14,15,16]. In contrast to the RCTs, most epidemiological studies in this field have pointed to an improved disease control [17,18,19,20,21] but also an increased risk of pneumonia [22] in patients treated with ICS delivered by high climate impact inhalers.

RCT studies on patients with asthma have shown equal effect and safety profile of ICS delivered by low and high climate impact inhalers in patients with asthma on bronchial hyperresponsiveness, quality-of-life scores, FeNO, spirometry test results, forced oscillation and time to first exacerbation [11,12,13,14,15,16], whereas epidemiological studies on patients with asthma showed an association between asthma control, reducing asthma exacerbations and FEV1 and treatment with ICS delivered by high climate impact inhalers [17,18,19, 38]. In patients with COPD, RCTs have shown similar effect on FEV1 [7], but epidemiological studies pointed to a superior effect of ICS delivered by high climate impact inhalers in controlling exacerbations of COPD [20, 21]. Some of the epidemiological studies comparing ICS delivered by low and high climate impact inhalers found a superior effect of ICS delivered by high climate impact inhalers in patients with either asthma or COPD, however this has not been shown for patients with both diseases. Further, several of these studies did not include clinical variables such as tobacco exposure, MRC, BMI and FEV1% in their analysis or match [17, 20, 21], one study switched patients from ICS delivered by low to ICS delivered by high climate impact inhalers at the discretion of the treating physician without matching or logistic regression analysis [18], and in one study the ICS treatment by high climate impact inhalers was compared to ICS treatment by low climate impact inhalers in unmatched groups [19]. In an epidemiological study showing an increased risk of pneumonia in patients with either asthma or COPD, matching was performed on only a limited number of variables, and not on important confounders like tobacco exposure, MRC, BMI and FEV1% [22].

Our study contributes to the accumulated evidence as both the first study on patients with concomitant asthma and COPD, but also as it is large and very well-matched in comparison to many of the other epidemiological studies done on ICS delivery by high and low climate impact inhalers.

Our study has strengths and limitations. Our study has a its large sample size, complete follow-up, and extensive data availability. The data available to us were extensive and for many patients complete including annually updated data on smoking status, FEV1%, MRC and BMI. Our analysis was based on all patients with COPD in an entire country; a population of > 100,000 patients with COPD, of whom almost 10,000 patients had concomitant asthma combined with relevant and quantifiable pharmacological treatment. In this context loss to follow-up was minimal (five patients in total), and the expected impact hereof on the results minimal.

As we were very keen on minimizing the risk of bias by indication, the propensity score matched groups for the primary analysis included only patients with all data available and the groups were matched very tightly on nine different variables. This was successful in generating very comparable groups even on unmatched variables such as co-morbidities evaluated in a hospital setting, which we also have full data availability on. Controlling for important confounders and successful propensity score matching is important to minimize bias by indication. However, our tightly matched groups might pose a challenge for the options for extrapolation of our study on a wider group of patients with asthma and COPD. Our results also appear robust on various outcomes and in both adjusted and unadjusted analyses.

As ICS use by low and high climate impact inhalers was only available to us as collected prescriptions, treatment adherence and inhalation technique were not captured, however since this is a large population, these factors (i.e. non-adherence and insufficient inhalation technique) are likely to be balanced between groups, and, in any case, our results reflect real-life use of both types of devices. We chose to include ICS dose in the matching variables based on collected ICS the previous year. Hence, patients who died just after they collected a first prescription of ICS, were not included and thus a risk of survivor bias cannot be dismissed.

When analysing ICS, it was not possible to consider the different molecule types, nor if delivery was ICS was delivered alongside other pharmaceuticals (LABA or long-acting muscarinic receptor antagonist (LAMA)), as matching or stratifying for these variables would have led to smaller populations for analysis and posed a risk for a type 2 error and a challenge with generalisability. Similarly, it was not possible to match or stratify for the many different types of low and high climate impact inhalers.

As our study is a retrospective cohort study, it has some inherent limitations regarding causality of the findings. However, our study is the first of its kind on the group of patients with both asthma and COPD, it is large and there was complete follow up. Despite the abundant availability of data, some variables were only available as semi-quantified data, and access to more complete data might have improved our analyses.

Future research could explore the development of environmentally friendly inhaler alternatives and assess their impact on patient outcomes and environmental sustainability. Longitudinal studies could also investigate the long-term effects of different inhaler therapies.

In conclusion, ICS delivered by low and high climate impact inhalers do not have different effect or safety profiles, even among patients with low pulmonary function (GOLD FEV1% stage 4 with FEV1% < 30%). In this light, there is no sign of a clinical benefit which could justify the climate footprint caused by high climate impact inhalers. Evidence-based decision-making should consider both patient outcomes and environmental considerations in inhaler selection.